Newtonian Microeconomics PDF

NEWTONIAN
MICROECONOMICS
A Dynamic Extension to
Neoclassical Micro Theory
Matti Estola
Newtonian Microeconomics
Matti Estola
Newtonian
Microeconomics
A Dynamic Extension to Neoclassical
Micro Theory
Matti Estola
University of Eastern Finland
Joensuu, Finland
ISBN 978-3-319-46878-5 ISBN 978-3-319-46879-2 (eBook)

DOI 10.1007/978-3-319-46879-2
Library of Congress Control Number: 2016961252
© The Editor(s) (if applicable) and The Author(s) 2017

This book was advertised with a copyright holder in the name of the publisher in error, whereas the author
holds the copyright.
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.
Cover illustration: © Carol Dembinsky / Dembinsky Photo Associates / Alamy Stock Photo
Printed on acid-free paper
This Palgrave Macmillan imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
Motivation for the Book

The theoretical framework of mainstream microeconomics, neoclassical
micro theory, is in many ways restrictive and sticking to it has prohibited
the development of economics into a quantitative science. The main
weaknesses of neoclassical analysis are: (1) The assumption of static opti-
mization does not allow modeling time dependent processes, such as time
dependent price and cost functions of firms and income of consumers.
(2) Static profit maximization does not allow increasing returns to scale
in firms’ production processes because in that case a profit-maximizing
flow of production would not exist. (3) The assumed optimal behavior of
economic units prohibits understanding changes in economic quantities,
because if every economic unit behaves in the optimal way from its point
of view, none will change its behavior and thus observed changes in
economies cannot be understood. (4) The static neoclassical theories of
a consumer and a firm and their dynamization by dynamic optimization
give equal results, if identical target functions are assumed for economic
units in both cases. Thus the dynamic theories of a firm and a consumer
derived by dynamic optimization as presented in economics textbooks are
inconsistent with the corresponding static ones.
These claims mean that economics is lacking a consistent dynamic
theory of microeconomics, and this is what we introduce in this book.
v
vi Preface
We change the assumed optimal behavior in the neoclassical framework

to the following: Economic units like to better their situation if possible.
This approach is shown to lead to an analogous framework of modeling
in economics with Newtonian framework in classical mechanics. We
define the ‘economic forces’ acting upon economic quantities in non-
equilibrium situations, and show that these forces cause an adjustment
towards an equilibrium state, or keep the system in motion with time.
The proposed framework gives game theoretic analysis and a dynamic
extension to neoclassical theory from a single principle.
An essential element of quantitative economics is a system of mea-
surement units by means of which economic events can be quantified.
Here we introduce one such candidate, and apply it throughout the
book in defining the dimensions of economic quantities and in solving
dimensional problems in functional forms common in economic analysis.
Defining the measurement units for economic quantities exactly does not
conflict with the neoclassical framework, however.
If the above-mentioned elements are added into current neoclassical
microeconomics, it meets the minimum requirements of quantitative
analysis. This was the ultimate reason for writing this book. However,
much work is still needed in the empirical evaluation of economic theories
before microeconomics can be considered as a reliable description of real-
world phenomena.
Innovations in the Book

The book has several innovations as compared with existing textbooks
on microeconomics. These are: (1) The scientific principles of modeling
microeconomic phenomena are defined in a new way. We define an axiom
of modeling for economic phenomena that is based on the decision-
making of rational economic units. This axiom can be used in modeling
static and dynamic behavior, and it covers the whole of microeconomics.
(2) A system of measurement units for economics is defined and applied
throughout the book. This guarantees that the equations written are
correct in measurement units, and that the theoretical and observed eco-
nomic quantities correspond to each other exactly. This way the empirical
Preface vii
testing of economic theories becomes rigorous. (3) The economic ‘forces’

acting upon economic quantities are defined, and these forces explain
the observed changes in economic quantities. This is a new framework
of modeling in economics, and it is analogous to Newtonian mechanics
in physics. (4) In every presented theory, the results are compared with
observed real-world behavior. In this way we get a view as to how accurate
current economic theories are as compared with observed data.
An old truth—everything affects everything—holds in economics.
Thus to be able to understand and model economic events, these have
to be simplified so that the created models do not get too complex. In
modeling, it is important to separate essential factors from unessential
ones, and create models by using only essential quantities. Many complex-
looking systems consist of various simple systems affecting each other, and
in modeling in this way the connections between the systems are revealed.
Economic phenomena can be understood in principle at three levels.
The first level is intuitive. We understand intuitively that, for example,
the more wanted and scarce a good is, the more people are ready to pay
for it, and the higher value it has. Another level of understanding eco-
nomic phenomena—that is extensively applied in economics textbooks—
is graphical presentation. In a graphical presentation, the relation between
economic quantities is presented as an increasing or decreasing graph, and
this visual presentation clarifies the situation. The third and deepest level
of understanding is the exact mathematical formulation of a situation so
that the explainable variables (or their time paths) are solved as functions
of explaining variables. This deepest level of understanding is required
in empirical testing of economic theories because graphical analysis is
not accurate enough for rigorous testing. In this book, all three levels of
understanding are applied.
A part of the results appearing in this book have been published
earlier in Estola and Hokkanen (1999, 2007, 2008) and in Estola (2001).
Empirical support for the theories presented in this book can be found in
Estola and Dannenberg (2012) and in Estola (2011, 2015). However, many
of the modeling situations presented in this book have not previously been
presented.
viii Preface
About the Audience

This book is aimed at students of undergraduate and graduate level
university courses in microeconomics. Directed especially to those inter-
ested in econophysics, the book presents a new methodological bridge
between modeling principles in economics and physics. The book starts
at an elementary level of economics, however, because the principles of
modeling in economics are defined in a new way. However, the core
material of the book is aimed at students having taken at least a basic
course in economics. The mathematical requirements are knowledge of
equations and differential calculus for the main parts of the book, which
also includes a mathematical appendix where the applied techniques
are presented. All sections in the book, where differential equations are
applied, are marked with asterisk to note that these sections are formally
more demanding.
I have used the contents of the book in my teaching at the University
of Jyväskylä during the period (1996–1999) and at the University of
Eastern Finland (former University of Joensuu) during (1999–2016) in
the following courses: Principles of Economics, Intermediate Microe-
conomics, and Advanced Microeconomics. This means that some parts
of the book are applicable at introductory and some at advanced level
microeconomics courses in universities. I have used the excellent inter-
mediate level textbook of physics by Hans C. Ohanian (1989) Physics,
Second Edition Expanded, as a model for the presentation of matters.
The mathematical level of this book represents in some parts that of
Ohanian; however, on average the book of Ohanian applies much more
complicated mathematics. With this argument I can claim that the
mathematical level of the book is not too high. The first version of this
book, Kansantaloustieteen perusteet, was published at 1996 in the Finnish
language at the University of Jyväskylä. This new English version of the
book includes a few new chapters as compared with the 1996 version, and
over the past 20 years all the material in the book has, of course, improved
and deepened.
Preface ix
Acknowledgements
The main results presented in the book were developed in my PhD
Dissertation Thesis in 1995, where Docent Veli-Matti Hokkanen had an
essential role in supervising the work. I am also grateful to the contributing
suggestions by Professor Jouni Suhonen during that work. In studying
these matters and writing this book, various colleagues and students
have made contributing comments to the work. Kaija Häkkinen wrote
a mathematical appendix for the Finnish version of the book in 1996, and
in this book I have translated it into English and added some extra minor
details. Parts of the Finnish version of the book has been reviewed by Tapio
Ruokolainen, Jaakko Pehkonen, Kari Heimonen and Kari Kinnunen. I
give my deepest thanks to these people for their contributions in that
work.
This new English version of the book has received input by Kalle
Määttä, Heinz Eckart Klingelhöfer, Thomas Hering, Michael Olbrich,
Kristiaan Kerstens, and Alia Dannenberg. I give my warmest thanks to all
of them for their contributing comments. Naturally, I am also grateful to
the several other students, whose names are not mentioned here, but who
have contributed to this work during my 20 years of teaching experience
at the University of Jyväskylä and the University of Eastern Finland.
Needless to say, I take full responsibility for all remaining errors and
inconsistencies existing in the book. Finally, I give my deepest thanks
to editor Laura Pacey for accepting this book in the Palgrave Macmillan
series.
Joensuu, Finland Matti Estola

Spring 2016
References
Estola, M. (2001). A dynamic theory of a firm: An application of economic
forces. Advances in Complex Systems, 4(1), 163–176.
Estola, M. (2011). Measuring the growth and the structural change in a
multi-sector economy. Hyperion International Journal of Econophysics & New
Economy, 4(1).
x Preface
Estola, M. (2015). Neoclassical and newtonian theory of production: An empir-

ical test. Hyperion International Journal of Econophysics & New Economy, 8(1).
Estola, M., & Dannenberg, A. (2012). Testing the neoclassical and the Newto-
nian theory of production. Physica A, 391(24), 6519–6527.
Estola, M., & Hokkanen, V.-M. (1999). A dynamic theory of consumer
behaviour. In S. B. Dahiya (Ed.), The current state of economic science. Rohtak:
Spellbound Publications Pvt. Ltd.
Estola, M., & Hokkanen, V.-M. (2007). Asset price dynamics by economic
forces. BetrieBswirtchaftliche Forschung und Praxis, 5.
Estola, M., & Hokkanen, V.-M. (2008). Consumer, firm, and price dynamics: An
econophysics approach. Modeling by economic forces. Saarbrücken: VDM Verlag
Dr. Müller.
Contents
1 Economics as a Science 1
1.1 Classification of Economics 3
1.2 The Axioms of Economics 5
1.2.1 Regular Needs of Human Beings 5
1.2.2 Business Competition and
Rational Behavior 6
1.2.3 The Birth of Organizations 9
1.2.4 The Principle of Modeling in Economics 12
1.3 Forecasting Human Behavior 14
1.3.1 Predictability of Economic Events 15
1.4 Frameworks of Economic Modeling 16
1.4.1 The Neoclassical Framework and
Its Critique 16
1.4.2 Econophysics 19
1.4.3 A Dynamic Extension to
Neoclassical Economics 24
1.4.4 Decision-Making Steelyard 26
1.5 A Summary of the Methodological Basis of
Economics 30
References 30
xi
xii Contents
2 Measuring in Economics 35
2.1 Principles of Dimensional Analysis 36
2.2 A Measurement System for Economics 41
2.2.1 Measurement of Volume of Goods 41
2.2.2 Measurement of Time 42
2.2.3 Measurement of Monetary Values 43
2.2.4 Measurement of Satisfaction 47
2.2.5 Primary Dimensions in Economics 49
2.3 Nominal and Real Quantities 52
2.3.1 How Do We Use Real Quantities? 54
2.3.2 Internal and External Value of a Currency 56
2.4 Discrete and Continuous Quantities 57
2.5 Measuring Changes in Scalars 59
2.6 Measuring Changes in Values 61
2.6.1 Changes in Values in Discrete Time 61
2.6.2 Changes in Values in Continuous Time 64
2.6.3 Interest Rate 65
2.7 Economic Kinematics 68
2.7.1 Average Velocity of Production 69
2.7.2 Instantaneous Velocity of Production 73
2.7.3 Velocity and Accumulated Production 74
2.7.4 Average Acceleration of Production 80
2.7.5 Instantaneous Acceleration of Production 81
2.7.6 Kinematics of a Two-Good
Production System 82
2.8 Index Numbers 88
2.8.1 Index Numbers as Average Quantities 89
2.9 The Production Function 91
References 96
3 Consumer Behavior 97
3.1 The Axioms of Consumer Behavior 99
3.2 A Consumer’s Budget Equation 100
3.3 A Consumer’s Preferences 103
3.4 A Consumer’s Optimal Choice 108
Contents xiii
3.5 Utility Function and Indifference Curves 115

3.6 The Equilibrium State of a Consumer 119
3.7 Dynamic Consumer Behavior 121
3.7.1 Lagrangian Formulation of
Consumer Behavior 125
3.7.2 Marginal Willingness-to-Pay and Demand 126
3.8 Newtonian Theory of a Consumer 134
3.9 Aggregate Food Consumption 143
3.10 Aggregate Demand Relation of Food 145
3.11 Mathematical Appendix 147
References 150
4 The Behavior of Firms 151

4.1 Principles of Firms’ Behavior 151
4.1.1 Planning Firms’ Behavior 153
4.2 The Forms of Business Organization 155
4.3 Revenues, Costs and the Profit of a Firm 156
4.3.1 Problems in the Calculation of a
Firm’s Profit 157
4.3.2 The Costs of Firms 160
4.3.3 The Revenues of Firms 166
4.4 The Production Decision of a Firm 172
4.4.1 Production Decision by
Revenues and Costs 172
4.4.2 Production Decision by Marginal
Quantities 174
4.5 Dynamic Analysis of a Firm’s Behavior 179
4.6 Newtonian Theory of a Firm 181
4.7 Firms’ Pricing Behavior 188
4.7.1 Dynamic Analysis of a Firm’s Pricing 190
References 194
5 Goods Markets 195

5.1 Different Market Situations 199
5.1.1 Why Do Different Market
Situations Exist? 200
xiv Contents
5.2 Perfect Competition in an Industry 202

5.2.1 A Firm in a Perfectly Competed Industry 202
5.2.2 Aggregate Production of an Industry 208
5.2.3 Equilibrium in Perfect Competition 209
5.2.4 Adjustment by Price Mechanism 212
5.2.5 Growth in a Perfectly Competed Industry 214
5.2.6 Further Observations on the
Adjustment Process 216
5.2.7 The Dynamics of Adjustment in
Perfect Competition 218
5.3 A Monopoly Firm in an Industry 226
5.4 A Firm in Monopolistic Competition 229
5.4.1 Firms’ Advertising 230
5.5 Oligopolistic Competition in an Industry 230
5.5.1 Pricing in Oligopoly: Contract
vs. Competition 231
5.5.2 Price Contracts in Oligopoly 232
5.5.3 Non-cooperative Oligopoly 233
5.5.4 Cournot’s Model of a Duopoly 233
5.5.5 Cournot’s Duopoly à la Newton 235
5.6 Price in Monopolistic Competition 240
5.6.1 Price Dynamics in Monopolistic
Competition 243
References 249
6 Labor as a Production Factor 251

6.1 A Firm’s Demand of Labor 251
6.2 Newtonian Theory of Use of Labor 256
6.3 Labor Supply of a Person 258
6.4 Newtonian Theory of Labor Supply 266
6.5 Atomistic Labor Market 267
6.5.1 The Demand of Labor 268
6.5.2 Adjustment of Labor Demand 270
6.5.3 The Supply of Labor 270
6.5.4 Adjustment of Labor Supply 271
Contents xv
6.5.5 The Force Acting upon Employment 271

6.5.6 Wage Adjustment 272
6.5.7 Equilibrium in Atomistic Labor Market 273
6.5.8 An Approximation of the Equilibrium 274
6.5.9 Labor Market Adjustment in Detail 276
6.6 Trade Unions in the Labor Market 280
6.6.1 Dynamic Trade Union Behavior 286
References 288
7 Capital Goods as Firms’ Inputs 289

7.1 Renting Capital Goods 290
7.1.1 A Dynamic Theory of Renting
Capital Goods 292
7.2 Investing in Capital Goods 293
7.3 Interest Calculation and Discounting 293
7.3.1 Discounting in Discrete Time 293
7.3.2 Interest Calculation in
Continuous Time 296
7.3.3 Parities Between Interest Rates 298
7.4 Present Values of Money Flows 302
7.4.1 Present Values in Discrete Time 302
7.4.2 Present Values in Continuous Time 305
7.5 Investment Decision of a Firm 308
7.5.1 Accumulation of Firms’ Capital Stocks 313
7.5.2 Investment Decisions in
Continuous Time 315
7.5.3 A Dynamic Theory of Investment 316
References 319
8 Money and Financial Markets 321

8.1 A Short History of Money 321
8.1.1 Money Systems in Brief 323
8.1.2 The Functions of Money 326
xvi Contents
8.2 Fundamentals of Financial Markets 328

8.2.1 Direct Finance 329
8.2.2 Indirect Finance 330
8.2.3 Financial Market Instruments 332
8.2.4 Primary and Secondary Markets 334
8.2.5 Exchanges and OTC Markets 336
8.2.6 Pricing Financial Instruments 337
8.2.7 Dynamics of Asset Prices 342
8.2.8 Aggregate Investor Behavior 345
8.2.9 Explicit Adjustment in Share Price 348
References 357
9 Saving, Borrowing, and Interest Rates 359

9.1 Saving Households 360
9.1.1 Dynamics of Savings 369
9.2 Borrowing Households 370
9.2.1 Dynamics of Borrowing 376
9.3 Loan Market Behavior 376
9.4 Aggregate Analysis 377
9.4.1 Adjustment in Aggregate Saving 379
9.4.2 Adjustment in Aggregate Borrowing 379
9.4.3 Adjustment in Interest Rate 380
9.4.4 Equilibrium State in the Loan Market 380
9.4.5 The Explicit Adjustment Process 382
9.4.6 Adding Investment in the Model 386
10 Mathematical Appendix 395

10.1 Introduction 395
10.2 Straight Line 396
10.3 Closed and Complete Sets 399
10.4 Vectors 399
10.4.1 Calculation Rules for Vectors 400
Contents xvii
10.5 Functions 401

10.5.1 The Definition of a Function 401
10.5.2 The Graph of a Function 402
10.5.3 Composite Function 403
10.5.4 Inverse Function 405
10.6 Limit Value 407
10.7 Derivative 407
10.7.1 The Definition of Derivative 407
10.7.2 Some Rules of Derivation 409
10.7.3 Examples of Derivatives 411
10.8 Applications of Derivatives 412
10.8.1 Increasing and Decreasing Functions 412
10.8.2 Extremum Values 413
10.9 Partial Derivatives 416
10.9.1 Partial Functions 416
10.9.2 Partial Derivatives 417
10.9.3 Rules of Partial Differentiation 418
10.9.4 Chain Rule of Partial Differentiation 419
10.10 Approximation 421
10.10.1 Taylor Series in Approximation 423
10.10.2 Total Differential 424
10.10.3 Taylor Formula for Scalar Fields 426
10.11 Extremum Values 427
10.12 Constrained Extremum Values 429
10.12.1 The Method of Lagrange 430
10.12.2 Uniqueness of Marginal
Willingness-to-Pay 430
10.12.3 The Interpretation of Lagrange’s
Multiplier 433
10.13 Implicit Differentiation 434
10.14 Integral Calculus 438
10.14.1 Integral Function 438
10.14.2 Definite Integral 439
10.15 Differential Equations 441
xviii Contents
10.16 Scales of Measurement 444

10.16.1 Nominal or Classification Scale 445
10.16.2 Ordinal Scale 447
10.16.3 Interval Scale 448
10.16.4 Ratio Scale 450
10.16.5 Absolute Scale 451
References 452
Author Index 453
Subject Index 457

1
Economics as a Science
Economics is a social science that analyzes the conflict between the almost
unlimited needs of human beings, and the limited amount of goods
available for the satisfaction of people’s needs. For example, the following
questions are studied in economics: (1) Why are certain goods produced in
different economies? (2) Whose needs does the production aim to satisfy?
(3) Which methods are applied in the production of different goods, and
what is the reason for this? (4) How are the prices of goods determined? (5)
What kind of institutions and economic units exist in societies, and how
they operate? (6) What are the factors that affect the welfare of a society,
and how are these factors measured? (7) Which factors cause economic
growth? and (8) How have the relations between economic units been
developed over time?
In the following we denote a definition by the symbol §.
§: By an economy we understand a society from the point of view of
its economic structure. ˘
§: A good is the term applied to material and non-material products. ˘
A good is thus a shirt, a radio set, a meal in a restaurant, a bottle of
wine, a haircut, a soccer game, etc.
© The Author(s) 2017 1

M. Estola, Newtonian Microeconomics,
DOI 10.1007/978-3-319-46879-2_1
2 Newtonian Microeconomics
§: By consumption we understand the eating or using up of something

of value, or the enjoyment of the services of a durable good. ˘
§: By a durable good we understand a good that produces services for
its holder during several time units, and which usually has a positive scrap
value. ˘
§: By production we understand the process where the inputs of
material and labor are used in the construction of goods. ˘
§: Labor consists of the physical and mental work of people in the
production of goods. ˘
§: An economic unit is a decision-making unit that consists of one
person or a group of people that behave together as one unit. An economic
unit has some goals it is willing to reach, and a limited amount of resources
for achieving these goals. ˘
§: By resources of economic units we understand the production
factors that are used in the production processes of an economy. Resources
are, for example, plants, machines, money, labor, raw materials, knowl-
edge of production methods, creativity, and so on. ˘
Economic units are, for example, households, investors, firms, trade
unions, non-profit making institutions, central bank, local govern-
ments, and the government of a country. All these can be seen as
independent decision-making units (see Sect. 1.3.1), the decisions of
which affect the behavior of the economy they operate. We assume that
economic units have defined for them certain goals they aim to achieve,
and in this book we do not study the conflicts of interest between
people belonging in one economic decision-making unit. We thus assume,
for example, that the goals of a trade union are high wages and full
employment for its members, even though union members have differing
views of the importance of these two conflicting goals.
The aim of the science of economics is to measure, understand, and
explain how the productive resources are allocated in an economy in a
time unit, which part of the resources is actively used, how these resources
are combined in various production processes, how the prices of goods are
determined, and what kind of a historical process has caused the current
situation. In economics, the aim is also to understand and explain changes
in these matters and to forecast their future development.
1.1 Classification of Economics

Economics can be classified in various ways; one way is the following:
positive and normative economics.
§: Positive scientific methodology, or positivism, can be character-
ized as follows: (1) Methodological monism, that is, one scientific method
is applicable in all scientific investigation; (2) Exact natural sciences, and
mathematical physics in particular, are accepted as a methodological ideal
that measures the degree of development and perfection of other sciences;
(3) Scientific explanation is causal and so in nature there exists general
laws of behavior including ‘human nature’, and individual cases can be
explained by causal reasoning using these laws; and (4) Observations have
a key role in making scientific hypotheses and statements (von Wright
1975, p. 4). ˘
In positive economics, formulated and tested economic theories are
used to describe the observed regularities in economic behavior.
§: Normative scientific methodology studies questions like how
things ought be for a system to function in the best possible way from
a certain point of view. ˘
Normative economics aims to answer questions about how an economy
should function. Normative economics studies questions such as what
actions should be taken in order that an economic unit, an economy
as a whole, or the whole world would operate in the best possible way.
Examples of problems in normative economics are: (1) Which tasks should
be carried out by the public and the private sector in an economy, and (2)
What kind of tax system should be used, and how it should be changed
in a certain situation to reach a preferred goal.
The classification into positive and normative scientific analysis is not
always clear. For example, by applying positive scientific methodology we
can measure how the welfare of an economy has developed within a time
unit. On the other hand, the concept of economic welfare can be defined
in various ways; which factors are measured and in which way. Thus this
measuring has a normative basis about how an economy should develop
so that its welfare increases.
Another classification in scientific analysis can be made between the-

oretical and empirical research. The beginning of formal theoretical
scientific analysis can be found in the works of Sir Isaac Newton when he
started to define mathematical models that mimic and explain observed
behaviors in nature.
§: By theoretical research we understand the construction of formal
models that mimic and explain observed events. ˘
§: By empirical research we understand scientific analysis based on
observations. ˘
§: By the real world of a science we understand the existing objects and
processes studied in the science. The real world of economics, for example,
consists of economic units in existing economies and the processes created
by them. ˘
Empirical analysis can be divided into descriptive analysis and testing
of hypotheses. In descriptive analysis, real world phenomena are mea-
sured and described by observed data. For example, measuring inflation,
i.e. an increase in the average price level in an economy, is descriptive
research. However, to explain why inflation exists we need a theory of
factors causing inflation, which is theoretical analysis. On the other hand,
testing the reliability of a theory by observations is scientific testing of
hypotheses.
Descriptive research and testing of hypotheses deviate from each other.
The former concentrates on describing and measuring phenomena with-
out a predetermined idea of how and why the phenomena has occurred.
In testing of hypotheses, on the other hand, a preliminary idea–a theory
or an assumed regularity–is needed, the reliability of which is tested
in the analysis. The deviation between descriptive research and testing
of hypotheses is not always clear, however, because the choice of the
measured variables in descriptive analysis is usually based on a theory of
the studied phenomenon.
In economics, the classification between theoretical and empirical
research is not as clear as in physics where university degree courses can
be taken both in physics and in theoretical physics. Economic theories
are still lacking the accuracy and universal acceptance as many physical
theories have, and thus economic theories cannot yet be taught in schools
as universal truths.
Like other sciences, economics can be divided into various subfields,

including labor economics, international economics, financial eco-
nomics, public sector economics and growth theory. However, the
main classification is microeconomics and macroeconomics. Microe-
conomics studies the behavior of individual economic units and macroe-
conomics studies economy-level phenomena. This classification also pre-
vails in physics (micro- and macrophysics), biology (micro- and macro-
organisms), and chemistry. In the above- mentioned branches of eco-
nomics, both the micro and macro frameworks are applied, depending
on the phenomenon. Thus, this classification covers the whole sphere of
economic events.
Examples of research problems in microeconomics are: (1) How an
increase in the price of electricity affects the costs of a firm, or (2) How an
increase in income tax affects the consumption of households. Research
topics in macroeconomics could be: (1) How an increase in interest rate
affects the aggregate investment in an economy, or (2) Why the US dollar
was devalued by 5% against the euro in one week.
1.2 The Axioms of Economics

1.2.1 Regular Needs of Human Beings
There are certain regularities in human sciences: people read newspapers,

drink coffee, tea or cocoa and brush their teeth in the morning, eat lunch
between 11.00–14.00, and so on. These regularities are almost as certain
as a stone falling to the earth if it is thrown upward. Even though a single
person might not brush his/her teeth on one morning, on average, most
people do. Thus a stable demand exists for toothpaste, newspapers, coffee,
tea, cocoa, and so forth, for a fixed time period.
According to observations, people in different countries are so similar
with respect to their physiological and mental needs that it is meaningful
to analyze the general features of human behavior. This can be seen in
firms’ marketing: there would be no sense in marketing if firms could
not forecast consumers’ behavior. Because firms apply marketing and have
sales at reduced prices, they must feel that these means are effective.
The needs of human beings in the twenty-first century have not

changed, essentially, since ancient times. Even though the satisfaction
of needs of different times takes place with different kinds of goods,
the desires guiding human behavior have remained the same during the
history of mankind. Thus we can state the first axiom of economics:
Axiom 1: The Regular Needs of Human Beings. People like to satisfy

their needs, which are partly physiological and partly cultural. ˘
The DNA -structure in the genes of the human being is so stable that
no essential change has taken place within it during the last 10,000 years,
(Hawking 1993, p. 119). The change in human beings during the near his-
tory has thus taken place due to cultural factors, and not because of genetic
changes. According to Charles Darwin, cultural factors strengthen certain
genetic features. For example, in modern societies, people’s need for phys-
ical power has been replaced by the need for mental ability; being able to
build a machine or use one is more important than being able to work with
the hands. Cultural evolution thus somewhat changes human beings.
1.2.2 Business Competition and Rational Behavior
In a market economy, people and organizations have to compete with

each other. The reason for this is that the aggregate needs of people
exceed the resources available to satisfy them. To be successful in a
market economy requires smart behavior, which we call individual or
organizational rationality.
§: By rational behavior we understand decision-making consistent
with the goals of the decision-maker. ˘
Rational economic units do not have to be egoistic in any sense, but
their actions must be consistent with their goals. The goals of economic
units are, however, restricted by their willingness to manage in the
competition for success. For example, the goal of an individual to become
unemployed, or of a firm to retain all its workers while making losses and
lacking customers, is in conflict with economic success.
Jeremy Bentham (1963) writes: “Nature has placed mankind under the
governance of two sovereign masters, pain and pleasure. It is for them to
point out what we ought to do, as well as to determine what we shall do.
: : : They govern us in all we do, in all we say, in all we think.” However,
according to Becker (1976), “Although Bentham explicitly states that
the pleasure-pain calculus is applicable to what we ‘shall’ do as well as
to what we ‘ought’ to do, he was primarily interested in ‘ought’ and
did not develop a theory of actual human behavior with many testable
implications” (Becker 1976, p. 8). “I am saying that the economic
approach provides a valuable unified framework for understanding all
human behavior. : : : If this argument is correct, the economic approach
provides a unified framework for understanding behavior that has long
been sought by and eluded Bentham, Comte, Marx, and others” (ibid.,
p. 14). “Only after long reflection on this work and the rapidly growing
body of related work by others did I conclude that the economic approach
was applicable to all human behavior” (ibid., p. 8).
We continue the efforts of Bentham and Becker in seeking a unified
framework to model economic phenomena. Our definition of rationality,
however, deviates from that of Becker (1976) who understands by rational
behavior that an individual maximizes a well-defined function, such as
utility or profit. We do not insist that a rational economic unit is
maximizing a function; only that he (she/it) has goals he is willing to
reach, and makes decisions consistent with these goals. The connection
between decisions and goals may be too complicated to be expressed by
a smooth function. Some decisions require strategic thinking in a game-
theoretic way (consider a chess player), while others are simple ‘pickings
of the best alternative from a set of options’ without considering other
people’s reactions.
On this basis, irrational behavior either means decisions that do not
support the goals of the economic unit, random decision-making without
any goals, or such behavior that is generally considered foolish. An exam-
ple of the first kind of irrationality is where a candidate for a professorship
announces that if he is appointed, he will stop his research and concentrate
on family life; or where an entrepreneur announces that his/her com-
petitors’ products are not only cheaper but also of better quality. This
behavior conflicts with success in these competitions. The second type of
irrationality turns up in a sequence of choices where in the first case an
economic unit prefers choice A to B, and in the second case choice B to A,
without any change in the situation. This economic unit will lose money
in a sequence of bargains. An example of the last kind of irrationality is a
person who likes to pay more for a used, broken car without any intrinsic
value, rather than an identical new one. When this becomes known by
his/her neighbors, they most probably start to consider him stupid. These
kinds of controls over other people force those of us who like to manage,
and be respected by, others to behave in a ‘rational’ way.
A necessary condition for a private firm to survive in a market economy
is profitability, because non-profitable firms run out of financial resources
and have to close down. This forces firms to behave rationally. An
individual interested in his material well-being must also make rational
consumption and education decisions. We can thus believe that the
average behavior of economic units is rational, even though irrational
people exist, and people make errors every now and then due to uncer-
tainties in the decision-making situations they face. As a result of this, we
can state the second axiom of economics:
Axiom 2: Competitive Market System. A market system is based

on competition between individuals and firms with the aim to achieve
economic success. ˘
Corollary to Axioms 1 and 2. People’s and firms’ willingness to manage

in the competition for economic success forces them to behave in a
rational way. ˘
The observed regularities in economies can be demonstrated by the

following example. We can believe that a rational person does not
consciously throw money away. This claim can be applied to the situation
where a rational person is ready to make an investment that offers him/her
great enough profit both to cover the risk of the investment and the returns
that could potentially be obtained for the invested money elsewhere. This
guides finance towards those investment projects people believe to be
profitable. This implies that there always exists people ready to start a
business that is predicted to be highly profitable. If it were otherwise,
many potentially profitable businesses would never get started. Investors
who have made wise choices will get interest on their investments, and
those who have made bad choices will lose their money. In the long run,
only those investors who make good decisions will survive. Thus, people’s
unwillingness to throw away money leads, in a market economy, to the
situation where only a profitable business, able to pay returns to its
investors, survives.
The factors described here can be considered as regularities that charac-
terize the market system. These regularities are caused by free competition
and people’s willingness ‘not to throw away money’. Because people and
organizations are similar in their needs, the willingness of people and
organizations to reach their goals creates certain regularities in economic
behavior. These regularities are observed, modeled, and explained in
economics. The regularities in economic behavior that arise from the
competition between firms and people for economic success are in many
ways similar to those occurring in biology between plants and animals, in
the battle for survival. The difference between regularities in economies
and in biological Nature is: Economic units make conscious decisions
to reach their goals while plants and animals mainly act based on their
instincts.
1.2.3 The Birth of Organizations
The opinions of Adam Smith, Leon Walras, and William Stanley Jevons
of a human being as a creature interested in its own wealth is quite
an accurate description of human beings (see Sect. 1.4.1). This claim
is supported by the observation that humans have won against ‘other
animals’ in the battle for existence. This has required selfish behavior
on the part of humans in respect to ‘other animals’. Humans must also
have behaved relatively rationally during the history of mankind, because
otherwise Darwinian evolution would have guaranteed the defeat of
humans by ‘other animals’. One factor that has helped human beings in
this battle is the social character of humans that allows us to cooperate.
These days we have evidence that socialist economies that were built on
the assumption of a non-egoistic human being have not functioned well.
If competition does not exist, and people do not receive compensation for
their efforts, it is difficult to motivate people to work effectively. Behind
all creative innovation is usually hard competition to achieve success, that

requires innovations, and compensation for people in return for their own
ideas. In this way the Smithian thinking is still modern, see Sect. 1.4.1.
The cooperation between somewhat egoistic people and the birth of
human organizations can be explained by people’s common interest. This
is evident in a labor union and, due to specialization, nowadays most tasks
in firms require cooperation between workers with various skills. If people
are willing to pay for cars, others are able to design them and organize
the necessary finance, and there are people who want to earn their living
by working in a car factory, then the common interest of these people
is to build a car factory. Cooperative skills are required in developing the
efficiency of production, and people with poor cooperative skills will have
problems in their working career. However, the goals of organizations are
dominated by the goals of individuals in these organizations. If a person
thinks that for him/her it is better to resign from an organization than to
stay in it, he/she most probably resigns. On this basis, organizations that
do not support the goals of their members will lose their members over
time.
However, there exists charity organizations that present as the motive
for their operation something other than the selfish interests of their
members. This raises doubt about the honesty of these organizations,
because their behavior is contrary to the clearly observed selfishness of
human beings. The members working in charity organizations with little
compensation must feel this meaningful because otherwise they would
stop these activities. Thus, these organizations create satisfaction for their
members so that via these organizations, their members achieve goals they
feel important. There are other types of organizations, the goals of which
their members consider more important than their own welfare. Some
are ready to sacrifice their life for the independency of a nation, for a
political party, or for a religious community. The aims and goals of an
organization may thus be other than its members individualistic goals.
An organization may therefore behave differently than ‘the sum of its
independent members’. We return to this in the case of a labor union
in Chap. 6.
The examples above demonstrate that we can understand the behavior

of an organization by the goals of its members. For example, one of the
goals of a person may be to help other people. This person behaves as
rationally as an egoistic one, and we can forecast his/her behavior on this
basis. Thus, if we know the goals of a rational person, we can understand
and forecast his/her decision-making in various situations on this basis.
The third axiom of economics can then be stated as:
Axiom 3: The Birth of Organizations. People build organizations to

fulfill their selfish or common interests. ˘
The goals of an organization may conflict with those of its members,

however. For a firm to stay in business it requires hard work on the part
of its employees, and it may sometimes have to lay them off; while most
workers prefer easy work, a high salary, and a sure work contract. However,
workers tend to accept the terms in their work contract because the
alternative is not being employed. Workers gain from their ‘membership’
in a firm by earning their living, and thus the common interest of the
employer and employees is to keep the firm in business. If workers are
not satisfied, they will change their employer when a better one is found.
According to the above, human organizations can be analyzed as
purposeful units, the goals of which may differ from those of their
members. This holds in other sciences too. Two atoms of hydrogen
and one of oxygen construct a water molecule that behaves differently
than hydrogen and oxygen separately. A human baby born from the
combination of a sperm and an egg cell can also be analyzed as a
decision-making unit independent of these cells. If a person is identified as
an atom, a group of people is a molecule, the ‘chemical formula’ of which
explains its behavior. This analogy between chemistry and economics is
not analyzed further, however. In the text that follows, organizations are
analyzed as purposeful decision-making units, the goals of which may
conflict with those of their members. We state this in the form of the
fourth axiom of economics.
Axiom 4: The Goals of Organizations. Organizations have goals they

aim to reach that may differ from those of their individual members. ˘
1.2.4 The Principle of Modeling in Economics
Economic phenomena originate from the decisions of economic units:

consumers, investors, entrepreneurs, workers, and so on. Some difficult
to forecast events, such as weather, earthquakes, or accidents, also affect
economic phenomena, and economic units’ expectations of these are
accounted for in their decisions. Insurance and other means can be used
to protect people from the harms these events may create. In every
decision-making situation, the benefits and harms (pleasures and pains,
revenues and costs) occurring from the decision may not be clearly dis-
tinguished. However, every decision must be made using the information
the decision-maker has.
The above means that the ability to forecast economic events depends
on the ability to forecast the decisions of economic units. If we know the
goals of economic units, our definition of rationality in Sect. 1.2.2 gives
a basis on which to model the decisions. Economic units are purposeful,
with a number of goals, which ideally they like to achieve in the most
favorable way, with the minimum effort.
von Wright (1975, p. 2) separates the two main traditions in science,
Aristotelian and Galilean, as follows: “As to the views of scientific expla-
nation, the contrast between the two traditions is usually characterized
as causal versus teleological explanation. The first type of explanation
is also called mechanistic, the second finalistic”. The galilean or causal-
mechanistic tradition is nowadays called analytic, and the aristotelian is
called hermeneutic scientific methodology. As part of a critique toward
the Galilean scientific tradition in human sciences, von Wright (1975,
p. 96) suggested the following statement, called ‘Practical Syllogism’ (PS),
as the basis for understanding and explaining events in human sciences:
Practical Syllogism. Person A intends to bring about B. A considers that he

cannot bring about B unless he does C. Therefore A sets himself to do C.
In economics, however, the two scientific methodologies, Galilean

and Aristotelian, support each other. We make assumptions about the
behavior of economic units in the spirit of hermeneutics: economic units
have have goals they like to reach. Based on this assumption, we make
forecasts of their actions. Thus we use hermeneutics when modeling
the behavior of economic units, and through the modeled behavior we
explain and forecast the regularities observed in economies in the spirit
of analytical scientific philosophy. Such regularities are, for example: (1)
A price increase decreases the demand for a good, and (2) An increase in
the profitability of the production of a good increases its supply, and so
on. This harmony between the two scientific philosophies prevails also
between Hamiltonian and Newtonian mechanics in physics. The former
is based on the ‘willingness of the Nature’ to minimize energy, and the
latter on the regularities (laws of Nature) resulting from the ‘willingness
of the Nature to minimize energy’. Both methods, Hamiltonian and
Newtonian, yield the same equations of motion for moving particles, and
thus the two frameworks are consistent.
According to these examples, the two philosophies of science are not
conflicting, and we can question the conclusion of von Wright. We
can even define a basis for modeling in economics—consistent with the
Galilean scientific tradition—by transforming the Practical Syllogism as
follows.
Axiom 5: The Principle of Modeling in Economics. Economic unit

A wants state B to occur. A knows that state B would not occur unless C
is done. Then it is rational for A to arrange for C to be done, if he (she/it)
considers that his willingness to get state B to occur exceeds the costs and
trouble from arranging for C to be done. ˘
The extension we make to PS takes care of the following matters: (1)

Many people like to have a luxury car (state B), and we know that working
hard (do C) we could buy one, but for most of us the trouble of doing the
extra work outweighs the pleasure of getting the car, and (2) Unit A does
not have to do C himself, but he (she/it) can employ someone to do it
for him. A rational economic unit thus does not fulfill all of his (her/its)
intentions; rather he does a benefit-harm analysis for his intentions before

acting. This is in accordance with Bentham’s idea referred in Sect. 1.2.2.
The modified form of PS (Axiom 5)—together with the assump-
tion that economic units behave rationally—provides a foundation for
modeling economic phenomena, similar to that Newton gave for clas-
sical mechanics. In every decision-making, a rational decision-maker
compares the benefits (pleasure) and costs (pain) he expects to result
from the decision, and makes the decision on this basis. The ‘force’
the decision maker directs upon the adjusting variable is thus: ‘plea-
sure minus pain’. The concept of static friction force can be added
into this as well. By this we can explain that often the acting force
must exceed a limit value before people start changing their accustomed
behavior. The preliminary definitions for economic energy concepts
corresponding to potential and kinetic energy in physics can be found in
Estola and Dannenberg (2016).
1.3 Forecasting Human Behavior

It is impossible to forecast accurately the behavior of individuals and firms,
but the law of large numbers helps in forecasting average and aggregate
behavior of a group of economic units.
§: The law of large numbers in statistics refers to the phenomenon
where the same experiment is repeated many times. In this kind of exper-
iment, the following holds: (1) The probability of an outcome stabilizes
to the fraction: number of observations of the outcome divided by
the number of experiments, and (2) The distribution of the sum and
average of many random quantities is close to the Gaussian or normal
distribution (Moore and McCabe 1993, pp. 325–327, 398). ˘
If different decision makers make their decisions independently, we can
think of these decisions as independent experiments of the same test. In
this kind of an experiment, for example, which goods a customer buys
when he enters a shop, the sales of all goods are random variables that
follow the Gaussian distribution. The average and aggregate decisions
of a group of people is thus much easier to forecast than that of
individual ones. The same holds in statistical mechanics. Even though
we cannot forecast the motion of an individual molecule, the average

behavior of a group of molecules can be forecasted by using statistical
distributions.
Because irrational people are a minority, the average behavior of a
group of independent people is rational. An increase in the number of
observations of people’s decisions improves the forecast of average and
aggregate decisions of the group. This occurs because the needs of people
are similar, or the population can be classified into groups of people with
similar needs: children under school age, teenagers, and so forth. If the
shares of these groups in the total population are known, this helps in
predicting the behavior of the population. Of course aging changes the
needs of a person, but if the age distribution of a population stays constant,
the distribution of needs of the population stays constant.
1.3.1 Predictability of Economic Events
Economic events are caused by the decisions of economic units. The

predictability of an economic event depends on whether we can predict
the decisions of people and organizations taking part in the situation.
The easier it is to predict the benefits and costs resulting from people’s
decisions, the more easy it is to predict their decisions. Let us, as an
example, suppose that a gasoline station decides to sell gasoline at half
price for one day. It is easy to forecast that the station will be able to sell
the gasoline for the whole day at full capacity. If, on the other hand, a new
product is launched in the market, it is much more difficult to forecast its
sales.
Economic quantities change via consumers’ ‘money-voting’ mech-
anism. The products in shops are candidates in a continuous voting
process. Every customer votes the shopkeeper and the producer of the
good he buys by the price of the good. The more money a consumer
spends, the more his decisions affect the economic quantities he ‘votes’.
The decisions of a single consumer do not usually matter; the average
consumer rules. On this basis, consumers can affect economic quantities
via group behavior. If a large group of consumers decide not to buy the
products of a firm, this firm will eventually collapse. We have been able
to observe examples of this kind of boycott.
Predicting a consumer’s decisions on a certain day is difficult. This is
not needed, however, because shopkeepers are not interested in forecasting
the purchases of every customer on every day. Rather, they are interested
in forecasting the sales of different goods at certain time units. The
situation resembles statistical mechanics. It is much easier to forecast the
average or aggregate behavior of consumers (molecules) than that of a
single consumer (molecule). The ability to predict an economic event
thus depends on whether we can forecast the decision-making of an
average economic unit in the situation. Compare, for example, an average
customer, in forecasting the sales of a hamburger bar and that of an opera.
This illustrates the problems in forecasting stock market behavior where
we should be able to forecast the buying and selling decisions of an average
investor. The interactions between investors complicate this forecasting,
see Sect. 1.4.2.
1.4 Frameworks of Economic Modeling

1.4.1 The Neoclassical Framework and Its Critique
Current mainstream neoclassical economics was established in the

following publications: Theory of Political Economy (1871) by William
Stanley Jevons, Elements of Pure Economics (1874) by Leon Walras, and
Principles of Economics (1890) by Alfred Marshall, although various
other research also made contributions. This neoclassical economic
thinking has been criticised by socialists. From this critique was born
marxian political economics. According to socialists, economics and
philosophy cannot be separated. Karl Marx (1818–1883) stated that a
person understands matters via his work; the position of a person in
his/her work organization affects the reality he/she faces. In spite of the
critique of socialists against the market mechanism (according to Marx,
capitalism)—that was supported by neoclassical economists—socialists
were not able to create an alternative economic system. Their arguments
concentrated on pointing out weaknesses in neoclassical analysis and
market mechanism rather than developing an alternative economic system

or an alternative way to analyze economic phenomena.
According to Marx, an alternative economic system will arise naturally
after capitalism has destroyed itself due to its impossibilities. This will
take place after capital (the ownership of corporations) has become
concentrated in the hands of a small number of multinational companies,
and the exploitation of employees has increased their tolerance. With
the increasing misery of employees, capitalism will no longer be able to
function and will be replaced by socialism, an equilibrating society with
collective ownership. A concrete version of the socialist economic system
was built by Vladimir Ilyich Lenin in 1917. This took place in a forceful
way, however, because at the time of the Russian revolution, capitalism
had not been collapsed in a natural way.
Karl Marx predicted the birth of multinational companies, and his
forecast of the increasing misery of employees was also well-grounded.
The birth of trade unions in Western economies may have been one
reason that the ‘increasing misery’ did not come about, and we do
not know whether Marx’ prognosis would have come true without the
improvements in working conditions brought out by trade unions. The
prognosis of Marx may also have changed the behavior of entrepreneurs,
if they realized that it was well-grounded.
In social sciences, forecasts may change the behavior of societies in two
ways: (1) A forecast fulfills itself so that, for example, a rumor that a firm
has financial problems encourages its stockholders to sell their shares, and
in this way actually causes the financial crisis of the firm; or (2) A reliable
description of a horrifying situation towards which current behavior will
end up in the future may change people’s behavior so that the forecast
loses its accuracy. Insofar as ‘capitalism’ has not entered the point in time
where it destroys itself in a natural way, the role of the theories of Marx
in the development of Western economies is a difficult matter to study.
Marxian thinking did not replace neoclassical thinking, however, but
it started to live alongside neoclassical thinking. Partly as a result of
the socialists’ critique against neoclassical economics, the science called
sociology was born, where the critique against the market system still
has a role. Nowadays, sociology mainly studies the poorness, inequality,
and fairness problems in market economies, while economics studies the

functioning of the market system.
The main challenger for the dominance position of the neoclassical
school in economics has been the tradition developed by John Maynard
Keynes in the 1930s. The world economy, at that time mainly based
on the market system, was then in a great depression, and the liberal
neoclassical thinking could not show a way out of the situation. The
General Theory of Employment, Interest and Money (1936) by Keynes
brought to economics, alongside with microeconomic thinking, a new
branch called macroeconomic thinking. Microeconomics is based on the
decision-making of economic units, and in microeconomics, economic
events are explained on the basis of the actions of economic units.
Macroeconomics, on the other hand, analyzes economic phenomena at
aggregate level by using aggregate quantities. The basic philosophy in
Keynesian thinking is that the market system is by nature unstable, and
it may require intervention from the public sector. Based on Keynesian
thinking, active public expenditure policies have been applied in various
countries with the aim of stabilizing business cycles.
From the critique against the Keynesian tradition, a macroeconomic
school of thought called monetarism was developed. In monetarism, the
main task of the public sector is to control the stability of the money
market; the goods market and public expenditure policy that are stressed
in Keynesian tradition, have a minor role. The most famous monetarist
is Milton Friedman. In The Methodology of Positive Economics Friedman
(1953) claims that economics should abandon normative recommen-
dations common in Keynesian tradition and concentrate on analyzing
observed economic events. Friedman thus criticized Keynesian thinking
of normatism in the spirit of positive scientific methodology.
A characteristic feature of monetarism is economic liberalism. Accord-
ing to monetarists, the public sector should not actively participate in the
market mechanism. Monetarists consider that active fiscal and monetary
policies by the public sector are harmful to the functioning of the market
system. For example, central banks should operate in a stable and neutral
way so that their actions do not increase the natural instability in financial
markets.
According to Philip Mirowski (1989b), neoclassical economics arose

from a combination of two separate structures of thinking: (1) The
assumption that human being is a creature that likes to satisfy his/her
needs suggested by Adam Smith, and (2) The mathematical methods of
classical mechanics. For example, the concept of equilibrium in economics
was borrowed from physics by Canard in 1801 (Mirowski 1989a). Since
then, neoclassical economics has successfully applied this concept.
In the neoclassical framework, economic units are assumed to behave in
an optimal way. Consumers, firms, labor unions, and so forth are assumed
to maximize their utility, profit or some other target function, restricted by
the resources available for the units. The result from this modeling is that
an equilibrium state prevails in the model economy where all economic
units behave in an optimal way from their point of view.
By applying the neoclassical framework, however, it is difficult to
explain observed changes in economic quantities because economic units
do not like to change their optimal behavior. Due to this, in the neoclassi-
cal tradition dynamic phenomena is modeled by assuming that economic
units solve the time paths for their decision variables by dynamic opti-
mization. There is one weakness in this. In existing models of economic
units’ static and dynamic optimization, different target functions are
assumed for the units; thus the two frameworks conflict with each other,
see Estola (2013)
1.4.2 Econophysics
Econophysics is a relatively new multidisciplinary branch of economics

where the models developed in physics are applied in modeling economic
processes. Econophysics originated in the 1990s, simultaneously in two
different places. Eugene Stanley in Boston, USA, and Imre Kondor in
Budapest, Hungary, as professors of physics allowed their students to
write their graduate theses on statistical physics by applying financial
data. Because physicists had developed models to explain the complex
behavior arising from the interactions of a huge number of microscopic
particles, students of physics were interested in studying whether similar
models could explain the complex behavior in stock and exchange rate
markets. This analog has a reasonable foundation, because asset prices are
determined according to the buying and selling behavior of thousands
of investors. If investors are identified as particles, asset prices can be
modeled as the result of ‘collisions’ of these particles. The methodological
tools in this research are computer-based simulation models where
the behavior of every ‘particle’ is modeled in a system of thousands of
particles. These programs are let to run over numerous time units, after
which the macro-level quantities characterizing the state of the system
at every time unit are reported. In econophysics, this methodology is
called agent-based modeling where the investors are the agents. This
multidisciplinary subject was given the name ‘econophysics’ because the
object of the research is economics and not physics.
Of course, throughout history there have been several pieces of research
where the methods of physics have been applied in economic problems.
Examples of this can be seen in (1) The book by astronomer Adolphe
Quatelet in 1842: Physics of Social Phenomena: An Essay on Human
Development;(2) Mathematician Louis Bachelier developed in his PhD
thesis in 1900 the ‘theory of speculation’, where he applied Brownian
motion in financial markets. Brownian motion gets its name according to
botanist Robert Brown from his studies of microscopic particles in 1827,
and Albert Einstein applied Brownian motion in his celebrated paper in
1905; (3) Mathematician Benoit Mandelbrot (1963) observed power-law
scaling in commodity markets. Thus, econophysics had its pioneers over
the course of scientific history, but these works were almost completely
neglected by neoclassical economists until the rise of econophysics in the
1990s gave them renewed interest.
Chen and Li (2012) state, however, that the standard models of
neoclassical economics have, in any event, been borrowed from physics.
Thus the connection between the two sciences is not a new phenomenon.
According to Mirowski (1989a), the pioneers of neoclassical economics
like Jevons and Walras consistently imitated classical mechanics in trying
to develop similar concepts for economics as physics has invented in the
development of Newtonian and Lagrangian mechanics. The forces acting
upon economic quantities as well as the kinetic and potential energy
concepts for economic quantities were defined in various ways by the

pioneers of neoclassical economics, see for example Fisher (2007, original
work 1892). Jevons (1905) for example, wrote that “Life seems to be
nothing but a special form of energy which is manifested by in heat and
electricity and mechanical force.”
The economic energy concepts defined by the pioneers of neoclassical
economics did not turn out to be useful in modeling economic dynamics,
however. Thus, the only fruit from this analogy was the concept of equi-
librium, which is still the core methodological principle in the neoclassical
framework. Mas-Colell et al. (1995, p. 620) state this as: “A characteristic
feature that distinguishes economics from other scientific fields is that, for
us, the equations of equilibrium constitute the center of our discipline.
Other sciences, such as physics or even ecology, put comparatively more
emphasis on determination of dynamic laws of change. : : : Certainly there
are intuitive dynamic principles: if demand is larger than supply then price
will increase, if price is larger than marginal cost then production will
expand. : : : The difficulty is in transforming these informal principles
into precise dynamic laws.”
According to Nicolescu (2010), the first step in the development of a
common methodology between different sciences is that a metatheoretical
isomorphism (formal analog) is found between the disciplines. This
can be a similar statistical behavior of events in different sciences, or
a similar form of mathematical model applied in the sciences. These
two kinds of analogies are relatively easily found in sciences that make
observations from the real world and model these events using statistical
or mathematical tools. A limited number of statistical models exist that
are applied in modeling observations, and every science applies these
models. Dynamic events are also modeled in every science by differential
or difference equations, and thus it is highly probable that the same
model is applicable in two phenomena in seemingly unrelated sciences.
The mathematical form of the famous Black-Scholes equation for option
prices, for example, is identical with a heat flow equation in physics, and
thus the two events are of an isomorphic nature.
We already mentioned the isomorphism between the behavior of a huge
number of molecules and numerous investors in asset market. Another
isomorphism, which has been extensively applied in econophysics, is the

Ising model named after the physicist Ernst Ising (1925), which is a model
of ferromagnetism in statistical mechanics. The Ising model consists of
discrete particles that represent magnetic dipole moments of atomic spins
that can be in one of two states: C1 or 1. The spins are arranged in a
lattice where each spin can interact with its neighbors. In econophysics,
the Ising model has been applied in modeling the group behavior of
interacting investors, consumers, voters in elections, and so on. Spin C1
can be identified for example, as ‘yes’ and spin 1 as ‘no’ in any decision
of a human being, like whether to invest or not invest money, vote or not
vote for a political party, buy or not buy a product, and so on. A good
survey of these models can be found in Sornette (2014). In Sect. 1.4.4 we
will define a third isomorphism between economics and physics where the
decision-making of a human being is shown to have an identical structure
to the behavior of a steelyard in physics.
The way that current econophysics began in the 1990s explains why
the research in this discipline has concentrated in financial markets. In
that field of economics, enough data is available for rigorous testing of
probability distributions and statistical hypotheses of financial behavior.
Several textbooks on econophysics exist, for example, Mantegna and
Stanley (2000), McCauley (2004), and Takayasu (2006), where the core
research topics and results in this research are presented; see also articles
Chen and Li (2012) and Ausloos (2013) Because econophysics research
requires knowledge of statistical physics, the majority of researchers from
economics backgrounds have not been interested in this topic.
There is some competition between researchers with economics or
finance background, and those with physics background. Jovanovic and
Schinckus (2013) explain this battle as follows. Economists apply mainly
normal (or Gaussian) distribution for asset returns, and they consider
expected return as the objective of investors and standard deviation as
the corresponding risk; see, for example Markowitz (1952, 1959). Econo-
physicists, on the other hand, apply Lévy or power law distributions for
asset returns because these give a higher probability for extreme values
like asset price collapses and sudden upward jumps. The problem in
applying Lévy or power law distributions is that their standard deviations
are in most cases unlimited, and thus their standard deviations cannot be
used as a measure of risk. Thus Lévy and power law distributions cause
a conflict between these two groups of researchers, and other areas of
financial economics, for instance, portfolio selection, capital asset pricing,
and the Black-Scholes model are also in conflict with these distributions.
This conflict is so serious that one of the leading econophysicists, Joseph
McCauley, states that “econophysicists are safer to ignore the lessons
taught in standard economics texts”, McCauley (2006, p. 602).
This conflict between economists and physicists has been considered
insolvable, but Jovanovic and Schinckus (2013) have proposed a solution.
If the Gaussian distribution applied in economics is changed to truncated
stable Lévy distribution, that distribution has a finite standard error. This
distribution can be accepted by both camps, and it has been shown to give
consistent results with observed behavior; see, for example Mariani and
Liu (2007), Mantegna (1991), and Lux (1996).
Needless to say, there still exists a deep gulf between people studying
financial markets with background in economics and in physics. Nat-
urally, some economists—that ‘have been there longer’—feel that the
newcomers from physics with new research tools are not welcome because
they threaten the position of economists as authorities in these matters.
However, scientific competition should be open for new ideas, and
thus economists should compete about superiority of research methods
rather than keep their own seminars of financial research. The new
concepts econophysicists have brought to financial markets research are
universality and scaling, power law distribution, and network theory
in modeling; see Ausloos (2013). There exists also examples of applying
thermodynamic framework in economic modeling; see, for example
Drăgulescu and Yakovenko (2000) and Kusmartsev (2011). Thus the
research in econophysics is very broad and applies almost all principles
of modeling invented in physics. An extension of econophysics to social
sciences has been made by Serge Galam (2004), called Sociophysics.
Sociophysics applies similar tools as econophysics in modeling the group
behavior of people, and it has been used in explaining the voting and con-
sumer behavior of a group of persons. Sociophysics, however, is beyond
the scope of this book, and thus it is not treated here in a detailed way.
As this section reveals, the neoclassical framework of economics is
lacking a consistent framework for modeling dynamic events. The gap
between static neoclassical modeling in economics and nonlinear dynamic

models in econophysics creates a wall between these two disciplines. In
this book we introduce a new framework to model dynamic economic
events in the neoclassical framework, and we have shown in Estola and
Dannenberg (2016) that this Newtonian framework of microeconomics
is consistent with the corresponding Lagrangian framework. Thus we
have defined the economic concepts corresponding to kinetic and poten-
tial energy in physics as well as the ’market forces’ acting in economies. By
using these concepts, the Lagrangian framework gives the same equations
of motion as the Newtonian equations to be presented in this book. Thus
we hope that we have completed the work of the pioneers of neoclassical
economics in finding the parallels between modeling principles in
these two sciences. If this link between the two sciences is accepted, all
physics that is based on Newtonian mechanics—like classical statistical
and quantum mechanics—can be obtained as direct extensions of this
framework along the lines developed in physics.
1.4.3 A Dynamic Extension to Neoclassical Economics
We stated earlier that the equilibrium concept in the neoclassical frame-

work has been borrowed from physics. On this basis, we can ask whether
we could apply similar principles as those used in physics to mod-
eling dynamic economic phenomena. This thinking is supported by
many economists who use the term ‘force’ when they explain why a
change occurred: see Lucas (1988) as an example. The principles of this
dynamization are given in this section, and in the rest of the book we
apply this framework.
Economics needs a framework for modeling by which we can model
static and dynamic behavior. Statics should correspond to a stationary
state in general dynamic behavior. Rational economic units change their
behavior if they believe that this would better their situation. A ratio-
nal decision maker changes his decision variables toward the preferred
direction until the equilibrium condition holds; that is, no further gain
is obtained. In this we define the concept of ‘economic force’ that
causes this dynamic. The ‘force’ by which an economic unit acts upon
an economic quantity is the ‘pleasure minus pain’ that results for the
decision maker from his decision, and the equilibrium state (the zero-
force condition) corresponds to the optimal situation. This framework
can be applied in the decision-making of every economic unit concerning
any quantity, as Becker (1976) and Bentham (1963) proposed.
According to the above, we change the assumption in the neoclassical
framework to the following: Economic units have goals they like to
reach and make decisions to obtain these goals. Assuming this, we can
explain the observed evolution of economic quantities as well as possible
equilibrium states that may change with time. This is in accordance with
the principle of modeling in economics given in Sect. 1.2.4, and also in
agent-based models applied in econophysics.
From the point of view of scientific thinking, the proposed framework
deviates from the neoclassical framework in that an economy is not
assumed to be in an equilibrium state. Rather, an economy consists of
various evolutionary processes driven by different factors. These factors
consist of the quantities economic units compare in their decision-
making. In an evolutionary economy, nothing stays fixed for long. New
products and technologies are created, the qualities of existing products
evolve with time, and people’s preferences change with time and wealth.
These elements complicate the forecasting of economic events.
We can think that dynamic economic events are caused by conflicting
‘economic forces’. In an equilibrium state, the forces acting upon opposite
directions cancel each other out. For example, we can think that employers
direct a negative, and employees a positive, force component upon wage
level, and the wage level changes according to the net or resultant force
acting upon it. An equilibrium wage balances these two force components.
The definition of economic forces is important because economic
forces are the reasons behind observed changes in economic quantities.
If, however, a control variable exists in the resultant force acting upon
an economic quantity, that an outside controller (an economic policy-
maker, a legislator, the manager of a firm, etc.) can adjust, the force is
controllable. In these cases, we can apply control theory in modeling
economic policy-making. Various kinds of inertial factors resist changes
in economic quantities. The easier it is to change an economic quantity,
the more it will fluctuate with time. This is consistent with observations
of the variability of economic quantities. Exchange rates, interest rates,
stock prices, goods prices, and growth rates of production of goods vary
with different daily and weekly rates.
In economics, we cannot hope to be able to reach the same degree of
predictability as can be obtained in certain physical processes by applying
Newtonian laws. The reason for this is that economic forces change
over time. The innovation of a new technology may change consumers’
preferences or firms’ costs so that some goods become more popular and
others become worthless. The longer the time unit, the more difficult it
is to accurately forecast economic events due to these reasons.
As a summary, we can state the foundation for modeling economic
phenomena as follows:
1. Define the economic units involved in the studied phenomenon, and

define the goals the units are aiming to reach in the situation.
2. Model the decision-making of economic units by their comparison
of the benefits (pleasure) and the costs (pain) they expect from their
decisions.
3. Define the force by which an individual unit is acting upon an adjusting
variable(s) in the situation as the difference between expected benefits
and costs arising for the decision maker from his (her/its) decision. In
the decision-making, the behavior of other units may or may not be
taken in account. In the former case the model is of ‘game theoretic
type’, and in the latter case of ‘perfect competition type’.
4. Model the dynamics of the adjusting variable(s) as a function of the
resultant of all acting forces (the vector sum of benefits and costs the
economic units expect to gain from their decisions) in the situation.
These modeling principles are applied throughout this book.
1.4.4 Decision-Making Steelyard
A common way to describe the decision-making process of an economic

unit is that he (she/it) must ‘weighs’ the benefits and harms resulting from
his decision. This is based on an isomorphism between two situations:
the decision-making of human beings and the comparison of weights
of two bodies by a steelyard. The words ‘comparing’ and ‘weighing’
alternatives are usually considered as synonyms. In physics, the resultant

force creating torque into the pivot point of the arm of a steelyard depends
on the difference in the weights of the two scales. In decision-making, the
benefits and harms are assumed to be set in the two scales of a steelyard,
and the decision is made on the basis of which of the scales is ‘heavier’.
To be able to compare the benefits and harms in a decision, they must
be measured in equal units; in units of pleasure (utility) or money. The
decision-making in economic problems always concerns whether to buy,
sell, produce, consume or change something; that is, do we increase or
decrease the amount of buying, selling, production, consumption, unit
price, and so on, of a good. Due to these decisions, the corresponding
quantities change, which are the economic events we observe.
Let us demonstrate this by one example. Suppose an entrepreneur is
considering whether or not to produce an extra unit of a good at time
moment t. In the decision-making, the entrepreneur sets his estimates of
the selling price and production costs of the good in the two scales of his
‘decision-making steelyard’. Suppose the steelyard is of the pulley type
as in Fig. 1.1, the length of the string is long enough (practically infinite)
so that the two masses never hit the pulley, and the pulley rotates only
counterclockwise with discrete steps. If the estimated unit price (mass m2 )
‘weighs’ more, the pulley rotates counterclockwise so that the deviation
s(t)
m1
m2
Fig. 1.1 Pulley type steelyard

of mass m2 from the pulley s(t)—this corresponds to the accumulated

amount of produced goods, for example, 1 cm = 1 unit—increases by one
unit. If the unit costs (mass m1 ) weighs more, mass m2 stays still.
Due to this decision, the position of mass m2 either changes or not,
and the cause of this change is the difference in the entrepreneur’s
estimated selling price and unit costs. If the measurement unit for the
amount of production is unit or kg, then that of the cause is e=unit or
e=kg, respectively. This decision-making steelyard operates according to
physical laws, and we use it here in modeling the decision-making of a
rational person.
Next, we assume a firm producing a good with constant daily velocity
(flow) with unit unit=day, and the entrepreneur is planning whether to
change this flow by a certain amount. Now the pulley rotates counter-
clockwise with constant velocity; assuming kinetic friction in the system
prohibits the permanent acceleration of the pulley in the case m2 weighs
more than m1 . The decision is made at the end of each day according
to daily sales. The entrepreneur puts into the two scales of his decision-
making steelyard the daily revenues and costs. If the revenue side weighs
more (less), mass m2 gains positive (negative) acceleration, and so the daily
flow of production changes as long as a new equilibrium is reached. This
decision-making steelyard explains the acceleration of production in units
unit=day2 , see Sects. 2.7.4–2.7.5.
Various inertial factors can be included in this modeling. First, we may
require that expected benefits must exceed expected harms by a certain
amount before the decision is made. This corresponds to the static friction
of the pulley. This way we can explain that usually firms (and people) do
not change their behavior unless the reasons become compelling enough,
that is, the acting force component exceeds a limit. Second, we may
assume that the scales of the steelyard are so large and light that air
resists their motion. This creates kinetic friction of the form ks0 (t) in
the system, where s0 (t) is the instantaneous velocity of mass m2 and k
is a positive constant. We may also add a delay in the model—the time
taken to produce one good—which reflects the inertia in the production
process.
In the above-described way, we can construct various decision-making

steelyards where uncertainty is present, that include frictional factors
and work analogously with physical laws, and that can be used in
explaining changes in economic quantities. These decision-making steel-
yards are based on the principle of modeling in economics given in
Sect. 1.2.4.
The decision-making steelyard was introduced to show one exact possi-
ble analogue between modeling in economics and in classical mechanics.
Some differences though exist. In Newtonian mechanics, a ‘deterministic’
force is the cause and the acceleration of a particle is the effect. In
economics, the modeling is based on causal relations too. However, the
rule force causes acceleration does not generally hold up in economics.
We have demonstrated earlier that ‘economic forces’ cause changes in
economic quantities, and ‘economic forces’ usually contain uncertainties.
‘Economic laws’ are thus probabilistic while physical laws in classical
mechanics are deterministic. However, physical laws in classical statistical
and in quantum mechanics are probabilistic too. In general, we can say
that economic forces cause changes in economic quantities. It is a
matter of taste whether we talk about reasons and effects, pressures for
changes and effects, or forces and effects. In this book we talk about forces
and effects because here we use the theoretical term ‘force’ exactly in the
same way it is used in physics.
The school of thought called econophysics applies the methods of sta-
tistical physics in modeling financial market phenomena, see Sect. 1.4.2.
In econophysics, the modeling is based on the application of models
constructed for physical processes—like the Ising model and agent-based
simulation models—to describe economic processes too. The decision-
making steelyard was introduced here to show one more isomorphism
between structures of events in the two sciences, and we argue the
applicability of similar models in these two sciences on this basis. By the
analogy between the decision-making of a rational economic unit and the
motion of a steelyard according to Newton’s laws, we motivate the reader
to study the expressions for economic forces to be introduced later.
1.5 A Summary of the Methodological Basis

of Economics
1. The willingness of people and organizations to stay ‘alive’ and be
successful causes them, via their organic body and mental thinking,
certain goals and needs they like to reach and satisfy.
2. People and organizations are rational and purposeful units with respect
to their needs and goals that may also be non-egoistic.
3. Organizations can be analyzed as decision-making units, the goals of
which may differ from those of their members.
4. Market mechanism is based on competition between people and firms.
5. In order to achieve a certain state of matters, an economic unit must
make a decision whether or not to take actions to achieve that state.
6. Every economic event, that in principle is explainable, is caused by the
actions of economic units to attain the states of matter they prefer.
7. The decision-making of an economic unit is based on the expected
benefits (revenues) and harms (costs) the decision causes for that unit.
8. The difference: benefit (revenue)–harm (cost) is the force a decision-
maker directs upon the quantity concerned in the decision.
9. An economic quantity changes due to the resultant force acting upon
it caused by various economic units. The resultant force may cause
an equilibrium state, an adjustment toward an equilibrium state, or
continuous motion like permanent growth, permanent increase in the
wealth of an economic unit, or some kind of cyclical motion as is
observed in stock markets.
10. The predictability of an economic event depends on whether the
decisions of people and organizations affecting the matter, which may
or may not depend on each other, can be predicted.
References
Ausloos, M. (2013, July). Econophysics: Comments on a few applica-
tions, successes, methods, & models. Society & Management Review, 2(2),
101–115.
Bachelier, L. (1900). Théorie de la spéculation. Annales Scientifiques de l’École

Normale Supérieure, 3(17), 21–86 [English translation in P. Cootnerr (Ed.)
The random character of stock prices. MIT Press, 1964, Reprinted Risk Books,
London 2000].
Becker, G. S. (1976). The economic approach to human behavior. Chicago: The
University of Chicago Press.
Bentham, J. (1963). An introduction to the principles of morals and legislation.
New York: Hafner.
Chen, S.-H., & Li, S.-P. (2012). Econophysics: Bridges over a turbulent current.
International Review of Financial Analysis, 23, 1–10.
DrMagulescu, A., & Yakovenko, V. M. (2000). Statistical mechanics of money.
The European Physical Journal B, 17, 723–729.
Einstein, A. (1905). On the movement of small particles suspended in stationary
liquids required by the molecular-kinetic theory of heat. Annalen der Physik,
Series 4, 17, 549–560.
Estola, M. (2011). Measuring the growth and the structural change in a
Economy, 4(1), 23–44.
Estola, M. (2013). Consistent and inconsistent ways to dynamize the neoclassical
theory. Hyperion International Journal of Econophysics & New Economy, 6 (1).
Estola, M., & Dannenberg, A. A. (2016). Newtonian and Lagrangian mechanics
of a production system. Hyperion International Journal of Econophysics & New
Economy, 9(2), 7–26.
Fisher, I. (2007). Mathematical investigations in the theory of value and prices.
New York: Cosimo Publications [Original work at 1892].
Friedman, M. (1953). The methodology of positive economics. In Essays in
positive economics. Chicago: University of Chicago Press.
Galam, S. (2004). Sociophysics: A personal testimony. Physica A, 336 (1–2),
49–55.
Hawking, S. (1993). Black holes and baby universes. London: Bantam.
Ising, E. (1925). Betrag zur Theorie des Ferromagnetismus. Zeitschrift für Physik,
31, 253–258.
Jevons, W. (1905). The principles of science (2nd ed.). London: Macmillan.
Jovanovic, F., & Schinckus, C. (2013). Towards a transdisciplinary econophysics.
Journal of Economic Methodology, 20(2), 164–183.
Keynes, J. M. (1936). The general theory of employment, interest, and money.
Cambridge: Macmillan Cambridge University Press.
Kusmartsev, F. V. (2011). Statistical mechanics of economics I. Physics Letters A,

375(6), 966–973.
Lucas, R. E. (1988). On the Mechanics of economic development. Journal of
Monetary Economics, 22, 3–42.
Lux, T. (1996). The stable Paretian hypothesis and the frequency of large returns:
An examination of major German stocks. Applied Financial Economics, 6 (6),
463–475.
Mandelbrot, B. B. (1963). The variation of certain speculative prices. The Journal
of Business, 36 (4), 394–419.
Mantegna, R. N. (1991). Lévy walks and enhanced diffusion in Milan stock
exchange. Physics A, 179(1), 232–242.
Mantegna, R. N., & Stanley, E. (2000). An introduction to econophysics. New
York: Cambridge University Press.
Mariani, M. C., & Liu, Y. (2007). Normalized truncated Lévy walks applied to
the study of financial indices. Physica A, 377, 590–598.
Markowitz, H. M. (1952). Portfolio delection. Journal of Finance, 7 (1), 77–91.
Markovitz, H. M. (1959). Portfolio selection; efficient diversification of investments.
Wiley: New York.
Marshall, A. (1890). Principles of economics. London: Macmillan and Company
Mas-Colell, A., Whinston, M. D., & Green, J. R. (1995). Microeconomic theory.
Oxford: Oxford University Press.
McCauley, J. L. (2004). Dynamics of markets: Econophysics and Finance. Cam-
bridge: Cambridge University Press.
McCauley, J. L. (2006). Response to “worrying trends in econophysics”. Physica
A, 371(1), 601–609.
Mirowski, P. (1989a). The rise and fall of the concept of equilibrium in economic
analysis. Recherches Economiques du Louvain, 55(4), 447–468.
Mirowski, P. (1989b). More heat than light, economics as social physics, physics as
nature’s economics. Cambridge: Cambridge University Press.
Moore, D.S. & G.P. McCabe, 1993. Introduction to the Practice of Statistics, W.
H. Freeman and Company, USA. pp. 325–327 and p. 398.
Nicolescu, B. (2010). Methodology of transdisciplinary - Levels of reality, logic of
the included middle and complexity. Transdisciplinary Journal of Engineering
and Science, 1(1), 19–38.
Quetelet, M. A. (1842). A treatise on man: The development of his faculties.
Edinburg: William and Robert Chambers.
Sornette, D. (2014). Physics and financial economics (1776–2014) puzzles, ising
and agent-based models. arXiv:1404.0243v1 [q-fin.GN] 1 Apr 2014.
Takayasu, H. (2006). Practical fruits of econophysics. In H. Takayasu (Ed.),

Proceedings of the third Nikkei econophysics symposium. Tokyo: Springer.
von Wright, G. H. (1975). Explanation and understanding. London: Routledge
and Kegan Paul.
Walras, L. (1874). Elements of pure economics (W. Jaffe, Trans., 1954). London:
Georg Allen and Unwin.
2
Measuring in Economics
An exact science of economics should fulfill the following requirements:
1. A system of measurement units by which economic events can be

measured.
2. A coherent theory of the behavior of people and organizations by which
economic events can be explained, which is changed if theoretical or
empirical inconsistencies are observed.
3. General principles for modeling economic events, where static and
dynamic phenomena are modeled by using a single framework.
4. Theoretical models expressed in principle with measurable quantities
to allow their rigorous testing.
5. All simplifications made in the modeling are reported and explained.
6. Macro-level models are based either on the micro-level behavior of
economic units, or on the relations that hold at aggregate level in
economies.
For example, in classical statistical fluid mechanics, macro-level events

are modeled by using either of the following principles: (1) Average
micro-level behavior of molecules (the kinetic theory of an ideal gas),

DOI 10.1007/978-3-319-46879-2_2
or (2) The principles of thermodynamics at macro-level that are lacking

a micro-basis. Thermodynamics is based on the conservation of energy
where the existence of atoms or molecules is not essential. The concept
of temperature, for example, is a macro-level quantity that cannot be
measured at micro-level. An increase in average velocity of molecules at
micro-level corresponds to an increase in temperature at macro-level. Also
in economics, macro-level phenomena can be modeled by using different
principles to those used at micro-level. However, in this book we do not
discuss macroeconomics.
In the following we try to fulfill the requirements for quantitative
microeconomics. We start with a system of measurement units for eco-
nomics.
2.1 Principles of Dimensional Analysis

In all measuring, a distinction must be made between qualities, mag-
nitudes, and quantities (Allen 1938, pp. 10–11). These all can be called
variables.
§: A variable is a defined concept with a specific content that may vary
with time and context. ˘
For example, with specific definitions the following concepts are vari-
ables: (1) The state of welfare of an economy, (2) The quality of a
product, and (3) The external beauty of a product. Common to all these
variables is that we cannot operate with them quantitatively, that is, we
cannot add, multiply, divide and so on with them. These variables are
qualitative descriptions of the state of the matter concerned, an economy
or a product.
However, if we can define a natural order for a variable, we call it
a magnitude. Magnitudes can be arranged in order, but their addition
or multiplication is not meaningful. An example of a magnitude is the
rank order of military persons. The situation changes when a specific
measurement unit can be defined for a variable. de Jong (1967, p. 7) gives
the following definition.
Definition: A quantity is a number of measurement multiplied by a
unit of measurement.
For example, the weight of a person 75 (kg) is a quantity; 75 is the

number and kg the unit of measurement.1 Also 10 (second) is a quantity;
10 is the number and second the unit of measurement.
To understand the relation between a dimension and a measurement
unit, we need to define some elementary concepts of dimensional
analysis. Although many economists have touched on the subject in
their works, de Jong (1967) is the only economist whose specific topic is
dimensional analysis in economics. Here we present only those concepts
and ideas of dimensional analysis relevant in this context. Those interested
in exploring this topic further can turn to the excellent book of de
Jong.
§: Dimensional analysis is an algebraic theory of measurement units.
A dimension defines a set of additive quantities. ˘
For instance, length is a dimension in the measurement system of
classical mechanics. Although the measurement units of kilometer and
mile both belong in the dimension of length, they are not directly
additive. One dimension may thus contain various measurement units.
Dimensional analysis insists that all measurement units belonging in
the same dimension have transformation rules by which a quantity
measured in certain units can be transformed to be measured in another
units.
For example, the transformation equation between miles and meters is
1 (mi) D 1609:38 (m). We can express this equation as:

1 mi
1 D 1609:38 (m=mi) , 1D : (2.1)
1609:38 m
Using the first form of Eq. (2.1) we can add 3000 meters to 5 miles as:
3000 (m) C 5 (mi) D 3000 (m) C 5 (mi) 1

D 3000 (m) C 5 (mi) 1609:38 (m=mi)
D 3000 (m) C 5 1609:38 (m) D 11;046:9 (m):
1
Measurement units are in brackets after the numbers of measurement.
Using the second form of Eq. (2.1) we can add the two lengths in miles as:
3000 (m) C 5 (mi) D 3000 (m) 1 C 5 (mi)

D 3000 (m) (1=1609:38) (mi=m) C 5 (mi)
D 6:86407 (mi):
Later on we will see that all monetary values measured in units of

any currency belong in the dimension of money, and are thus additive
quantities. With a transformation equation—for example 1 (GBP) D
2:5 (USD), the nominal exchange rate between the currencies—we can
add x (USD) and y (GBP) in the units of USD or GBP. On the other
hand, adding quantities with unit USD to those with unit kg is not
meaningful.
In every science where measurements are made some elementary or
fundamental dimensions must be chosen. These are called primary
dimensions. In modern mechanics, for instance, mass, length and time
are chosen as primary dimensions (de Jong 1967, p. 12). The basic units of
these primary dimensions are, for example, the kilogram, the centimeter
and the second. These units define a measurement system that covers
all physical quantities measured as products of powers of these units. For
example, the velocity of a moving object — distance divided by time — is
measured in centimetres per second: cm=sec. Velocity is thus a secondary
dimension derived from the primary dimensions as length/time.
Primary dimensions are the basic elements of a measurement sys-
tem. To be able to measure all phenomena of a science occurring in
its sphere of events, the primary dimensions must be such that they
‘span a measurement system’ that covers the whole sphere of events.
In mechanics, all quantities can be measured by the above given three
primary dimensions and secondary dimensions, that are products of
powers of the primary dimensions. Later we will see that in economics
every unit price measured, for example in units USD=kg D USD kg1
belongs in the secondary2 dimension Œmoney=real D Œmoney real1 .

Analogously, in the metric system of mechanics the measurement unit
of force is obtained as the product of powers of the primary units as:
1 Newton D 1 kg m sec2 .
A special case of a secondary dimension occurs if a quantity belonging
in dimension ŒA is divided by another such quantity. The resulting ratio
is an abstract or pure number because the unit of measurement in the
numerator is cancelled by that in the denominator. The dimension of such
quantity can be expressed as: ŒA=ŒA D ŒA11 D ŒA0 D Œ1. Symbol Œ1
thus indicates a dimensionless quantity. For example, 2 (cm)/5 (mm) =
2 (10 mm)/5 (mm) = 2 10 (mm)/5 (mm) = 4 is a dimensionless quantity.
The statement that dimensionless quantities are abstract numbers may not
be reversed. There do exist abstract numbers that are not dimensionless
quantities simply because they are the result of counting and not of
dividing two measured quantities. This second kind of abstract numbers
are called tallies or counts (ibid. p. 14).
The velocity of money in circulation V can be defined as the tally
of the number of times a unit of money changes hands in a time
unit. The dimension of V is then [‘tally’=T], whereby ŒT is denoted
the dimension of time. We can also call V the frequency of money in
circulation. In Sect. 2.6.3 we will see that the dimension of interest rate r
is [‘dimensionless’=T]. Dimensions [‘tally’=T] and [‘dimensionless’=T]
are not identical, even though they have equal measurement units when
time has an equal unit. Although the identification of the dimensions of
V and r is legitimate from the point of view of the algebraic theory of
dimensional analysis (ibid. p. 168), it makes no sense adding interest rate
and the frequency of money in circulation.
Dimensional analysis is based on Bridgman’s axiom that can be
presented as follows. The ratio of two quantities with the same dimension
must be independent of the unit of measurement of the dimension
(de Jong 1967, p. 141).
2
In dimensional analysis, a dimension is denoted by brackets.
Suppose that weighing two masses in kilograms gives results x and y.

Multiplying the measurement unit kilogram by the number ˛, ˛ ¤ 0; 1,
we get two new quantities, x=˛ and y=˛. This does not change the ratio
of the measured masses (x=˛)=(y=˛) D x=y, however, which is insisted
on by Bridgman’s axiom. In Sects. 10.18.3–10.18.4 we will see that all
quantities are measured at least at interval scale, and a unique zero point
means that the quantity is measured at ratio scale.
Bridgman’s axiom holds for power function, (x=˛)ˇ =(y=˛)ˇ D (x=y)ˇ ,
ˇ integer, and various other transformations, but it does not hold
for transcendental functions. The reason for this is that in general
ln(x=˛)= ln(y=˛) ¤ ln(x)= ln(y). This can be understood by the
following power series
x3 x5 x7 x2 x3
Sin(x) D x C C ; ex D 1 C x C C C ;
3Š 5Š 7Š 2Š 3Š
(x 1)2 (x 1)3 (x 1)4
ln(x) D (x 1) C C ;
2 3 4
where 2Š D 2 1; 3Š D 3 2 1 and so on. These power series are

meaningful only with a dimensionless argument x; different powers of x
are non-additive in the case x has a measurement unit. For this reason,
the arguments of transcendental functions applied in physics are always
dimensionless quantities and this should hold for economics too.
Using dimensional analysis we can check that theoretical concepts cor-
respond to their empirical counterparts, and that dimensional equations
are written in a correct form. Before we enter into the application of
dimensional analysis in economics, we need a few more definitions.
§: Abstract numbers, dimensionless quantities and quantities with
measurement units, together, are called scalars. ˘
Scalars are, for example, 5 (cm), 10 (USD), 4 (USD/GBP) as well as all
numbers such as 3 or 4.6, etc.
§: Constants appearing in fundamental equations are called dimen-
sional constants, if they are not abstract numbers. An example is the
Newton’s constant of gravity: 6:67 1011 (m3 kg1 sec2 ). ˘
§: If a nonzero quantity can be found, multiplying by which both sides

of an equation makes the equation dimensionless, the equation is called
dimensionally homogeneous. ˘
Practically, an equation is dimensionally homogeneous if both sides
of the equation have equal dimension. Dimensional homogeneity does
not prove that the equation is true or meaningful in any sense; it only
guarantees that it is dimensionally well-defined. On the other hand, any
dimensional equation that is not dimensionally homogeneous, cannot be
correct. This gives us a means of checking our calculations because all
well-defined transformations of a dimensionally homogeneous equation
retain the homogeneity.
2.2 A Measurement System for Economics

The basic quantities we need in measuring in economics are: the amount
of goods, the monetary values of various things, and time. The primary
dimensions of a measurement system for economics then consists of
these additive quantities denoted by ŒR, ŒM, and ŒT, respectively. This
notation is taken from de Jong (1967), and it refers to the words Real,
Money and Time. In economics, we can still define a fourth primary
dimension called Satisfaction (or utility or ‘happiness’) denoted by ŒS,
see Sect. 2.2.4.
2.2.1 Measurement of Volume of Goods
Depending on the good we are measuring, the measurement unit of the

volume of the good can be kilogram, meter, meter2 , meter3 , liter or
purely real numbers. These units are denoted as in physics: kilogram by
kg, meter by m, square meter by m2 , cubic meter by m3 , liter by l, and real
numbers by unit. The reason for these units is that they are actually used
in measuring the amounts of goods. Treating real numbers as a separate
unit emphasizes that counting is one way to measure the amount of a
good.
Now, every good produced or sold in an economy defines an own

dimension for the quantity units of this good. Similarly, as it is not
meaningful to directly add a certain amount of sugar measured in units
kilogram to that measured in units liter, it is also not meaningful to add
a quantity of potatoes measured in units kilogram to a quantity of carrots
measured in units kilogram if the aim is to measure the amount of these
goods. If we are interested in the common weight of a certain amount of
potatoes and carrots, then the above adding is meaningful.
Even though all quantity units kg; m; m2 ; m3 ; l; unit, and so forth,
used in measuring the amount of good i belong in dimension ŒRi , they
are not directly additive. A dimension was defined as a set of additive
quantities, however. We stated earlier that with fixed transformation rules,
lengths measured in units meter and mile can be added together and thus
belong in the same dimension. When we know the good, the amount of
which we are measuring, we can find transformation rules that make the
various quantity units of this good additive. We can weigh a quantity of
cotton measured in units m to get its kg measure, for example, or we can
pack it in boxes of known volume to get its m3 measure, and so on. With
these transformation rules, we can add quantities of cotton measured in
units m, kg, and so forth. This shows that these quantity units belong in
the same dimension, even though they are not directly additive. Although
we later talk about the dimension of the volume of goods (denoted by ŒR),
this dimension contains as many separate dimensions as there are different
goods in the economy.
2.2.2 Measurement of Time
It is essential to notice that time differences, not time moments, are

additive quantities that belong in the dimension of time differences. For
example, adding time moment 11.05 to moment 21.45 gives time moment
32.50 that does not exist. On the other hand, adding 5 minutes to 1 hour
by using the transformation rule 1 (hour) D 60 (min) gives 65 minutes,
a quantity that belongs in the dimension of time differences.
In some analyses, it is practical to measure time differences (time), by

several units. Labor input, for example, can be measured by the amount
of hours worked in a year with unit hour=year, which is a dimensionless
quantity. Because of this, in the following we measure time by several
units: seconds s, hours h, days d, weeks week, months mn, years y, and
so on. The transformation rules between these units are: 1 (y) = 12 (mn),
1 (week) = 7 (d), 1 (d) = 24 (h). The problem with these units is that the
number of days in one month varies. This problem is solved in financial
accounting by defining 1 (mn) = 30 (d), which implies 1 (y) = 12 (mn) =
12 (mn) 30 (d=mn) = 12 30 (d) = 360 (d). Another possibility is 1 (y)
= 52 (week), which gives 1 (y) = 52 (week) = 52 (week) 7(d=week) =
52 7 (d) = 364 (d). It is a matter of agreement as to which one of these
transformations is used.
2.2.3 Measurement of Monetary Values
History has shown that it is practical to use a specific good called ‘money’
in measuring the values of traded goods, see Sect. 8.1. The good to be
used as money can be any good that is accepted by all. Independent of
the good used as the reference good (money), it is not possible to obtain
generally accepted values for goods. This is due to personal differences in
(1) the valuation of goods, and in (2) the valuation of the reference good
we use in measuring. Some people pay to listen to classical music, and
some to heavy metal music. Some pay to see a boxing match while others
wouldn’t go there even if they were paid. These examples demonstrate
the first claim. The second claim can be understood by realizing that in
all trading we use one good in valuing the other goods. When we are
trading two goods A and B without a third good giving the prices, we
value good A in units of good B or vice versa. Let us study some examples
to demonstrate the second claim.
Suppose you are planning to buy a consumer good. If you are rational,
you buy the good if you can afford it, and you value the good higher than
or equal to its price. To be able to do the comparing, you have to transform
your valuations into monetary units, or vice versa. A rational person must,
therefore, have in his/her mind a rule by which the comparison is made.

Here are some suggestions for this rule.
Rule 1. Value one unit of money according to the trouble you expect to have in
acquiring it. Use this valuation to transform the prices of goods to the amount
of trouble it causes you to pay their prices. Then buy a good if it is worth the
trouble; that is, if the pleasure you get from consuming the good exceeds or
equals the trouble of acquiring the money to pay its price. ˘
In Rule 1, prices and personal valuations are made comparable by

assuming that acquiring money is troublesome, as it usually is. Another
way to solve the comparability problem is to use the concept of alterna-
tive cost.
Rule 2. Buying a good involves a loss of pleasure in the form of losing the
best alternative use of the money: buying another good or saving it. Find the
alternative cost of acquiring a good by estimating the loss of pleasure due to
giving up the best alternative use of the money. Then buy a good if the pleasure
you get from consuming it exceeds the pleasure you could get from the best
alternative use of the money. ˘
In Rules 1 and 2 the comparison is made in terms of satisfaction units

because pleasure can be thought of as positive, and trouble as negative
satisfaction. In the next rule, the comparison is made in monetary units.
Rule 3. Define your marginal willingness to pay for a good by dividing your
marginal utility of the good by your marginal utility of money. Then buy the
good if your marginal willingness to pay of the good exceeds its price.˘
This rule is a practical one because it is measurable in monetary

terms. We will show in Sect. 3.7 that in applying this rule we can model
consumer behavior in a measurable way. The three rules imply that the
valuation of one unit of money differs between individuals depending on
their ability to acquire money, their standard of wealth, and the enjoyment
they get from consuming goods. Due to these reasons, one of two rational
consumers may buy a good and the other may not, even though they both
enjoy consuming the good and can afford it. Suppose we use squirrel furs
in valuing goods. A skillful hunter probably values one squirrel fur less
than an unqualified hunter because of the less trouble he has in catching
it. A skillful hunter may, then, be more ready to go shopping due to his
lower valuation of ‘money’. These examples demonstrate the second claim
stated earlier.
In physics, it is required of a measurement unit that it does not change
with time, with location, or with the person who is doing the measuring.
Now, we have seen that we need at least one good in valuing the other
goods—any commonly accepted one will do—and a historical process
has led to money being used as the reference good, see Sect. 8.1. Above
we demonstrated that the personal valuation of one unit of money differs
from one person to another, and we know that the purchasing power of
money varies with time, see Sect. 2.3. We can thus conclude that a good
with an equal value for all people and a constant value with time—that
could be used in measuring the values of goods—does not exist. We thus
have to give up any hope of exact measuring of values because valuation
is based on personal judgments.
A practical way to value a good is to let the market do it. This is what
people do in real life. The market value of a good measured by any other
good—squirrel fur, gold, money, and so on—increases with its necessity,
popularity, and scarcity. Due to these reasons, the market value of a good
depends on the market in which the good is offered for sale. The market
value of winter clothes, for example, depends on whether they are offered
for sale in Kenya or in Finland. Therefore, we cannot determine whether
an increase in the value of a good is caused by its increased necessity or
popularity, or by a decrease in the purchasing power of the measuring good
(money). This demonstrates the difficulties we have in explaining changes
in values of goods. Independently of these problems, we can always get a
measure—possibly zero or negative—for the market value of a good. The
market value of a good is thus a measurable quantity, and we can state the
following definition:
§: The market value of a good is the maximum monetary offer made
for it when it is offered for sale. ˘
This definition can be applied for negative values too; that is, for
garbage goods. The owner of garbage likes to pay the minimum amount of
money to get rid of the garbage. This corresponds to the highest (nearest
to zero) of all negative offers people require for taking care of the garbage.
Following on from the above, we suggest the euro as the basic unit of
the dimension of monetary values (or money for short). The arguments
for this choice are that any commonly accepted good will do, and money
is used for this purpose in everyday life. With respect to measurement
units, the most widely used currencies are denoted by their common
abbreviations, euro by e, US dollar by $, Great Britain pound by £,
and so on. However, these types of abbreviations do not exist for some
currencies, and we therefore denote these by their official abbreviations,
such as Swedish crown by SEK, and so forth. With regard to the exchange
rates of currencies, we apply the official abbreviations: USD for US dollar,
GBP for Great Britain pound, and so on.
Now, all monetary quantities measured in units euro, US dollar,
Swedish crown, and so on, belong in the dimension of money and are
thus additive. Different currencies have transformation rules like those
in adding quantities belonging in the dimension of length measured in
different units, and in adding the amounts of a good measured in different
units. Suppose the exchange rate between US. dollar and Great Britain
pound is 1 (GBP) D 2:5 (USD). We can write this transformation
equation as

1 GBP
1 D 2:5 (USD=GBP) , 1D :
2:5 USD
With these two forms of the transformation equation, we can add 2 (GBP)
to 5 (USD) in US dollars as3
2 (GBP) C 5 (USD) D 2 (GBP) 1 C 5 (USD)

D 2 (GBP) 2:5 (USD=GBP) C 5 (USD) D 10 (USD);
3
See the transformation equation between two units of length in (2.1).
and in British pounds as
2 (GBP) C 5 (USD) D 2 (GBP) C 5 (USD) 1

D 2 (GBP) C 5 (USD) (1=2:5) (GBP=USD)
D 4 (GBP):
All transformations between currencies can be made analogously. The

only problem is that the transformation rules between two currencies
change with time, while in physics the transformation rules between
measurement units are fixed. The reason for this is the lack of a basic unit
for the dimension of monetary values that would have a fixed value. We
chose the euro as the basic unit of money, but we know that its purchasing
power and its exchange rates with other currencies change with time. This
is a fact we economists have to accept, and it makes the monetary values
of goods expressed in different currencies uncertain.
2.2.4 Measurement of Satisfaction
An old problem in economics is how to measure a person’s level of

satisfaction or ‘utility’. Allen (1956, pp. 670–673) concludes that utility
is a measurable quantity subject only to the arbitrary placement of the
zero point and the unit of measurement used. The reasons are the
following. What we can observe is marginal utilities, that is, the increases
in consumers’ satisfaction due to their consumption of goods. Knowing
his marginal utilities, a person can rationalize his consumption by ranking
the goods. This preference ordering may be unique, but it is indeterminate
with respect to a measurement unit. Any positive transformation of the
ordering gives the same ordering. Another problem is the integration
constant we get when we integrate the level of utility (satisfaction) of
a person from his marginal utilities. This makes the level of utility
unmeasurable.
de Jong’s (1967, pp. 98–99) solutions to these problems are the fol-
lowing: choose a specific constant and transform all marginal utilities in
terms of that, and set the value of the integration constant zero. With
these choices the ‘level of utility’ becomes a measurable entity, that is, a
‘quantity’. After this de Jong gives the following definition:
§: The set of all conceivable levels of satisfaction is called the dimension
of satisfaction symbolized by ŒS. ˘
This definition assumes that the satisfaction consumers receive from the
consumption of different goods is of a homogeneous type. All consumer
goods do not satisfy the same need, however. The needs of human beings
can be roughly classified into physiological, social, and mental categories.
Satisfying the need of hunger by eating does not decrease the need of
social relations, and the satisfaction of social relations does not decrease
the need of sleeping, and so on. The level of satisfaction of a consumer
is thus a holistic feeling that depends on his/her satisfaction of various
kinds of needs.
One solution to the comparability problem of various types of needs
is the following: consumers classify their needs in different categories.
One category of need, for example hunger, can be satisfied by eating
different kinds of food; another, such as social relations, can be satisfied
by meeting people in various ways, for example at home, in a restaurant,
or via the Internet. To be able to make a rational choice, a consumer must
be aware of his/her needs, and be able to classify the satisfaction he gets
from the consumption of every available good in one of his categories
of needs. Consumers may have several categories of needs depending on
how sophisticated they are with the types of satisfaction they recognize.
The satisfaction a consumer gets from his consumption of all goods that
belong in one of his categories of needs are directly comparable (goods are
close substitutes), and the satisfaction from goods belonging in different
categories is not directly comparable; they are either distant substitutes or
complements.
If a consumer can directly compare the satisfaction he receives from the
consumption of all goods belonging in one of his categories of needs, and
the unit prices of these goods are known, then the consumer can make a
rational choice between these goods. When consumers allocate funds for
their consumption, they first budget funds for their different categories of
needs for a time unit: housing, food, hobbies, and so on. Exact decisions
are made between the goods in these categories within the limits of
budgeted funds. If a consumer can estimate the satisfaction he gets from
the consumption of every possible combination of existing goods at his

current level of satisfaction, and he knows the combination of different
kinds of satisfaction that pleases him most, then the consumption of all
possible combinations of existing goods is an ordered set according to
the consumer’s preferences. These assumptions are required for rational
consumption decisions, see Chap. 3.
From these remarks we can conclude that utility—the holistic degree of
satisfaction of a person—is a measurable quantity measured by its own
basic unit called ut. The concept of ‘utility’ includes the fact that the
degree of satisfaction of a person depends on all his needs; for example a
well-to-do person may thus be unhappy due to a lack of satisfaction of his
social needs. Unit ut is needed to be able to write rigorous equations where
the concept of utility exists. It turns out, luckily, that in order to model
economic phenomena, the exact measuring of the level of satisfaction
is not needed. Satisfaction is an auxiliary dimension that allows us to
write equations containing the concept of utility in a rigorous way. Every
time we model consumer behavior on the basis of his seeking satisfaction,
we transform the behavioral equations in the form where the consumer
compares his marginal willingness to pay for a good and its price. These
are measurable quantities with units of other primary dimensions. This
way we can omit all problems in the exact measuring of the satisfaction
of a person.
2.2.5 Primary Dimensions in Economics
The primary dimensions in a measurement system for economics are

ŒM, ŒT, ŒRi , i D 1; : : : ; n, and ŒS, where n is the number of goods
in the economy. All economic phenomena we study in this book can
be measured by units belonging in these dimensions, or by secondary
dimensions that consist of products of powers of the basic dimensions. All
unit prices of good i measured, for example, in the units of e=unit, e=kg,
and so on, belong in the secondary dimension ŒM=Ri , the flow (velocity)
of production of good i with unit kg=week belongs in the secondary
dimension ŒRi =T, and so forth.
Note. The difference between the monetary value of a good and its price
is that monetary values belong in the primary dimension ŒM and prices
in the secondary dimension ŒM=R. For example, the market value of a
computer, say, 1000 (e), is the highest monetary amount that someone is
ready to pay for it. The price of the computer can be solved as 1000 (e)
D 1 (unit) ) 1000 (e/unit) = 1, where number 1 in the latter form of
the equation represents the equality of the numerator and denominator
in the price on the left hand side. This equation expresses the relationship
between the price and the monetary value of the good. The value of the
computer can be obtained from its price as: 1000 (e) = 1000 (e/unit)
1 (unit). ˘
Examples of Dimensional Calculations
The following examples demonstrate the rules of mathematical operations

with dimensional quantities. Addition and subtraction is allowed only
with quantities of equal units; and in multiplication and division, we mul-
tiply (divide) the numbers of measurement and the units of measurement
separately:
5 (kg) C 3 (kg) D 8 (kg);
2 500 (e=y) D 1000 (e=y);
2 (e=kg) 100 (kg=unit) 8 (unit=week) D 1600 (e=week);
100 (e=mn)
D 5 (kg=mn);
20 (e=kg)
100 ($=mn) 12 (mn=y) D 1200 ($=y);
1200 ($=y) 4 (y) D 4800 ($);
200 ($)
D 20;
10 ($)
10 ($=week) C 5 ($=unit) 4 (unit=week) D 30 ($=week);
x (kg=unit) y ($=kg) D xy ($=unit):
All the above equations are homogeneous with respect to measure-

ment units; that is, dimensionally homogeneous. Thus, if both sides of
the last equation are multiplied by 1 (unit=$), the equation becomes
dimensionless.
Examples of Measurement in Economics
Example 1
Suppose 10 tons of paper has been produced in a paper factory in two days.
The dimension of this flow of production, ŒR=T D ŒRT 1 , is the product of
the powers of basic dimensions ŒR and ŒT. The flow of production is then
measured as: 10 (tn)/2 (d) = 5 (tn=d). ˘
Example 2
The production in Example 1 has been sold at unit price 1 ($=kg) to a
newspaper corporation. The dimension of the unit price, ŒM=R D ŒMR1 ,
is the product of the powers of basic dimensions ŒM and ŒR. ˘
In the following we abbreviate the unit prices of goods to ’prices’,

because prices are always related to a quantity unit: kg, unit, 1000 unit,
m2 , and so on.
Example 3
The value of production of the paper factory in Example 1 (or the value
flow of production) can be calculated asa : 5 (tn=d) 1 ($=kg) = 5 (1000kg=d)
1 ($=kg) = 5000 (kg=d)($=kg) = 5000 ($=d). The dimension of the value of the
flow of production, ŒM=T D ŒMT 1 , is the product of powers of primary
dimensions ŒM and ŒT. ˘
a
Note that the number of measurement and the unit of measurement are
multiplied together, and so we can multiply the number of measurement
by the number in the measurement unit.
2.3 Nominal and Real Quantities

The difference between nominal and real quantities in economics orig-
inates from using a separate good in measuring monetary values—the
money or the currency—the purchasing power of which varies with time.
§: The (real) purchasing power of a currency measures the amount
of goods that can be obtained by one unit of the currency. ˘
The nominal purchasing power of a currency means that by one
unit of the currency you can buy the amount of goods of value of one
unit of the currency. Thus the concept of nominal purchasing power
is meaningless, and in the following we use only the concept of real
purchasing power of money, or for short, the purchasing power of money.
Proposition. If the average price level increases in an economy, the purchas-

ing power of the home currency decreases.
Proof. Suppose the country is USA and the amount of every good in
the economy is measured in units kg. Assuming this, we can express
the price of every good in units $=kg. Let us, for instance, calculate the
kilogram price of a computer. The physical weight of the computer is
mC (kg=unit). Dividing the price of the computer pC ($=unit) by its
mass, we get its kilogram price as pC =mC ($=kg). The average
P kilogram
price of all goods in the economy is then pN D (1=n) njD1 pj , where
pj is the kilogram price of good j and n is the number of goods in the
economy. With one US dollar you can then, on average buy the amount
1=(Np ($=kg)) = 1=Np (kg=$) of goods in the economy; quantity 1=Np thus
measures the purchasing power of one US dollar. If the average price level
in the economy increases, that is 1=Np decreases, the purchasing power
of US dollar decreases. By one dollar you can then on average buy less
goods. ˘
If the average price level increases in an economy, that is, the purchasing
power of the home currency decreases, this raises the nominal values of
most goods. Real quantities, such as the number of goods or the amount of
production of a good in a time unit, may then stay constant even though
their (nominal) values increase. This is the topic in this section.
§: Nominal quantities, such as the values of goods, their unit prices,

and so on, measure the exchange rates between goods and money. The
corresponding real quantities, on the other hand, measure the amounts
of goods or the exchange rates between goods. ˘
If the measurement unit of value, the money, would meet the require-
ments for a measurement unit in physics, namely that its purchasing
power stays constant with time, we would not have problems concerning
nominal and real quantities. Without the invention of money we would
value goods on the basis of their pairwise exchange rates, which are real
quantities.
Let us denote, for instance, the liter price of oil by po (USD=l) and
the kilogram price of silver by ps (USD=kg). These (nominal) prices may
increase due to the price inflation in the US economy, but the price ratio
of the two goods po =ps (kg=l) is a real quantity that measures the exchange
rate between these goods. The development of the purchasing power
of the US dollar thus does not affect the exchange rate of these goods.
However, if the price of oil is expressed in units pQ o (EUR=l) and the price
of silver in US dollars, then the price ratio pQ o =ps (EUR=USD) (kg=l)
depends on the development of purchasing powers of the two currencies.
The purchasing power of US dollar can be calculated as the inverse of
the average price level in the USA. When transforming nominal prices to
‘real’, we operate analogously. For example, the nominal liter price of ice
cream pj ($=l) is obtained from the transformation equation between ice
cream and US dollars pj ($) D 1 (l) as 1 D pj ($=l). The number 1 in
the latter equation shows the equality of the numerator and denominator
on the right-hand side. The real price of ice cream, on the other hand,
measures the amount of other goods for which one liter of ice cream can
be exchanged in the economy. We obtain this as pj ($=l)/p ($=kg), where
p is the average price level in the economy. The real price of ice-cream
pj =p (kg=l) thus measures the exchange rate between one liter of ice cream
and other goods in the economy.
Example 1
Let the price of ice cream be 10 ($=l) and the average price level of goods in
the economy be 100 ($=kg). The exchange rates between goods and money
are then: 10 ($) D 1 (l) and 100 ($) D 1 (kg). The transformation equation
between ice cream and other goods in the economy is obtained by dividing
these equations: 1 (kg) D 10 (l). The last equation defines the real price of
ice cream as 1/10 (kg=l) D 1. Thus, on average, one liter of ice cream can be
exchanged to 1=10 (kg) of other goods. If, however, the average price level is
expressed as ‘liter price’, then the ‘real price of ice cream’ is a dimensionless
quantity that measures the exchange rate between liters of other goods
and one liter of ice cream in the economy. ˘
Example 2
Let a nominal wage be W ($=h) and the price of milk p ($=l). The corre-
sponding transformation equations between goods and money are then:
W ($) D 1 (h) and p ($) D 1 (l). The ‘real wage’ W=p (l=h) D 1 is obtained
by dividing these equations, and it measures the exchange rate between
liters of milk and one hour of work. The corresponding transformation
equation is (1=W) (h) D (1=p) (l). The higher the nominal wage W, the less
work time is required in exchange for one liter of milk. If, however, the
nominal wage is divided by the average kilogram price of goods in the
economy p ($=kg), then ‘real wage’ W=p (kg=h) measures the exchange rate
between the amount of goods in the economy and one hour of work. This
corresponds to the purchasing power of wage W. ˘
The above means that when we transform nominal quantities to real,

we divide nominal quantities by some price; usually by an average price
level in the economy. We call the transformation of nominal quantities
to real as deflation or, in other words, the removal of inflation. This
terminology comes from the passage of time because usually we transform
nominal time series of an inflationary economy to real ones.
2.3.1 How Do We Use Real Quantities?
Real prices can be used in comparing the development of exchange rates

between goods, and real wages measure the purchasing powers of wages
in an inflationary economy. Besides real prices and wages, real values can
be defined for various forms of wealth. Real values inform whether these
forms of wealth have kept their exchange rates with respect to other forms
of wealth.
Suppose the value of the flow of production of good k, Vk (e=mn) and
its price pk (e=kg) are known from one month. Then we can calculate
the amount of production of the good during that month. The value of
production consists of the flow of production qk (kg=mn) multiplied by
the price pk (e=kg). The value of the production can then be expressed
as Vk (e=mn) D pk (e=kg) qk (kg=mn) from which we get qk D
Vk =pk (kg=mn).
The flow of production can then be calculated by dividing the value of
the flow of production by the price. We call this ‘deflation of the value
flow of production by the price of the good’. If, however, the deflation is
made by an average kilogram price of goods in the economy, p (e=kg), we
get Vk =p = pk qk (e=mn)=(p (e=kg)). Quantity Vk =p (kg=mn) measures
the amount of other goods the monthly production of good k can be
exchanged for in the economy. ‘The real value of monthly production’
of good k thus either measures the flow of production of the good (if
deflated by the price of the good), or the average amount of goods the
monthly production can be exchanged for in the economy (if deflated by
an average price level in the economy).
We can, similarly, approximate the flow of production of a group of
goods on the basis of their aggregate value in a time unit and their prices.
This is based on the idea that the amount of production of these goods
can be transformed to be measured in equal units; for example, in mass
units. We operate as follows. Define the kilogram price for all goods,
as described earlier for a computer. Then define a weighted average of
kilogram prices of the goods where the weights can be; for example, the
shares of the goods from the aggregate value flow. This way the defined
average kilogram price is denoted by p. Divide the value of the aggregate
flow of production V (e=mn) into price and volume flow components
as: V D p (e=kg) q (kg=mn). If we then divide V by p, we get an
estimate for the aggregate flow of production q (kg=mn) of the goods.
This approximation is required if we have aggregated value data from the
production of many goods, and we like to approximate the aggregate flow
of production of these goods.
At the economy level aggregated data, average price level is usually

approximated by some dimensionless average price index pO of the econ-
omy. If we denote by V (e=y) the annual value flow of production of an
economy (the annual nominal Gross Domestic Product), then quantity
V=Op has unit e=y even though our aim was to approximate the annual
flow of production. In this case, ‘nominal euros’ are deflated to ‘real’ by a
dimensionless price index. We thus have to notice the measurement units
when we formulate real quantities by deflating nominal ones by average
price levels or price indexes, see Sect. 2.8.1.
2.3.2 Internal and External Value of a Currency
Nominal Exchange Rate
The nominal exchange rate between two currencies, for instance,

Swedish crown and US dollar, defines the external value of one currency
in terms of another currency. For example 5 (SEK) D 1 (USD), from
which we get the nominal exchange rate in two possible forms as:

SEK 1 USD
5 D1 , D 1:
USD 5 SEK
The nominal exchange rate is thus 5 (SEK=USD) or 1=5 (USD=SEK).
Purchasing Power and Real Exchange Rate
Let the average kg-prices of all goods in USA and Sweden be denoted
as pu and ps . The internal value or the purchasing power of money
in the two countries is then 1=(pu (USD=kg)) D 1=pu (kg=USD), and
1=(ps (SEK=kg)) = 1=ps (kg=SEK), respectively.
Next, we denote certain amounts of goods in USA as xu (kg) and
in Sweden as xs (kg). Suppose that the values of these amounts are,
respectively, pu (USD=kg)xu (kg) and ps (SEK=kg)xs (kg), where pu ; ps
are average price levels of the two amounts of goods. Next we choose the
amounts xu ; xs so that with current nominal exchange rate S (SEK=USD),
the values of the amounts of the goods are equal. Then we get
xs Spu
Spu xu (SEK) D ps xs (SEK) ) D : (2.2)
xu ps
The left hand side of the last form of the equation expresses the exchange
rate between the amounts of the goods, and the right hand side shows that
this depends on the ratio of average unit prices multiplied by the nominal
exchange rate. Ratio xs =xu is called the real exchange rate. In contrast
with the nominal exchange rate—which expresses the exchange rate
between two currencies—the real exchange rate expresses the exchange
rate between amounts of goods of equal value in the two countries.
Equation (2.2) shows that the higher the US prices on average, the
smaller the Swedish prices, and the more valuable one US dollar is in
terms of Swedish crowns, the higher the real exchange rate; that is, the
more Swedish goods are needed on average in exchange for one kilogram
of US goods.
2.4 Discrete and Continuous Quantities

Basically two kinds of quantities exist: continuous and discrete. A
discrete quantity is non-continuous in the way that its value changes with
varying or fixed ‘jumps’ or steps. For example, a discrete quantity may take
values only in natural numbers 1; 2; 3; : : : . An example is the number of
produced goods. Another reason for the discreteness of a quantity is that
its values are measured from fixed time units or time moments, or at a
certain interval scale. Examples are the daily production of a firm, the
temperature of air measured every morning at eight o’clock, or the weekly
production of sugar measured in full kilograms.
Figure 2.1 presents the values of quantity x(t) at time moments t0 , t1 ,
and so on. By t is denoted time, and the subindex refers to a fixed time
moment. The first figure displays a continuous quantity and the other two
are discrete ones with time. The graphical presentation of a continuous
quantity is a continuous curve, and that of a discrete quantity is a step
function or a set of separate points.
From continuous variables with respect to time we can always get a
discrete quantity by taking observations from fixed time units (daily,
x(t ) x(t ) x(t )
●
●
●
t0 t1 t t0 t1 t2 t t0 t1 t2 t
Fig. 2.1 One continuous and two discrete quantities with time
weekly, etc.), or at certain time moments. Another way to transform a

continuous quantity to a discrete one is to divide the value set of the
quantity in discrete intervals where the value is constant. For example,
we can measure the production of sugar only in full kilograms Q (kg);
1 Q < 2 is then measured as 1 (kg), 2 Q < 3 as 2 (kg), and so
forth.
Example
Suppose the production of a firm is measured by the number of finished
products: 1; 2; 3; : : : . Let the daily production of the firm be 3 goods. The
velocity (flow) of production of the firm is thus 3 (unit=d). This discrete
velocity of production can be expressed as
3 3
3 (unit=d) D 3 (unit=24h) D (unit=h) D (unit=60min)
24 24
1 1 1
D (unit=min) D (unit=60sec) D (unit=sec) etc.;
480 480 28800
if the production process is in operation day and night. The momentary

flow of production 1/28800 (unit=sec) thus corresponds to the daily flow
of 3 (unit=d). If the firm operates for 8 hours per day, the hourly flow of
production is 3/8 (unit=h). This is analogous with the instantaneous speed
of a car expressed in units km=h, even though the measurement is made by
a radar in less than a second. The instantaneous speed of a car 100 (km=h)
implies that with this speed the car will reach 100 kilometers in one hour.
The flow of production 3 (unit=d) thus means that on the average the speed
of production was 3/8 (unit=h), if the firm operated 8 hours per day. ˘
In economics, marginal quantities are usually measured from a rela-

tively short period of time, or with a small change of the argument of a
function. By using the marginal cost of production we can estimate the
costs of a larger amount of production, and marginal costs per hour can
be used in estimating the production costs from a longer period of time.
However, marginal quantities frequently deviate from the corresponding
average ones. For example, the average tax rate of an employee may be
35% while his marginal tax rate is 50% due to progressive taxation.
The profitability of overtime work should then be analyzed by using
the marginal, and not the average, tax rate. Similarly, the profitability
of producing the ‘next’ boat should be based on the marginal cost of
increasing the weekly production by one boat, because overtime work and
other such factors may increase the costs of extra production. Due to these
reasons, various kind of marginal quantities are important in economic
analysis.
2.5 Measuring Changes in Scalars

In the following we denote time moments by t0 , t1 , t2 ; : : : , and their
distances t are assumed equal, that is, t D t1 t0 D t2 t1 D t3 t2 D
. Thus t0 C t D t0 C t1 t0 = t1 , t0 C 2t = t0 C t C t = t1 C t =
t2 etc., and the time units are named according to their ending moments.
In discrete time, the length of time interval t does not matter; essential
is that the quantities are measured at time moments t0 , t1 , t2 ; : : : (or at
time units t1 , t2 ; : : : ) and not between these moments. This holds for all
time series in economics. Continuous time is constructed from discrete
time by letting t ! 0. The time units of zero length obtained in this
way correspond to time moments, and continuous time is constructed by
connecting the adjacent time moments.
Now, we have four possible ways to measure the change in quantity x
during time unit t D (t0 C t) t0 , where t0 is a fixed moment.
From Table 2.1 we see that absolute change is measured in the units
of x, relative change is dimensionless, average velocity (flow) is measured
in units x=t, and growth rate in units 1=t. Now, if the length of the
time unit is one, that is t = 1 (time unit), the numerical values of the
Table 2.1 Discrete time quantities measuring

change
Absolute change x(t0 C t) x(t0 )
x(t0 Ct)x(t0 )
Relative change x(t0 )
x(t0 Ct)x(t0 )
Average velocity or flow t
Œx(t0 Ct)x(t0 )=t
Growth rate x(t0 )
Table 2.2 Continuous time quantities measuring change

Instantaneous absolute change limt!0 Œx(t0 C t) x(t0 ) D dxjtDt0
h i
Instantaneous relative change limt!0 x(t0 Ct)x(t
x(t0 )
0)
D dxx jtDt0
h i
Instantaneous velocity or flow limt!0 x(t0 Ct)x(t
t
0)
D dx j
dt tDt0
h i
Œx(t0 Ct)x(t0 )=t dx=dt
Instantaneous growth rate limt!0 x(t0 )
D x jtDt0
first and third, and the second and fourth quantities are equal but their
measurement units differ. The common assumption t = 1 (time unit)
in economics is the reason that, for example, an interest rate is often
erroneously identified as relative change, and not as the growth rate of a
monetary quantity, which is the correct identification. Knowing the exact
measurement units of economic quantities is a necessity for well-defined
mathematical expressions with these quantities, as we will see later.
The continuous time correspondents for the above discrete time quan-
tities are obtained as their limits with t ! 0, see Table 2.2.
In mathematics, the instantaneous absolute change of x, dx, is called
the differential of x, and the instantaneous velocity of x with respect to
time, dx=dt D x0 (t), is called the time derivative of x; see Sect. 10.7.
Mathematics differs from ‘real sciences’ in that mathematics does not
operate with dimensional quantities or with measurement units.
Proposition. If x=t > 0, x is increasing with time.
Proof. We denote x D x(t1 ) x(t0 ) and let us suppose first that t D

t1 t0 > 0; that is t1 > t0 . If we then multiply both sides of the inequality
x=t > 0 by t, we get x > 0; that is, x(t1 ) > x(t0 ).
Suppose next that t D t1 t0 < 0, that is, t1 < t0 . If we now
multiply both sides of the inequality x=t > 0 by t (remember the
rules of multiplying an inequality by a negative number), we get x < 0;

that is x(t1 ) < x(t0 ). Thus x(t) is increasing with time if x=t > 0. ˘
Note 1. In continuous time, proposition x=t > 0 ) x(t) increases

with time. If x0 (t) > 0, then x(t) is increasing with time. ˘
Note 2. According to dimensional analysis, the measurement units of the

continuous time quantities in Table 2.2 are identical with those of their
discrete correspondents because taking the limit does not affect the unit.
In physics, time derivatives are used in measuring instantaneous velocities
of particles or flows of materials. An example of the former is the velocity
of a car measured by the speedometer in units km=h, and of the latter is
the instantaneous flow of water measured in units liter=h. ˘
Note 3. Growth rates can also be expressed in per cent. Growth rates in
per cent deviate from growth rates in that the former measures the strength
of change as ‘parts of a hundred in a time unit’, and the latter as ‘parts
of a unity in a time unit’. The former equals the latter multiplied by 100.
Growth rate in per cent ‘10 parts of a hundred in a year’ equals the growth
rate ‘0.1 parts of unity in a year’: 10 (%=y) = 10 ((1=100)=y) = 0.1 (1=y).
§: The symbol % for per cent is not a dimension nor a measurement

unit; it is only an abbreviation for number 1/100. ˘
2.6 Measuring Changes in Values

2.6.1 Changes in Values in Discrete Time
Let V0 D p0 q0 and V1 D p1 q1 be the values of production of a firm at two

adjacent months; V (e=mn) is the value flow of production, q (kg=mn)
the flow of production, and p (e=kg) the price. The relative change in
the value flow of production is then
V1 V0 p1 q1 p0 q0 p1 q1
vD D D 1:
V0 p0 q0 p0 q0
From this we get:

p1 q1
1Cv D :
p0 q0
Relative change in the flow of production is

q1 q0 q1
zD D 1;
q0 q0
from which we get:

q1
1CzD :
q0
Relative change in price is

p1 p0 p1
iD D 1;
p0 p0
from which we get:

p1
1CiD :
p0
We can then write:

V p q
1 C v D (1 C z)(1 C i) , 1 C D 1C 1C : (2.3)
V0 p0 q0
Formula (2.3) is known as Fisher equation (Fisher 1930), and from the
equation we can solve one of the three quantities x=x, x D V; p; q when
the other two are known.
Logarithms in Measuring Value Changes
By taking a logarithm we can transform multiplication to adding. Because
V1 p1 q1
D ; (2.4)
V0 p0 q0
taking the e-based natural logarithms (Sect. 10.7.2) of both sides of

Eq. (2.4) we get:

V1 p1 q1
ln D ln C ln : (2.5)
V0 p0 q0
Now, if x 1 0, that is x 1, for the logarithmic function holds
x 1 ln .x/
because ln(1) D 0. Then, denoting x D V1 =V0 we can write:

V1 V0 V1 V1
D 1 ln :
V0 V0 V0
Applying this approximation to formula (2.5) we get:

V1 V0 p1 p0 q1 q0
C : (2.6)
V0 p0 q0
Relative change in the value of production can thus be approximated by

adding relative changes in the price and in the flow of production. The
growth rate expression of Eq. (2.6) is
ŒV1 V0 =t Œp1 p0 =t Œq1 q0 =t
C ; (2.7)
V0 p0 q0
and if, for example, t D 1 (mn), the numerical values of relative changes
and growth rates are equal, even though their measurement units differ.
Equation (2.7) shows that the growth rate of the value of production
approximately equals the growth rate of the flow of production plus the
growth rate of price.
§: The growth rate of price level is called price inflation. ˘
Note. Logarithms are needed in the derivation of approximative equations

(2.6) and (2.7). However, approximating a relative change by using
logarithmic function as

V1 V0 V1
ln
V0 V0
is not required; relative changes and growth rates are more clear measures
for changes in quantities than logarithms of their ratios. We have to
remember that the value of a logarithmic function is a pure number,
and the argument of a logarithmic function must be a pure number for
the function to be dimensionally well-defined. This holds in Eq. (2.5)
because x(t)=x(t0 ) is always a pure number independent of the unit of x,
x D V; q; p. ˘
2.6.2 Changes in Values in Continuous Time
Let us denote the value flow of production of a good as V(t) D p(t)q(t),

where p(t) (e=kg) is the price of the good and q(t) (kg=mn) the flow
of production. The time derivative of this value flow V 0 (t) with unit
e=mn2 —the acceleration of the value of production—is then
V 0 (t) D p0 (t)q(t) C p(t)q0 (t); (2.8)
where the measurement units of p0 (t)q(t) and p(t)q0 (t) are e=mn2 . If we
divide Eq. (2.8) by V(t), we get the growth rate of the value flow as:
V 0 (t) p0 (t)q(t) p(t)q0 (t) p0 (t) q0 (t)

D C D C : (2.9)
V(t) p(t)q(t) p(t)q(t) p(t) q(t)
Thus, in continuous time the approximative equation (2.7) becomes

exact. From Eq. (2.9) we can solve the growth rate of the flow of
production as
q0 (t) V 0 (t) p0 (t)
D ;
q(t) V(t) p(t)
where price inflation p0 (t)=p(t) is subtracted from the growth rate of the
value flow of production. This is a general method of estimating the
growth rate of the flow of production from the growth rate of the value
flow of production; that is, the growth rate of the real flow from the
nominal one.
2.6.3 Interest Rate
Let x measure the amount of money deposited in a bank account on which

the bank pays interest. The next example shows how the amount of the
deposited capital affects the interest revenues.
Example 1
Let the annual interest rate the bank pays on its deposits be 4 (%=y). The
annual interest revenues for capitals 100 (e) and 1000 (e) are then 4 (e/y)
and 40 (e=y), respectively. ˘
Due to this dependence of interest revenues on the amount of deposited

capital, it is practical to measure the strength of change in the capital value
so that the amount of the capital does not affect this measurement. This
can be done by relating the earned money flow to the invested capital,
which corresponds to the growth rate of the capital.
Example 2
The interest earnings in Example 1 are 4 (e=y) and 40 (e=y). Dividing these
by the corresponding capitals, we get the growth rates of the capitals as: 4
(e=y)/100 (e) = 0.04 (1=y) and 40 (e/y)/1000 (e) = 0.04 (1=y). Thus the growth
rates are equal: 0.04 (1=y) or 4 (%=y). ˘
§: Interest rate is the growth rate of a monetary quantity measured

in units 1=t. By interest rate we measure the strength of the growth of
deposited, borrowed, or invested money. ˘
It is common in economics to compare interest rates and the rates of
return of financial and real investments, inflation, the growth rate of Gross
Domestic Product (GDP) of an economy, and the like. This is legitimate
because the growth rates of quantities measured from time intervals of
equal length are comparable.
Example 3
Let W (e=h) be a nominal wage and p (e=kg) an average price level in an
economy. The real wage W(t)=p(t) (kg=h) then measures the exchange rate
between one hour of work and the amount of goods in the economy. The
growth rate of the real wage during time unit t1 t0 is:

W(t1 )
W(t0 )
=(t1 t0 )
p(t1 ) p(t0 ) 1 W(t1 )p(t0 )
D 1
W(t0 )
p(t0 )
t1 t0 W(t0 )p(t1 )

1 W(t1 )p(t0 ) W(t0 )p(t1 )
D : (2.10)
t1 t0 W(t0 )p(t1 )
If, on the other hand, we calculate the growth rate of the nominal wage
during the time unit and subtract from this the growth rate of the average
price level, we get almost the same result:
ŒW(t1 ) W(t0 )=(t1 t0 ) Œp(t1 ) p(t0 )=(t1 t0 )

W(t0 ) p(t0 )

1 W(t1 )p(t0 ) W(t0 )p(t1 )
D : (2.11)
t1 t0 W(t0 )p(t0 )
The difference in these two ways of measuring is in the denominator. The

smaller the change in the average price level, the more accurately Eq. (2.11)
approximates that in Eq. (2.10). If the growth rate of the average price level
and that of the nominal wage are equal, the real wage stays constant.
In continuous time, this approximation becomes exact (see Sect. 10.7):

d W(t) W 0 (t) W(t)p0 (t)
dt p(t) p(t)
p2 (t) W 0 (t) p0 (t)
W(t) D W(t) D : ˘
p(t) p(t)
W(t) p(t)
This example shows that the development of the real wage can be approx-
imated by subtracting price inflation from the growth rate of the nominal
wage, when these two are measured from the same time unit. This result
also holds for other real quantities that are obtained from nominal ones
by dividing them by an average price level; this result holds particularly for
interest rates. ˘
§: Interest rate measured as the growth rate of a monetary quantity

is called nominal interest rate. Real interest rate, on the other hand,
measures the growth rate of the purchasing power of the currency. ˘
Let us denote the amount of money in a bank account at time moments

t0 < t1 by x(t0 ) (e) and x(t1 ) (e), respectively, when only interest revenues
are assumed to be added in the capitals. The interest revenues from time
unit t1 t0 are then x(t1 )x(t0 ) (e). The nominal interest rate r measured
in units 1=t during time unit t1 t0 is then:
x(t1 )x(t0 )
t1 t0 1 x(t1 ) x(t0 )
rD D :
x(t0 ) t1 t0 x(t0 )
Let the average price level in the economy be p(t) (e/kg) at time moment
t. The growth rate of the average price level, namely the inflation i with
unit 1=t at time unit t1 t0 , is then:
p(t1 )p(t0 )
t1 t0 1 p(t1 ) p(t0 )
iD D :
p(t0 ) t1 t0 p(t0 )
The real interest rate rR with unit 1=t measures the growth rate of the
x(t)
purchasing power p(t) (kg) of money x(t) (e) as follows:
x(t1 ) x(t0 )
p(t1 ) p(t0 )
t1 t0 1 p(t0 )x(t1 ) p(t1 )x(t0 )
rR D D : (2.12)
x(t0 )
p(t0 )
t1 t0 p(t1 )x(t0 )
The real interest rate can be approximated by subtracting inflation from

the nominal interest rate:

1 p(t0 )x(t1 ) p(t1 )x(t0 )
riD : (2.13)
t1 t0 p(t0 )x(t0 )
The deviation between the accurate measure rR and its approximation ri
is in the denominator; the smaller the difference p(t1 ) p(t0 ) the more
accurate is the approximation. As in the case of the real wage, the reader
can check that this approximation becomes exact in continuous time.
Nominal lending and saving interest rates are positive because the
lender (saver) requires a compensation for his lending (saving). In an
inflationary economy, the purchasing power of the domestic currency
decreases with time. For the purchasing power of the lent (saved) money
not to decrease with time, the nominal interest rate must be at least as
great as inflation. If this does not hold, lenders (savers) are not interested
in lending (saving) money.
2.7 Economic Kinematics

Kinematics is based on the innovations of Isaac Newton in describing
the motion of a particle in mathematical terms. Newton developed
differential and integral calculus for this purpose. Without knowing this,
Gottfried Wilhelm Leibniz developed the same calculus in mathematics
in order to measure changes in function values. Later on it was realized
that the two techniques are identical. In this way, a connection between
mathematics and theoretical physics was found, which is still active today.
Various concepts defined in mathematics have a correspondent in physics.
For example, a vector (the position vector of a particle), the algebra
of vector calculation (the calculation of the resultant force vector), the
Euclidean norm of a vector (the speed of a particle in a multidimensional
space), and so on.
The earlier presented connections between economics and physics
imply that there exists a link between these sciences too. Later we will
see that in modeling dynamic economic phenomena, we can also apply
similar principles to those applied in physics by using differential and
integral calculus.
§: Kinematics is the study of the geometry of motion: it deals with the
mathematical description of motion in terms of position, velocity, and
acceleration. (Ohanian 1989, p. 25). ˘
Economic kinematics can be analyzed, as in physics, by the motion
of an idealized particle (no size and no internal structure), see Ohanian
(1989, p. 25). In physics, an ideal particle is identified as a point with
mass, the motion of which is studied in different coordinate systems. Even
though a concrete moving object does not exist in economic kinematics as
it does in physics, the description of how an economic quantity changes
with time can be presented in an identical mathematical way as the motion
of an object is described in physics. The moving point describes the
position of an object in physics, and it describes the value of an economic
quantity in economics. Changes in these two quantities with time are

described identically. Economic kinematics is a necessary prelude for
economic dynamics, which will be presented later.
2.7.1 Average Velocity of Production
We describe, mathematically, the development of the amount of produc-

tion of the good k of a firm. The starting moment of production is t0 ,
and the accumulated amount of production at moment t is denoted by
Qk (t) (kg). The development of the accumulated amount of production
with time is described by the motion of a point along line S in Fig. 2.2. At
every moment of time, t0 < t1 < t2 , the accumulated production
up until that moment is represented by one point on line S. When
the accumulated amount of production increases, the point measuring
it moves along the line S. This is an example of economic kinematics.
The movement of the point along line S represents the kinematics of
accumulated production, while in physics the movement of a point along
a line represents the kinematics of a body.
The accumulated amount of production of good k of a firm can also
be graphed in a two-dimensional coordinate system. On the horizontal
axis is measured time, and on the vertical axis the accumulated amount
of production of good k of a firm. The accumulated production up to a
certain time moment corresponds to one point in this coordinate system,
see Fig. 2.3.
In Fig. 2.3, the relationship between the accumulated production and
time is presented so that time is divided in units of equal length. In
physics, the graphical presentation of the time path of a particle is
called its worldline, (Ohanian 1989, p. 27). If the points in Fig. 2.3 are
connected, we get a continuous curve that represents the relationship
between the accumulated production and time. As an example, consider
• • •
Qk (t0 ) Qk (t1 ) Qk (t 2 ) S
Fig. 2.2 The accumulated amount of production of good k

● ●
Qk (t ) ●
●
●
t0 2 4 6 8 10 12 14 t (mn)
Fig. 2.3 The worldline of good k
the production process of a mass (like cellulose) as a continuous flow

process. Even though in most real production processes the products are
finished in a discrete way, we can still analyze the production processes as
continuous ones. The argument for this claim is that parts of goods under
production are finished at every time moment, even though to finish a
complete good (for example a car) takes time. If a factory operates for 8
hours per day, we can consider that the resting of workers at night, as well
as their lunch and coffee breaks, are parts of the process. The production
process continues during nights and lunch breaks in the form of the
workers’ required rest and gathering of energy. In this way, we can analyze
any production as a continuous process even though finished goods are
completed in a discrete way. The idea of a continuous production process
allows us to define the instantaneous velocity of production (see the
next section). The average velocity of production is defined as follows.
§: The average velocity of production of good k of a firm is:
the amount of production of good k of a firm during time unit t

:˘
t
The dimension of the average velocity of production of good k is

ŒRk =T, and it is greater the more is produced during time unit t.
Example 1
Suppose 10 cars were produced in a factory in one day. The velocity
of production of the factory is then: 10 (unit=d) = 10 (unit=24h) = 10/24
(unit=60min) = 1/144 (unit=60sec) = 1/8640 (unit=sec), if the factory operates 24
hours per day. This average velocity of production can also be expressed as
10 (unit=(1=30)mn) D 300(unit=mn) assuming 1(mn) D 30(d) and transforming
by time units in the opposite direction. ˘
According to Fig. 2.3, the production of good k of the firm starts at

time moment t0 , the velocity of production increases during the first 8
months (the slope of the curve turns upwards), the velocity decreases
during the next 4 months (the slope of the curve turns downwards), and
the production ends after 12 months. We can thus identify the curve as the
worldline of good k. Next, we give a graphical definition for the average
velocity by using the worldline.
We denote the accumulated production of good k till moment ti as
Qk (ti ), i D 1; 2. Then Qk (t2 ) Qk (t1 ) is the change in the accumulated
amount of production during time unit t2 t1 . The average velocity of
production of good k during time unit t2 t1 is then
Qk (t2 ) Qk (t1 ) Qk

vk D D ; (2.14)
t2 t1 t
where Qk D Qk (t2 ) Qk (t1 ) and t D t2 t1 > 0. Notation v comes

from the word ‘velocity’, and the bar above the variable refers to average.
If the amount of production of good k is measured in units kg and time
in units week, the measurement unit of v k is kg=week; see Fig. 2.4.
The average velocity of production is positive, if the accumulated
amount of production increases during t. Negative velocity of pro-
duction, which could mean that a certain amount of produced goods
disappears for instance in a fire accident, is not needed in economics. In
physics, the direction of motion of a particle may change, and negative
velocity implies that the particle is moving in the opposite direction as
was defined positive.
Qk (t )
Qk (t 2 )
ΔQ k
Qk (t1 )
Δt
t1 t2 t
Fig. 2.4 Average velocity of production of good k
In Fig. 2.4, a line is drawn through points .t1 ; Qk (t1 )/ and .t2 ; Qk (t2 )/.
Expression (2.14) then corresponds to the slope of this line, see Sect. 10.2.
The graphical interpretation of the average velocity of production of good
k is the slope of the line in Fig. 2.4. The steeper the line, the greater the
average velocity of production and vice versa.
Example 2
A factory produces 500 (kg) cement in 7 hours. The average velocity of
production is then: 500 (kg)/(7(h)) = 500=7 (kg=h) = 500/7 (kg=((1=24)d)) =
24 500=7 (kg=d), if 24 hours in a day are worked. ˘
Example 3
Suppose 50 cars have been produced in a factory in two weeks by working
8 hours a day and 5 days in a week. The average velocity of production
of the factory was then: 50 (unit)/(2(week)) = 25 (unit=week) = 25 (unit=(5d)) =
25=5 (unit=d) = 5 (unit=(8h)) = 5=8 (unit=h). ˘
2.7.2 Instantaneous Velocity of Production
We have argued, earlier, that we can analyze a firm’s production as a

continuous process in time. Let us study the average velocity of production
during time unit (t1 ; t2 ), and let the length of the time unit t D t2 t1
approach zero, that is, t ! 0 or t2 ! t1 .
§: The instantaneous velocity of production of good k at moment t1
is the slope of the tangent of the worldline of the good at the moment.
This corresponds to the time derivative of the accumulated production of
good k,
Qk (t2 ) Qk (t1 ) Qk dQk ˇˇ

vk D lim D lim D ˇ D Q0k (t1 ): ˘
t2 !t1 t2 t1 t!0 t dt t1
The instantaneous velocity of production of good k is demonstrated in

Fig. 2.5. The slope of line S1 is Q
t
, and it cuts the worldline of good k at
two points. Decreasing the change in the argument (t ! 0 or t2 ! t1 )
turns line S1 to position S2 where it touches the worldline of the good.
The instantaneous velocity of production thus corresponds to the slope of
the tangent of the worldline at time moment t1 . When tangents are drawn
Qk (t )
S1
Qk (t 2 )
S2
Qk (t1 )
t1 t2 t
Fig. 2.5 Instantaneous velocity of production of good k

to the worldline of good k at different time moments, the instantaneous

velocities correspond to the slopes of these tangents.
2.7.3 Velocity and Accumulated Production
The amount of production of good k during time unit (t0 ; tn ), t0 < tn , is

measured by the change in the accumulated production during the time
unit. Because the accumulated production of good k up to moment t
is denoted by Qk (t), the amount of production during time unit (t0 ; tn )
denoted by Qk (t0 ; tn ) is calculated as Qk (tn ) Qk (t0 ). Another way to
calculate this amount of production is to split the time interval (t0 ; tn ) in
time units of length t and add the productions during these time units:
Qk (t0 ; tn ) D Qk (tn ) Qk (t0 )

D Qk (t0 C t) Qk (t0 ) C Qk (t0 C 2t) Qk (t0 C t)

C C Qk (tn t) Qk (tn 2t) C Qk (tn ) Qk (tn t)
D Qk (t0 C t) C Qk (t0 C 2t)
X
n
C C Qk (t0 C (n 1)t) C Qk (tn ) D Qk (t0 C it);
iD1
P
where niD1 xi D x1 Cx2 C Cxn1 Cxn and t0 Ct D t0 Ct1 t0 D t1 ,
t0 C 2t D t0 C 2(t1 t0 ) D t1 C (t1 t0 ) D t1 C t D t2 , etc.
Let the amount of production of good k be measured in units kg and
time in units week, and let us analyze the average velocity of production
as a function of time. The situation is described in Fig. 2.6, where time
on the horizontal axis is divided into units of length t. On the vertical
axis is the average velocity of production of good k at every time unit
t0 C it, i D 1; : : : ; n, and the time units are named according to their
ending moments. The assumption that time is divided into intervals of
equal length is not necessary, but it simplifies the analysis. The amount
of production during time unit (t0 ; tn ) can be measured by adding the
productions at time units t0 C it, i D 1; : : : ; n. However, the same
result is obtained by multiplying the average velocities of production by
the time units:
vk (t n − Δt )
vk (t n − 2Δt)
vk (t n )
vk (t0 + 2Δt)
vk (t0 + Δt)
t0 t 0 + Δt t0 + 2Δt t n − 2Δt t n − Δt tn t
Fig. 2.6 Measuring the amount of production by its velocity
Qk (t0 ; tn ) D Qk (t0 C t) C Qk (t0 C 2t) C C Qk (tn )

Qk (t0 C t) Qk (t0 C 2t) Qk (tn )
D t C t C C t
t t t
Xn
Qk (t0 C it) X n
D t D v k (t0 C it)t; (2.15)
iD1
t iD1
where v k (t0 C it) D Qk (t0 C it)=t is the average velocity of

production at t0 C it.
Example 1
Suppose the average velocity of production is 40 (kg=d). The amount of
production during 10 days can then be approximated as: 40 (kg=d) 10 (d)
= 400 (kg). ˘
Because in Eq. (2.15) average velocities of production v k are measured

in units kg=week at every time unit, and time is measured in units week,
the measurement unit of Qk (t0 ; tn ) is kg. According to formula (2.15), the
amount of production during time unit (t0 ; tn ) can be identified as the
shaded area in Fig. 2.6. If then the length of the time unit is decreased
as t ! 0, the amount of production during time period (t0 ; tn ) can be
expressed as the following definite integral (see Sect. 10.14.2)
X
n Z Z
Qk (t0 C it) tn
dQk tn
Qk (t0 ; tn ) D lim t D dt D Q0k (t)dt
t!0
iD1
t t0 dt t0
D Qk (tn ) Qk (t0 ): (2.16)
This definite integral measures the area between the curve of instanta-
neous velocities of production and the horizontal axis during time unit
(t0 ; tn ). With t ! 0, time units t turn to time moments, and average
velocities turn to instantaneous velocities at these moments.
We can measure the accumulated amount of production by using
formula (2.16) for any time interval. For example, setting Qk (t0 ) D 0
and Qk (t) 0 with t > t0 , formula (2.16) measures the accumulated
production that was started at time moment t0 . We can thus call the
definite integral in Eq. (2.16) the accumulation function of production
of good k.
In Eq. (2.16), the upper limit of the integral is fixed. However, because
tn is a fixed moment of time, we can define it as a measure for the flowing
time that increases without limit. The integral can thus be analyzed as a
continuous function of its upper limit. Denoting tn D t and using the
principles of differential and integral calculus (see Sect. 10.14.2), we can
write:
dQk (t0 ; t)
D Q0k (t):
dt
The instantaneous velocity of production at time moment t thus corre-
sponds to the time derivative of the accumulation function of production.
By the notation t we underline that t is a continuous variable that measures
time, while tn is a fixed time moment.
According to the principles of differential and integral calculus
(Sect. 10.14.2), the accumulation function of production of good k
corresponds to the time integral of the instantaneous velocity of
production
Z t
Qk (t) D Qk (t0 ) C Q0k (s)ds; (2.17)
t0
where by t0 is denoted a time moment, by t flowing time, and by s

running time during time interval (t0 ; t). This notation is applied later
with definite time integrals, and with indefinite time integrals running
time is denoted by t instead of s. For a continuous (flowing) variable with
time, the accumulation function of the flow—or the accumulated stock of
the flowing ‘material’—can be obtained by using Eq. (2.17). The flowing
variable can be, for example, the position of a car on a road, the price of a
good, and so for. In these cases, Qk (t) in Eq. (2.17) measures the position
of the car and the price at moment t, and Qk (t0 ) is the initial position
of the car and the initial price of the good. On the other hand, Q0k (s) is
the instantaneous velocity of the car and the instantaneous velocity of the
price at time moment s.
Together with the velocity of production and price, other economical
flows are, for example, the flow of consumption of a good, the flow of
investment of a firm, and the flows of saving and borrowing of people.
The corresponding stocks are the accumulated stock of consumption
of the good up to a time moment, the accumulated stock of invested
capital of the firm, the accumulated savings and loan capitals of people.
Even though the velocities of production and consumption cannot be
negative, the flows of net investments, net savings, and prices may be
negative.
Example 2
Let the velocity of production of a firm be 6 (unit=d). How much does the
firm produce in 14 days?
Answer. We denote the initial time moment by t0 . The amount of produc-

tion of the firm over 14 days is then
Z ˇt0 C14
t0 C14 ˇ
6 ds D ˇˇ 6 s D 6 (t0 C 14 t0 ) D 84 (unit);
t0 t0
where the running time with unit d is denoted by s. The same result is
obtained by adding the daily amounts of production:
X14
Q
Q(t1 ; t14 ) D t D 6 (unit=d) 1 (d) C C 6 (unit=d) 1 (d)
iD1
t
D 6 (unit=d) .1 (d) C C 1 (d)/ D 6 (unit=d) 14 (d)

D 84 (unit): ˘
Example 3
Let the velocity of production of a firm be 10 (kg=h) when all workers are
working. How much does the firm produce in a week, if every worker does
8 hours per day and the firm operates 5 days a week?
Answer. Let us denote the initial moment by 0 (we could also denote it by
t0 ). The production of the firm in one day is then
Z ˇ8
8 ˇ
10 ds D ˇˇ 10 s D 10 (8 0) D 80 (kg);
0 0
and in one week the production is:

Z ˇ5
5 ˇ
80 ds D ˇˇ 80 s D 80 (5 0) D 400 (kg):
0 0
Notice that the unit of running time s is h in the first integral and d in the
second. The measurement unit of both integrals is kg, even though in the
first is calculated on the daily, and in the second the weekly, production.
In these calculations, we have to remember in which units the velocity of
production and time are measured. According to the measurement units,
the calculated daily and weekly amounts of production are stock quantities,
even though they are measured from fixed time units.
The above result can also be obtained by calculating first the amount
of working hours in a week 40 (h) and then integrating as:
Z ˇ40
40 ˇ
10 ds D ˇˇ 10 s D 10 (40 0) D 400 (kg);
0 0
where running time s is measured in units h. ˘

Example 4
Suppose the average velocity of production of a firm is constant q (kg=week).
How much does the firm produce in a year?
Answer. The annual amount of production of the firm is

Z 52 ˇ52
ˇ
q ds D ˇˇ q s D q (52 0) D 52 q (kg);
0 0
where running time s has unit week. ˘
Example 5
Let the velocity of production qk (kg=week) of a firm producing good k be
constant. The accumulation function of production of the firm is then
Z
Qk (t) D qk dt D qk t C bk (kg);
where the constant of integration bk has unit kg and time t is measured in

units week. Thus production accumulates with time with constant velocity
qk . In this example, the notation corresponds to the earlier ones with bk D
Qk (t0 ) qk t0 . We can thus write:
Qk (t) D Qk (t0 ) C qk (t t0 ); Q0k (t) D qk : ˘
Example 6
Let the instantaneous velocity of production of a firm producing good k
be constant 100 (kg=week), and let the accumulated amount of production
of the firm from the beginning of the year be 1500 (kg). What is the
accumulated amount of production of the firm in this year, in five weeks
from this moment?
Answer. We denote the initial moment by t0 , running time by s, and time

is measured in units week. The solution is then:
Z t0 C5
Qk (t0 C 5) D Qk (t0 ) C 100 ds D 1500 C 100 (t0 C 5 t0 ) D 2000 (kg): ˘
t0
Example 7
Let the loan capital of a person be 10;000 (e) at time moment t0 , and suppose
he pays back the loan with constant velocity 500 (e=mn). (a) What is the loan
capital in six months from now, if no interest is paid on the loan? (b) When
is the loan completely paid off?
Answer. (a) We denote the initial moment by t0 , the loan capital by A(t),
and running time by s with unit mn. The solution is then:
Z t0 C6
A(t0 C 6) D 10000 500 ds D 10000 500 (t0 C 6 t0 ) D 7000 (e):
t0
(b) The length of the time period is denoted by x (mn). Then we get
Z t0 Cx
0 D 10000 500 ds , 0 D 10000 500x , x D 20 (mn):
t0
The reader can check that the measurement units of x (mn) and dimen-
sional constants 10,000 (e) and 500 (e=mn) make the equations well
defined with respect to dimensions; notice that ds has unit mn. ˘
2.7.4 Average Acceleration of Production
§: A motion with changing velocity is called accelerated motion (Oha-

nian 1989, p. 32). ˘
Let us denote the instantaneous velocity of production of a firm of good
k at moment t1 by vk (t1 ), and at moment t2 by vk (t2 ), t2 > t1 .
§: The average acceleration of production ak of a firm of good k (a
refers to ‘acceleration’) during time unit t2 t1 is then:
vk (t2 ) vk (t1 ) vk
ak D D ;
t2 t1 t
where vk D vk (t2 ) vk (t1 ) and t D t2 t1 . ˘

Average acceleration of production of a good measures average rate of
change in the velocity of the production in a time unit. Average accel-
eration of production is positive, if the velocity of production increases
vk (t2 ) > vk (t1 ), and negative, if the velocity decreases vk (t2 ) < vk (t1 )
during time unit t2 t1 > 0. As described earlier, in economics negative
velocity of production does not exist, which simplifies the concept of

acceleration of production. If the amount of production of good k is
measured in units kg and time in units week, then the velocity of
production has unit kg=week and average acceleration of production has
unit kg=week2 .
Example 1
Let the velocity of production of a firm be 17 (kg=d) at time moment t0 and
22 (kg=d) at moment t0 C 10, where time is measured in units d. What is the
average acceleration of production of the firm?
Answer. The length of the time unit is t0 C 10 (d) t0 D 10 (d), and the
average acceleration of production is:

v 22(kg=d) 17(kg=d) 5 kg 1 kg
D D D :˘
t 10(d) 10 d2 2 d2
Example 2
Let the instantaneous velocity of production of a firm producing good k be
6 (unit=h) at moment t0 , and 4 (unit=h) after six hours from moment t0 . What
is the average acceleration of production of the firm?
Answer. The average acceleration of production of the firm is:

vk 4 6(unit=h) 1 unit
D D :
t 6(h) 3 h2
Average acceleration of production is thus negative, if the velocity of

production decreases with time. ˘
2.7.5 Instantaneous Acceleration of Production
Assuming time continuous, we can state the following definition:

§: The instantaneous acceleration of production ak of a firm of good
k is the limit value of average acceleration when the length of the time unit
t approaches zero:
vk (t2 ) vk (t1 ) vk dvk ˇˇ

ak D lim D lim D ˇ D vk0 (t1 ): ˘
t2 !t1 t2 t1 t!0 t dt t1
The instantaneous velocity of production of a firm of good k was defined

earlier as: vk D dQdt
k
. The instantaneous acceleration of production of a
firm of good k is then:

d dQk d2 Qk
ak D D D Q00k (t);
dt dt dt2
that is, the instantaneous acceleration of production is the second order

time derivative of the accumulation function of production. Notice that
we could also call ak the velocity of velocity of production.
Example
Let the instantaneous acceleration of production of good k be constant ak
(kg=week2 ). The velocity of production of good k is then
Z
vk D ak dt D ak t C bk ;
where the constant of integration bk has unit kg=week, and time t is mea-
sured in units week. The velocity of production thus increases (decreases)
with constant positive (negative) velocity ak with time. The accumulation
function of the production with unit kg is then
Z
ak 2
Qk D (ak t C bk )dt D t C bk t C ck ;
2
where ck with unit kg is the constant of integration. ˘
2.7.6 Kinematics of a Two-Good Production System
The kinematics of a production system of two goods can be analyzed by

using a vector (valued) function. This is done exactly in the same way
as the motion of a particle in two dimensions in physics. For example,
the motion of a thrown ball in a two-dimensional coordinate system
where the horizontal axis is on the level of the earth. The horizontal
axis is pointing in the direction the ball is moving, and the vertical axis
measures the height of the ball from the earth. A generalization to a
production system with three, four, five, or more goods is straightforward.
It only requires the definition of the corresponding vector function and
to be operating analogously, see Estola (2011). In this book, however, we
restrict the analysis in the two-dimensional case that can be demonstrated
graphically.
In physics, the free motion of a rigid body can be demonstrated by
the motion of its mass point. The particle may be in a rotational motion
around its mass point (for example the motion path of a thrown hammer),
but the path of its motion can be described roughly by the motion of its
mass point. In the gravitational force field, if we forget the resistance of the
air, the time path of the mass point of a moving particle is a continuous
curve with the shape of a parabola. If we are only interested in the time
path of a particle—and not its rotational motion around its mass point—
the motion path of the particle can be characterized according to that of
its mass point which is identified as a point particle with a mass.
Also in economic kinematics, we can apply the motion of a point
particle with a ‘mass’. An economy with many industries, or a firm with
many products, can be identified at one time moment as a point in the
coordinate system of accumulated production values of all its products.
Here we describe the motion of this point in the space of accumulated
production values in order to define the velocity vector and speed for an
economy and for a multi-product firm.
The accumulated production values of goods 1 and 2 are denoted
as V1 (t) D p1 (t)Q1 (t) and V2 (t) D p2 (t)Q2 (t) at time moment t,
where Q1 ; Q2 with units kg; unit, respectively, are the accumulated
amounts of production of the two goods, and p1 ; p2 with units e=kg,
e=unit, respectively, the corresponding prices. The accumulated values
are thus measured in units e. The corresponding vector function for the
accumulated production values is
V(t) D .V1 (t); V2 (t)/; (2.18)

where V1 (t) and V2 (t) are the coordinate functions and the vector
function is denoted by a bold letter, see Sect. 10.5.1. This vector function
describes the state of a two-good production system at time moment
t. The function maps the state of the production system at every
time moment as a point (vector) in the two-dimensional space of the
accumulated production values of the two goods. The position vector
changes when the accumulated value of production of either of the goods
increases (accumulated values of production cannot decrease). Vector
function (2.18) can be presented as
Z t Z t
0 0
V(t) D V1 (t0 ) C V1 (s)ds; V2 (t0 ) C V2 (s)ds
t0 t0
Z t Z t
D .V1 (t0 ); V2 (t0 )/ C V10 (s)ds; V20 (s)ds
t0 t0
D V(t0 ) C .V1 (t0 ; t); V2 (t0 ; t)/ D V(t0 ) C V(t0 ; t); (2.19)
where the coordinate functions are the accumulated values of productions

of the two goods, and t0 is the starting moment of production of the
good, the production of which was started earlier. The value set of a
two-dimensional vector function is a one dimensional set of points (a
curve) in a two dimensional space; see the calculation rules for vectors
in Sect. 10.4.1.
The graph of a two- dimensional vector function deviates from that
of a real valued function with one argument so that the value of a vector
function is a point in the two-dimensional space (the points of the curve),
and not a point on the vertical axis. In the coordinate axes of a two-
dimensional vector function are the values of the coordinate functions,
and all coordinate functions have a common argument (in the present
case, time) so that at every moment of time, each coordinate function has
a certain value. Thus the value of the vector function is exactly defined at
every time moment. A change in time is seen as a change in the length of
the curve, which is the graphical demonstration of the values of the vector
function.
The following connection can be presented between a one variable real-
valued function and a certain two-dimensional vector function. Let us
denote a one variable real function as x D f (t). In the coordinate system

(t; x) (t on the horizontal and x on the vertical axis), the graph of the
function x D f (t) equals with that of the following vector function
y(t) D (t; x) D .t; f (t)/:
The reader can check this by giving function f (t) a specific form (for
example x(t) D 5t), and making the graphs of both functions in the
coordinate system (t; x). The graph of a one variable real-valued function
can thus be identified as that of a vector function, the coordinate functions
of which are the argument of the real function and the real function
itself.
A two-dimensional vector function is a generalization of a two-
dimensional vector that can be described either as a point in a
two-dimensional space, or as an arrow drawn from the origin to the
point. A vector function deviates from a vector so that the value of the
vector function is not a fixed point in the vector space, but the value point
of a vector function moves when the values of the coordinate functions
change.
Because the coordinate functions V1 and V2 of vector function (2.18)
are both functions of time t, the value point of vector function (2.18)
changes with time. If time is assumed continuous, the value set of
the vector function (2.18) defines a continuous curve in the coordinate
system, see Fig. 2.7. If, however, the coordinate functions depend on
a discrete variable (for example discrete time), the graph of the vector
function is a set of separate points in the coordinate system.
In Fig. 2.7 is shown the graph of the vector function
F(t) D .F1 (t); F2 (t)/ D .3t; t C t2 / (2.20)
with 0 t 5. The first coordinate function is a linear function of

time, and the second is a second order polynomial of time. Figure 2.7
shows that the second coordinate function increases with time at a higher
velocity than the first. This makes the graph—that describes the values of
the vector function—curving as is shown in Fig. 2.7.
Fig. 2.7 The graph of vector function F(t) D (3t; t C t2 )
The derivative of a vector function is defined so that the derivative

of every coordinate function is taken separately according to the stan-
dard rules of derivation. For example, the time derivative of function
(2.18) is

V V1 V2
lim D lim ; lim , V0 (t) D .V10 (t); V20 (t)/;
t!0 t t!0 t t!0 t
and that of function F(t) in Eq. (2.20) is
F0 (t) D .3; 1 C 2t/: (2.21)
The time derivative of a vector function is its instantaneous velocity

vector, and its coordinate functions are the instantaneous velocities of
the coordinate functions. For example, in velocity vector (2.21) the
instantaneous velocity of F1 is constant and that of F2 increases linearly
with time.
From a vector function we can derive scalar-valued functions by
calculating its norms. Every norm measures in different ways the length
of the arrow corresponding to the vector function. The most common
norms are the Euclidean norm—that exactly measures the length of the
arrow—and the absolute value norm. These norms taken from function
V(t) are
q
kV(t)k D V12 (t) C V22 (t) and kV(t)k1 D jV1 (t)j C jV2 (t)j:
The norms of a velocity vector are scalars that measure the velocity (speed)
of the point the vector function describes. The speed of the value of
production in a two-good production system can thus be measured by the
norms of the corresponding vector function. As an example, the Euclidean
norms of the vector function F0 (t) in Eq. (2.21) at moments t D 1 and
t D 2 are:
p p p p
kF0 (1)k D 9 C 9 D 18 D 4:24 and kF0 (2)k D 9 C 25 D 34 D 5:83:
The increase in the velocity of the latter coordinate function makes the
length of the arrow corresponding to the velocity vector increase with
time; that is, the speed of the point increases with time. An analogous
result is obtained by using the absolute value norm. The absolute value
norm of vector function F0 (t) in Eq. (2.21) at time moments t D 1 and
t D 2 is:
kF0 (1)k1 D 3 C 3 D 6 and kF0 (2)k1 D 3 C 5 D 8:
If the coordinate functions of a vector function have a measurement

unit—for example, the coordinate functions of V have unit e—the norm
of the vector function has the same unit as the coordinate functions.
Next we study the production system of a whole economy and
assume, for simplicity, that only two goods are produced in the economy.
The accumulated values of the two goods at moment t are denoted as
above, V1 (t) (e), V2 (t) (e), and we assume that time t is measured in
units y. The velocity (speed) of the accumulated value of production in
the economy can then be measured by norms of the velocity vector that
corresponds to the vector function of accumulated production values. The
absolute value norm is
kV0 (t)k1 D jV10 (t)j C jV20 (t)j D V10 (t) C V20 (t); (2.22)
because here Vi0 (t) with unit e=y are nonnegative, i D 1; 2.

If the velocities of production values Vi0 (t) in Eq. (2.22) are replaced by
annual value added of industries in the economy, this norm corresponds
to the annual gross domestic product (GDP) of the economy. The GDP
of an economy thus measures the speed of a point in the space of
accumulated value added of industries of the economy, see Estola (2011).
2.8 Index Numbers

An index number presents the value of a quantity with respect to some
base value. We study the principle of index number calculation by one
example. Let us denote the price of a good at time unit 0 by p0 , and at time
unit 1 by p1 . The corresponding price index value x0 at time unit 0 is set as
100 (a large enough positive base value is chosen to avoid negative values
in the case the index decreases). Next we transform the original prices
to the corresponding index numbers. Let us denote the index number at
time unit 1 by x1 . The ratio between the two prices is then set equal to
the ratio of the corresponding index numbers, and the obtained equation
is solved with respect to x1 :
p1 x1 p1
D , x1 D 100 :
p0 100 p0
From the above equation we see that index number x1 is dimensionless;

the measurement units of the two prices cancel each other out. The next
observation x2 in the index number time series is obtained in an identical
manner. We denote the price at time unit 2 by p2 and use the above
derived value for x1 :
p2 x2 p2 p1 p2 p2
D ) x2 D x1 D 100 D 100 :
p1 x1 p1 p0 p1 p0
Continuing in this way we get the index number time series xi , i D

0; 1; : : : that corresponds to the price time series pi , i D 0; 1; : : : . The
index number time series has the same information of the development
of the price as the original time series, but it is scaled at the level starting
at value 100.
Table 2.3 The prices of goods and price indexesa

Year 1989 1990 1991 1992
Price of cheese, Emmental (FIM/kg) 43.1 46.8 49.7 50.1
Price of jeans (FIM/unit) 357.8 374.7 398.5 405.9
Cheese price index (1989 = 100) 100 108.6 115.3 116.2
Jeans price index (1989 = 100) 100 104.7 111.4 113.4
a
Source: The statistical yearbook of Finland 1993, p. 404
The usefulness of index numbers is based on the following: (1) dimen-

sionless index numbers are additive even though the original quantities
are not, and (2) index numbers of any quantities are directly comparable
when they have the same base time unit and base number (for exam-
ple 100). As an example, Table 2.3 demonstrates that price indexes show
more easily than prices that the price of cheese has increased faster than
that of jeans.
2.8.1 Index Numbers as Average Quantities
§: By the weighted average of observationsPxi , i D 1; : : : ; n of quantity

X wePunderstand the following sum: X D niD1 ai xi , where 0 ai 1
and niD1 ai D 1. Arithmetic average is a weighted average where every
observation has the same weight, i.e. ai D 1=n, i D 1; : : : ; n. ˘
In the above definition, the ‘weight’ of one observation means the effect
of the observation in calculating the average. Name weight comes from
physics where weighted average is used in calculating the central mass
point of a body. If a body is constructed of separate mass points laying on
a straight line, its central mass point is calculated as the sum of deviations
of the individual mass points from a fixed point. These deviations are
multiplied (weighted) by the shares of the mass points of the total mass
of the body. In this calculation, the deviations of the mass points from a
fixed point are the observations, and the weights are the shares of the mass
points from the total mass of the body. These relative weights add up to
unity as is required.
§: In probability calculus, the expected value of a discrete quantity
is calculated by weighting the observations of the quantity by their
probabilities that add up to unity. ˘
Expected value is thus a weighted average of the observations of a

quantity, and the probabilities of the observations can be identified as
the shares of the observations of the total probability ‘mass’. The expected
value of a quantity is its most probable value calculated on the basis of its
observations. The higher the probability of an individual observation, the
greater the weight it has in the calculation of the expected value.
Example
Suppose a player tosses a fair die once and wins the amount of euros the
die expresses. What is the expected win in this game?
Answer. Every possible value 1; : : : ; 6 of the die has probability 1=6. The
expected win in the game (E comes from ‘Expected’ and W from ‘Win’) is:
1 1 1
E(W) D 1 (e) C 2 (e) C C 6 (e)
6 6 6
1 21 1
D (1 C 2 C C 6) (e) D (e) D 3 (e): (2.23)
6 6 2
The expected win in the game is a weighted average of all outcomes

weighted by their probabilities. Because the probability of every side of
the die is 1=6, the expected win in the game corresponds to the arithmetic
average of possible outcomes. ˘
One problem, that can be solved by using indexes, is the additivity of

quantities with different measurement units. Oftentimes we are interested
in the average development of prices of a group of goods. As an example,
we make an average price index of the prices of sausage with unit e=kg
and milk with unit e=l. The constructed average price index measures the
average development of these two prices. We operate as follows. Collect
observations of sausage and milk prices from a time unit; for example,
monthly observations in one year. Then construct the corresponding
monthly price indexes for the two goods, as we showed earlier. Choose
January as the base time unit for both index time series, and set the index
values at the base month to 100. Then weight the two indexes by any
suitable method (for example with equal weights, or sausages with 2/3
and milk with 1/3, etc.) and calculate a weighted average of the two index
values every month. This is a weighted price index of the two goods.
The most commonly used weighted price index is the consumer price
index (CPI) that measures the average living costs of households. The CPI
is a weighted average of price indexes of consumption goods. The weights
are the shares of the goods in the aggregate expenditures of households.
These weights sum up to unity as they should, and they are determined by
interviewing consumers after fixed time units. For example, if on average
consumers spend 10% of their monthly expenditures on milk, the price
index of milk has weight 0.1 in CPI.
The CPI is not the only commonly used weighted price index. The
average price development in export and import goods are calculated on a
similar basis. For example, the prices of export goods of a country are
transformed to the corresponding price indexes, and an average price
index of export goods is calculated by weighting the individual price
indexes by the value shares of the goods of the total exports of the country
at a time unit. This type of calculated average price index is called an
export price index.
Besides price indexes, volume indexes of amounts of goods are defined
because the amounts of different goods are not additive. For example,
the annual production of sugar may be measured in units kg=y, that of
milk in units l=y, and so forth. Thus this addition has the same problem
with measurement units as prices have. Volume indexes are dimensionless
numbers as price indexes. By weighting the volume indexes of production
flows of individual goods with a certain principle, we get average volume
indexes that measure the development of the flow of production of a group
of goods. Commonly used average volume indexes are the volume index
of domestic production, and export and import volume indexes.
2.9 The Production Function

Production function describes in a mathematical form the connection
between the amount or the flow of production and that of the applied
production factors.
§: Let the flow of production of good k of a firm depend on the labor

input used in the production as qk D fk (Lk ), where qk is the average flow
of production of good k in a time unit and Lk the labor input used in the
production of good k at the time unit. Function fk —that represents the
relationship between these quantities—is called the production function
of good k of the firm. ˘
The production function of a firm producing good k describes the
production method of the firm at a very abstract level. The exact form
of the production function of a firm can be estimated on the basis of
the measured labor input and the flow of production. However, even
though we get an accurate estimate of the production function, it does
not tell how these quantities actually depend on each other; that is, how
the employees construct the final products. In spite of these limitations,
production function is a useful theoretical term; with it we can define
useful concepts for modeling the behavior of a firm.
§: The average productivity of factor x in the production of good k is
measured by dividing the amount of production of good k in a time unit
by the use of factor x in this production at the time unit. ˘
§: According to the production function, the average productivity of
labor can be measured as: qk =Lk D fk (Lk )=Lk . ˘
If the flow of production of a firm is measured in units kg=mn, and
the labor input by the number of workers working in a month, the
average productivity of labor with unit (kg=mn)=(unit=mn) = kg=unit
measures the average monthly production of one worker. If, however,
labor input is measured by the number of hours worked in a month, and
the flow of production as before, the average productivity of labor with
unit (kg=mn)=(h=mn) = kg=h measures the average hourly production in
a month.
§: The marginal productivity of factor X in the production of good k
is the ratio of a change in the flow of production of good k and the change
in the use of factor X in this production in a time unit. ˘
In theoretical modeling, we can assume that the production function
qk D f (Lk ) is differentiable, and then the labor input can be changed
in ‘small bits’ (see Sect. 2.7.1 on the continuity of production processes).
The marginal productivity of labor can then be expressed as the following

limit value (see Sects. 10.7–10.8):
ˇ
qk (Lk ) qk (Lk0 ) qk dqk ˇˇ
lim D lim D D f 0 (Lk0 ): (2.24)
Lk !Lk0 Lk Lk0 Lk !0 Lk dLk ˇLk DLk
0
The marginal productivity of labor is obtained by differentiating the

production function with respect to the labor input. The measurement
unit of marginal productivity is the same as that of average productivity,
and it depends on the units of the flow of production and labor input.
Derivative f 0 (Lk ) expresses the same thing as quantity qk =Lk but is
measured at a smaller change in the argument. If qk and Lk are measured,
for example, in units kg=y and h=y, respectively, then, for instance,
marginal productivity of labor 100 (kg=h) can be transformed as follows:
100 (kg=h) = 100 (kg=3600sek) = 1=36 (kg=sek) and so on. Marginal
productivity can thus be expressed in the units we prefer.
A firm’s marginal productivity of any production factor is measured
by using observations from a fixed time unit, and in this measuring, the
firm’s use of other productive factors do not usually stay constant. Because
this measurement cannot be done in laboratory circumstances, we have
to notice the risk in this that a change in another production factor has
actually caused the change in the flow of production we use in measuring
the marginal productivity of this factor. We can believe that the marginal
productivity of every factor is positive at every ‘reasonable level’ of use of
the factor; that is, an increase in use of any production factor increases the
flow of production and vice versa.
If the production function of a firm is assumed continuous, the use of
the inputs of the firm can be assumed to be changed by small amounts.
In economics textbooks, it is usually assumed that inputs are changed
by one unit at a time. This restriction is not necessary, however, because
even though labor input is measured by the number of workers or number
of working hours in a time unit, we can still change the labor input by
‘a part’ of a worker or by a part of an hour at the period (Example 1
below). In measuring marginal productivity it is not essential how much
the production factor is changed, but in which units the quantities are
measured. When we measure marginal productivity in a real production

process, we use observations of realized changes; in theoretical analysis,
on the other hand, we can assume that the corresponding functions are
differentiable and therefore differentiate them.
Example 1
The labor input of a worker is changed in a certain production by the share
1=100 of his weekly work time (he does mostly other works but changes
this work this much). Labor input in working hours can also be changed,
for example by 3=8 hours in a day, the use of computer time in production
can be changed by 3 minutes in a week, and so on. ˘
Example 2
If the use of labor of a firm is analyzed as a function of time t denoted
by L(t), the following connection exists between average and marginal
productivity of labor, and ‘instantaneous velocity of productivity of labor’
(see Sect. 10.7.3):

q.L(t)/
d L(t)
L0 (t) 0 q.L(t)/
D q .L(t)/ : (2.25)
dt L(t) L(t)
Formula(2.25) implies that the instantaneous velocity of productivity of
q.L(t)/
labor, d L(t) =dt, is positive, if the instantaneous growth rate of use of
labor L (t)=L(t) is positive, and the marginal productivity of labor q0 .L(t)/ D
0
dq=dL
is greater than the average productivity q .L(t)/ =L(t). However,
d L(t) =dt is also positive if L0 (t)=L(t) < 0 and q0 .L(t)/ < q .L(t)/ =L(t). We
q.L(t)/
thus have to be careful with respect to which variable we study changes,

that is, with respect to which variable we differentiate. ˘
§: The law of non-increasing marginal productivity. If the use of

factor x in the production of good k is increased in a time unit, and the
use of other production factors are kept constant, the flow of production
of good k in the time unit increases at a non-increasing rate. ˘
Example 3
One worker working with a computer does a certain amount of production
in one month. Adding another worker to work with the same computer
increases the monthly production with that computer, but hardly as much
as the first worker increased from zero production. Adding a third worker
working with the same computer still increases the monthly production
with this computer, but the fifth worker working with the same computer
hardly increases the monthly production at all. ˘
Let the measurement unit of labor input be h=mn and suppose that
the labor input in the production of good k is increased by one hour per
month at a time. The law of non-increasing marginal productivity can
then be expressed as:
ˇ ˇ ˇ
qk ˇˇ qk ˇˇ qk ˇˇ
;
Lk ˇLk D0 (h=mn) Lk ˇLk D1 (h=mn) Lk ˇLk D2 (h=mn)
which is demonstrated in Fig. 2.8.
Δqk
ΔLk
Δq k
ΔLk Lk = 100
Δq k
ΔLk Lk = 300
100 300 Lk (h / mn)
Fig. 2.8 Non-increasing marginal productivity of labor

Example 4
Suppose the production function of a firm is of the form q(L) D aL bL2 ,
where the measurement units of q; L are kg=week and h=week, respectively,
and a; b are positive constants with units kg=h and (kgweek)=h2 , respectively.
The marginal productivity of labor is then
a
q0 (L) D a 2bL 0 ) L ;
2b
where q0 (L) has unit kg=h. Suppose a D 100; b D 1=2. Then non-increasing
marginal productivity of labor holds with this function (L < a=2b D 100):
q0 (L)jLD1 D 100 1 D 99; q0 (L)jLD2 D 100 2 D 98; q0 (L)jLD20 D 80; etc. ˘
References
Allen, R. G. D. (1938). Mathematical analysis for economists. London: MacMillan.
Allen, R. G. D. (1956). Mathematical economics. New York: St. Martin’s Press.
de Jong, F. (1967). Dimensional analysis for economists. Amsterdam: North-
Holland.
Estola, M. (2011). Measuring the growth and the structural change of a
Economy, 4(1), 23–44.
Fisher, I. (1930). The theory of interest. New York: Macmillan.
Ohanian, H. C. (1989). Physics. Second Edition Expanded. New York: W.W.
Norton & Company.
3
Consumer Behavior
We can analyze basically two kinds of consumption decisions of

consumers: a consumption decision (1) for a given time unit and (2)
for the rest of the life of a consumer. The latter contains the choice
for the length of education and the estimation of the income for the
working and pension times, depending on the age of the consumer. This
analysis contains various uncertain elements that make an exact life-long
consumption plan impossible. On the other hand, the prices of goods
change with time, and people’s needs change with aging. These elements
make an exact life-long consumption plan irrational. Thus, a rational
person makes exact consumption plans only for a fixed time unit,
such as one day, week, month, and so on. Concerning the empirical
evaluation of a theory, the lifetime consumption plans of consumers
cannot be measured as long as consumers are alive. If we thus want to
construct a theory of consumer behavior that can be tested empirically,
we have to choose the first of the two possibilities because from people’s
daily, weekly, monthly, etc, consumption we get observations.
§: A consumer is a person that consumes goods in an economy. For
example, an entrepreneur is a consumer when he/she buys goods for
his/her personal consumption. ˘

DOI 10.1007/978-3-319-46879-2_3
In the following we assume that the studied consumer is male to avoid

using he/she every time. Now, every consumer has some kind of a plan for
his future consumption and its financing. For example, people make plans
for their education and working career, to buy a home, and so on. The
lifetime income plan of a consumer defines the framework for his lifetime
consumption plan. The short-term consumption of a rational consumer
follows the framework of his lifetime consumption plan. A consumer may
save money to buy a home in the future and so currently consume less than
he can afford. On the other hand, a consumer may consume currently
more than he can afford by way of a loan. We can therefore think that
consumers have an idea of their lifetime income schedule and they use it to
budget money for their consumption for every time unit of their life. If the
lifetime income schedule changes essentially, for example due to winning
in a lottery, the lifetime consumption plan is adjusted accordingly.
In this chapter, we model a consumer’s consumption decision for
a relatively short time unit. With respect to the consumer’s lifetime
consumption plan we assume that at a relatively robust level he has one,
and he uses it to budget money for his consumption for every time unit
of his life. The studied time unit is assumed to be one week because
most people make their daily consumption decisions for a few days or
a week. We assume that the consumer chooses his weekly consumption
by one decision where he decides his weekly amounts of consumption (or
flows of consumption) of goods with known prices and a fixed amount of
money the consumer has budgeted himself for the week. For simplicity,
the consumer is assumed not to be able to borrow money; borrowing is
analyzed separately in Chap. 9.
People make their consumption decisions every now and then, and so
real word behavior deviates from the idealized one we will model in the
following. The reason for this idealization is to keep the model as simple
as possible. The weekly consumption flows of goods are determined
from the daily ones, however, and we can think the decision for the
weekly consumption as a weighted average of daily decisions. This is
demonstrated in the next example.
Example
Let the average daily flow of meat consumption of a consumer be 0.1 (kg=d);
this corresponds to the weekly flow 0.7 (kg=week). Weekly consumption is
thus obtained as an average of daily decisions. ˘
3.1 The Axioms of Consumer Behavior

The theory of consumer choice is based on the following four axioms that
are assumed to hold during the studied time unit (one week).
Axiom 1: The Decision-Making Situation The prices of goods are

assumed fixed and known by consumers who have budgeted a fixed
amount of money for their consumption for the time unit.
Axiom 2: The Space of Consumption Flows The consumption space

of a consumer is a closed convex set (Sect. 10.3).
Axiom 3: Nonsatiation of Needs Consumers prefer a situation where

they consume more. For a rational consumer, the consumption of any
good does not reach the satiation level; that is, an increase in consumption
of any good increases the satisfaction of the consumer.
Axiom 4: Optimality of Choice A consumer chooses the bundle of

consumption flows of goods that gives him the greatest possible satisfac-
tion for the time unit.
Corollary to Axioms 2 and 3. If three combinations of consumption

flows of goods X1 ; X2 ; X3 exists in the consumption space of a consumer,
and the consumer prefers X3 to X2 and X2 to X1 , then a unique combina-
tion of consumption flows of goods exists in the consumption space the
consumer considers as good as X2 which is located on the line segment
connecting points X1 and X3 .
The role of these axioms for modeling consumer behavior is analyzed

later. Axiom 2 is needed only in the mathematical solution of the
optimization problem of a consumer, and it has no role in real world
consumer behavior. Axiom 3 implies that a rational consumer never
consumes any good so much that it reaches the satiation level where the
consumer would no longer enjoy consuming the good. The reason for this
is that consumers have many opportunities for using their scarce money,
and a rational consumer allocates his funds for consumption so that before
approaching the satiation level of a good, he stops spending money on
that good and starts spending on other goods. The corollary is presented
in the connection of the axioms because it is needed later in the modeling.
The proof of the corollary would require an exact set-theoretic definition
of the consumption space, and for this reason the proof is omitted. The
principles of the proof will be explained later. The proof can be found in,
for example, Debreu (1959, pp. 56–58).
3.2 A Consumer’s Budget Equation

Suppose a consumer consumes n different goods in a week. His weekly
consumption expenditures E (e/week) can then be expressed as
E D p1 q1 C p2 q2 C C pn qn ;
where qi are the flows of consumption of the n goods and pi their prices,
i D 1; : : : ; n. The measurement units of the quantities could be, for
example, q1 (kg=week), q2 (unit=week), q3 (l=week), : : : , and p1 (e/kg),
p2 (e=unit), p3 (e=l), and so on.
In this chapter, we assume a consumer’s decision-making situation as
simple as possible. The consumer can choose his weekly consumption of
only two goods the consumer consumes every week. For clarity, let good
1 be ‘food’ and good 2 ‘playing video games’ according to the traditional
choice between ’food or fun’. This simplification is made because choices
between two goods can be demonstrated in a two-dimensional space. A
generalization to n goods (n > 2) is analogous, but it requires functions
of several variables and it cannot be demonstrated graphically if n 4.
As stated above, a consumer is assumed to have budgeted a fixed

amount of money T (e=week) for his consumption for the week, and
the consumer knows the fixed prices of food and playing video games pf
(e=kg) and pv (e=h), respectively. The weekly budget of the consumer
is then:
T (e=week) D pf (e=kg) qf (kg=week) C pv (e=h) qv (h=week); (3.1)
where the consumption flows of the two goods are denoted by qf and qv .
In the following we do not present the measurement units of the quantities
in equations as above to simplify the notation; units are treated separately
when necessary. The combinations of consumption flows the consumer
can afford with his budgeted funds are shown in Fig. 3.1, where on the
coordinate axes are the consumption flows of the two goods. Only the
positive quadrant of the space is presented—negative consumption flows
are impossible—and on the horizontal axis qv D 0 and on the vertical
axis qf D 0.
The line in Fig. 3.1 is called the budget line or the frontier of the
set of consumption possibilities of the consumer. The points on the
line represent the combinations of consumption flows of the two goods
qv (h / week )
T / pv
T= p q + pq
f f v v
T/p f
q f (kg / week )
Fig. 3.1 The weekly budget equation of a consumer

where the budget holds as an equality. Thus the line is the graphical
presentation of the weekly budget of the consumer. Axiom 1 implies
that the graphical presentation of the weekly budget of a consumer is
unique. If the consumer does not play video games at all, he operates
on the horizontal axis in Fig. 3.1 and can at most consume food qfmax D
T=pf (kg=week). If the consumer does not eat at all during the week, he
operates on the vertical axis and can at most play video games qvmax D
T=pv (h=week). These cases are obtained from Eq. (3.1) by setting first
qv D 0 and then qf D 0, and solving the other quantity from the
equation. In the combinations of consumption flows of the two goods
inside the area restricted by the budget line and the two coordinate axes,
the consumer does not use all the money T (e=week).
The slope of the budget line in Fig. 3.1 is derived as follows. Suppose
two combinations of consumption flows of the two goods are located on
the budget line; let these be (qf0 ; qv0 ) and (qf1 ; qv1 ). Then
T D pf qf0 C pv qv0 and T D pf qf1 C pv qv1 :
Subtracting these two equations we get
T T D pf qf1 C pv qv1 pf qf0 pv qv0 ,

0 D pf (qf1 qf0 ) C pv (qv1 qv0 ) , 0 D pf qf C pv qv ; (3.2)
whereby is denoted the change in a quantity. Because pf and pv are

positive, the last form of the equation shows that if qf > 0 then qv <
0 and vice versa for the right hand side of the equation to equal zero.
Assuming qf ¤ 0, we can solve from Eq. (3.2):
qv pf
D :
qf pv
This slope of the budget line can also be obtained by solving Eq. (3.1)
with respect to qv and taking the derivative with respect to qr because
limqf !0 q
qf
v
D dq
dqf
v
, see Sect. 10.7.
The slope of the budget line is the ratio of the prices of the two goods
with a minus sign. The measurement unit of the slope h=kg shows that
it expresses the exchange rate between hours of video games and one
kilogram of food. The negativity of the slope shows that if the consumer
operates on his budget line, he must decrease his food consumption
(qf < 0) if he wants to increase his playing of video games (qv > 0),
and vice versa. These are the alternative costs of increases in consumption
of the two goods, because on the points of the budget line, all money T
(e=week) is used.
Example
Suppose a consumer has budgeted himself T D 400 (e=week) for consump-
tion, and the prices of the two goods are: pf D 2 (e=kg) and pv D 4 (e=h),
respectively. The budget equation is then
400 D 2 qf C 4 qv :
p
The slope of the budget equation is pvf D 24 D 12 , and the maximal
consumption flows are: qfmax D 200 (kg=week) and qvmax D 100 (h=week). ˘
3.3 A Consumer’s Preferences

We continue analyzing a consumer’s choice in the two good situation.
Suppose the consumer consumes at a randomly chosen point A with the
combination of consumption flows (qfA ; qvA ) in Fig. 3.2. Now, Axiom 3
implies that the consumer considers every point in area II (above and
right of A) better than A (the flow of consumption of either of the goods
is greater), and every point in area IV (below and left of A) worse than A
(the flow of consumption of either of the goods is smaller). On the other
hand, areas I (above and left of A) and III (below and right of A) cannot
be directly considered as better or worse than A. In this we need Axioms 2
and 3.
Axiom 3 implies that a consumer considers every point in Fig. 3.2 on
the horizontal line going through point A on the left of A (like B) worse
qv (h / week )
q vC
●C
D ● I II
q vA
B ● ●A
IV III
q fB
q fA q f (kg / week )
Fig. 3.2 The consumption space of a consumer
than A, because in these points his flow of consumption of video games

is as great as in A, but that of food is smaller. Accordingly, the consumer
considers every point on the vertical line above A better than A because
in these points (like C), his flow of consumption of video games is greater
than in A and that of food is equal. According to Axiom 3, the consumer
ranks the points as B A C, where B A means that the consumer
considers point A at least as good as point B.
Note. Note that denotes the preference relation and not the ‘at least
as great as’ relation, that is denoted by . Thus we cannot operate
mathematically with relation A B as we can with A B. ˘
Mathematically, the consumption space of a consumer in Fig. 3.2

can be identified with a two-dimensional vector space. The corollary
presented with Axioms 1–4 implies that there exists a point in the
consumption space the consumer considers as good as point A that is
located on the segment of the line connecting points B and C. Let this
point be D. Every point on the line segment BC can be presented as the
following convex combination of points B and C:
D D aB C (1 a)C; 0 a 1: (3.3)
The end points of line segment BC are obtained as: a D 0 ) D D C,

and a D 1 ) D D B. If 0 < a < 1, point D is an inner point of
line segment BC. Axiom 2 guarantees that point D is in the consumption
space.
Example
Let us identify the combinations of consumption flows in Fig. 3.2 as points
in a two-dimensional vector space. Let B D (qfB ; qvB ) D (4; 8) and C D
(qfC ; qvC ) D (12; 16), where we have omitted measurement units for short-
ness. Then A D (qfA ; qvA ) D (12; 8). The rules for multiplication a vector by
a scalar and adding two vectors are given in Sect. 10.4.1. With these rules,
the following results can be derived:
(1) a D 1=2 ) D1 D 1=2 (4; 8) C 1=2 (12; 16) D (2; 4) C (6; 8) D (8; 12).
(2) a D 1=4 ) D2 D 1=4 (4; 8) C 3=4 (12; 16) D (1; 2) C (9; 12) D (10; 14).
The reader can check that points D1 , D2 are on the line segment BC. ˘
Axiom 2 guarantees that no ‘holes’ exist in the consumption space

that would complicate finding the optimal consumption point for the
consumer. Axiom 2 implies that if the consumer can choose combination
50 (kg=week) food and 5 (h=week) video games, and combination 20
(kg=week) food and 10 (h=week) video games, he can also choose any
convex combination D of these two bundles of consumption flows: D D
a(50; 5)C(1a)(20; 10), 0 a 1. If a D 1=2, D D (25; 5=2)C
(10; 5) D (35; 15=2) and if a D 1=5, D D (10; 1) C (16; 8) D (26; 9),
and so on.
When two combinations of consumption flows are chosen (for example
B and C in Fig. 3.2), every point on line segment BC can be expressed as
a convex combination of these points. When a continuously increases
from 0 to 1, the convex combination D in Eq. (3.3) gets every point of
line segment BC. The location of point D—that the consumer considers
as good as point A on line segment BC—depends on his preferences.
Two consumers with different preferences find a different such point on
line segment BC. The location of point D giving an equal satisfaction
as point A on line segment BC is unique for one consumer, but not

for two consumers with different preferences because the exchange rates
between different kinds of satisfaction consumers gain from consuming
goods differ; see Sect. 2.2.4.
The idea of the proof of the corollary presented with Axioms 1–4 can be
explained by using Fig. 3.2. According to Axiom 3, a consumer’s weekly
satisfaction increases if he changes the combination of his consumption
flows along line segment BC from point B to point C. In the beginning
his weekly satisfaction is smaller, and in the end higher than at point A.
Axiom 2 guarantees that the consumer’s weekly satisfaction increases at
every step during this change, because every point in the consumption
space is a possible combination of consumption flows. Thus if satisfaction
continuously increases during this motion, starting at a lower level and
ending at a higher level than at point A, then at some point it must be
equal with that in A.
Next we show that according to the corollary, the consumer can rank
every point in areas I and III in Fig. 3.2 with respect to point A. The points
equally good as point A construct a unique one-dimensional set of points
(a curve) in areas I and III. This curve can be derived as follows. Draw a
straight line OS through the origin of the consumption space, and twist it
with the origin as the fixed point so that during its turning it crosses the
whole consumption space; see Fig. 3.3. Line OS cuts the horizontal and
qv (h / week )
I II
q vA
●A C●
IV ● D III
●B
Fig. 3.3 An indifference curve of a consumer

vertical lines going through point A so that for every cutting point on the
vertical line, a unique cutting point exists on the horizontal line.
Let one such pair of cutting points be B, C in Fig. 3.3, where the
segment of line OS combining points B and C belongs in area III. Now,
Axiom 3 implies that a unique point exists on line segment BC that the
consumer considers equally good as point A. Because the same holds for
every pair of cutting points B, C defined by line OS when it turns over the
consumption space, the twisting of the OS line over the space defines a
unique set of points (a curve) that represents equal weekly satisfaction
as point A. This curve can be proved continuous, that is, the points
formulate a continuous curve in the space. The proof of this would require
an exact set-theoretic description of the situation, which is the reason we
omit it; see Debreu (1959, pp. 55–58). This way derived curve is called an
indifference curve of the consumer.
§: An indifference curve represents points of equal satisfaction (later
utility) for a consumer. ˘
One continuous indifference curve going through point A divides the
consumption space of a consumer into two subspaces; points closer to
the origin than the curve and points further away. The first subspace
represents smaller and the latter higher weekly utility than point A.
The uniqueness of an indifference curve proved above implies that the
consumer can rank every point in the consumption space with respect
to point A. Points further away from origin than the indifference curve
going through point A represent higher, those on the curve equal, and
those closer to the origin smaller weekly utility than at point A. Because
point A was chosen randomly, any other point could have been chosen as
the reference point, and a curve could have been defined that represents
points of equal weekly utility as that point. Axioms 2 and 3 thus guarantee
that every point in the space can be ranked in preference relation with
respect to other points. Thus the consumer can say of any two points in
the consumption space whether he prefers one to another, or considers
them equally good.
The indifference curves representing different levels of weekly utility
of a consumer construct a family of curves that covers the whole
consumption space. Every point in the space represents certain weekly
utility, and equally good points define an indifference curve. Thus the
space is covered with indifference curves lying next to each other. The
further away from origin a curve lies, the higher weekly utility it repre-
sents.
The diagram of a family of indifference curves is analogous to the
contour curves of hills in large-scale topographical maps. The third
dimension, that measures the level of weekly utility, is assumed to be
rising upward from the two dimensional plane of the page of the book.
Indifference curves are the contour curves of the ‘hill of weekly utility’.
Indifference curves are obtained by cutting the ‘weekly utility hill’ with
horizontal planes at various heights, and dropping these ‘cutting point
curves’ on the ground plane. The method is exactly the same as how
contour curves of hills are presented in large-scale topographical maps, as
track-finding runners know. The lowest point of the ‘weekly utility hill’ is
in the origin (qf ; qv ) D (0; 0), and the hill rises when moving away from
the origin.
In Fig. 3.4 is graphed two-variable function u D A(aqf )c (bqv )1c ,
where the units of the positive constants A; a; b are ut=week, week=kg,
and week=h, respectively, and 0 < c < 1 is dimensionless. The numerical
values of the constants are: A D 10; a D 2; b D 2; c D 0:5. Utility u
is thus measured in units ut=week, and the values of the function create
a surface in the three-dimensional space. The arguments of the function
are measured on the horizontal axes, and the values of the function on
the vertical axis (Sect. 10.5.2). The contour curves corresponding to the
surface in Fig. 3.4 are presented in Fig. 3.5. This example demonstrates
the indifference curves corresponding to this utility function.
3.4 A Consumer’s Optimal Choice

We continue analyzing the two-good situation. Axiom 4 given earlier
states that a consumer chooses the bundle of consumption flows that gives
him the highest possible weekly utility. Next we study how this point is
found.
Fig. 3.4 The surface of values of a two-variable function
In the previous section we showed that if Axioms 1–3 hold, a consumer

can rank different bundles of consumption flows in a preference order.
This ranking does not, however, show how much the consumer’s utility
changes when he changes his bundle of consumption flows, because the
ranking is measured on order ordinal scale (Sect. 10.16.2). If a variable is
measured on an order scale, we can only say whether it increases, decreases,
or stays constant. To be able to say how much a consumer’s weekly utility
changes when he changes his consumption flows, we should be able to
measure the weekly utility of the consumer at least on an interval scale
(Sect. 10.16.3). We should therefore be able to measure the satisfaction of
human beings on an interval scale, and so far we do not know how to do
this.
The problems in measuring the satisfaction of a consumer do not,
however, prohibit us from modeling a consumer’s behavior on the basis
Fig. 3.5 Contour plot of the surface in Fig. 3.4
of his preferences measured on an order scale. We can, namely, prove that

if the order relation between bundles of consumption flows is complete
(Sect. 10.3) and Axioms 2 and 3 hold, we can define a continuous scalar
valued function u D u(qf ; qv ) for which holds u(qf1 ; qv1 ) u(qf2 ; qv2 )
when the consumer considers combination (qf2 ; qv2 ) at least as good as
(qf1 ; qv1 ), that is, (qf1 ; qv1 ) (qf2 ; qv2 ). The proof of this proposition
requires an exact set-theoretic formulation of the situation, for which
reason we omit it. The proof is given in Debreu (1959, pp. 55–59).
Function u(qf ; qv ) is called the weekly utility function of the con-

sumer; it expresses the preference order of the consumer in a unique way
as a continuous function of its arguments. In this chapter, we assume
that the utility function of a consumer depends only on two variables
the mathematical theory of which is given in Sect. 10.5.
The measurement of the consumer’s weekly utility by function
u(qf ; qv ) is not unique, however. Any positive transformation of function
u, f (u); f 0 (u) > 0—for example a u(qf ; qv ) where a is a positive
constant or Œu(qf ; qv )2 —describes the same preference order as u(qf ; qv ).
If a > 1, function a u(qf ; qv ) expresses a higher absolute change in
utility than u(qf ; qv ) for a fixed change in the consumption bundle
(qf ; qv ). Thus the level of utility measured by a utility function is
not unique, even though the preference order expressed by proper
transformations of a utility function is unique.
To be able to write dimensionally well-defined mathematical expres-
sions with a utility function, we need to give a measurement unit for
utility. In Sect. 2.2.4 we introduced unit ut for the level of satisfaction
of a person. The measurement unit ut is not unique, however, because it
depends on the utility function. Every utility function—that expresses
the same preference order—defines an own measurement unit for
utility according to the values of the function. If, however, in modeling a
certain situation we consistently use one utility function for a consumer,
and we define measurement unit ut according to the values of this
function, we have defined one consistent way to measure the level of
satisfaction of the person in this situation.
The explicit measuring of utility in units ut is not needed in mod-
eling consumer behavior, however. Utility is only an auxiliary quantity
required in defining the marginal willingness-to-pay of a consumer for
various things. All utility functions that express the same preference
order of a consumer give equal marginal willingness-to-pay values for
goods near the consumer’s optimum. We prove this in Sect. 10.12.2. The
measuring problems of utility can thus be omitted with these remarks.
In a consumer’s decision-making situation, the length of the time
horizon is assumed to be one week. Because the consumer spends all
the money he has budgeted for his consumption for this week during
the week, the satisfaction he gains from his consumption takes place
in that week. For this reason, the values of utility function u(qf ; qv )
are measured in units ut=week. Utility u is thus a flow quantity that
measures the average level of satisfaction of the consumer in a week. Utility
function itself is a useful theoretical term in formulating quantities—such
as a consumer’s marginal willingness-to-pay for goods—that are helpful
in modeling consumer behavior.
§: The Average utility of a good for a consumer is measured by
dividing the utility of the consumer at a time unit by his flow of
consumption of the good at the time unit. ˘
The average utility of food and playing video games in a week are thus
u(qf ; qv ) u(qf ; qv )
and
qf qv
with units (ut=week)=(kg=week) D ut=kg and (ut=week)=(h=week) D

ut=h, respectively. They measure the average satisfaction the consumer
gains from one kilogram of food and one hour of playing video games in
the week.
§: By marginal utility of a good we understand the ratio between a
change in the flow of utility of a consumer and a marginal change in his
flow of consumption of the good at a time unit. Marginal utility measures
average utility from a marginal increase in the flow of consumption of a
good. ˘
The consumer’s marginal utilities of the two goods are
u(qf ; qv0 ) u(qf0 ; qv0 ) u @u(qf0 ; qv0 )

lim D lim D ; (3.4)
qf !qf0 qf qf0 q f !0 qf @qf
u(qf0 ; qv ) u(qf0 ; qv0 ) u @u(qf0 ; qv0 )

lim D lim D ; (3.5)
qv !qv0 qv qv0 qv !0 qv @qv
where qf0 ; qv0 are fixed flows of consumption.

Marginal utilities are obtained as partial derivatives (Sect. 10.9) of
a utility function with respect to the consumption flows. A partial
derivative—like an ordinary derivative—is the ratio between a change in
the function value and a marginal change in one argument of the function.
A change in the argument may be positive or negative (Sect. 10.9), and
the value of the function either changes or stays constant due to this
change. If an increase in the argument increases the value of a function,
and thus a decrease in the value of the argument decreases the value of
the function, the partial derivative like an ordinary derivative is positive,
and vice versa. Consumption flows qr0 ; qh0 are shown as arguments of the
utility function to demonstrate that marginal utilities depend on current
flows of consumption. Later, we do not usually show them in order to
simplify the notation.
The measurement unit of the marginal utility of a good is the same as
that of average utility. According to Axiom 2, for a rational consumer
the marginal utilities of goods are positive; that is, an increase in the
weekly flow of consumption of a good increases the weekly utility of the
consumer, and vice versa. If qf > qf0 , then u(qf ; qv0 ) > u(qf0 ; qv0 ) and
vice versa, namely, both the denominator and numerator in Eq. (3.4) are
simultaneously positive or negative, and the same holds for playing video
games.
§: Law of non-increasing marginal utility. Non-increasing marginal
utility means that an increase in the flow of consumption of a good
increases the consumer’s utility (satisfaction) flow at a non-increasing
rate. ˘
Example
Eating one orange increases the daily utility of a consumer. Eating another
orange during the same day further increases the daily utility, but at most
by as much as the first did. Eating a fifth orange during the same day,
however, does not increase the daily utility as much as the first did, and
eating the twentieth orange during one day hardly increases the daily
utility at all. The twenty-fifth orange may even make the consumer feel
bad, but a rational consumer never reaches that level of consumption
(Axiom 2). ˘
The above example demonstrates the law of non-increasing marginal

utility that holds for all goods. The more we consume any good in a time
unit, the less we enjoy increasing the consumption of the good in the
time unit. In the case of a differentiable utility function, the law of non-
increasing marginal utility can be expressed mathematically as:
@2 u(qf ; qv ) @2 u(qf ; qv )
0; 0;
@q2f @q2v
that is, the second order partial derivatives of the utility function with
respect to same quantity are non-positive. According to the law of non-
increasing marginal utility, the weekly utility of a consumer increases
with the consumption flows of goods, but a fixed increase in the flow
of consumption causes at most an equal utility increase as the previous
identical increase in the argument.
From the point of view of the consumer, finding the optimal combina-
tion of consumption flows can be thought of as moving along the budget
line starting from either of the coordinate axes. Suppose a consumer is
consuming at point qfmax and he considers whether to move along the
budget line towards playing more video games, see Fig. 3.1. The consumer
considers whether to change x (e=week) from food consumption to
playing video games. The consumer is assumed to do this change if it
increases his weekly utility. With x (e=week) the consumer can buy the
amount (u=qv ) (x=pv ) (ut=week) of utility from playing video
games. Analogously, with x (e=week) the consumer can buy (u=qf )
(x=pf ) (ut=week) of utility using this money on food.
In the optimal situation, the consumer cannot increase his weekly
utility by changing his combination of consumption flows. In the optimal
situation, the amount of weekly utility to be obtained with a fixed amount
of money is the same for all goods. This situation corresponds to
u x u x u 1 u 1
D , D ; (3.6)
qf pf qv pv qf pf qv pv
where the amount of weekly utility obtained by x (e=week) is set equal

for both goods; in the latter form of the equation, quantity x is cancelled
away. Equation (3.6) is called the condition for efficient consumption.
If Eq. (3.6) does not hold, the consumer can increase his weekly utility by
changing his bundle of consumption flows. The ratios
u 1 u 1
and
qf pf qv pv
with unit ut=e are called the efficiency ratios of consumption of the
goods, because they measure the amount of utility to be obtained from
these goods by one euro at the prevailing flows of consumption.
Note. Even though measuring the level of utility of a consumer is not

unique, the above efficiency ratios measured by the same utility function
are comparable quantities; that is, the above comparing is meaningful. ˘
The above means that the consumer chooses that point on his budget
line where he ‘most efficiently buys utility’. If the prices, budgeted funds,
and goods stay the same during several weeks, we can believe that with
time the consumer ends up consuming so that his weekly utility gets
maximized. The optimal situation can thus be considered as the point
where the consumer ends up if the decision-making situation stays the
same long enough.
The condition for efficient consumption in Eq. (3.6) can be generalized
to a situation where the consumer is choosing between n goods,
u 1 u 1 u 1
D D D ; (3.7)
q1 p1 q2 p2 qn pn
where i D 1; : : : ; n refer to goods. If Eq. (3.7) holds, the consumer cannot

increase his weekly utility by changing his bundle of consumption flows.
3.5 Utility Function and Indifference Curves

We continue the analysis in the previous section. The weekly utility of a
consumer depends on his consumption flows of the two goods, and other
factors affecting his satisfaction are omitted. We study the change in utility
when the consumer changes his weekly playing of video games from qv0 to
qv1 , that is qv D qv1 qv0 . The weekly utility of the consumer is in the
beginning u D u(qf0 ; qv0 ), and we analyze how it changes with qv ¤ 0.
The marginal utility of playing video games u=qv measures in units
ut=h the change in weekly utility due to a change in playing video games.
Because qv is a certain amount of playing video games in a week, its
u
effect on the weekly utility can be expressed as: u D q v
qv (ut=week).
When the consumer changes his playing of video games by qv ¤ 0,
his weekly utility changes due to the positive marginal utility. For the
weekly utility to stay constant after this change, the consumer must
simultaneously change his food consumption in the opposite direction.
Let us denote the change in the flow of food consumption as qf D
u
qf1 qf0 . The effect of this on utility is u D q f
qf (ut=week). For
the weekly utility to stay constant when the composition of consumption
changes from (qf0 ; qv0 ) to (qf1 ; qv1 ), the following condition must hold
(qf ; qv ¤ 0):

u u qv u u
qf C qv D 0 , D = : (3.8)
qf qv qf qf qv
On the left hand side of the latter form of the equation is the ratio
of changes in consumption flows, and on the right hand side the ratio
of marginal utilities with minus sign. This ratio expresses the exchange
rate between the two consumption flows that keeps the weekly utility
of the consumer constant. Because marginal utilities are positive at all
consumption flows, the right hand side of the latter equation in Eq. (3.8)
is negative. Then, the left hand side is also negative, that is, if qf is
positive then qv is negative and vice versa.
The consumer is assumed to be able to estimate the numerical values
of his marginal utilities at all levels of his consumption flows. From
Eq. (3.8) we can solve how much the consumer is ready to decrease his
food consumption (qf < 0) to increase his playing of video games
by qv > 0 for his weekly utility to stay constant. The exchange
rate between the two consumption flows, which keeps the consumer’s
weekly utility constant, is determined by marginal utilities. The greater
the marginal utility of food, and the smaller that of playing video games,
qv (h / week )
I II
q vA
●A
IV III
Fig. 3.6 An indifference curve of a consumer
the greater the absolute value of the exchange rate; that is, the more
playing of video games the consumer is ready to exchange for a certain
consumption of food, and vice versa.
We analyze the situation by using Fig. 3.6, where a certain combination
of consumption flows is denoted by A. The bundles of consumption flows
the consumer considers equally good as point A are in areas I and III.
Only in these areas the consumption flow of one good increases and that
of the other decreases as compared with A. Earlier on we showed that
all bundles of consumption flows equally good as point A formulate a
continuous curve in the plane. Next we show the connection between
an indifference curve and the analysis in this section. The smaller
. the
qv u u
changes qf , qv are, the more exactly ratio qf D qf qv
corresponds to the slope of the tangent of an indifference curve at a point.
By the limiting process we get the following exact correspondence (see
Sects. 10.7.1 and 10.10.2):
.
dqv @u @u
D : (3.9)
dqf @qf @qv
Result (3.9) can be derived also by using differential calculus. The change
in the value of a two variable function u D u(qf ; qv ) (the total differential
of u) is defined in differential calculus as (see Sect. 10.10.2):
@u @u
du D dqf C dqv :
@qf @qf
A change in the function value, du, depends on changes in the arguments

dqf ; dqv multiplied by their partial derivatives. When we measure the
change in u by its total differential, changes in all arguments of the
function are taken in account. On an indifference curve, the weekly utility
is constant, that is, du D 0. Then, setting du D 0 in the above formula
and considering dqf ; dqv ; @u=@qf ; @u=@qv as scalars by which we can
multiply and divide, we get result (3.9) by solving quantity dqv =dqf from
the above equation.
Because the slope of the tangent of an indifference curve is negative,
indifference curves are decreasing in coordinate system (qf ; qv ). Non-
increasing marginal utility means that the absolute value of u=qv is
the greater the smaller is the consumer’s playing of video games, and
vice versa. Thus an indifference curve is the more steep the smaller the
consumer’s playing of video games, and the greater his flow of food
consumption. The form of the indifference curve in Fig. 3.6 results from
the law of non-increasing marginal utility of consumption. This can be
proved mathematically as:
! 2
@2 u @2 u dqv @u @ u dqv @2 u @u
C C
d2 qv @q2f @qv @qf dqf @qv @qv dqf
2 @qf @qv @qf
D 2 :
dq2f @u
@qv
(3.10)
According to Eq. (3.9), dqv =dqf < 0. Then Eq. (3.10) implies that
d2 qv =dq2f > 0, that is, the shape of the indifference curve is as in Fig. 3.6,
if @u=@qi > 0, @2 u=@q2i 0, i D f ; v, and @2 u=@qv @qf > 0. We return
to this last condition in the next section.
3.6 The Equilibrium State of a Consumer

A consumer is assumed to aim at the highest possible level of weekly utility
his budget allows (Axiom 4). Thus, the consumer aims to operate on an
indifference curve furthest away from the origin his budget allows. At the
optimal point, one indifference curve touches the budget line. This indif-
ference curve represents the highest weekly utility the consumer can attain
with his budget. At the optimal point, the slope of the budget line and
that of a tangent of an indifference curve are equal. This corresponds to

pf u u u 1 u 1
D = , D : (3.11)
pv qf qv qf pf qv pv
Equation (3.11) is the condition for efficient consumption we derived

earlier in Eq. (3.6). It implies that the exchange rate between the two
goods expressed by the price ratio (the slope of the budget line) equals
with the subjective exchange rate between the goods that keeps the weekly
utility of the consumer constant; that is, the slope of an indifference curve.
If the consumer likes to exchange the two goods at a different rate than
they can be exchanged according to their prices; the consumer can increase
his weekly utility by changing his bundle of consumption flows. In the
optimal situation, this is not possible.
A consumer’s optimization problem can be solved by solving Eq. (3.11)
together with the budget equation with respect to quantities qf and qv .
Mathematically, this can be formulated as a constrained optimization
problem where the weekly utility function u D u(qf ; qv ) is maximized,
restricted by the budget equation T D pf qf C pv qv . This is shown in
Sect. 10.12.1. The optimal point is shown graphically in Fig. 3.7, where
the optimal consumption flows are denoted by qf and qv .
Our theory assumes that a consumer changes his bundle of consump-
tion flows of goods so that he ends up consuming in his optimal point
called his equilibrium state. This terminology is borrowed from physics.
In physics, the equilibrium state of a moving body is the one where the
resultant sum of forces acting upon the body is zero. Outside equilibrium,
the motion of the body is modeled mathematically by using differential
qv (h / week )
T/p v
q *
v
u 2
u 1
u 0
q *
f
T/p f
q f (kg / week )
Fig. 3.7 The equilibrium state of a consumer
equations. In economics, so far, the exact modeling of this adjustment

has been replaced by the following verbal explanation: if a consumer is
not in his optimal situation, he will adjust his bundle of consumption
flows so that he ends up at his optimum. In Sect. 3.8 we show, however,
that a dynamic modeling similar to Newtonian mechanics can be made
in economics too.
In Sect. 3.8 we define a force field that forces a utility-seeking con-
sumer towards his equilibrium state with time, if the initial state is not
the optimum. A consumer’s equilibrium is thus stable; that is, a state
where the consumer returns if he happens to deviate from it. If a factor
uncontrollable by the consumer (like budgeted funds after winning in a
lottery, or the price of a good) changes, we show in Sect. 3.8 how the
adjustment of a consumer from an ‘old’ equilibrium to a ‘new’ one can be
modeled.
In the modeling of dynamic phenomena, we need to understand
differential or difference equations, which cannot be assumed of a reader
at the beginning of his studies of economics. For this reason, all sections
where differential equations are treated are marked with a an asterisk.
These sections demonstrate the applicability of the dynamic framework
introduced in this book. Hopefully this motivates serious students of

economics to learn the mathematical (and physical) techniques applied
in them.
3.7 Dynamic Consumer Behavior

In this section, we model consumer behavior in a dynamic form so that the
above-presented optimal behavior corresponds to an equilibrium state in
this. We continue analyzing the two-good situation. The budget equation
is as earlier, but now the consumer’s weekly utility is expressed by function
u D u.qf (t); qv (t)/, where the consumption flows of the two goods are set
to depend on time t. The consumer is assumed to adjust his consumption
flows of the two goods so that his weekly utility increases with time.
From the weekly budget equation we get qv (t) D .T pf qf (t)/=pv ,
where, according to the Axiom 1, all other quantities besides the consump-
tion flows are assumed fixed. Substituting this in the utility function, we
get:

T pf qf (t)
u(t) D u.qf (t); qv (t)/ D u qf (t); : (3.12)
pv
The consumer can now affect his weekly utility only by quantity qf ,
because other quantities in the function are constants and qv is substituted
by the budget equation. Differentiating the utility function with respect
to time by using the chain rule (Sect. 10.9.4), we get:

du 0 @u pf @u
u0 (t) D q (t) D q0f (t); (3.13)
dqf f @qf pv @qv
@u @u p
where the two quantities subtracted in @q f
pvf @q v
both have unit ut=kg.
In Eq. (3.13), the unit of u (t) is ut=week and that of q0f (t) is kg=week2 ;
0 2
u0 (t) and q0f (t) are thus the instantaneous acceleration of utility and food
consumption, while u and qf are the corresponding velocities.
The consumer is assumed to adjust his consumption flows with time

so that his weekly utility increases. The adjustment rules for food con-
sumption are: q0f (t) > 0 if dq
du
f
> 0, q0f (t) < 0 if dq
du
f
< 0, and q0f (t) D 0
du
if dq f
D 0. These adjustments make the last form of Eq. (3.13) positive,
because the sign of a product of two quantities where they both have an
equal sign is positive. Then, also, the left hand side is positive; that is
u0 (t) > 0, and thus the weekly utility increases with time.
Next, we check under which conditions the equilibrium state q0f (t) D 0
of the consumer is optimal. The necessary condition for the consumer’s
optimum is (see Sect. 10.12.2):
du @u @u pf @u 1 @u 1
D0 , D0 , D : (3.14)
dqf @qf @qv pv @qf pf @qv pv
This corresponds to the optimal situation of the consumer presented

earlier. The sufficient condition for the maximum is that the second order
derivative
2
d2 u @2 u @2 u pf @2 u pf @2 u pf
D C (3.15)
dq2f @q2f @qv @qf pv @q2v pv @qf @qv pv
is negative. Non-increasing marginal utilities of goods makes the second

order partial derivatives of the utility function with respect to the same
quantity non-positive. The first and third additive terms in Eq. (3.15) are
thus non-positive. It has been proved in mathematics that if the partial
functions of a multivariable function are continuous (Sect. 10.9.1), then
second order cross partials @2 u=(@qf @qv ) and @2 u=(@qv @qf ) are equal
(Apostol 1969, pp. 277–281). Now, assuming the partial functions of the
utility function to be continuous, the sufficient condition for maximum
is that the unique second order cross partial is positive. This cross partial
measures the effect on the marginal utility of food consumption caused
by a change in playing video games, or vice versa. In practice, this effect
can be very small, but if either sign must be assumed, positive is more
plausible. The greater the flow of food consumption, the more we can
assume a consumer enjoys increasing his playing of video games when he
consumes within the limits of his budget.
According to the above, a consumer compares the positive change

@u
in utility due to a marginal increase in food consumption @q f
, and the
@up
negative change in utility pvf @q v
due to a decrease in playing video
games. This alternative cost is caused by the binding budget constraint.
The greater the marginal utility of food and the price of playing video
games, and the smaller the marginal utility of playing video games and
the price of food, the more certainly the consumer increases his flow of
food consumption and vice versa.
In Sect. 2.2.3 we claimed that a consumer’s choice can be simplified to
a comparison between the price of a good and the consumer’s marginal
@u @u
willingness to pay for it. Because quantities @q f
, @qv , pf and pv are all pos-
itive, multiplying the inequalities describing the consumer’s adjustment
@u
of food consumption by positive factor pv = @q f
, we can express them as:
pv @u
q0f (t) > 0 if pf > 0;
@u @qf
@qv
pv @u
q0f (t) < 0 if pf < 0;
@u @qf
@qv
pv @u
q0f (t) D 0 if pf D 0:
@u @qf
@qv
Analogous adjustment rules can be derived for playing video games. This
is done as follows. Solve the budget equation with respect to qf (t), use this
to substitute qf (t) in the utility function, and differentiate it with respect
to time. Then, define the adjustment rules as we did above. Quantities

@u @u @u @u
hf D = pv ; hv D = pf
@qf @qv @qv @qf
derived in this way have units ((ut=kg)=(ut=h))(e/h) D e=kg and

((ut=h)=(ut=kg)) (e=kg) D e=h, respectively, and we can interpret
them as this consumer’s marginal willingness-to-pay for one kilogram

of food and for one hour of playing video games, respectively. The
explanation is as follows. A utility-seeking consumer compares the above
quantities and the prices of the goods, and increases the flow of consump-
tion of that good for which the above quantity is greater than the price,
and decreases the flow of consumption of that good for which the quantity
is smaller than the price. The consumer thus pays the price of the good
the consumption of which he increases, and does not pay the price of the
good the consumption of which he decreases. This consumer behavior
is empirically testable by creating a questionnaire concerning consumers’
marginal willingness-to-pay for different goods and comparing these with
their prices.
A consumer’s marginal willingness-to-pay for food is the greater the
@u
higher the marginal utility of food, and the smaller the quantity @q v
=pv
with unit ut=e. Next we show that the latter quantity measures the
consumer’s marginal utility of budgeted funds for the week, @u=@T.
We get this result by differentiating the form of the utility function in
Eq. (3.12) with respect to T,
@u @u 1
D :
@T @qv pv
If we eliminate qf from the utility function by the budget equation, we

@u @u
get for the marginal utility of budgeted funds: @T D @q f
=pf . In the
consumer’s optimum, these two quantities measured in units ut=e are
equal; see Eq. (3.14) and Sect. 10.12.3.
§: By marginal utility of budgeted funds for a time unit we under-
stand the ratio between a change in a consumer’s flow of utility at a time
unit and a marginal change in his budgeted funds for the time unit. ˘
A consumer’s marginal willingness-to-pay for a good depends positively
on his marginal utility of the good, and negatively on his marginal
utility of budgeted funds. The wealthier the consumer is (the smaller his
marginal utility of budgeted funds), and the more he enjoys the extra
consumption of the good (the higher his marginal utility of the good),
the greater his marginal willingness-to-pay for the good. A consumer’s
marginal willingness-to-pay for food positively depends on his marginal
utility of food and the price of playing video games, and negatively on his
marginal utility of playing video games.
In Sect. 3.8 we show that quantity Ff ,

@u @u
Ff D hf pf D = pv pf
@qf @qv
can be interpreted as the force this consumer directs upon his food
consumption. The explanation for this interpretation is that the greater
this quantity, the higher the acceleration of food consumption of this
consumer. The force acting upon the food consumption of this consumer
consists of the pleasures and pains a marginal increase in the flow of
food consumption causes for him. The consumer compares his marginal
willingness-to-pay and the price of food and changes his flow of food
consumption on this basis. The acceleration of food consumption of this
consumer is positive, if the above-defined force is positive, and vice versa.
Thus we use the concept of force in this modeling analogously as it is used
in Newtonian mechanics.
In Sect. 10.12.2 we show that a consumer’s marginal willingness-to-
pay for a good is independent on the chosen utility function: any
continuous function expressing the same preference order defines an
equal marginal willingness-to-pay for a good in the neighborhood of
a consumer’s optimum. The ambiguity in measuring utility thus does
not affect our modeling of consumer behavior, because different values
for marginal utilities by different utility functions divided by marginal
utilities of budgeted funds by the same utility function give equal marginal
willingness-to-pay values.
3.7.1 Lagrangian Formulation of Consumer Behavior*
It is shown in Sect. 10.12.2 that a consumer’s optimization problem can

be solved by using the so-called Lagrangian function. The Lagrangian
function in this restricted optimization problem is

L(t) D u.qf (t); qv (t)/ C z T pf qf (t) pv qv (t) ;
where the measurement unit ut=e of Lagrangian coefficient z makes

the function well-defined with respect to measurement units. The time
derivative of the Lagrangian function is:
@u 0 @u 0
L0 (t) D q (t) zpf q0f (t) C q (t) zpv q0v (t)
@qf f @qv v

@u 0 @u
D zpf qf (t) C zpv q0v (t);
@qf @qv
where T; pf ; pv ; z are assumed to be constant. The changes in the flow of

food consumption, which makes the time derivative of the Lagrangian
@u
function positive, are: q0f (t) > 0 if @q f
zpf > 0, and vice versa.
Changes in the flow of consumption of video games, which increases
the Lagrangian function with time, are correspondingly: q0v (t) > 0 if
@u
@qv
zpv > 0, and vice versa.
In Sect. 10.12.3 we show that in a consumer’s optimum, the value of
Lagrangian coefficient z equals with the marginal utility of budgeted
@u @u
funds: z D @q v
=pv D @q f
=pf (* refers to an optimal value). We can
thus conclude that in the neighborhood of the consumer’s optimum, z is
positive because marginal utilities and prices are positive. The adjustment
rules for the consumer’s food consumption can then be expressed as:
@u
q0f (t) > 0 if 1z @q f
pf > 0, and vice versa, and the adjustment rules
for the consumption of video games can be transformed analogously.
The consumer’s marginal willingness-to-pay for food and for playing
@u @u
video games can thus be expressed as: 1z @q and 1z @q , respectively. These
. f .v
@u @u @u @u
are equal with quantities @q f @qv
pv and @q v @qf
pf , respectively,
and in the following we use the latter ones to avoid using the Lagrangian
formulation.
3.7.2 Marginal Willingness-to-Pay and Demand
A consumer’s marginal willingness-to-pay for food,

.
@u(q ; q ) @u(q ; q )
f v f v
hf D pv ; (3.16)
@qf @qv
where time t is left out of the consumption flows to simplify the nota-
tion, is a multivariable function. In Sect. 10.5 is given the mathematical
principles of two-variable functions. The above multivariable function
can always be simplified to a one or a two-variable function by assuming
that other variables in the analysis stay constant. This way we can operate
with only one or two variables at a time, even though we know that a
consumer’s marginal willingness-to-pay for any good depends on various
quantities.
Next we study how a consumer’s marginal willingness-to-pay for food
depends on the quantities included in the model. We study this by
differentiating hf with respect to qf ; T; pv :
!
@2 u @2 u pf @u @u @2 u @2 u pf
2
@hf @q2f @qf @qv pv @qv @qf @qv @qf @qv pv
D pv ; (3.17)
@qf @u 2
@qv
@2 u @u @u @2 u

@hf @qv @qf @qv @qf @q2v
D ; (3.18)
@T @u 2
@qv
2
@ u @u @2 u @u T pf qf @u

@hf @qv @qf
2 @qv @qf @qv pv @qf
D 2 C : (3.19)
@pv @u @u
@qv @qv
The law of non-increasing marginal utility
@2 u @2 u
0; 0;
@q2f @q2v
and the positiveness of the second order cross partial

@u @u
@ @
@qv @u2
@u2
@qf
D D D
@qf @qf @qv @qv @qf @qv
make @hf =@qf < 0 and @hf =@T > 0. In @hf =@qf , the first term in
parentheses in the numerator is negative, the latter term in parentheses is
positive, and the denominator is positive. The condition for @hf =@qf <
0; @hf =@T > 0 is thus the same as that the equilibrium point of the
consumer is a maximum. A unique negative relation thus exists between
the consumer’s marginal willingness-to-pay for food and his flow of food
consumption. In @hf =@pv , the first additive term is negative and the latter
is positive; thus the sign is ambiguous.
The following equation corresponds to the consumer’s optimum,

@u(qf ; qv ) @u(qf ; qv )
pf D hf , pf D = pv : (3.20)
@qf @qv
Equation (3.20) is called the demand relation for food of this consumer.
The demand relation is similar to that of the marginal willingness-to-
pay, but their slopes in coordinate system (qf ,e=kg) differ. We prove this
next. By totally differentiating Eq. (3.20) and using the utility function in
Eq. (3.12), we get:
0 1
@2 u @u @2 u @u
B 2 qf C
B @qv @qf @qv @qv @qf C
B1 C C dpf
@ @u 2 A
@qv
0 ! 2 1
@2 u @2 u pf @u @u @2 u pf @u
B C
B @q2f @qv @qf pv @qv @qf @qv @q2v pv @qf C
DB
B pv C
C dqf
@ @u 2 A
@qv
0 1
@2 u @u @2 u @u
B @q @q @q @q2 @q C
B v f v v f C
CB C dT
@ @u 2 A
@qv
0 1
@2 u @u @2 u @u T pf qf @u
B @q2 @q @q @q @q @qf C
B v f v f v pv C
CB C C dpv ; (3.21)
@ @u 2 @u A
@qv @qv
see Sect. 10.10.2. We can present Eq. (3.21) also as
a1 dpf D a2 dqf C a3 dT C a4 dpv ; a1 > 0; a2 < 0; a3 > 0; (3.22)
whereby ai , i D 1; : : : ; 4 are denoted the coefficients of the differentials

of which a4 is of ambiguous sign. From Eq. (3.22) we can solve:
@pf ˇˇ a2 @pf ˇˇ a3 @pf ˇˇ a4

ˇ D < 0; ˇ D > 0; ˇ D ;
@qf dTDdpv D0 a1 @T dqf Ddpv D0 a1 @pv dTDdqf D0 a1
where the sign of the last partial is ambiguous. Because pf , hf both have
unit e=kg, they can be measured on the same coordinate axis. The slope
@pf
@qf
D aa21 < 0 of the demand relation in Eq. (3.20) in coordinate system
@h
(qf ;e=kg) deviates from that of the marginal willingness-to-pay: @qff D
a2 < 0. Because a1 > 1, the latter of the curves is steeper. The reason for
this is the income effect a change in price has on the marginal willingness-
to-pay. If the price of food decreases, a consumer’s utility maximizing flow
of food consumption increases. However, a price decrease increases the real
budgeted funds of the consumer, and this moves his marginal willingness-
to-pay relation away from the origin. A price increase, analogously, moves
the marginal willingness-to-pay relation towards the origin.
Equations (3.16), (3.20) give similar results concerning how quantities
qf ; T; pv affect the optimal flow of food consumption, and both are
useful. The demand relation is estimable from the real world by statistical
methods with observed prices and flows of consumption of a consumer,
and the marginal willingness-to-pay relation can be quantified by creating
a questionnaire.
Example 1
Let the weekly utility function of a consumer be u D aqf qv , where a with
unit (ut week)=(kg h) is a positive constant and the budget equation is as
earlier. With this utility function, the marginal utilities are
@u @u
D aqv > 0; D aqf > 0;
@qf @qv
and the sufficient conditions for utility maximization hold,
@2 u @2 u @2 u @2 u
2 D D 0; D D a > 0:
@qf @qv2 @qf @qv @qv @qf
Solving qv from the budget equation and setting in the utility function
gives:
aqf
uD T pf qf :
pv
The necessary condition for optimization is then
du a T T
D0 , T 2pf qf D 0 ) q
f D , pf D ; (3.23)
dqf pv 2pf 2qf
and the sufficient condition for maximum holds: d2 u=dq2f D 2apf =pv < 0.
The utility maximizing flow of food consumption q f in Eq. (3.23) positively
depends on T and negatively on pf . We call function q f this consumer’s
demand function for food, and the last form of the equation his inverse
demand function for food. Price pv does not affect q f in this case, which
result is caused by the assumed form for the utility function. The consumer’s
marginal willingness-to-pay for food is
@u
@qf @u @u
hf D pv ; where D aqv and D aqf :
@u @qf @qv
@qv
Thus
pv qv T
hf D D pf ; (3.24)
qf qf
where the latter form is obtained by solving the budget equation as

pv qv D T pf qf and substituting this in Eq. (3.24). Another way to derive
@u @u
the marginal willingness-to-pay is to divide @qf
D aqv by @T D aqf =pv . In
the consumer’s optimum, his marginal willingness-to-pay equals with price
pf . ˘
Example 2
Let a consumer’s weekly utility function be
u D A(aqf )c (bqv )1c ; (3.25)
where the quantities are as earlier, constants A; a; b > 0 have units ut=week,
week=kg, week=h, respectively, and 0 < c < 1 is a pure number. Utility is thus
measured in units ut=week and the terms in parentheses are dimensionless
as they should be for dimensional consistency. Marginal utilities of the two
goods with units ut=kg, ut=h, respectively, are then:
@u
D Aac(aqf )c1 (bqv )1c > 0; (3.26)
@qf
@u
D Ab(1 c)(aqf )c (bqv )c > 0; (3.27)
@qv
and the second order partials are:
@2 u
D Aa2 c(c 1)(aqf )c2 (bqv )1c < 0;
@q2f
@2 u
D Ab2 c(1 c)(aqf )c (bqv )c1 < 0;
@q2v
@2 u
D Aabc(1 c)(aqf )c1 (bqv )c > 0:
@qv @qf
Marginal utilities are thus decreasing and the unique second order cross
partial is positive; sufficient conditions for utility maximization thus hold.
Substituting the earlier assumed budget equation in the utility function
gives:

bŒT pf qf 1c
u D A(aqf )c :
pv
The necessary condition for the consumer’s optimum is:

du bŒT pf qf 1c
D0 , Aca(aqf )c1
dqf pv

bŒT pf qf c bpf
A(1 c)(aqf )c D 0: (3.28)
pv pv
From Eq. (3.28) we can solve this consumer’s demand and inverse demand
functions for food as:
cT cT
q
f D , pf D : (3.29)
pf qf
An increase in T increases, and an increase in pf decreases this consumer’s

optimal flow of food consumption q
f . Price pv also does not affect qf in this
case. If we multiply the first order condition in Eq. (3.28) by factor
c
bŒT pf qf
(aqf )c pv
pv
> 0;
Ab(1 c)
we get
c T
pf pf D 0;
1c qf

where hf D 1c
c T
qf
pf is the consumer’s marginal willingness-to-pay for
food. Notice that we could also have derived the marginal willingness-to-
pay as
@u
@qf cpv qv
hf D pv D ;
@u (1 c)qf
@qv
@u @u
where @q f
, @qv
are as in Eqs. (3.26), (3.27), respectively, and substituting there
the budget equation in the form pv qv D T pf qf .
Solving the budget equation with respect to qf , substituting this in the
utility function and optimizing with respect to qv , we can solve the optimal
playing of video games as: q v D (1 c)T=pv (h=week). Another way to get
this result is to substitute q f in the budget equation and solve it for qv . The
condition for efficient consumption holds with q
f ; qv :
c 1c
@u 1 ac b(1 c) @u 1
DA D ;
@qf pf pf pv @qv pv
which is obtained substituting q

f , qv into Eqs. (3.26) and (3.27). Notice that
the measurement unit of (ac=pf ) is (week=e)c and that of (b(1 c)=pv )1c is
c
(week=e)1c . Thus their product has unit week=e, and multiplying this unit
by that of A, ut=week, we get ut=e. ˘
Assuming the following values for the constants: c D 0:7; T D 100

we can present the demand and the marginal willingness-to-pay relations
in Example 2 with two values for pf : pf 0 D 10 and pf 1 D 20. These are
in Fig. 3.8. Notice that the demand relation (the thick curve) is graphed
in the figure in the form of inverse demand. Figure 3.8 shows how the
demand and the marginal willingness-to-pay relation are related to each
other. Both are decreasing with increasing flow of food consumption, and
the demand relation stays constant with a price change while the marginal
willingness-to-pay relation moves so that the two curves cross each other
at current price.
€
kg
40
30
20
10
0 qf
0 2 4 6 8 10
Fig. 3.8 Demand and marginal willingness-to-pay relations of food

The optimal flow of food consumption of this consumer can be pre-

sented graphically as the crossing point of the horizontal line representing
the price of food and the demand relation in Eq. (3.29). In these points,
the marginal willingness-to-pay and the demand schedule cross, and they
both define equal optimal flow of food consumption qf , see Fig. 3.8.
Note 1. A consumer may not always operate on his demand relation.

However, we have showed that a utility-seeking consumer adjusts his
consumption flows of goods with time so that he ends up consuming
on his demand relation. The demand relation of a consumer is thus the
long-term relation toward which he adjusts his flow of consumption with
time. ˘
Note 2. We cannot observe the demand relations of consumers as such. At

a certain price level, we get only one observation of the demand relation
in Fig. 3.8, and via one point we can draw an infinite number of curves.
The observed point may not even be on the demand relation. However,
we can estimate the demand relation from various weekly observations
by statistical methods. This is based on the assumption that on average
the observed (flow of consumption, price) points are located on the
demand relation of a consumer. The price must also change during
the observations so that at least two points of the curve are observed.
This is the minimum requirement for the correct shape of the demand
relation, and more points give a more accurate estimate of the shape of
the function. ˘
3.8 Newtonian Theory of a Consumer*

The dynamic consumer behavior studied in Sect. 3.7 can be modeled
mathematically as follows. Let q0f (t) depend positively on quantity dq
du
f
so
@u @u pf
that q0f (t) D 0 when du
dqf
D @qf
@qv pv
D 0. This corresponds to
du
q0f (t) D f (Ff ); f 0 (Ff ) > 0; f (0) D 0; Ff D ; (3.30)
dqf
where f is a function with the above characteristics. The first order Taylor
series approximation (Sect. 10.10.1) of function f in the neighborhood of
@u @u pf
the optimum point Ff D @q f
@q v pv
D 0 is:
f (Ff ) D f (0) C f 0 (0)(Ff 0) C D f 0 (0) Ff C :
Assuming D 0, we can approximate Eq. (3.30) as follows

@u @u pf
q0f (t) 0
D f (0) Ff , q0f (t) 0
D f (0) ; (3.31)
@qf @qv pv
where f 0 (0) > 0 is a constant. The unit of q0f (t) is kg=week2 , that of
@u @u pf
@qf
@q v pv
is ut=kg, and the unit of f 0 (0) equals with that of f 0 (Ff ) D
dq0f (t)=dFf , which is (kg=week)2 =ut. Equation (3.31) is therefore dimen-
sionally homogeneous.
Now, q0f (t) is the instantaneous acceleration of food consumption of
the consumer. If the reason for the acceleration of food consumption,
@u @u pf
@qf
@q v pv
, is named as the force acting upon the food consumption
of this consumer, we can denote f 0 (0) D 1=mfd (subindex d refers to
demand) and name positive constant mfd as the inertial ‘mass’ of food
consumption of this consumer. Equation (3.31) is then of the same form
as Newton’s equation in physics, a D (1=m) F , F D ma, where a
is acceleration, F force, and m the mass of the moving particle.
Constant mfd is the ratio between force and acceleration, and it
measures the sensitivity of the flow of food consumption of this consumer
with respect to the force. The factors affecting mfd are those that retard
changes in the flow of food consumption of this consumer, like limited
knowledge of compensating goods, time to find such goods, and so on.
The inertial ‘mass’ of food consumption can be measured via the force and
@u @u pf
acceleration as mfd D @qf @qv pv =q0f (t), if these quantities are known
and deviate from zero. This corresponds to the definition of inertial mass
in physics.
q f ' (t )
∂q f ' (t )
q f ' (t ) = f ( F f ), <0
∂q f
q f ' (t ) > 0
q f ' (t ) = 0 •
q f 1(t )
*
q f (t ) q f 2 (t ) q f (t )
q f ' (t ) < 0
Fig. 3.9 The relation between flow and acceleration of food consumption
Next, we show that a utility-seeking consumer ends up with time

consuming on his demand relation of food. From Eq. (3.30) we get
@q0f (t) @Ff d2 u

D f 0 (Ff ) D f 0 (Ff ) 2 ;
@qf @qf dqf
and according to Eq. (3.15), d2 u=dq2f < 0. Equation (3.30) can thus
be graphed as a decreasing relation in coordinate system (qf ; q0f (t)), see
Fig. 3.9.
In the points of the decreasing relation in Fig. 3.9, q0f (t) > 0 (qf
is increasing with time) if qf < qf , (for example qf D qf 1 ), and
q0f (t) < 0 if qf > qf (e.g. qf D qf 2 ). Situation q0f (t) D 0 (qf constant)
corresponds to qf D qf . The arrows show the direction the consumer
changes his flow of food consumption when he is at some other point on
the decreasing relation than in the equilibrium point on the horizontal
axis, where q0f (t) D 0. In Fig. 3.9, point qf is one point on the inverse
demand function of the consumer. Outside the equilibrium point qf , the
consumer adjusts his flow of food consumption toward the equilibrium.
.
@u @u
Because every point in the inverse demand function pf D @q f @qv
pv
is the equilibrium state of the consumer at a certain price, the stability
demonstrated in Fig. 3.9 holds for every point on the inverse demand
function. The demand relation of the consumer is thus stable so that a

utility-seeking consumer enters towards it if he is not on it. Notice that
the decreasing relation @q0f (t)=@qf (t) in Fig. 3.9 must cut the horizontal
axis; otherwise a stable equilibrium point would not exist.
In all economic behavior, various inertial factors exist. For example,
being in the habit of always using a particular good makes us reluctant
to change it; practicing new things is often disagreeable, even though we
know we would gain from that; and so forth. Various costs may also be
related to a change in a consumer’s bundle of consumption flows of goods,
such as changing the nearest grocery store to a supermarket further away.
Due to these reasons, the bundle of consumption flows of goods of a
consumer may stay constant, even though he directs non-zero forces upon
his consumption of some goods. This phenomena can be added in the
model in the form of static friction.
§: Static friction is a force that resists all changes in motion. It works
in the opposite direction as the active force, and it cancels the active force
as long as this does not exceed a limit. ˘
Textbooks of economics usually talk about adjustment or transaction
costs instead of static friction. Static friction is, however, a more general
concept because it contains also other factors resisting changes than the
costs related to them. If we add static friction to the model, we can use
it to explain that often a consumer changes his bundle of consumption
flows of goods only when the reasons become compelling enough. This
way a derived model of dynamic consumer behavior corresponds to the
principle of modeling in economics given in Sect. 1.2.4, and it is
@u @u pf
mfd q0f (t) D C FSf I (3.32)
@qf @qv pv
the static friction force with unit ut=kg is denoted by FSf .

Static friction FSf contains factors resisting changes in this consumer’s
food consumption, not included in his utility function or budget equa-
tion: laziness, stubborn habits, costs from changing the flow of food
consumption, and so forth. Measuring the static friction of a consumer
requires the measuring of these factors and making a weighted average of
them with unit ut=kg. This measuring is, however, omitted in this book
and static friction is treated as an unknown quantity, the numerical value
of which can be estimated by using Eq. (3.32).
@u @u pf
According to Eq. (3.32), q0f (t) > 0 if @q f
@q v pv
C FSf > 0,
@u @u pf
and vice versa. Further, FSf < 0 if @qf
@qv pv
> 0, vice versa, and
@u @u pf
jFSf j j @q f
j.
Thus the consumer changes his flow of food
@qv pv
consumption only if the net benefit from this exceeds his static friction.
Static friction does not affect the dynamics of food consumption of this
consumer after the adjustment has begun, that is, after the active force
component has exceeded the static friction. Static friction only explains
that the flow of food consumption may not always be changed when the
active force component deviates from zero.
Example
Let the utility function of a consumer be u D aqf qv , where a with unit (ut
week)=(kg h) is a positive constant. The budget equation is assumed as in
Example 1 in Sect. 3.7.2. This functional form is assumed for utility because
it gives a simple form for the Newtonian equation of food consumption.
The reader should realize that if we applied, for example, function (3.25)
for utility, a very complicated model would result.
Solving the budget equation with respect to qv and substituting this in
the utility function, we get:
aqf
uD T pf qf :
pv
Assuming a; T; pv ; pf to be constants, the force acting upon the food con-

sumption of this consumer is
du a
D T 2pf qf ;
dqf pv
and the Newtonian equation of food consumption with this force is:
du a
mfd q0f (t) D , mfd q0f (t) D .T 2pf qf (t)/: (3.33)
dqf pv
We first study qualitatively the content of Eq. (3.33) without solving it.
Because a; pv > 0, the right hand side (rhs) of the equation is positive if
T
T 2pf qf (t) > 0 , qf (t) < ;
2pf
and vice versa. If the rhs of Eq. (3.33) is positive, then also its lhs is positive,
that is, q0f (t) > 0. Because qf D T=(2pf ) is the utility maximizing flow of
consumption derived earlier, Eq. (3.33) implies that the consumer increases
qf if qf < q
f , and decreases qf if qf > qf . Thus the system is stable and will
converge with time into the optimal state.
The solution of differential equation Eq. (3.33) (Sect. 10.15) confirms
these qualitative results:
2apf
T t
qf (t) D C C0 e pv mfd ; (3.34)
2pf
where e D 2:71 : : : is the base of the natural logarithm, C0 D qf (0)

T=2pf (kg=week) the constant of integration, and time with unit week is
denoted by t. According to Eq. (3.34), qf (t) approaches its optimal value q
f D
T=2pf with time because e(2apf =(pv mfd ))t ! 0 with t ! 1. The asymptotic
equilibrium state thus corresponds to the zero force situation assumed in
the neoclassical theory.
In this example, the force was presented in the form du=dqf , and not in
the form hf pf . The latter form of the force would be
T
pf ;
2qf
and if the Newtonian equation for food consumption is constructed with

this force, the following non-linear differential equation results
T
pf D mfd q0f (t) , T 2pf qf D 2mfd qf q0f (t);
2qf
the solution of which is much more difficult. Because quantities
a T
(T 2pf qf ) and pf ;
pv 2qf
are simultaneously positive and negative—they have identical zero points

with positive values of qf —both of them can be applied as the force
acting upon the food consumption of this consumer. The advantage of
the former is a more simple Newtonian equation, and that of the latter
is measurability; it has unit e=kg while the former has unit ut=kg. We can
remark here that force du=dqf was derived so that the budget constraint
was included in the utility function. Without this, the derivative would not
function as a force.
Substituting Eq. (3.34) in the budget equation, the consumer’s weekly

consumption of video games can be solved as:
T pf 2apf t
qv (t) D C0 e pv mfd :
2pv pv
The asymptotic equilibrium state thus corresponds to the consumer’s opti-

mum: q
f D T=2pf , qv D T=2pv . ˘
The ‘free body’ diagram of the forces acting upon food consumption
of the consumer is shown in Fig. 3.10, where the static friction force
is omitted for simplicity. The direction of motion on ‘right’ is defined
positive, and on ‘left’ negative. The positive force component acting upon
the food consumption of the consumer is hf , and the negative force
component is pf . Quantity qf (t) with unit kg=week on the horizontal axis
measures the flow of food consumption of the consumer (the velocity of
the ‘particle’), and quantity mfd with unit eweek2 =kg2 resists changes
in this movement. The shape of the ‘particle’ has no economic meaning,
and actually the particle should be drawn as a point on coordinate axis qf .
However, the ‘box-shape’ for the variable the flow of food consumption of
the consumer visualizes better the analogy we make here with Newtonian
mechanics.
The dynamic consumer behavior presented in this section has still one
advantage as compared with the static utility maximization assumed in
the neoclassical framework. Because in the neoclassical analysis time is
omitted, in that framework we cannot study how a consumer’s increasing
wealth with time affects his consumption. In the proposed framework
this can be done. Suppose a consumer gains wealth so that he can steadily
ℎ
( )
Fig. 3.10 Free body diagram of the forces acting upon consumption
increase funds for his consumption. The budgeted funds for his weekly
consumption are then a function of time, and we assume a linear form for
the function: T(t) D T0 C bt, where T0 > 0; b are constants with units
e=week and e=week2 , respectively, and time t has unit week. Assuming
the utility function as in the previous example, the following Newtonian
equation results:
a a
mfd q0f (t) D .T(t)2pf qf (t)/ , mfd q0f (t) D .T0 Cbt 2pf qf (t)/:
pv pv
The solution of this is

2apf T0 bpv mfd bt 2apf
pv m t
qf (t) D C C C1 e fd ; (3.35)
4ap2f 2pf
where C1 (kg=week) is the constant of integration. Now if b > 0, then

with t ! 1, qf (t) ! 1 because even though the exponential term
vanishes, the linear time trend (b=2pf )t keeps qf (t) increasing without
limit, with time. However if b < 0, then qf (t) ! 1 with t ! 1.
Of course qf cannot be negative, and in case b < 0 the decrease in qf
will stop at level qf D 0. This case shows how decreasing wealth decreases
consumption with time. In Fig. 3.11 is a graphed function (3.35) with the
following numerical values: C1 D 10, a D 2, T0 D 500, pf D 5, pv D 2,
mfd D 1; first b D 0:01 and then b D 0:01.
Notice that for example b D 0:01 (e=week2 ) causes bt D
0:01 (e=week2 ) 1(week) D 0:01 (e=week) increase in budgeted
funds. Thus we can model the growth in consumption due to increas-
ing wealth in the proposed framework that cannot be done in the
neoclassical one. For this consumer, limited utility maximizing flows
of consumption do not exist, and thus this kind of situation cannot
be modeled in the neoclassical framework. Time-dependent prices and
consumer preferences can be added to the model analogously. This
implies that the neoclassical framework is not general enough to cover
the observed real-world behavior.
142
qf
qf 49.995
50.020
50.018
50.016 49.990
50.014
Newtonian Microeconomics
50.012
49.985
50.010
50.008
50.006
time time
6 8 10 12 14 16 18 20 6 8 10 12 14 16 18 20
Fig. 3.11 The graph of Eq. (3.35) with b D 0:01 and b D 0:01
3.9 Aggregate Food Consumption

Earlier on we discussed that an individual consumer reacts to a price
change of a good by changing his consumption flows of goods. Because
every consumer faces the same price change, all consumers change their
consumption flows according to their preferences. As an example of this,
we study a situation where the utility function of consumer j is of the
form in Eq. (3.25), and the budget equation is as earlier, Tj D pf qfj C
pv qvj . By Tj is denoted the budgeted funds of consumer j for his weekly
consumption, and by qfj ; qvj his flows of consumption. The measurement
units of the quantities are as earlier. According to Eq. (3.29), the demand
function for food of consumer j is:
cTj
qfj D :
pf
Next, we assume m consumers and derive their optimal aggregate flow of

food consumption. For simplicity, the utility function of every consumer
is assumed to be of the form in Eq. (3.25), but the budgeted funds may
differ. The optimal aggregate flow of food consumption of m consumers
is then
X
m X
m
cTj c X
m
cTA
qfA D qf1 C qf2 C C qfm D qfj D D Tj D ;
jD1 jD1
pf pf jD1 pf
(3.36)
Pm
where TA D jD1 Tj is the aggregate of budgeted funds of the consumers.
The demand relation of a good is a unique, monotonous relationship
between the price and the aggregate flow of consumption of the good. It is
a matter of taste whether we present it in the form that the aggregate flow
of consumption depends on price, or its inverse function, the dependence
of price on the aggregate flow of consumption of the good. The first
form expresses the aggregate flow of consumption that corresponds to
the equilibrium states of all consumers at a certain price, and the latter
expresses the highest possible price at which a certain aggregate flow of
consumption gets sold. The former is traditionally called the demand

function and the latter the inverse demand function. In the inverse
demand function, the idea that price depends on the aggregate flow
of consumption can be understood so that every aggregate flow of
consumption defines a unique price at which this flow takes place. We
state this in the form of a definition.
§: The demand or the market demand relation of a good defines the
(aggregate flow of consumption, price) combinations that correspond to
the equilibrium states of all consumers consuming the good. ˘
According to Eq. (3.36), the aggregate demand and aggregate inverse
demand functions of food are, respectively:
cTA cTA
qfA D , pf D : (3.37)
pf qfA
The optimal aggregate flow of food consumption thus positively depends

on TA and negatively on pf .
Note. The assumption that consumers’ utility functions are of identical

form is unrealistic; it was made here only to simplify the aggregation. In
the real world, varying utility functions representing different preferences
of consumers make the aggregation more complicated. In the next section,
we show that in the case of varying utility functions, we can approximate
the demand relations of consumers by linear ones in the neighbor-
hood of their equilibrium states and in this way solve the aggregation
problem. ˘
To forecast the consumption of a good in the real world, we need to

model the aggregate consumption of a group of consumers. It is possible
that this can be done with a relatively simple form of a utility function,
but a lot of work is required before this question is solved in a satisfactory
way.
3.10 Aggregate Demand Relation of Food

In the Appendix of this chapter, we derive approximations for aggregate
demand and inverse demand functions of food as
b3
qfA D b0 C b1 pf C b2 pv C TA ; (3.38)
m

1 b3
pf D qfA b0 b2 pv TA ; b1 < 0; (3.39)
b1 m
where bi , i D 0; : : : 3 are constants with units kg=week, kg2 =(eweek),

(kg h)/(eweek), and unit=e, respectively,
P b0 P
0; b1 < 0, and b2 ; b3
may be positive or negative; qfA D m q
jD1 fj , T A D m
jD1 Tj .
Assuming b2 D b3 D 0, that is, Tj D Tj0 8j (8 means ‘for
all’) and pv D pv0 (see the Appendix of this chapter), we can write
Eqs. (3.38), (3.39) as:
a 1 a 1
qfA D pf , pf D abqfA ; D b0 ; D b1 > 0; (3.40)
b b b b
where the units of constants a; b > 0 are e=kg and (eweek)=kg2 ,

respectively. The reason for this notation of the constants is that these
make the inverse demand function in Eq. (3.40), that is used more often,
in the most simple form.
Fig. 3.12 presents the volume of aggregate private consumption expen-
ditures (the value of consumption at year 2010 prices) and available
income of consumers at current prices in million euro in Finland. The
figure shows that the aggregate volume of consumption has been growing
at a slower rate than available income because price inflation has increased
available income measured in current prices.
In Fig. 3.13 is the estimated linear model where aggregate private
consumption volume is explained by available income in Finland. The
model explains over 99% of observed variation in private consumption,
and the two downturns in consumption trend are explained by
simultaneous downturns in available income. The obtained equation
is of the form C(t) D 35402:5 C 0:41I(t), whereby C is denoted the
Fig. 3.12 Households’ consumption expenditures and income in Finland
Fig. 3.13 Estimated model for consumption by available income
aggregate volume of private consumption, by I available income, and t

denotes time. Thus the model for aggregate consumption in Eq. (3.38)
gets support from the data, even though we have not used any price
variables as explanators as Eq. (3.38) suggests. However, from the graph
of the residuals of the model we see that the errors are positively correlated,
and thus the model is not statistically well defined. It is presented here
only to demonstrate that aggregate consumption can be explained highly
accurately by aggregate income.
In the earlier sections, the consumption flows of consumers were
derived by assuming that every consumer behaves in the optimal way from
their points of view. In Sect. 3.8, on the other hand, the behavior of a
consumer was presented in a dynamic form consistent with the optimal
behavior. There we assumed that a consumer may not always behave in an
optimal way, but he will adjust his consumption flows of goods with time
to increase his weekly utility. This explains how consumers find a new
equilibrium when changes in prices and budgeted funds take place. Thus
we can think that consumers actively react to changes in their decision-
making situation, and they adjust their consumption flows of goods with
time to reach their optimum.
The relatively complicated modeling in this section was made because it
is needed in the next chapter in modeling firms’ behavior. For simplicity,
in modeling firms’ behavior we assume that consumers choose between
good k and other goods that are treated as a basket good. This basket good
is constructed as was described in Sect. 2.3.1 by assuming that the amounts
of these goods are measured in units kg, and their prices in units e=kg.
The average price of these goods is denoted by pG (e=kg) and the flow
of consumption of the basket good of consumer j by qGj (kg=week). In
this chapter when we modeled food consumption, and quantities qvj ; pv
correspond to qGj ; pG , respectively.
3.11 Mathematical Appendix

Here we derive the aggregate demand relation for food. The demand
functions for food of m consumers can be solved from the following
equations that hold in their optimum:
@uj
@qfj
pf D p ; j D 1; : : : ; m; (3.41)
@uj v
@qvj
where subindex j refers to consumer and f ; v to goods. According to the

utility function in Eq. (3.12), the demand functions that can be solved
from Eq. (3.41) can be presented in general form as:
qfj D Dfj .pf ; Tj ; pv /; j D 1; : : : ; m; (3.42)
where symbol D refers to ‘demand’. Next, we take the first order Taylor
series approximation (Sect. 10.10.3) of the demand function of consumer
j in the neighborhood of his equilibrium point zj0 D (pf 0 , pv0 ; Tj0 ) as
@Dfj @Dfj
qfj D Dfj (zj0 ) C zj0 .pf pf 0 / C zj0 (pv pv0 )
@pf @pv
@Dfj
C zj0 (Tj Tj0 ) C j ; (3.43)
@Tj
where j is the residual term.

The idea of a locally linear approximation of a nonlinear relation is
demonstrated in Fig. 3.14. This approximation works properly in the
neighborhood of the point the approximation is made, and a price change
means that a new approximation of the demand relation must be made
around the new point.
€
kg
∂u ∂u
pf = pv
∂qf ∂qv
pf 1
pf 2
qf 1 qf 2 qf
Fig. 3.14 Two local approximations of the demand relation of food

Assuming j D 0, j D 1; : : : ; m in Eq. (3.43), we get an approximation

for the aggregate demand relation of food by adding the linearized
demand relations of the m consumers as
Xm h i
@Dfj @Dfj @Dfj
qfA Dfj (zj0 ) (zj0 )pf 0 (zj0 )pv0 (zj0 )Tj0
jD1
@pf @pv @Tj
X
m
@Dfj X
m
@Dfj X
m
@Dfj
Cpf (zj0 ) C pv (zj0 ) C (zj0 )Tj
jD1
@pf jD1
@pv jD1
@Tj
b3
b0 C b1 pf C b2 pv C TA ; (3.44)
m
where1
Xm h i
@Dfj @Dfj @Dfj
b0 D Dfj (zj0 ) (zj0 )pf 0 (zj0 )pv0 (zj0 )Tj0 ;
jD1
@pf @pv @Tj
X
m
@Dfj X
m
@Dfj X
m
@Dfj
b1 D (zj0 ); b2 D (zj0 ); b3 D (zj0 )
jD1
@pf jD1
@pv jD1
@Tj
are constants with units kg=week, kgP 2

=(eweek), (kg Pm h)=(eweek)
and unit=e, respectively, and qfA D m q
jD1 fj , T A D jD1 Tj .
Because the flow of food consumption qfj of consumer j is non-negative
at every pf ; pv ; Tj , then b0 0 (let pf ; pv ; Tj ! 08j in Eq. (3.44). For
normal goods, b1 < 0 and b3 > 0, for Giffen goods, b1 > 0, and
for inferior goods, b3 < 0. Giffen goods are such that their demand
increases when their price increases, and inferior goods are such that their
demand decreases when consumers’ incomes increase (see Chap. 5). A
local approximation of consumers’ aggregate flow of food consumption
thus linearly depends on the two prices and the aggregate funds consumers
have budgeted for the period.
Pm Pm Pm Pm
1
Because jD1 cj Tj D c jD1 Tj C jD1 (cj c)Tj where c D (1=m) jD1 cj , the approximation
is the more accurate the less cj D @Dfj =@Tj vary, j D 1; : : : ; m.
Note. Assuming pv D pv0 , Tj D Tj0 and j D 0, j D 1; : : : ; m, we can

approximate the demand relation for food [see Eq. (3.43)] as:
1 b0
qfA D Df (pf ) b0 C b1 pf , pf D Df 1 (qfA ) D Bf (qfA ) qfA ;
b1 b1
(3.45)
where the general forms of the demand and inverse demand functions are
denoted by Df and Bf D D1 0 0
f , respectively. Because Df (pf ) D 1=Bf (qfA ),
then D0f (pf ) < 0 implies B0f (qfA ) < 0 and vice versa. ˘
References
Apostol, T. M. (1969). Calculus (Vol. II, 2nd ed.). New York: John Wiley & Sons.
Debreu, G. (1959). Theory of value. An axiomatic analysis of economic equilibrium.
A Cowles foundation monograph 17. New Haven: Yale University Press.
4
The Behavior of Firms
4.1 Principles of Firms’ Behavior

From the beginning of neoclassical economics, it has been assumed that
the goal of a firm is to maximize its profit. This assumption has been
criticized by claiming that firms rather try to maximize their owners’
revenues, or that firms try to fulfill some personal goals of their managers.
However, these goals are based on profitability because a non-profitable
firm hardly pleases its owners or managers. It has also been claimed that
firms try to maximize their turnover rather than their profit. In the 1980s,
some Finnish banks actually operated according to this principle. When
the banks defaulted, it was realized that if firms do not take care of their
profitability, they do not manage in business competition and have to
close down. This happened to Säästöpankki brand in Finland and the
Finnish government had to support the Finnish banking sector by roughly
50 billion Finnish marks. This was approximately 10% of the annual GDP
of Finland (see (Vihriälä 1999, p. 40)).
In Sect. 1.4 we argued that organizations can be analyzed as purposeful
decision-making units, the goals of which may be other than those of
their members. Every organization has a goal for which it was founded.

DOI 10.1007/978-3-319-46879-2_4
A private firm has to compete for its existence with other firms; it must
be profitable and not run out of financial resources, so that banks and
investors would be interested in investing in the firm and giving loans to
it. If the owners and managers of a firm know how its profitability can
be increased, it would be irrational not to use these means. That would
correspond to ‘throwing money away’ as was described in Sect. 1.2.2. It has
also been claimed that firms have some social goals they aim to achieve.
These kinds of activities can, however, in most cases be explained as a part
of their marketing strategy.
Competition between firms, if there are no rules, can easily cause
harm to society; for example, polluting of the environment, causing
dangerous working conditions for workers and other people, misuse of
monopoly power of a firm, and so on. Due to these reasons, the market
system requires business laws that define the limits within which firms
are allowed to operate, and how they must report to officials of their
operation. Such laws include environmental laws, laws of working time
and working conditions, tax laws, and the laws of accounting principles
of firms. The task of political leaders is to state the necessary laws and
make sure that firms follow them. The lack of clear business laws has been
one obstacle in the transformation of former socialist countries to market
economies.
To understand the role of business laws for business competition we can
use an analogy between the rules of sporting games and sporting competi-
tion, and between business laws and business competition. Competition
in business and sport is similar in many ways. We can think of the firms
in one industry as teams playing in the same series. The rules of sporting
games are clear, and if one team does not follow them, it can be ruled
out of the series or punished in some other way. For example, in 2006
the Italian soccer team, Juventus, was transferred to a lower series as a
punishment for arranging the results of games by bribery. Similarly, the
rules for competition between firms are stated in the laws of societies, of
which many are international.
If all teams obey the rules of a sporting game, the teams can be ranked
according to their playing against each other, and we can consider this
ranking fair. Similarly, if firms compete for customers according to the
existing laws, the most effective firms gain most customers and we can
consider this fair. Error-correcting coaching and learning from other

teams (firms) are good strategies for success, and the players’ (workers’)
and coaches’ (managers’) cooperative skills help in this. If one team (firm)
does not hire good players and coaches (workers and managers), and
other teams (firms) do, this team (firm) will not manage. Competition
guarantees the development of players’ (workers’) skills, playing tools
(production technologies), and strategies (ways to organize production).
Important in this is that the rules (laws) are meaningful for the game
(business) and are equal in every country.
In various kinds of sport, development is limited by the physical
ability of the human being. On the other hand, in some sporting games
development can be achieved by better equipment: new rackets, balls, and
so forth. In business, development in technology and knowledge as well as
organizational innovations allow firms to decrease their costs and improve
the quality of their products. The ‘level of the economic game’ can thus
be raised by the development of players’ skills and the means the game is
played with.
According to these arguments, the goal of a firm to become as highly
profitable as possible is a meaningful principle to understand and
model firms’ behavior. This corresponds to the principle of modeling
in economics given in Sect. 1.2.4. Firms operating with other principles
have to compete with those that take care of their profitability, and in the
long run non-profitable firms run out of financial resources and must close
down. Another form of this principle can thus be stated as follows: Firms
try to avoid bankruptcy. On this basis, public authorities should set
equal obligations for all firms, or at least for all firms in the same industry.
The internationality of business laws is important to prevent firms taking
advantage of less stringent laws in some countries. For example, child
labor is forbidden in most countries but lack of control in this area allows
misbehavior in some countries. One advantage of the expansion of Euro-
pean Union is that the same business laws cover most European countries.
4.1.1 Planning Firms’ Behavior
A firm’s planning can be classified as follows. The longest term planning

treats matters such as whether to continue or close down the firm, start
the production of a new product or quit the production of some product,

buy a competing firm or one that extends the product selection of the
firm, sell the firm or parts of it, and start, extend or quit an international
operation. Middle term planning treats strategies for developing pro-
duction methods, quality of products, and marketing. The shortest term
planning covers decisions for the amounts of produced goods and their
prices for a relatively short time; for example, one day, week or month
during which the production factors, methods, and marketing strategies
are fixed.
§: By production method we understand the combination of produc-
tion factors a firm uses in producing its products. ˘
In calculating the costs of a firm, we do not need to know how
the production factors and raw materials are actually connected in the
process. To calculate the costs of production, we only need to know the
prices of production factors and raw materials and their amounts used in
production. This resembles the difference between the recipe of a cake and
its baking. The costs of a product (the costs of a cake) can be calculated
by knowing the amount of production factors (flour, sugar, baking time,
and so on) used in it. In making a product, however, technical knowledge
of the production process is required, similarly as baking a cake requires
baking skills.
On this basis, the degrees of freedom in planning the operation of a
firm (the number of factors that can be changed) increase with the length
of time. With a long enough time horizon, firms have no factors they
cannot change. For this reason, it is very complicated to model the long-
term behavior of a firm. For example, what kind of a model would describe
the change in Nokia Corporation from electric power, rubber, and cable
producer to a producer of mobile phones and wireless networks; or the
move of a Finnish farmer to France as a vine producer after Finland joined
in the EU. For these phenomena, a modeling framework corresponding
to the Darwinian evolution would be needed, and that is not included in
this book.
For the above-mentioned reasons, in this book we analyze only such
behavior of firms where their products stay the same; the price, the
production method, the qualities of products, and the flows of production
may change. In this book, firms’ planning is divided into two categories:
short- and long-term planning. In management, these are called tactical
and strategic planning, which terminology is borrowed from military
language. These terms are also common in sport. It depends on the type of
the product and the industry as to the length of time tactical and strategic
planning covers; a hamburger bar and a nuclear power plant naturally
operate with different timescales.
According to the above, the length of the time period for tactical and
strategic planning depends on the firm. In this book, we assume that
the tactical time horizon of a firm is one week and strategic planning
covers a few years. The time horizon for tactical planning could just as
well be one month, a quarter of a year or any other suitable time unit,
but in this book we assume it to be one week. In economics, short- and
long-term analyses have traditionally been separated as follows.
§: In economics, we call long-term a time period during which all
economic units have completely adjusted their behavior to a change in
circumstances. On the other hand, we call short-term a time period
during which all economic units have not completely adjusted their
behavior. ˘
4.2 The Forms of Business Organization

Although the legislation in various countries differs somewhat in this
matter, three major forms of business organization can be classified as:
(1) sole proprietorship, (2) partnership, and (3) corporation.
§: A sole proprietorship is a firm owned and managed by an individual
called an entrepreneur. ˘
In a sole proprietorship, the owner of the firm is completely authorized
in the profits of the firm, and is responsible for the liabilities of the firm by
his personal wealth. Sole proprietorships are usually relatively small firms
where the entrepreneur controls all activities of the firm.
§: A partnership is an unincorporated firm with more than one owner.
The owners share the responsibility for financing and managing the firm,
and are personally responsible for the liabilities of the firm. ˘
In a partnership, more than one owner allows for a larger scale of

operation than in a sole proprietorship. The common responsibility
for debts requires a good trust between the partners who share the
management of the firm.
§: A corporation is a legal entity that can enter into contracts, sue and
be sued, and own property with its owners’ limited liability. ˘
Corporations deviate from the two other types of firms so that their
existence is not dependent on the living of an individual owner. The
owners of corporations are stockholders who own a certain share of the
company. For example, if 100 shares have been issued and you own 1,
you own 1% of the firm and are entitled to 1% of its profit. Stockholders
hire managers to run the firm under their command. The financial
structure of corporations and the separation of owners from management
distinguishes corporations from the two other types of firms. Stockholders
are responsible for the liabilities of a corporation by the money they have
invested in the shares of the firm; if a corporation goes into bankruptcy,
stockholders will lose this money. As a type of firm, the corporation is the
more common the larger the firm. The reason for this is that large firms
need much financial capital, and corporations can flexibly issue new shares
to raise funds from the stock market.
A new corporation issues stock to raise funds for buying plants,
machinery, and equipment, and an established corporation can finance its
expansion by issuing additional shares of stock. Stockholders get revenues
from corporations in three forms: (1) Annual dividends, (2) Additional
shares given to shareholders in special cases, and (3) Capital gains when
shares are sold at a higher price than they were bought.
§: Corporations distribute a part of their annual profit to their stock-
holders in the form of a dividend payment per one share. ˘
4.3 Revenues, Costs and the Profit of a Firm

Before we start modeling the behavior of a firm, we need to define some
concepts. The revenues of a firm are the payments it gets from its sales, and
its costs are the rents of its plants, the expenditures for its raw materials,
and the compensations paid to its production factors and debtors.
§: The official profit of a firm is the difference between its revenues

and costs in the fiscal period of the firm, which is usually one year. ˘
§: We call the plants, machinery and equipment that create revenues
during several fiscal periods the physical capital of the firm. ˘
§: We call the money of a firm in its bank accounts, together with its
receivables, the financial capital of the firm. ˘
A corporation can gain financial capital from its sales and from equity
and debt financing.
§: We call equity financing the funding provided by a firm’s owners
through their purchases of the shares of the firm. ˘
§: We call debt financing the funding provided by the firm’s creditors.
Debts are financial obligations that firms must pay back to their creditors
under an agreed time period. ˘
§: A corporate bond is a long-term debt instrument incurred by a
corporation. It commits the corporation to pay its bondholders a fixed
amount of interest each year (the interest coupon payment), and to pay
back the loan capital at the date of maturity. ˘
The maturity of a bond is usually between 5 and 30 years. Like
stockholders, bondholders face no personal responsibility for the firm’s
liabilities, but the amount owed to bondholders is fixed at the time the
bonds are sold. Bondholders are creditors of firms, and in the case of
bankruptcy, bondholders are privileged as compared with stockholders
who get the residual if there is anything left after creditors have taken
their share.
A firm pays dividends to its stockholders and taxes to the government
on the basis of its official profit. However, in this chapter we study the
determination of the weekly profit of a firm as the difference between
its weekly revenues and costs. This profit concept is unofficial, but the
official profit of a firm accumulates from its weekly profits during the
fiscal period.
4.3.1 Problems in the Calculation of a Firm’s Profit
The payments for raw materials and machines, as well as the revenues
from sales, are not always obtained at the same fiscal period the real
transactions take place. In accounting laws in different countries, it is

a common requirement that payments should be booked in the fiscal
period that the real transaction occurs. This creates outstanding bills in
firms’ accounting. Because some production factors—for example plants
and machinerys—create revenues during various fiscal periods, their costs
are booked in firms’ accounting during their entire ‘life’ in the form of
depreciation.
Example
A firm buys two computers at a value of 1000 (e) in the same fiscal period.
At the end of the period, both computers can be sold at 600 (e). The
depreciation of the computers during this fiscal period is then 2 400
(e=t) D 800 (e=t), where t is the length of the fiscal period. The firm
can thus book these costs at this fiscal period, and the rest of the investment
is depreciated during the following two or three fiscal periods. ˘
Treating depreciation rather than the purchasing price as the true

economic cost spreads the initial cost over the whole ‘life’ of capital goods.
This is not the most important reason for depreciation, however. The firm
in the above example could have sold both computers at the end of the
fiscal period, and then the true cost of the computers would have been
800 (e=t).
§: The acquiring of a production factor, which is expected to create
revenues during various fiscal periods, is called investment. For example,
plants, machinery, and equipments are such production factors. ˘
The monetary value of the plants, machinery and equipment owned
by a firm decreases basically in two ways: due to physical wearing out
and due to technological aging. These matters are taken into account in
firms’ accounting so that firms can depreciate their plants and machines
both on their expected wearing out and technological aging. For example,
buildings are depreciated over 30–40 years, computers over 3–5 years, and
so on.
§: A firm’s inventoryis its produced goods and raw materials held in
stock waiting for their selling or use in production. ˘
If some products are not sold in the fiscal period they were produced,
their production costs cannot be subtracted from the revenues of the
firm at that period because the corresponding revenues have not been
obtained. This increase in inventory value of a firm is called inventory
investment from which the revenues are obtained in the following fiscal
periods. Inventory investments may be intentional or not, because firms
can stock their products and raw materials if they believe that their price
will rise in the future.
The Income Statement or Profit-and-Loss Account of a firm shows
the difference in revenues and costs of the firm in its fiscal period; that is,
the net amount of money that flowed into the firm in the time unit. The
Balance Sheet, on the other hand, describes the net wealth of a firm as the
difference between its assets and liabilities. The Balance Sheet measures
the accumulated profitability of a firm during its existence, because the
share of the firm’s profit, that is not distributed to the owners of the firm,
is booked in the assets of the firm. The losses of a firm at one fiscal period
can be covered by its assets. Because outstanding bills are booked in the
Balance Sheet, the Balance Sheet shows the net wealth position of a firm
after its fiscal period.
An example of a hairdresser’s Income Statement is shown in Table 4.1,
and an example of its Balance Sheet is shown in Table 4.2. Net worth—
also called own capital—is the difference between the assets and liabilities
of the firm. Net worth balances the two sides of the Balance Sheet.
Table 4.1 The income statement of a hairdresser

Revenues (e=y)
5000 (unit=y) at price 10 (e=unit) 50;000
Expenses (e=y)
Wages 20;000
Advertising 5000
Office rent 8000
Other expenses 2000
Total expenses (e=y) 35;000
Profits before tax (e=y) 15;000
Taxes paid (e=y) 3000
Profits after tax (e=y) 12;000
Table 4.2 The Balance Sheet of a hairdresser

Assets (e) Liabilities (e)
Cash 5000 Accounts payable 800
Inventories 4000 Salaries payable 200
Equipments 8000 Bank loan 10;000
Net worth 6000
Total 17;000 Total 17;000
4.3.2 The Costs of Firms
In Sect. 4.1 we argued that firms aim to operate as profitably as possible.

This means that firms try to use such production methods that create the
least possible cost at every flow of production. The length of the planning
horizon affects a firm’s choice of production method so that in long-term
planning, the production method can be changed within the limits of the
financial resources of the firm. In short-term planning, fewer degrees of
freedom exist.
§: By the costs of a firm from one product we understand all costs the
production of this good causes for the firm in a time unit. These costs
contain a normal compensation for entrepreneurs so that a zero-profit
situation is the minimum profitability that keeps a firm in business. ˘
§: The Average unit costs of a firm from the production of a good are
calculated by dividing the costs of production of the good at a time unit
by the firm’s production of the good at the time unit. ˘
§: By marginal costs we understand the ratio between a change in
the costs of a firm from a good and a marginal change in the flow of
production of the good at the time unit. Marginal costs measure unit
costs from a marginal increase in the flow of production. ˘
§: The costs of a firm can be classified as follows. A part of the costs do
not depend on the flow of production of the firm; these are called fixed
costs in a time unit. The costs that depend on the flow of production
are called variable costs. Dividing fixed costs by the flow of production
in the time unit we get average fixed unit costs. Dividing variable costs
by the flow of production in the time unit we get average variable unit
costs. ˘
In the following we study a one-good firm producing good k.

The weekly costs of the firm are denoted as Ck (qk ) (e=week) where
qk (unit=week) is the flow of production of the firm. The measurement
unit implies that we could call Ck (qk ) also as the flow of costs of the firm.
The production process is assumed continuous, although it can be seized
at any time. If the firm operates 8 hours per day with one lunch and two
coffee breaks, we can still consider the production process continuous by
assuming that resting at night and eating during workdays are parts of
the process. Although the production process is continuous, products are
finished in a discrete way.
The time path of accumulated production of a firm is analogous to that
of the speedometer of a car registering full kilometers in discrete units,
and parts of a kilometer every second when running. The speed of a car
varies when it moves, and is zero when sitting at traffic lights, a gasoline
station, in a garage, and so on. If four radio sets were completed in 7.15
hours in a factory, the average velocity of production at the time unit was
4=7:15(unit=h). We can also express this as (4 24)=7:15(unit=d), if one
workday consists of 24 hours. The measured velocity of production thus
depends on the length of one workday. This is analogous to a car having
driven 200 kilometers in 2.5 hours with the driver having had one lunch
break of 30 minutes during the trip. The average speed during the trip was
then 200=2:5 (km=h), but on the road the average speed was 100 (km=h).
The accumulated production of a firm till time moment t (the accu-
mulated kilometers the car has driven) denoted by Qk (t) and measured in
units unit (a marginal change in time ds is measured in units week) is
Z t
Qk (t) D Qk (t1 ) C qk (s)ds; Q0k (t) D qk (t); Q00k (t) D q0k (t);
t1
where Qk (t1 ) is the accumulated production of the firm from its foun-
dation until moment t1 , Q0k (t) D qk (t)(unit=week) the momentous
velocity of production, and Q00k (t) D q0k (t)(unit=week2 ) the momen-
tous acceleration of production at time moment t (see Sect. 10.14). The
reason for a separate symbol for the flow of production, qk , is that in
economics the flow of production is a basic quantity, while in physics,
C(q k)
45
40
35
30
25
20
15
qk
1 2 3 4 5
Fig. 4.1 The cost function of a firm producing good k
the accumulated length of motion is a basic quantity. This kinematics of

production is a necessary prelude for production dynamics to be analyzed
in Sects. 4.5–4.6.
The weekly costs of the firm producing good k are graphed in Fig. 4.1.
On the horizontal axis is the flow of production, and on the vertical axis
are the costs. The shape of the beginning part of the cost function can
be explained as follows. (1) Fixed costs, like rents of plants and marketing
costs per one product decrease with increasing flow of production. (2)
An increase in the flow of production allows using a more effective
combination of production factors; for example, working in two periods
in a day and workers’ specialization. (3) When larger quantities of raw
materials are acquired, material costs are decreased per one product. These
elements decrease unit costs when the flow of production increases.
§: If unit costs decrease (increase) with rising flow of production,
increasing (decreasing) returns to scale prevail in the production. ˘
The three above-described matters create increasing returns to scale in
production. Later on we will show that a firm can gain increasing returns
to scale in production in the long term by applying mass production
methods.
§: By full-capacity production we understand the flow of production

of a firm that can be achieved in a time unit when all production factors
are in full use, and no overtime work is done. ˘
When the flow of production of a firm approaches its full capacity, the
firm’s unit costs start to increase. This causes the ‘upward curving’ of the
cost function in Fig. 4.1. With fixed resources, an increase in the flow of
production of a firm requires, at some level, overtime work, new space for
operation and greater inventories, which increase unit costs. The shape of
the end part of the cost function in Fig. 4.1 results from these matters.
We demonstrate the cost concepts by using specific functions. Let
the fixed costs of a firm be C0 (e=week), and variable unit costs g(qk )
(e=unit) depend on the flow of production qk (unit=week). The weekly
costs are then
Ck (qk ) D C0 C g(qk )qk ; (4.1)
where g(qk )qk (e=week) are the variable costs. Then we get
Ck C0 C g(qk )qk C0
Average unit costs: D D C g(qk );
qk qk qk
Ck dCk
Marginal costs: lim D D Ck0 (qk ) D g0 (qk )qk C g(qk ):
qk !0 qk dqk
Now, C0 =qk are average fixed unit costs, g(qk ) average variable unit
costs, and function g(qk ) is assumed differentiable; if this does not hold,
marginal costs are measured by C qk
k
. Marginal costs and all different
unit costs are measured in units e=unit. The shape of the marginal cost
function depends on the function
Ck00 (qk ) D g00 (qk )qk C 2g0 (qk ); where g00 (qk ) > 0:
Marginal costs are constant (C00 (qk ) D 0) if qk D 2g0 (qk )=g00 (qk ), and
decreasing (Ck00 (qk ) < 0) if qk < 2g0 (qk )=g00 (qk ), and vice versa. Thus
with small (large) values of qk , marginal costs are decreasing (increasing).
Example
Suppose g(qk ) is of the form of second order polynomial
g(qk ) D c1 c2 qk C c3 q2k ;
where c1 ; c2 ; c3 are positive constants with units e=unit, (eweek)=unit2 , and

(eweek2 )=unit3 , respectively. Then
g0 (qk ) D c2 C 2c3 qk ; g00 (qk ) D 2c3 > 0;
and the condition for marginal costs to be decreasing is:

g0 (qk ) c2 C 2c3 qk c2
qk < 2 , qk < 2 ) qk < :
g00 (qk ) 2c3 3c3
Thus with values qk < c2 =(3c3 ), marginal costs are decreasing and with
larger values increasing. The assumed functional forms for the cost concepts
demonstrate how an increase in the flow of production decreases units
costs in the beginning, but after a certain flow of production, unit costs
start to increase. These cost curves demonstrate in a simple mathematical
form how an increase in the flow of production affects the costs of a firm.
The weekly, average unit, and marginal cost functions are then:
Ck (qk ) D C0 C (c1 c2 qk C c3 q2k )qk D C0 C c1 qk c2 q2k C c3 q3k ; (4.2)

Ck (qk ) C0
D C c1 c2 qk C c3 q2k ;
qk qk
Ck0 (qk ) D c1 2c2 qk C 3c3 q2k :
Assuming values C0 D 10; c1 D 15; c2 D 6; c3 D 1 for the constants, the

above cost curves are graphed in Fig. 4.2. The curve in Fig. 4.1 is obtained
by using Eq. (4.2) with these parameter values. ˘
From Fig. 4.2 we see that the only constantly decreasing (the thinnest
one) cost curve with increasing qk is C0 =qk . The marginal cost curve (the
thickest one) is first decreasing and then increasing with qk , and it cuts
the average unit cost curve as well as the average variable unit costs curve
(the latter of these is below the former) in their minimum points. We
Fig. 4.2 Marginal and unit cost curves of a firm
prove this next (the rules of derivation are in Sect. 10.7.2; zero derivative
corresponds to the horizontal tangent line of a function):

d Ck (qk ) Ck0 (qk )qk Ck (qk ) Ck (qk )
D0 , 2
D 0 , Ck0 (qk ) D
dqk qk qk qk
and g0 (qk ) D 0 ) Ck0 (qk ) D g0 (q)qk C g(qk ) D g(qk ):
When the derivatives of both average unit cost curves with respect to the
flow of production are zero, marginal costs equal with both unit costs.
Because marginal costs are calculated by taking the derivative of the
cost function, the shape of the marginal cost function was explained in
connection with the cost function. Thus if the flow of production is
increased, marginal costs decrease when the flow of production is small
with respect to full capacity level. Marginal costs may be constant or
decreasing as long as free production capacity is available. However, near
the full-capacity level, overtime work and other such factors increase
marginal costs.
4.3.3 The Revenues of Firms
We analyze the revenues of a firm producing good k as a relation between

the price of the good and the sales of the firm. Good k is assumed to
deviate from all other goods so that none of these can be considered as a
perfect substitute for good k (see Chap. 5). Good k could be, for example,
a special type of food produced only by this firm. The planning horizon
of the firm is assumed to be one week, and the firm aims to sell its whole
weekly production at the highest possible unit price. This takes place if
the firm prices its product according to its sales function.
We assume m consumers interested in consuming good k but, of course,
a certain consumer may not consume the good every week. Consumers
have utility functions uj D uj (qkj ; qGj ) and budget equations Tj D
pk qkj CpG qGj , where j refers to consumer and by qGj (kg=week) is denoted
the basket of other goods consumer j consumes during the week. The
aggregate demand relation of the m consumers for good k corresponds
to their equilibrium states at prices pk ; pG . According to Sect. 3.10 and
the Appendix of Chap. 3, the aggregate demand relation of good k can
be approximated in the neighborhood of point (pk0 ; pG0 ; Tj0 ) with Tj D
Tj0 ; pG D pG0 8j as
a 1
qkdA D pk , pk D a bqkdA a; b > 0; (4.3)
b b
where qkdA (unit=week) is the aggregate weekly flow of consumption of

good k, subindex d refers to demand, and pk (e=unit) is the price of
good k. Positive constants a; b have units e=unit and (eweek)=unit2 ,
respectively.
According to Sect. 3.10, the demand relation of good k is a function
of several variables the mathematical basis of which is given in Sect. 10.5.
In this chapter, however, we denote the demand and the inverse demand
function of good k as in Eq. (4.3), and here we do not analyze dependen-
cies other than that between the price of good k and its aggregate flow of
consumption.
We assume here that the firm cannot charge its customers different
prices. Consumers decide on their weekly consumption flow of good k
at the existing price, and the firm must sell its products to all customers
at the price it has chosen. The firm has an estimate of its sales at various
prices, and it uses this to plan its weekly flow of production. This estimate
may be based on market research or on the experience of the firm of its
sales at different prices.
§: The sales function of a firm expresses the sales of the product of the
firm during a time unit at different prices. These sales correspond to the
equilibrium states of all consumers consuming the good. ˘
Because the demand relation of good k is derived on the basis of all
consumers interested in consuming this good, and good k is produced
only by one firm, the demand relation of good k equals the sales
function of the firm producing good k. On this basis, the maximal
weekly revenues of the firm from its sales of good k can be approximated
as:

a 1
Rk D pk qkdA D (a bqkdA )qkdA D pk pk ; (4.4)
b b
where R with unit e/week refers to ‘Revenues’. Because the demand of

a good is a unique relation between the price and its aggregate flow of
consumption, either of quantities pk or qkdA in Rk can be substituted by
the demand relation. This is done in Eq. (4.4). We can thus express the
weekly revenues of a firm as a function of only one quantity—the price
or the aggregate flow of consumption.
The approximate demand relation of good k in Eq. (4.3) is presented in
Fig. 4.3. On the horizontal axis is the aggregate flow of consumption and
on the vertical axis is the price. The weekly revenues of the firm, Rk D
pk qkdA , are shown in the figure as the area of the rectangle. The lengths of
the sides of the rectangle are qkdA and pk , and thus the measurement unit
of the area is (unit=week)(e=unit) = e=week. At price pk D 0 (e=unit)
the area is 0 (e=week), a price rise increases the area until a certain point,
after which the area starts to decrease. Thus the maximal weekly revenues
Fig. 4.3 Weekly sales function and the revenues of a firm
are obtained at a certain (qkdA ; pk ) combination. The reader can confirm

this by figuring out the corresponding rectangles.
§: The average revenues of a firm from a product, that correspond to
the equilibrium states of all consumers, is obtained as: (revenues in a time
unit)/(the flow of production of the product at the time unit). ˘
The average revenues of the firm from good k are then:
Rk pk qkdA
D D pk D Bk (qkdA ) a bqkdA :
qkdA qkdA
Thus the demand relation expresses average revenues from one sold good.
The firm plans its flow of production and the price of good k so that
its weekly profit gets maximized. Now, demand relation pk D a bqkdA
expresses the highest price at which the flow of consumption qkdA takes
place. If the firm sets a lower price than that expressed by the demand
relation at a certain flow of production, the whole weekly production gets
sold, but a higher price could have been obtained for each unit. Thus, if a
firm aims to sell its whole production at the highest possible price, it prices
its product according to its sales (demand) function. A firm behaving in
this way can either decide the price of its product or its flow of production,
but not both. If the firm sets a certain price, consumers decide how much
they will consume at that price in a week. If the firm want’s to sell a certain
flow of production, consumers decide at which price they will buy the
whole weekly production of the firm.
In the following we assume that the firm knows its sales function
and decides its flow of production according to it. This assumption is
made to simplify the modeling. In the real world, a difference between
estimated and the true sales function makes this situation uncertain. The
assumption, that a firm prices its product according to its sales function,
allows us to set the flows of production and consumption equal qk D qkdA ,
because behaving in this way the whole production of the firm gets sold.
The revenues of a firm from producing a good can then be expressed as
a function of the price of the good, or as a function of the firm’s flow
of production. In Chap. 5 we will analyze the behavior of a perfectly
competed industry where firms cannot affect the price of their product.
Because firms can always decide their flow of production, but not always
the price of their product, in the following we analyze the revenues of a
firm as a function of its flow of production. A deviation from this is made
in Sect. 4.7 where we study firms’ pricing.
§: The ratio between a change in the revenues and a marginal change in
the flow of production of a one-good firm is called the marginal revenues
of the firm. Marginal revenues measure average revenues from a marginal
increase in the firm’s flow of production. ˘
The marginal revenues of a firm producing good k are:
Rk Rk (qk ) Rk (qk0 ) Bk (qk )qk Bk (qk0 )qk0

lim D lim D lim
qk !0 qk q k !q k0 qk qk0 q k !q k0 qk qk0
Bk (qk )qk Bk (qk0 )qk C Bk (qk0 )qk Bk (qk0 )qk0
D lim
qk !qk0 qk qk0

.Bk (qk ) Bk (qk0 )/qk Bk (qk0 ).qk qk0 /
D lim C
qk !qk0 qk qk0 qk qk0

.Bk (qk ) Bk (qk0 )/qk
D lim C Bk (qk0 )
qk !qk0 qk qk0
D B0k (qk0 )qk0 C Bk (qk0 ): (4.5)
Equation (4.5) could also have been obtained by differentiating function

Rk D Bk (qk )qk with respect to qk at qk0 , see Sect. 10.7. From Eq. (4.5) we
see that marginal revenues change with the flow of production. Because
qk has unit unit=week and Rk has unit e=week, marginal revenues have
unit e=unit.
Example
Let the sales function of a firm producing good k be as in Eq. (4.3). Then
constant a (e=unit) shows the price at which the first produced unit can be
sold (qkdA D 0). Constant b with unit (eweek)=unit2 measures the strength of
the relationship between pk and qkdA . Suppose the firm is the only producer
of good k and the firm prices its product according to its sales function. The
weekly revenues of the firm, which correspond to the equilibrium states of
all consumers, are then Rk D pk qk D aqk bq2k , where the consumption and
production flows are set equal (qk D qkdA ). The revenues of the firm are thus
a parabolic function of qk , and marginal revenues dR k
dqk
D a 2bqk decrease
linearly with increasing qk . ˘
Because the marginal revenues of a firm are measured in equal units

as the price of the good, and marginal revenues and sales function both
depend on the flow of production, we can graph them in the same
coordinate system. Figure 4.4 is drawn by using the functions in the
previous example. The marginal revenue function cuts the vertical axis
at the same point as the sales function, and it cuts the horizontal axis
at a half distance from origin than the sales function. These results are
obtained setting first qk D 0 and then pk D dRk =dqk D 0 in the two
relations, and solving them with respect to the other quantity.
The weekly revenues of a firm have previously been demonstrated as the
area of a rectangle in coordinate system (qk ; pk ). Another way is to present
the relationship between weekly revenues and the flow of production.
With the assumed functions in the previous example, the weekly revenues
can be described as Rk D aqk bq2k in coordinate system (qk ; Rk ). This
is shown in Fig. 4.5, where the null points of the parabola are:
a
Rk D 0 , qk D 0 or qk D :
b
= −
= −2
⁄(2 ) ⁄
Fig. 4.4 Sales and marginal revenue function of a firm
2
= −
⁄(2 ) ⁄
Fig. 4.5 Weekly revenues of a firm
The flow of production that maximizes the weekly revenues of the firm is
found at the highest point of the parabola. We can solve it as
dRk a
D0 , a 2bqk D 0 , qk D :
dqk 2b
From Figs. 4.4 and 4.5 we see that marginal revenues are zero at the flow
of production a=2b that maximizes the weekly revenues.
4.4 The Production Decision of a Firm

4.4.1 Production Decision by Revenues and Costs
The profit of a firm can be demonstrated graphically by presenting the

weekly revenue and cost functions in the same figure. In Fig. 4.6 on the
horizontal axis are the flow of production of the firm and the flow of
consumption of the good; these both are measured in units unit=week.
On the vertical axis are the weekly revenues and the weekly costs of the
firm measured in units e=week. The flow of production that maximizes
the weekly profit of the firm is found at the point where the difference
between these two curves is at its maximum. In this point, the slopes of the
tangents of the two curves are equal, which condition is further analyzed
in the next section.
Fig. 4.6 Weekly revenues and costs of a firm

Example
We analyze the finding of the optimal flow of production of a firm by
a numerical example. The planning time horizon is assumed to be one
week, and the weekly sales and costs estimated by the firm are shown in
Table 4.3. The measurement units are: flow of production: unit=week, price:
1000 e=unit, revenues, costs, and profit: 1000 e=week.
In Table 4.3 in column (1) are the firm’s estimate of its weekly sales at the
prices in column (2); together they construct the sales function of the firm.
Column (3) is formulated by multiplying the numbers in columns (1) and (2)
so that the numbers in the same row are multiplied, and the product is set
in column (3). In column (4) are the firm’s estimate of its weekly costs at the
flows of production in column (1). Columns (1) and (4) formulate the cost
function of the firm. The weekly profit of the firm in column (5) is obtained
by subtracting column (4) from column (3) so that the numbers in every row
are subtracted and the result is set in column (5). From Table 4.3 we see that
the profit is maximized at the flow of production 6 (unit=week), and weekly
revenues are maximized at the flow of production 11 (unit=week). ˘
Table 4.3 The sales, revenues, and costs of a firm

(1) Sales (2) Price (3) Revenues (4) Costs (5) Profit
0 – 0 10 10
1 21 21 25 4
2 20 40 36 4
3 19 57 44 13
4 18 72 51 21
5 17 85 59 26
6 16 96 69 27
7 15 105 81 24
8 14 112 95 17
9 13 117 111 6
10 12 120 129 9
11 11 121 135 14
12 10 120 140 20
4.4.2 Production Decision by Marginal Quantities
The optimal flow of production of a firm can be solved either on the

basis of weekly revenues and costs, or on the basis of marginal revenues
and costs. We show this arithmetically. We write the weekly profit …k
(e=week) of the firm producing good k as:
…k D Rk (qk ) Ck (qk ) D Bk (qk )qk Ck (qk ); B0k (qk ) < 0; Ck0 (qk ) > 0;
(4.6)
where the weekly revenues and costs are denoted by Rk (qk ) and Ck (qk ),
respectively, and Bk (qk ) is the sales function of the firm, see the Appendix
in Chap. 3. In Eq. (4.6) we assumed that the firm can estimate its
weekly sales at various prices and costs, and that the firm prices its
product according to its sales function. The necessary condition for profit
maximization is
d…k
D 0 , R0k (qk ) Ck0 (qk ) D 0 , B0k (qk )qk C Bk (qk ) D Ck0 (qk );
dqk
(4.7)
where R0k (qk ) D B0k (qk )qk C Bk (qk ) are marginal revenues and Ck0 (qk )
marginal costs of the firm. The sufficient condition for maximum is
d 2 …k
D B00k (qk )qk C 2B0k (qk ) Ck00 (qk ) < 0:
dq2k
Because B0k (qk ) < 0, the above condition holds if Ck00 (qk ) 0 and
B00k (qk ) 0. Thus marginal costs should be non-decreasing and the sales
function non-increasing with qk ; we return to these matters with specific
cost and sales functions. At the profit maximizing flow of production,
marginal revenues and costs are equal. This corresponds to the situation
in the previous section where the slopes of the tangents of the weekly
revenue and cost functions are equal. If the forms of functions Bk (qk ) and
Ck (qk ) are known, we can solve the profit maximizing flow of production
qk from Eq. (4.7).
€
unit
Ck ' (qk )
p k*
C k (qk* ) Ck (qk )
qk* qk
q k* a /(2b) a/b qk
Fig. 4.7 The optimal flow of production of a firm
The situation can be demonstrated graphically as follows. Suppose the

weekly sales function is of the form as in the example in Sect. 4.3.3,
and weekly costs are as in the example in Sect. 4.3.2. Because marginal
revenues and costs, unit costs, and the sales function are all measured
in units e=unit—and they all depend on the flow of production qk
that equals with the aggregate flow of consumption—their graphs can
be drawn in the same coordinate system. This is done in Fig. 4.7.
In Fig. 4.7, marginal revenues (not labeled in the figure) and costs
are equal at the flow of production qk , which is the profit maximizing
flow of production. The price at the aggregate flow of consumption qk is
obtained from the sales function, which shows the highest price at which
the amount qk gets sold in a week. The weekly profit of the firm is the
shaded area in the figure. It can be calculated by subtracting average unit
costs Ck (qk )=qk from price pk at qk , and multiplying this average profit
from one product by qk .
Example 1
Suppose the weekly sales function of a firm is of the form pk D a bqk as
in Sect. 4.3.3, where the units of qk and pk are as earlier. The weekly cost
function of the firm is Ck (qk ) D C0 C c1 qk , where C0 and c1 are positive
constants; C0 (e=week) is fixed costs and c1 (e=unit) marginal costs (= variable

unit costs). The weekly profit of the firm is then:
…k D pk qk Ck (qk ) D aqk bq2k C0 c1 qk : (4.8)
The profit maximizing flow of production is obtained as
d…k a c1
D0 , a 2bqk D c1 ) q
k D ; (4.9)
dqk 2b
and the sufficient condition for maximum holds too because
d 2 …k
D 2b < 0:
dq2k
In Eq. (4.9), marginal revenues are a 2bqk and marginal costs are c1 ;
the condition for optimality thus holds. For the profit maximizing flow
of production to be positive, a > c1 must hold. Thus the price at which
the first produced good can be sold must exceed variable unit costs.
Equation (4.9) implies that the optimal flow of production increases when
the sales function moves away from origin (a increases), and decreases
when c1 increases or the sales function gets ‘steeper’ (b increases). The price
corresponding to qk is:

a c1 2a a C c1 a C c1
p
k D a bqk D a b D D > 0: (4.10)
2b 2 2
This derivation must still be completed by checking that profit is positive at

q
k and pk . The weekly profit with qk is:

a C c1 a c1 a c1
…k jqk Dqk D p
k qk Ck (qk ) D C 0 c1
2 2b 2b

a c1 a C c1 2c1 (a c1 )2
D C0 D C0 :
2b 2 4b
The condition for non-negativity of the profit is
(a c1 )2 (a c1 )2
C0 0 , C0 ; (4.11)
4b 4b
which shows the upper limit for fixed costs C0 for the profit to be positive.
Next we show how we can use measurement units in checking our
calculations. We defined the units of the constants as: C0 (e=week), a (e=unit),
c1 (e=unit) and b ((eweek)=unit2 ). With these units, the last inequality is

dimensionally homogeneous. This supports correct calculations because the
profit function in Eq. (4.8) is dimensionally homogeneous, and all proper
algebraic transformations for such an equation remain its homogeneity.
Errors in calculation are then seen as dimensional errors unless two errors
occasionally cancel each other out. Changes in measurement units, though,
occur; profit function in Eq. (4.8) is measured in units e=week, the middle
form of Eq. (4.9) in units e=unit, and so on.
The quantity that is eliminated from the profit function by the sales
function does not affect the result. If we solve the sales function as qk D
(a pk )=b and substitute this in the profit function, we get:

pk a p2k a pk
…k D C 0 c1 :
b b
The profit-maximizing price can then be obtained as
d…k a 2pk c1 a C c1
D0 , C D0 ) p
k D ;
dpk b b 2
which result we got earlier. The optimal flow of production with price p
k is
obtained by using the sales function:
a C c1 a c1
p
k D D a bqk ) q
k D :
2 2b
The sufficient condition for maximum holds also in this case,
d 2 …k 2
D < 0:
dp2k b
This example shows that if a firm operates with the (qk ; pk ) combinations
of its sales function, the firm has actually only one quantity by which it
can affect its profit. If the firm changes the price of its product, consumers
react to this by changing their flows of consumption according to the
sales function. On the other hand, if the firm reduces its weekly flow of
production from that which was previously completely sold, the firm can
sell its whole production at a higher price as is shown by the sales function.
It is a matter of taste which quantity—the price or the flow of production—
is considered as the policy variable of the firm, because a firm that aims to
sell its whole production has to accept that consumers decide at which price
they will buy it.
The average unit costs of the firm at q

k are:
1
Ck (qk)
C0 C c1 ac 2bC0 C c1 a c21
D ac1
2b
D : (4.12)
qk 2b
a c1
The non-negativity condition for profit in Eq. (4.11) can now be derived by
using Eqs. (4.10) and (4.12) and demanding that p k must exceed average
unit costs at q
k :
a C c1 2bC0 C ac1 c21 Ck (qk)

p
k D D :
2 a c1 qk
The checking of this is left to the reader as well as that the mathematical
expressions in this example are dimensionally well-defined. ˘
Example 2
We continue the numerical example in Sect. 4.4.1. Suppose the flow of
production is 3 (unit=week) and the firm is considering whether to increase
its flow of production by one unit in a week. We assume that the firm
has estimated its weekly sales and cost functions correctly. Increasing the
weekly flow of production by one unit causes extra costs 51 44 D 7(1000
e=week), and revenues increase by 72 57 D 15 (1000 e=week). Thus profit
increases by 8 (1000 e=week) and the change should be done.
By comparing the marginal revenues (MR) and marginal costs (MC)
we see that condition MR D MC holds at the profit maximizing flow of
production 6 (unit=week). If MR > MC, the firm should increase its flow of
production, and if MR < MC, decrease. The flow of production should thus
be increased as long as 6 (unit=week) is obtained (see Table 4.4).
Note 1. The increase in the flow of production in Table 4.4 takes place
in units 1 (unit=week). In marginal revenues and costs, the denominator
is thus constant 1 (unit=week). This does not hold in general, however,
because in the real world the effects on the revenues and costs of a
firm can be measured with varying amounts of changes in the flow of
production. ˘
Note 2. If marginal analysis is used in defining the optimal flow of

production of a firm, it is necessary to check that profit is non-negative
Table 4.4 Marginal analysis of a firm’s production decision

(1) Flow of prod. (2) MR (3) MC (4) MR–MC (5) Profit
0 – – – 10
1 21 15 6 4
2 19 11 8 4
3 17 8 9 13
4 15 7 8 21
5 13 8 5 26
6 11 10 1 27
7 9 12 3 24
8 7 14 7 17
9 5 16 11 6
10 3 18 15 9
at the flow of production where marginal revenues and costs are equal.
The reason for this is that in marginal analysis fixed costs are eliminated.
For example, in Table 4.4 fixed costs 10 (1000 e=week) at the flow of
production 0 (unit=week) are not seen in the marginal analysis. Also with
differentiable functions, a constant representing fixed costs vanishes from
the necessary condition for optimum because the derivative of a constant
is zero. ˘
4.5 Dynamic Analysis of a Firm’s Behavior

Let the flow of production qk of a firm producing good k be measured in
units unit=week. The weekly profit of the firm …k (e=week) is then
…k (t) D Bk .qk (t)/qk (t) Ck .qk (t)/;
where we have assumed that the firm knows its sales and cost functions,
and the flow of production is set to depend on time t. Now qk (t) is
the only quantity by which the firm can affect its profit, because we
have substituted price by the sales function pk (t) D Bk .qk (t)/. The time
derivative of the profit function is (Sect. 10.7.3):
d…k 0
…0k (t) D qk (t) D B0k .qk (t)/qk (t) C Bk .qk (t)/ Ck0 .qk (t)/ q0k (t):
dqk
Analogously as with the behavior of a consumer, the firm is assumed

to change its policy variable qk with time to increase its target function
(weekly profit). The adjustment rules that make the acceleration of the
profit …0k (t) (e=week2 ) positive are: q0k (t) > 0 if d…
dqk
k
> 0, and q0k (t) < 0
if d…
dqk
k
< 0. The rule q0k (t) D 0 if d…
dqk
k
D 0 can be understood so that it is
not reasonable to change qk if this does not affect the profit. This behavior
can be expressed mathematically as:
d…k
q0k (t) D f (Fk ); f 0 (Fk ) > 0; f (0) D 0; Fk D ; (4.13)
dqk
where f is a function with the above characteristics.

Now, q0k (t)(unit=week2 ) is the instantaneous acceleration of produc-
tion of the firm. If the reason for this acceleration, d…
dqk
k
(e=unit), is named
as the force acting upon the production of the firm, this behavior can
be interpreted to mean that the flow of production of the firm increases
(decreases) when a positive (negative) force is acting upon it.
We can remark here that we assumed good k to deviate from its
imperfect substitutes so that we can analyze its production independently
of other goods. Due to this, the force acting upon the production of good
k consists only of the decisions of this firm. In Chap. 5 we will study a
force acting upon the production of a homogeneous good produced by
many firms, and there we return to these matters.
The interpretation of d… dqk
k
as the force acting upon the production of
good k of a firm can be explained as follows. The greater this quantity, the
more profitable it is for the firm to increase its flow of production, and
the more eager we can believe the firm is to do this. The force consists of
marginal revenues and marginal costs that the firm’s managers consider
in their decision-making, as was required in the principle of modeling
in economics (see Sect. 1.2.4). This theory is testable with the applied
measurable quantities, even though measuring the marginal quantities in
d…k
dqk
is difficult.
In the next section we will show that the dynamic behavior of a
firm described here can be modeled by using the defined force so that
the profit-maximizing situation—that is assumed in the neoclassical
framework—is a special case of it: the zero-force situation that corre-

sponds to the equilibrium state of the firm.
4.6 Newtonian Theory of a Firm*

The first order Taylor series approximation (see Sect. 10.10.1) of function
f in Eq. (4.13) in the neighborhood of the optimum point Fk D d… dqk
k
D0
is:
f (Fk ) D f (0) C f 0 (0)(Fk 0) C D f 0 (0) Fk C ;
where is the error term. Setting D 0 we can approximate Eq. (4.13) as
d…k
q0k (t) D f 0 (0) ,
dqk

q0k (t) D f 0 (0) B0k .qk (t)/qk (t) C Bk .qk (t)/ Ck0 .qk (t)/ ; (4.14)
where f 0 (0) > 0 is a constant with unit (unit=week)2 =e; this unit
comes from f 0 (Fk ) D dq0k (t)=dFk , where Fk D d… dqk
k
(e=unit) and
0 2
qk (t)(unit=week ).
Next we denote f 0 (0) D m1ks and name mks (> 0) the ‘inertial "mass"
of production of good k’ because in the above formulation, mks has the
same role as inertial mass has in Newtonian mechanics. Subscript s refers
to ‘supply’. With these definitions, Eq. (4.14) exactly corresponds to the
Newtonian equation of motion a D F=m, where a is acceleration, F
force, and m the mass of the moving particle. Factor mks (e/(unit=week)2 )
contains the factors resisting changes in the flow of production, like time
required to find new workers and teach them, the laws restricting the firing
of employees, and so on.
If the force and acceleration of production of good k are known, factor
mks can be calculated as: mks D d… dqk
k
=q0k (t). The smaller the mks , the
faster the flow of production adjusts with a fixed force, and vice versa.
According to Eq. (4.14), the flow of production increases (q0k (t) > 0)
if B0k .qk (t)/qk (t) C Bk .qk (t)/ Ck0 .qk (t)/ > 0, and vice versa. The
flow of production continuously adjusts with force B0k .qk (t)/qk (t) C
Bk .qk (t)/Ck0 .qk (t)/, and the profit maximizing flow of production—that
corresponds to the neoclassical theory—can be solved from the zero-force
situation: B0k .qk (t)/qk (t) C Bk .qk (t)/ D Ck0 .qk (t)/. The characteristics of
the cost and sales functions guarantee that the force is positive at small
and negative at high flows of production, which makes the equilibrium
state stable.
Factor mks explains that the adjustment of the flow of production
takes time. We can also add static friction to the situation to explain
that a firm may not always change its flow of production when the
defined force deviates from zero. Economics textbooks usually talk about
adjustment costs instead of static friction. Static friction is, however, a
more general concept that covers all factors resisting changes in firms’
flows of production not included in sales and cost functions. Such factors
are: the building costs of new capacity and the time it takes, the hiring
and firing costs of labor, the costs of changing production methods,
uncertainties in the sales and cost functions, and so forth. Adding static
friction into the model we get
d…k
mks q0k (t) D C FSk ; (4.15)
dqk
where the static friction force is denoted by FSk (e=unit). The direction
of static friction force is opposite to that of d…
dqk
k
, and jFSk j j d…
dqk
k
j. Static
friction keeps the resultant force zero as long as j dqk j does not exceed the
d…k
friction force. According to Eq. (4.15), q0k (t) ¤ 0 if d… dqk

k
C FSk ¤ 0. Thus
by adding static friction into the model we get the situation exact with
the principle of modeling in economics (see Sect. 1.2.4).
The ‘free body’ diagram of the forces acting upon the production of
good k of a firm is illustrated in Fig. 4.8, where the static friction force is
omitted for simplicity. The direction of motion on the ‘right’ is defined
positive, and on the ‘left’ negative. The positive force component acting
upon the flow of production is B0k (qk )qk C Bk (qk ), and the negative
force component is Ck0 (qk ). Quantity qk (t) with unit unit=week on the
horizontal axis measures the flow of production of good k of the firm (the
Ck’(qk) B’k (qk)qk + Bk (qk)

mks
qk(t)
Fig. 4.8 Free body diagram of forces acting upon production
velocity of the ‘particle’), and quantity mks with unit e=(unit=week)2

resists changes in this motion. Similarly, as in the consumer’s case, the
shape of the ‘particle’ has no economic meaning, and the particle should
be drawn as a point on coordinate axis qk . However, the ‘box shape’ for
the variable the flow of production of good k of the firm visualizes better the
analogy we make here with Newtonian mechanics.
Example
Let the weekly sales and cost functions of a firm producing good k be
pk (t) D a bqk (t); Ck .qk (t)/ D C0 C c1 qk (t); (4.16)
where pk (e=unit) is the price of good k, qk (unit=week) the flow of production,

Ck (e=week) the weekly costs of the firm and a; b; C0 ; c1 positive constants
with units: e=unit, (eweek)=unit2 , e=week and e=unit, respectively. The profit
of the firm with unit e=week is then:
…k (t) D pk (t)qk (t) Ck .qk (t)/ D aqk (t) bq2k (t) C0 c1 qk (t): (4.17)
Suppose next that no static friction exists in production. The Newtonian

equation of production is then
d…k
D mks q0k (t) , a 2bqk (t) c1 D mks q0k (t); (4.18)
dqk
where mks is the inertial ‘mass’, q0k (t) the instantaneous acceleration, and
d…k
dqk
the force acting upon production. The equation of motion in Eq. (4.18)
shows that the firm increases its flow of production (q0k (t) > 0) if qk (t) <
(a c1 )=2b D q
k , and decreases its flow of production if qk (t) > qk , where qk
is the profit maximizing flow of production. This can be interpreted and
compared with physics, in that in this situation, together with constant
positive force component a c1 > 0, there also exists a negative component

2bqk (t) that depends of the flow of production qk (t). These kinds of
situations occur in physics when an object is moving in a resisting medium:
a boat in water or an airplane in air. The force component 2bqk (t)—that
in physics corresponds to the kinetic friction of the moving object caused
by the resisting medium—originates from the negative relation between
pk and qk in the sales function. Thus Eq. (4.18) shows how an increase in qk
decreases the force d…dqk
k
.
The solution of the differential equation in Eq. (4.18) is
a c1 2b t
qk (t) D C A0 e mks ; (4.19)
2b
where A0 (unit=week) is the constant of integration and m2bks t a dimensionless

quantity because time t is measured in units week. According to Eq. (4.19),
the flow of production approaches the firm’s profit maximizing level qk (t) D
q ac1
k D 2b with time; this situation corresponds to zero force. The flow of
production increases or decreases depending on whether in the beginning
(t D 0), qk (0) D ac
2b
1
C A0 ) A0 D qk (0) ac
2b
1
, qk (0) is smaller or greater
than the optimal one, that is, whether A0 is negative or positive. ˘
Next we show one advantage of the proposed framework as compared

with the static neoclassical one. We can model the growth of a firm and
its possible bankruptcy in this framework too. Suppose the marginal costs
of a firm decrease with time due to, for example, employees’ learning in
work (c2 > 0), or increase due to, for example, an increasing wage level
(c2 < 0). The sales function is assumed as in the previous example but
the cost function is now:
Ck (qk (t); t) D C0 C (c1 c2 t)qk (t) D C0 C c1 qk c2 qk (t)t;
where constant c2 has unit (eweek)=unit. The resulting Newtonian

equation is
d…k
mks q0k (t) D , mks q0k (t) D a c1 C c2 t 2bqk (t); (4.20)
dqk
and its solution is

2ab 2bc1 mks c2 c2 t 2b t
qk (t) D 2
C C A1 e mks ; (4.21)
4b 2b
where A1 (unit=week) is the constant of integration. The exponential time

trend vanishes with time as in the previous case, but now a linear time
trend (c2 =2b)t exists in qk (t) too. If c2 > 0, that is, marginal costs
are decreasing with time, this causes a permanent growth in the flow of
production of the profit-seeking firm. However, if c2 < 0, then qk (t)
decreases without limit over time and the firm goes into bankruptcy. In
these two cases, a bounded profit maximizing flow of production does
not exist for the firm, and thus the neoclassical framework cannot explain
these real world events. However, if c2 D 0, the equation is the same as
in Eq. (4.18), and so the neoclassical equilibrium is obtained as a special
case from Eq. (4.20). ˘
Fig. 4.9 shows the graph of function (4.21) with the following param-
eter values: A1 D 10, a D 100, mks D 1, b D 0:1, c1 D 2, and first
c2 D 0:01 and then c2 D 0:01.
Note. The assumption of linearly decreasing marginal costs was made

here only to get a simple Newtonian equation of motion for production.
More complicated time trends in qk (t) are obtained by assuming that the
demand of good k is an increasing or decreasing function of time, or by
assuming that increasing returns to scale exists in production, see Estola
(2001). This example shows the limits of the neoclassical framework in
modeling the observed behavior of firms, see Estola (2014). ˘
In order to evaluate the empirical performance of the presented New-

tonian theory of production, in Estola and Dannenberg (2012) and
Estola (2015) we have compared the static neoclassical theory against the
Newtonian one by using Finnish and Swedish data from several industries.
Annual industrial flows of production at several industries were applied,
and the explanation power of neoclassical and Newtonian theories were
compared with each other and in Estola (2015) also with the simple first
order autoregressive model (AR1)
y(t) D ay(t 1) C b;
where y(t) is the flow of industrial production and a; b parameters to be

estimated from the data. The first order autoregressive model explains
variable y(t) at time unit t by its previous value y(t 1) together
Fig. 4.9 The graph of function (4.21) with c2 D 0:01 and c2 D 0:01
Fig. 4.10 The graphs of neoclassical and Newtonian models in Sweden
with constant b. Our results in Estola (2015) were that the AR1 model
outperformed the neoclassical one in explaining the flows of production
in every tested industry in both countries, and the Newtonian theory
outperformed the AR1 in 10 out of 13 cases in Finnish industries,
and in 14 out of 18 cases in Swedish industries. Finally, in Estola and
Dannenberg (2012) and in Estola (2015) the Newtonian theory was shown
to outperform the neoclassical one in every tested industry. Figure 4.10
shows examples of these results from industries C20C21: Chemicals and
chemical products, and C31 C32: Furniture and other manufacturing
in Sweden. The graphs of the estimated models show that the Newtonian
model follows the real data more accurately than the neoclassical one in
all cases. Thus we have empirical evidence that the Newtonian theory
performs better than the neoclassical one in empirical tests.
4.7 Firms’ Pricing Behavior

Good k, that somewhat deviates from all other goods, is assumed to be
produced by only one firm. This firm is assumed to know its weekly sales
and cost functions, and the weekly profit of the firm …k (e=week) is
…k (t) D pk (t)qk (t) Ck .qk (t)/; qk (t) D Dk .pk (t)/; Ck0 (qk ) > 0; D0k (pk ) < 0;
where Rk D pk qk (e=week) and Ck .qk / (e=week) are the weekly

revenues and costs of the firm, pk (e=unit) the price of good k,
qk (unit=week) the flow of production, and qk D Dk (pk ) the sales function
of the firm. Now, dqk =dpk D D0k (pk ) measures the effect on the weekly
sales of the firm by a marginal change in price. If the flow of production is
substituted by the sales function, the profit can be expressed as a function
of price only:

…k (t) D pk (t)Dk .pk (t)/ Ck Dk .pk (t)/ :
The time derivative of the weekly profit is then

d…k 0 dRk dCk
…0k (t) D p (t) D p0k (t)
dpk k dpk dpk

D Dk (pk ) C pk D0k (pk ) Ck0 (qk )D0k (pk ) p0k (t);
where dpk =dt D p0k (t) ((e=unit)=week) is the velocity of the price. A
profit-seeking firm changes the price of its product to increase its weekly
profit. The adjustment rules, that increase the firm’s profit, are:
p0k (t) > 0 if Dk (pk ) C D0k (pk )Œpk Ck0 (qk ) > 0;
p0k (t) < 0 if Dk (pk ) C D0k (pk )Œpk Ck0 (qk ) < 0;
p0k (t) D 0 if Dk (pk ) C D0k (pk )Œpk Ck0 (qk ) D 0:
The quantity
d…k dRk dCk

D D Dk (pk ) C D0k (pk )Œpk Ck0 (qk )
dpk dpk dpk
with unit unit=week can then be named as the force acting upon the
price of good k. In this decision-making, the firm compares changes in
its weekly revenues dRk =dpk and costs dCk =dpk due to a price change.
Term qk D Dk (pk ) measures the increase in the firm’s revenues from sold
goods due to a price raise, and term pk D0k (pk ) < 0 measures the decrease
in revenues due to reduced sales after a price increase. These factors define
the force component dRk =dpk that measures demand-based price effects,
and factor Ck0 (qk )D0k (pk ) > 0 measures cost-based price effects.
The demand-based force component acting upon the price may be
positive or negative. In the latter case, the revenues of the firm can be
increased by decreasing the price (remember that the revenues of a firm
are measured by the area of a rectangle as shown in Fig. 4.3). Also, the
costs-based force component acting upon the price may be positive or
negative. According to it, all factors that increase marginal costs Ck0 (qk )
positively affect the price, and vice versa.
Because D0k (pk ) < 0, the force Dk (pk ) C (pk Ck0 (qk ))D0k (pk ) is positive
if pk Ck0 (qk ). Thus if price is below marginal costs, the firm always
benefits by a price increase. In situations pk > Ck0 (qk ), however, we cannot
say for sure in which direction the price should be changed to increase
profit. The zero force situation Dk (pk ) C pk D0k (pk ) D Ck0 (qk )D0k (pk )
defines the price that maximizes the weekly profit of the firm.
Example
Let the weekly sales and cost functions of a firm be qk (t) D b1 b2 pk (t) and
Ck D C0 Cc1 qk (t), respectively, where qk (unit=week) is the flow of consumption
(= flow of production), pk (e=unit) the price of the product of the firm, and
positive constants b1 ; b2 ; C0 ; c1 have units unit=week, unit2 =(eweek), e=week
and e=unit, respectively. The weekly profit of the firm is then:
…k (t) D pk (t)qk (t) Ck (t) D pk (t)Œb1 b2 pk (t) C0 c1 Œb1 b2 pk (t):

The force acting upon the price of the product of the firm is
d…k
D b1 C b2 c1 2b2 pk (t):
dpk
The constant force component b1 Cb2 c1 > 0 consists of demand and cost
factors; the greater b1 and c1 are, the greater is the force. The negative force
component 2b2 pk (t) depends on the price; the greater the price the greater
the absolute value of this factor. The equilibrium price, that maximizes the
profit of the firm, is pk (t) D (b1 C b2 c1 )=2b2 , and the sufficient condition for
maximum holds too:
d 2 …k
D 2b2 < 0: ˘
dp2k
The demand- and cost-based changes in the equilibrium price are

demonstrated in Fig. 4.11. The sales and cost functions in the diagrams
are as in the previous example. The first figure shows the effect of a move
of the sales function away from the origin (b1 increases). An increase in
marginal costs c1 , on the other hand, moves the relation dCk =dqk D
c1 ‘upward’ in the second figure. The equilibrium state of the firm is
denoted in the beginning as (qk0 ; pk0 ) and after changes as (qk1 ; pk1 ).
The difference in the two situations is that a move of the sales function
away from origin increases both the equilibrium price and the flow of
production. An increase in marginal costs, on the other hand, increases
the equilibrium price but decreases the equilibrium flow of production.
4.7.1 Dynamic Analysis of a Firm’s Pricing*
We continue the analysis from the previous section. The modified Newto-
nian equation of motion for the price is formulated analogously as earlier:
mkp p0k (t) D Dk .pk (t)/ C D0k .pk (t)/Œpk (t) Ck0 .Dk (pk (t))/; (4.22)
where constant mkp > 0 with unit unit2 =e is the inertial ‘mass’ of
price pk . Notice that p0k (t) is the flow, and not the acceleration, of
price pk . Thus Eq. (4.22) does not exactly correspond to the Newtonian
Fig. 4.11 Demand- and cost-based increases in equilibrium price

Fig. 4.12 Free body diagram of the forces acting upon price pk
equation F D ma. However, Eq. (4.22) follows the principle of modeling

in economics, and thus we can call Dk .pk /Cpk D0k .pk /Ck0 .Dk (pk )/D0k (pk )
the ‘force’ acting upon the price.
The ‘free body’ diagram of the forces acting upon price pk in Fig. 4.12
is made according to the example in the previous section. The direction
of motion on the ‘right’ is defined positive, and on the ‘left’ negative. The
positive force component acting upon the price is b1 C b2 c1 , and the
negative force component is 2b2 pk (t). Quantity pk (t) on the horizontal
axis measures the price level, and quantity mkp with unit unit2 /e resists
changes in the motion of the ‘particle’. Similarly to the consumer’s case,
the shape of the ‘particle’ has no economic meaning and the particle
should be drawn as a point on coordinate axis pk . However, the ‘box shape’
for the variable the price of good k visualizes better the analogy we make
here with Newtonian mechanics.
Example
We continue the analysis in the example in the previous section, and we
assume, for simplicity, that the price has no static friction. The modified
Newtonian equation of motion for the price is then:
d…k
D mkp p0k (t) , b1 C c1 b2 2b2 pk (t) D mkp p0k (t); (4.23)
dpk
where mkp with unit unit2 =e is the ‘inertial mass’ resisting changes in the
price. The solution of this differential equation is
b1 C c1 b2 2b
2t
pk (t) D C Ae mkp ; (4.24)
2b2
where A (e=unit) is the constant of integration: A D pk (0) (b1 C b2 c1 )=2b2 .

According to Eq. (4.24), price adjusts with time toward pk (1) D (b1 C
c1 b2 )=2b2 because 2b2 =mkp < 0. The greater the constants b1 and c1 , the
higher the equilibrium price, and constant b2 affects the equilibrium price
as well. Demand- and cost-based factors thus affect the equilibrium price
that maximizes the weekly profit of the firm. ˘
In Fig. 4.13 is graphed annual industrial flows of production and

price levels of Finnish manufacturing industries: DA: Food products,
beverages and tobacco; DB+DC: Textiles, textile products, leather and
leather products; DF: Refined petroleum products, coke and nuclear fuel;
DG: Chemicals and chemical products. Industrial prices are computed as
p(t)q(t)=(p(0)q(t)) D p(t)=p(0), that is, current value time series p(t)q(t)
Fig. 4.13 Industrial flows of production and prices in Finnish industries

are divided by a fixed price series p(0)q(t). Prices are still multiplied by
1000 to get their numerical values close to those of production flows
so that they can be graphed in the same figure. Figure 4.13 shows that
industrialized prices may have positive and negative time trends, or prices
may fluctuate around a relative constant value like in industry DF. In
some industries production flows follow the price development and in
others not.
References
Estola, M. (2001). A dynamic theory of a firm: An application of economic
forces. Advances in Complex Systems, 4(1), 163–176.
Estola, M. (2014). The neoclassical theory of a firm; corrections for its errors.
Hyperion International Journal of Econophysics and New Economy, 7 (1), 7–25.
Estola, M. (2015). Neoclassical and Newtonian theory of production: An empiri-
cal test. Hyperion International Journal of Econophysics and New Economy, 8(1),
7–22.
Estola, M., & Dannenberg, A. (2012). Testing the neoclassical and the Newto-
nian theory of production. Physica A, 391(24), 6519–6527.
Vihriälä, V. (1999). Banks and the Finnish credit cycle 1986–1995. Bank of Finland,
BOF Series E (Vol. 7). Helsinki: Suomen Pankki.
5
Goods Markets
In this chapter we analyze the behavior of different kind of goods markets.

The first problem in this is the classification of goods. For example, every
car model differs in various respects from other models. Cars are thus
heterogeneous goods. If all cars are classified in one class of goods, a
more detailed classification can be made within this class. Cars can be
classified according to their size, type, cubic meter of engine, and so forth.
We can thus classify the class ‘cars’ according to various characteristics,
which gives a more detailed classification of cars. When the distinctions
have been made according to all possible differences in cars, we end up
with a situation where the constructed classes contain only homogeneous
cars. One class of homogeneous cars may include, for example, cars that
were produced in 2003, have a 2000 cm3 diesel engine, four wheels, four
doors, and carry five persons. The same holds for all other goods; for
example, bananas can be classified according to their size and country of
origin, and so on.
§: By perfect substitutability between two goods we understand that
consumers consider these goods equally good in satisfying a certain need.
Imperfect substitutability means that the goods do not satisfy exactly
the same need, or they satisfy the same need with different efficiency. ˘

DOI 10.1007/978-3-319-46879-2_5
Example 1
Perfect substitutes could be electricity and oil in heating houses, loans
issued by different banks, and same kinds of insurances from insurance
companies. Imperfect substitutes could be, for example, coffee and tea,
Edam cheeses from different producers, and traveling by car or bus. ˘
Example 2
The carrots of different producers can be considered as almost perfect
substitutes. On the other hand, for the need to eat, which a carrot satisfies,
various imperfect substitute goods exist. These are, for example, turnips,
Swedish turnips, and other vegetables from which consumers get the same
vitamins and almost the same taste experience. ˘
§: The firms producing close substitute goods that satisfy a consumer’s

particular need creates an industry of these goods. ˘
The classification of industries can be made in a rough or in a detailed
way. In 1968, the United Nations defined an international classification
of goods called ‘A System of National Accounts’ (SNA), which has been
adjusted in 1993 and 2008 (see http://unstats.un.org/unsd/sna1993/).
The roughest level in this classification contains the following 16 classes: A:
Agriculture, forestry and hunting; B: Fishing; C: Mining and Quarrying;
D: Manufacturing; E: Electricity, gas and water supply; F: Construction;
G: Trade, repair of motor vehicles and household Goods; H: Hotels
and restaurants; I: Transport, storage and Communication; J: Financial
intermediation and insurance; K: Real estate and business activities; L:
Administration, compulsory social security; M: Education; N: Health
and social work; O: Other community, social and personal services; P:
Household service activities.
Within these 16 main classes, more detailed classifications are made.
For example, manufacturing is classified into 17 main classes, which
themselves contain more detailed subclasses. Examples of manufacturing
industries are: DA–Manufacture of food products, beverages and tobacco;
and DH–Manufacture of rubber and plastic products. With a detailed
5 Goods Markets 197
enough classification we can construct industries that contain goods that

are roughly homogeneous.
Even though the classification of industries can be made so specific that
every industry contains only homogeneous goods, these industries do not
function independently. The reason for this is that various heterogeneous
goods satisfy the same needs of consumers. The given examples show
that we can always find a close or a distant substitute for all goods
satisfying a consumer’s particular need, the existence of which affects the
functioning of the industries of these goods. Every consumer defines his
personal degree of substitutability between two goods, as we described in
Sect. 2.2.4. For example, one person may consider apples and oranges as
close substitutes; another person may not eat apples at all but likes oranges,
and so on.
In this chapter, we study the determination of the flow of production,
consumption, and price of a good that satisfies a particular need of
consumers. The goods that satisfy this need can be perfect or imperfect
substitutes. If, for example, we analyze the car industry, we have to realize
that different car types and models are imperfect substitutes. The car
industry is also affected by other types of vehicle industry, and also the
influence of status symbols, depending on whether a car is a vehicle for
the consumer, a status symbol, or both.
Goods can be classified according to their characteristics and con-
sumers’ attitudes towards them. One classification is between normal and
Giffen goods.
§: We call a good normal, if the aggregate flow of consumption of the
good decreases when its price increases. ˘
§: We call a good Giffen, if the aggregate flow of consumption of the
good increases when its price increases. ˘
Giffen goods get this name according to British statistician Sir Robert
Giffen (1837–1910), who discovered their existence. Sir Giffen made
observations of Irish people who used to eat potatoes and lamb. When
the price of potatoes increased, many people could no longer afford to buy
both lamb and potatoes. They therefore substituted lamb by eating more
potatoes, which, although more expensive, was still cheaper than lamb.
Thus the consumption of potatoes increased in spite of the price increase.
If we forget the observations of Sir Giffen, generally taken, some Giffen
goods can be status goods, which are bought due to their scarcity. A
price increase raises the status value of a good, and this may increase
some people’s willingness to pay for them. A price increase may also
be perceived as a sign of an improved quality of the good, which may
increase its demand. Giffen goods are, however, exceptional and in the
text that follows, therefore, we assume the studied goods to be normal
unless otherwise indicated.
In Sect. 3.9 we defined the demand for good k as a relation between
its price and the aggregate flow of consumption that corresponds to the
equilibrium states of all consumers consuming the good. We define supply
analogously.
§: By the market supply relation of a good we understand the
(aggregate flow of production, price) combinations that correspond to
the equilibrium states of all firms producing the good. ˘
It is a matter of preference whether we present the supply function
in a form where the aggregate flow of production of the good depends
on its price, or in its inverse function, where the dependence of the
price is on the aggregate flow of production. The former expresses the
aggregate flow of production that corresponds to the equilibrium states of
all firms producing the good at a certain price, and the latter expresses the
minimum possible price at which a certain aggregate flow of production
takes place. The former is traditionally called the supply function and
the latter the inverse supply function. In the inverse supply function,
the idea that price depends on the aggregate flow of production of a good
can be understood so that every aggregate flow of production defines a
unique price at which this flow takes place.
In Sect. 5.2 we will define a relation between the price and the flow of
production of a firm corresponding to the equilibrium states of the firm at
different prices in a perfectly competed industry. The supply relation of an
industry is aggregated from those of all firms in the industry. In imperfect
competition, firms operate as price setters and not as price takers like
firms in a perfectly competed industry. The supply relation of a good can
be defined only when the price of the good is determined ‘in the market’,
and firms adjust their production on this basis. Usually, no such place like
the ‘market of a good’ exists. The term ‘market’ is used to underline that
price is determined ‘outside firms’ according to the aggregate demand and
supply of the corresponding good.
5 Goods Markets 199
5.1 Different Market Situations

In this chapter, we analyze the short-term behavior of firms in different
market situations. The firms’ planning time horizons are assumed to be
one week, and the flows of production are measured in units kg=week.
Earlier on we assumed that the amount of production of a firm is
measured in units ‘unit’. By changing the measurement unit we show
that this has no effect on the analysis.
Different market situations can be classified on the basis of (1) the
characteristics of goods and (2) the number of sellers (producers) and
consumers. The characteristics of a good defines its degree of substitutabil-
ity and possible returns to scale in production. The more homogeneous
different firms’ products are and the less returns to scale exist in produc-
tion, the more perfect competition exists in the industry, and vice versa.
The reason to separate producers and sellers is that many times producer
firms give other firms (for example department stores) the right to sell
their products. Even though only a few producers exist in an industry, the
competition in retail trade may be almost perfect if numerous department
stores sell these goods.
According to the number of sellers or producers, two extreme situations
are: perfect competition and monopoly. In the former, all firms are
relatively small and their number is great enough so that no firm has
price-setting power on the product. On the other hand, a monopoly is
a one-producer market situation where that firm can unilaterally set the
price of its product. However, even though a monopoly firm can set the
price of its product, the firm has no means to force consumers to buy its
products.
According to the number of consumers, perfect competition is a
situation where none of the consumers can affect the price of the good.
A one-consumer situation is called monopsony, and in this situation
the only consumer can affect the price. An example of a one-consumer
situation is a firm that produces weapons and the laws of the country
deny the firm’s selling of its products to anyone other than the military
force of the home country. A common situation is, however, that so many
customers exist that none of them can affect the price. For this reason, we
assume in the following that numerous consumers exist for the goods we
are studying.
All market situations that are not perfect competition are called imper-
fect competition. In imperfect competition, either individual sellers
(producers) or consumers can affect the prices of the products. Three
forms of imperfect competition, which can be separated from the rest,
are: monopoly, monopolistic competition, and oligopoly.
5.1.1 Why Do Different Market Situations Exist?
A homogeneous product produced by different firms, increasing returns

to scale in production, and a relatively small market (a small number of
customers and/or a small turnover in the industry) together mean that
one big firm can produce the aggregate production of the industry with
mass production methods at smaller unit costs than various small firms
could do. In this kind of a situation, one firm can conquer the market
over time by price competition. The monopolization of an industry due
to increasing returns to scale in production is called a natural monopoly.
If a monopolized industry is relatively small, it does not attract new firms.
For these reasons, the monopoly firm in the industry can price its product
in the limits the prices of imperfect substitute goods set for it, and the
profitability of the monopoly firm does not attract new firms in the
industry.
Some industries can be monopolized also due to social reasons. Such
are: (1) The product is a public good; (2) Society likes to control the
production of the good for political reasons, for example, nuclear power,
education, or health care; (3) The good is important for the functioning
of the state, like postal services or energy and water supply.
§: A ‘pure’ public good is such that if one consumer buys it, other
consumers get it simultaneously at no charge. ˘
The above means that in consuming a pure public good, ‘intimacy’ is
lacking. Pure public goods are, for example, public defence, police, and
fire protection. The services of state-owned research institutes are also
usually publicly available, and thus they are public goods. When these
5 Goods Markets 201
goods have been produced, every inhabitant can enjoy their services and
access cannot be denied from any individuals. For this reason, consumers
are not willing to buy pure public goods, and private firms are not
interested in producing such goods because no person likes to buy them
for himself. The only way to finance the production of pure public goods
is to use tax money.
Some goods can be partly public and partly market goods. For example,
most parks and sport fields in cities are public goods. Even though every
citizen can freely use these goods, people are in different position with
respect to the services produced by them. Those living at the opposite
side of the city must travel to enjoy these services, and in this way they
have to pay for these services. Similar examples are public entertainment
areas that benefit most those people living next to them.
Increasing returns to scale in production may define a minimum scale
of production after which unit costs do not remarkably decrease any
further. This together with a relatively small market may lead to an
oligopoly situation.
§: Oligopoly is a market situation where a few relatively large firms
produce a roughly homogeneous good with increasing returns to scale in
production (see Sect. 5.5). ˘
In oligopoly, small firms cannot manage in cost competition, and the
size of the market does not allow profitable operation of many medium-
size firms. The high costs of small firms prevent their entry into the
market, which explains how the market situation stays constant. In an
oligopoly, the products of firms are similar. The difference between
oligopoly and perfect competition (Sect. 5.2) is that in the former a
smaller number of firms exist with a greater size. Increasing returns to
scale explains this difference.
The difference between perfect and monopolistic competition is in the
heterogeneity of the goods. In monopolistic competition, firms compete
with product differentiation and quality rather than price (see Sect. 5.4).
The quality difference may be real—for example cars and television
sets—or fictitious—for example candies and beverages. In monopolis-
tic competition, increasing returns to scale are not so remarkable that
this would restrict the entry of small firms. In this way, monopolistic
competition deviates from oligopoly. The entry of new firms guarantees

competition and prevents price cartels between firms.
§: By a price cartel we understand a situation where all sellers of a
certain good agree a common price, and so the firms do not compete on
price. Usually, price cartels are illegal, but it is often difficult to prove their
existence. ˘
A price cartel between the producers of heterogeneous goods is not
meaningful, however. With heterogeneous goods, price cartels can be
made between the sellers (department stores) of the product of one
producer. Examples are refrigerators, television sets, and so forth. Product
differentiation allows more freedom for firms in pricing their products
in monopolistic competition than in oligopoly. The price differences are
limited by the substitutability of existing goods.
The market situations in the real world may not exactly correspond
to any of the four idealized cases described above, but various features of
these can be seen in real life. The market situations we will study in the
following are thus a few idealized cases. By analyzing these idealized cases
we gain an understanding of the real world market behavior. The higher
the barriers of entry, and the more heterogeneous the products are that
satisfy a certain need of consumers, the more freely firms can price their
products.
5.2 Perfect Competition in an Industry

5.2.1 A Firm in a Perfectly Competed Industry
A firm in a perfectly competed industry—like all other firms—aims to

operate as profitably as possible. In perfect competition there is a special
feature whereby firms cannot affect the price of their product, because
that adjusts in the market according to the demand of all consumers con-
suming, and the supply of all producers producing the good. Price adjusts
with time according to the deviation between aggregate production and
consumption towards the level that the production of the industry gets
sold. This occurs because in the long run firms do not produce more than
they can sell.
5 Goods Markets 203
If the weekly production of a firm is greater than its weekly sales, it is

rational for the firm to decrease the price of its product in the case where
the price exceeds the firm’s marginal costs. If price is below marginal costs,
it is rational for the firm to decrease its flow of production. In a perfectly
competed industry, firms’ products are almost perfect substitutes. Thus if
one firm sells at a lower price than others, customers will buy from this
firm. To keep their customers, other firms must then decrease their prices
accordingly.
If the aggregate weekly production of an industry is smaller than gets
sold at current price, those firms facing excess demand can raise their
product prices. In an excess demand situation, consumers must compete
to buy the scarce goods, and then some people are willing to pay more. For
this reason, every firm can increase its product price because consumers
buy from those firms having goods left. Thus in a perfectly competed
industry, firms cannot decide the price of their product independently: it
is determined on the basis of the demand of all consumers consuming, and
the supply of all producers producing the good. Let us study this process
further.
Next we derive the supply relation for a firm in a perfectly competed
industry. The weekly profit of a firm producing homogeneous good k in
a perfectly competed industry is
…k (t) D Rk .qk (t)/ Ck .qk (t)/ D pk (t)qk (t) Ck .qk (t)/;
where the flow of production of the firm is denoted by qk (kg=week) and

the price of good k by pk (e=kg); profit …k , revenues Rk and costs Ck all
have unit e=week. The dependence of the flow of production and price
on time t is assumed, because later we analyze their adjustment with time.
The time derivative of …k (Sect. 10.9.4) is
@…k 0 @…k 0
…0k (t) D pk (t) C qk (t) D qk (t)p0k (t) C pk (t) Ck0 .qk (t)/ q0k (t);
@pk @qk
where marginal revenues equal with pk and by Ck0 .qk (t)/ is denoted
marginal costs. Because a firm in a perfectly competed industry cannot
affect the price of its product, the only variable by which the firm can
affect its profit is qk . A profit-seeking firm adjusts its flow of production

as follows: q0k (t) > 0 if pk (t) Ck0 .qk (t)/ > 0, and vice versa, and
q0k (t) D 0 if pk (t) D Ck0 .qk (t)/, where q0k (t)(kg=week2 ) is the acceleration
of production of the firm.
These adjustment rules can be explained as follows. Earlier on we
explained that the price of good k adjusts with time at the level the
production of the industry gets sold (including the production of this
firm). If the price is greater than the marginal costs of this firm, this
firm can increase its profit by increasing its flow of production. The firm
knows, however, that if it increases its flow of production, the aggregate
flow of production in the industry increases, and this has a decreasing
effect on price pk . Thus the firm has to take account that if it increases
its flow of production, this may require it to decrease the price of its
product to get its production sold. In spite of this, we identify the quantity
Fsk D pk (t) Ck0 .qk (t)/ as the force acting upon the flow of production
of good k of the firm.
The defined force measures the firm’s marginal profitability at current
price and prevailing flow of production. The argument for the force
interpretation is as before; the more profitable the production of one
extra kilogram, the more eager a profit-seeking firm is to increase its flow
of production. This force consists of the benefits and costs in the firm’s
production decision, and the profit maximizing situation corresponds
to zero-force:
@…k
D 0 , Fsk D 0 , pk (t) D Ck0 .qk (t)/: (5.1)
@qk
The relationship between the flow of production and price in Eq. (5.1)
defines those flows of production at different prices that correspond to
the equilibrium states of the firm. A profit-seeking firm changes its flow
of production with time so that eventually Eq. (5.1) holds.
If the form of the cost function Ck is known, the optimal flow of
production of the firm at moment t can be solved from Eq. (5.1) as
1
qk (t) D Ck0 .pk (t)/; (5.2)
5 Goods Markets 205
where Ck0 1 is the inverse function of Ck0 . To be able to find this solution,
function Ck0 must be at least partially monotonic so that it has a unique
inverse at every flow of production (see Sect. 10.5.4).
In a perfectly competed industry, every firm decides its flow of produc-
tion with the aim to sell this production at the price determined in the
market. Firms know the price when they make their production decision,
but they do not know the production decisions of other firms. If, then,
the aggregate flow of production is greater than was sold in the previous
time unit, consumers may not buy the increased production at current
price. Some firms then have unsold goods, and they can decrease their
production to diminish the aggregate flow of production towards the level
that gets sold at current price. On the other hand, if price is greater than
marginal costs of these firms, these firms can decrease their product price
to get their production sold.
Now, a price decrease by one firm forces other firms to follow this,
because otherwise their products would not get sold. In this way the price
determination in perfect competition leads to marginal cost pricing with
time. At a higher price than that—assuming that firms’ marginal costs are
higher than their unit costs—firms can increase their profit by decreasing
their price and producing and selling more. This would increase the
aggregate flow of production in the industry. If one firm decreases the
price of its product lower than that of other firms, the firm knows that
other firms will follow this. This decreases firms’ interest to lower their
product price to increase their profit and market share. However, a higher
price than marginal costs of firms may attract new firms in the industry.
Thus in a perfectly competed industry, the aggregate flow of production
will increase with time if price is higher than firms’ marginal costs. An
increase in supply decreases the price, and thus price adjusts with time at
the level that equals with the marginal costs of firms.
In several textbooks on economics, it is claimed that competition
between firms forces the equilibrium price in perfect competition at the
level where marginal costs of firms equal their unit costs. This occurs
because those firms, for which this condition does not hold, may adjust
their production so that they can decrease their product price at that
level. This way they gain customers from other firms, which forces other
Fig. 5.1 The supply relation of a firm in perfect competition
firms to follow the price decrease and produce at the flow of production
that creates the minimum possible unit costs. If a firm produces at its
minimum unit costs, the firm can sell its products at the lowest possible
price (see Fig. 5.1).
The supply relation of a firm in a perfectly competed industry can be
derived as is shown in Fig. 5.1. The unit and marginal costs of the firm
are described as in Sect. 4.3.2. Suppose that price is pk0 in Fig. 5.1. Then
every produced kilogram until the flow of production qk0 increases the
weekly profit of the firm, and the firm is motivated to increase its flow of
production until qk0 . However, the firm knows that its production affects
the aggregate supply in the industry, and this way the market price of
good k. For this reason, the production decision of the firm cannot be
analyzed separately from the price adjustment mechanism, and the market
mechanism must be analyzed as a complete system in Sects. 5.2.3–5.2.7.
Figure 5.1 shows that the force by which the firm affects the aggregate
production of the industry is positive at price pk0 if qk < qk0 , and negative
if qk > qk0 . An increase in the flow of production of the firm from qk0
5 Goods Markets 207
creates more costs than revenues at price pk0 , which would decrease the
profit of the firm. At price pk0 , the weekly profit of the firm is maximized
at the flow of production qk0 , and this corresponds to the zero-force
situation.
If price increases from pk0 to pk1 in Fig. 5.1, every sold kilogram until
the flow of production qk1 increases the profit of the firm. If the whole
production of the firm gets sold, the optimal flow of production at price
pk1 is qk1 . We can thus think that the optimal flow of production of the
firm is derived by its marginal costs. By the uniqueness of the marginal
cost function above the unit cost function, the marginal cost function
uniquely defines the flow of production the firm is willing to produce at
every price in the case its production gets sold. Thus it is profitable for the
firm to adjusts its flow of production with time at the level its marginal
costs equals with the price.
The supply relation of a firm shows the flows of production at different
prices that correspond to the optimal states of the firm. If the price
decreases, the optimal flow of production of a firm decreases according to
its marginal cost function. However, if price pk decreases below the unit
costs of a firm, the firm starts making losses. In that situation, the firm
can either stop its operation permanently or temporarily, or continue by
making losses. The last option is rational if the firm can cover its variable
costs and the firms’ managers believe that the firm can reduce its costs, or
that the price will increase in the future. On this basis, we can define the
marginal cost function as the supply relation of a firm in a perfectly
competed industry. Marginal costs show the optimal flow of production
at every price greater than unit costs of the firm.
Example
Let the cost function of a firm in perfect competition be Ck (qk ) D C0 C
g(qk )qk , where the flow of production qk has unit kg=week, g(qk ) D c1 C c2 qk
with unit e=kg is the variable unit cost function, and C0 ; c1 ; c2 are positive
constants with units e=week, e=kg and (e week)=kg2 , respectively. The
weekly profit of the firm is then
…k D pk qk Ck (qk ) D pk qk C0 c1 qk c2 q2k ; (5.3)

and the supply relation of the firm can be derived as
d…k pk c1
D0 , pk D c1 C 2c2 qk , q
k D : (5.4)
dqk 2c2
The marginal revenues of the firm equal with price pk , and marginal
costs c1 C 2c2 qk linearly increase with the flow of production. The supply
relation of the firm is the last form of Eq. (5.4). The optimal flow of
production increases with price and decreases if constants c1 ; c2 increase.
If pk < c1 , the firm makes losses and it is optimal to decrease or stop the
production. ˘
5.2.2 Aggregate Production of an Industry
In the previous section we showed that a firm in a perfectly competed

industry reacts to a price change by changing its flow of production.
Because every firm in the industry faces the same price change, firms
change their production flows according to the difference between price
and their marginal costs. As an example of this, we study a situation where
the profit of firm i is of the form in Eq. (5.3), and its flow of production
is denoted by qki . Thus
…ki D pk qki Cki (qki ) D pk qki C0i c1i qki c2i q2ki ;
and the supply relation of the firm is

d…ki pk c1i
D0 , pk D c1i C 2c2i qki , qki D : (5.5)
dqki 2c2i
Next we assume n firms in the industry and derive the optimal aggregate
flow of production of these firms. For simplicity, the cost functions are
assumed of identical form for every firm, but the constants in them may
vary. The optimal aggregate flow of production is then
X
n X
n
pk c1i
qkA D qk1 C qk2 C C qkn D qki D
iD1 iD1
2c2i
Xn
1 X
n
c1i
D pk D pk A B; (5.6)
iD1
2c2i iD1 2c2i
5 Goods Markets 209
Pn Pn c1i
where A D iD1 2c2i , B D
1
iD1 2c2i are positive constants. The
aggregate supply function thus positively depends on price pk ,
qkA D Apk B; (5.7)
and qkA 0 if pk B=A, where B=A defines the minimum price at

which the firms are interested in producing good k.
Note. The assumption that firms cost functions are of identical form, was
made to simplify the aggregation. In the real world, varying cost functions
of firms make the aggregation more complicated. In the next section
we show that in the case of varying cost functions, we can approximate
the supply relations of firms by linear ones in the neighborhood of
their equilibrium states, and in this way we can solve the aggregation
problem. ˘
To describe the behavior of a perfectly competed industry, we need to

model the aggregate production of the firms in the industry. It is possible
that this can be done by using a relatively simple average cost function
for the firms. Above we proposed one such candidate, but much work is
needed before we have solved this question in an empirically satisfactory
way.
5.2.3 Equilibrium in Perfect Competition
In this section we omit subindex k referring to the good to simplify the

notation, and we separate supply and demand by subscripts s; d. Thus
pk D p, qksi D qsi , qkdj D qdj , Cki D Ci and hkj D hj , where hj is
the marginal willingness-to-pay of consumer j of good k. According to
Sects. 3.8 and 5.2.1, if every firm and consumer have adjusted optimally,
we have:
p D Ci0 (qsi ); i D 1; : : : n and p D hj .qdj ; Tj ; p; pG /; j D 1; : : : ; m: (5.8)

Adding the n and m equations in Eq. (5.8), separately, and dividing the
results by n and m, respectively, we get
1X 0 1X
n m
pD Ci (qsi ) D hj .qdj ; Tj ; p; pG /I (5.9)
n iD1 m jD1
the middle term in Eq. (5.9) is the average

Pn of firms’ marginal costs Pat the
m
aggregate flow of production qs D q
iD1 si , and term (1=m) jD1 hj
is the average marginal willingness-to-pay of consumers P for one kilogram
of good k at the aggregate flow of consumption qd D m jD1 qdj . In the
equilibrium, the price equals the average of firms’ marginal costs and
consumers’ marginal willingness-to-pay, and no economic unit likes to
change its behavior. This corresponds to the neoclassical equilibrium.
In the Appendix of this section we show that we can approximate the
average of firms’ marginal costs as
1X 0 X
n n
0 a0 a1
C .qs / Ci .qsi / C 2 qs ; qs D qsi ; (5.10)
n iD1 n n iD1
where constants a0 > 0; a1 have units e=kg, (eweek)=kg2 , respectively.

If a1 > 0, decreasing returns to scale prevail in the industry and vice versa.
In the Appendix of this section, we show that we can approximate
average marginal willingness-to-pay for one kilogram of good k by m
consumers as
1X
m
b0 b1 b2 b3 b4
h(qd ; p; T; pG ) hj C 2 qd C p C 2 T C pG ; (5.11)
m jD1 m m m m m
where the aggregate

Pm flow of consumption of the m consumers is denoted
by qd D q , their aggregate budgeted funds for one week by
Pm jD1 dj
T D jD1 Tj , and constants b0 0; b1 < 0; b3 have units
e=kg, (e week)=kg2 and week=kg, respectively; b2 < 0 and b4 are
dimensionless. In the following in this section, we assume that pG D pG0 ,
Tj D Tj0 ; this way we can eliminate quantities T; pG from Eq. (5.11); see
5 Goods Markets 211
the Appendix of this section. An approximate average of the consumers’

marginal willingness-to-pay for one kilogram of good k is then
1X
m
b0 b1 b2
h(qd ; p) hj (qdj ; p) C 2 qd C p: (5.12)
m jD1 m m m
The equilibrium state in the industry can then be approximated as

a0 a1
Aggregate supply relation: pD C 2 qs ; (5.13)
n n
b0 b1 b2
Aggregate demand relation: pD C 2 qd C 2 p: (5.14)
m m m
The aggregate supply and demand relations can also be presented as
na0 n2
qs D C p; (5.15)
a1 a1
2
mb0 m b2
qd D C p; (5.16)
b1 b1
where a0 ; b0 ; n; m > 0, and b1 ; b2 < 0. In this section we assume a1 > 0,

that is Ci00 (qi ) > 08i, and thus the supply relation is increasing with price
(see the Appendix of this chapter).
Note. The aggregate demand relation in Eq. (5.16) equals with that
derived in Sect. 3.10 with a D mb0 =(m2 b2 ) > 0, b D b1 =(b2 m2 )
> 0. ˘
Setting qd D qs in System (5.15)–(5.16), we can solve the equilibrium

price p . The equilibrium flows of production and consumption can then
be solved by substituting p in Eqs. (5.15), (5.16). These give
a0 b1 n a1 b0 m nŒa0 (m2 b2 ) b0 mn

p D > 0; qd D qs D :
b1 n2 C a1 (b2 m2 ) a1 (b2 m2 ) C b1 n2
(5.17)
Fig. 5.2 Equilibrium state in a perfectly competed industry
The reader can check the measurement units of the constants to see
that the given solutions are dimensionally well defined. The condition
for qd D qs > 0 is that p1 D a0 =n < b0 m=(m2 b2 ) D p2 , where p1
is the price at which the aggregate supply, and p2 the price at which the
aggregate demand, is zero. The equilibrium state is displayed in Fig. 5.2.
Note. In Fig. 5.2, on the horizontal axis are the aggregate flows of pro-
duction and consumption of good k; these are both measured in units
kg=week. On the vertical axis is the price of good k, the average marginal
costs of firms, and the average marginal willingness-to-pay of consumers
for one kilogram of good k; these all are measured in units e=kg. ˘
5.2.4 Adjustment by Price Mechanism
Traditionally, the adjustment in a perfectly competed industry has been

assumed to take place according to price dynamics via excess demand as
p0k (t) D fp (qkd qks ); fp0 (qkd qks ) > 0; fp (0) D 0; (5.18)
5 Goods Markets 213
where qkd qks is the excess demand (supply) when it is positive (neg-
ative). This mechanism was suggested by Paul Samuelson (1941, 1942).
Assuming the demand and supply relations as in the previous section,
2
and denoting the constants in them as A0 D na a1
0
> 0, A1 D na1 > 0,
> 0, B1 D m bb
2
B0 D mbb1
0
1
2
> 0, the excess demand becomes the
following:
qkd qks D B0 B1 pk (t) C A0 A1 pk (t) D B0 C A0 (B1 C A1 )pk (t):
Taking the Taylor series approximation of function fp in Eq. (5.18) in the

neighborhood of point qkd qks D 0 and assuming the error term zero,
we get fp fp0 (0).qkd qks /, and then we can approximate Eq. (5.18) as
p0k (t) D fp0 (0) ŒA0 C B0 (A1 C B1 )pk (t); (5.19)
where fp0 (0) is a positive constant with unit e=kg2 . From Eq. (5.19) we
see that the quantity in brackets is positive if pk < AA01 CB
CB1
0
D pk , and
vice versa. Thus price increases (p0k (t) > 0) if pk < AA01 CB0
CB1
D pk , and
decreases if pk > AA01 CBCB1
0
D pk . The equilibrium is thus stable, and
price adjusts with time toward its equilibrium value pk . According to
Eqs. (5.15), (5.16), a price raise increases supply and decreases demand,
and vice versa. Thus firms and consumers react to price changes, and the
demand and supply relations show how both parties adjust their flows of
production and consumption.
Equation (5.19) can be explained as follows. If the whole production
in the industry gets sold, and even more could have been sold at current
price, the lack of goods forces consumers to compete to buy the scarce
goods. This allows some firms to raise their product price, and other firms
can follow suit because their whole production will also get sold. In the
opposite situation, any firm can assure that its production gets sold by
decreasing its product price, even though the whole production of the
industry does not get sold. The price reduction of one firm forces other
firms to follow its example if they want to sell their whole production.
Difference qkd qks is thus the cause that changes price pk , and every
consumer and producer affects this quantity by their decisions.
The differential equation in Eq. (5.19) has solution
A0 C B0 0
pk (t) D C C0 efp (0).A1 CB1 /t ;
A1 C B1
where C0 (e=kg) is the constant of integration. Because A1 C B1 > 0,

then with t ! 1, pk (t) will adjust toward the equilibrium price in the
industry pk D AA01 CB0
CB1
D b1an02bCa
1 na1 b0 m
2 (see Sect. 5.2.3). The process is
1 (b2 m )
thus stable.
5.2.5 Growth in a Perfectly Competed Industry
The growth of a perfectly competed industry can be analyzed by the

given model as follows. Suppose the equilibrium state prevails in the
industry so that firms and consumers do not want to change their
flows of production and consumption. Suppose then that the average
marginal costs of firms decrease due to some reason. This may result from
technological development in the production of one or several firms, from
the learning of workers, or from a decrease in factor prices in production.
At least one firm is then motivated to increase its flow of production
at the prevailing price. A decrease in the average marginal costs moves
the supply relation of the industry further from the origin (a0 decreases)
(see Fig. 5.3). This changes the equilibrium state in the industry from
(qk0 ; pk0 ) to (qk1 ; pk1 ).
If, on the other hand, the average marginal willingness-to-pay of
consumers for one kilogram of good k increases due to some reason, then
at least one consumer is willing to increase his/her consumption of good k
at the prevailing price. An increase in the average marginal willingness-to-
pay may originate from an increase in consumers’ income, from changes
in the prices of other goods, or from a change in consumers’ preferences.
An increase in consumption of one consumer increases the aggregate
flow of consumption of this good, and thus excess demand occurs. This
increases the price, which motivates firms to increase the aggregate flow
5 Goods Markets 215
Fig. 5.3 A decrease in average marginal costs of firms
of production. The new equilibrium state (qk1 ; pk1 ) shown in Fig. 5.4 is
obtained at a higher price and higher aggregate flows of production and
consumption than at (qk0 ; pk0 ).
Note. Notice the difference in these two cases of growth of production. A

decrease in marginal costs increases the flow of production and decreases
the price level, while an increase in consumers’ marginal willingness-to-
pay increases the flow of production and the price level too.
By the presented model we can also analyze the possibilities for the
public sector to promote economic growth. The public sector can affect
firms’ marginal costs by decreasing taxes, giving technological support,
and/or by improving education in society. The public sector may also
promote consumers’ marginal willingness-to-pay for one kilogram of the
good. We do not know, however, which goods consumers buy if they get
Fig. 5.4 An increase in average marginal willingness-to-pay of consumers
support. For example, financial support from the state may increase the
consumption of imported goods, which does not contribute to domestic
production. Thus if the public sector wants to support the production of
domestic firms in a perfectly competed industry, the best way to do this
is to try to decrease the marginal costs of the firms.
5.2.6 Further Observations on the Adjustment Process
The price adjustment mechanism described in Sect. 5.2.4 assumes that

consumers and firms react to price changes infinitely fast, because firms
are assumed to produce and consumers consume at every price with their
equilibrium flows. In the real world, however, the adjustment of firms and
consumers takes time, and next we add this element into the model.
According to Sect. 3.8, the force m consumers direct upon the con-
sumption of good k at moment t is derived as the average behavior of the
consumers:
5 Goods Markets 217
1X
m
Fd D hkj (qkdj ; pk ; pG ; Tj ) pk ; (5.20)
m jD1
where hkj is the marginal willingness-to-pay for one kilogram of good

k of consumer j. According to Eq. (5.20), a positive force is acting
upon the consumption of good k if the average marginal willingness-
to-pay of consumers exceeds price pk at the prevailing aggregate flow of
consumption, and vice versa. The equation of motion for the aggregate
flow of consumption is then:
q0kd (t) D fd (Fd ); fd0 (Fd ) > 0; fd (0) D 0; Fd D hk .qkd (t); pk (t); T; pG / pk (t);
(5.21)
where function fd obeys the above characteristics and by hk .qkd ; pk ; T; pG /
is denoted the average marginal willingness-to-pay of consumers for one
kilogram of good k. The equilibrium situation of consumers hk D pk
is asymptotically stable because according to Sect. 5.2.3, @hk =@qkd
b1 =m2 < 0 and thus @q0kd (t)=@qkd fd0 (Fd ) @hk =@qkd < 0. The
graphical analysis of stability of Eq. (5.21) thus exactly corresponds to that
in Fig. 3.9.
The equation of motion for the aggregate flow of production is defined
analogously as for consumption:
q0ks (t) D fs (Fs ); fs0 (Fs ) > 0; fs (0) D 0; Fs D pk (t) Ck0 (qks ); (5.22)
where fs is a function with the above characteristics and Ck0 (qks ) is

the average marginal costs of firms. The equilibrium state of all firms,
Ck0 .qks / D pk , is asymptotically stable, if average marginal costs of
firms are increasing with the aggregate flow of production (a1 > 0),
because according to Sect. 5.2.3, Ck00 (qks ) a1 =n2 and so @q0ks (t)=@qks
fs0 (Fs ) Ck00 (qks ) < 0 if a1 > 0.
The resultant force acting upon the production of good k consists
of the force components in Eqs. (5.21), (5.22). The reason to include
consumer behavior in the resultant force is that firms are not interested
in producing goods that do not get sold. A firm increases its flow of
production if it believes that the extra production gets sold. The forces in
Eqs. (5.21), (5.22) are measured in units e=kg and so we can add them.
The resultant force is
Fres D hk .qkd ; pk ; T; pG / pk C pk Ck0 .qks / D hk .qkd ; pk ; T; pG / Ck0 .qks /:

(5.23)
The rationale for the resultant force is that the two force components are
caused by independent economic units that both affect production. This
derivation is based on the ‘superposition’ principle applied in physics.
§: According to the superposition principle, the force acting in a
situation can be derived as the resultant sum of independent force
components acting in the situation (Ohanian 1989, pp. 107–108). ˘
According to Eq. (5.23), a positive force is acting upon the production
in the industry, if the average marginal willingness-to-pay of consumers
for one kilogram of good k is greater than the average marginal costs
of firms at the existing aggregate flows of consumption and production.
In the zero-force situation, the average marginal willingness-to-pay of
consumers equals the average of marginal costs of firms, and these both
equal with price pk .
In a perfectly competed industry, the price of good k adjusts according
to excess demand as:
p0k (t) D fp (qkd qks ); fp0 (qkd qks ) > 0; fp (0) D 0; (5.24)
where fp is a function with the above characteristics. The equation of

motion for price in Eq. (5.24) was explained in Sect. 5.2.4. In the next
section, we analyze the dynamics of the adjustment described in this
section.
5.2.7 The Dynamics of Adjustment in Perfect

Competition*
In the previous section we presented the equations of motion for aggregate

flows of production, consumption, and price of a homogeneous good k in
a perfectly competed industry. By taking the Taylor series approximations
5 Goods Markets 219
Fig. 5.5 The forces acting upon consumption, production, and price
of functions fc in Eqs. (5.21)–(5.22) in the neighborhood of the equilib-

rium points Fc D 0, c D d; s, and assuming the error terms zero, we can
approximate the functions as fc (Fc ) D fc0 (0) Fc , where fc0 (0) are positive
constants. Denoting constants fc0 (0) D 1=mkc , c D d; s, we can interpret
these as ‘inverses of inertial “masses” mkc of the adjusting quantities’.
These inertial ‘masses’ measure factors resisting changes in the adjusting
quantities, and their measurement unit is e(week=kg)2 ; this unit makes
the equations dimensionally homogeneous.
Price equation (5.24), however, corresponds to a spring equation in
physics, and so we denote fp0 (0) D kp , where kp with unit e=kg2 is the
spring constant measuring its strength.1 This analogy becomes visible in
Fig. 5.5. The factors resisting price changes are the costs from rewriting
new prices in products and advertisements, fear of the reactions from
customers, existing products in inventories, and so on. The value of
constant kp depends on these factors. We could also have assumed static
friction in price, which would explain that in the real world excess demand
must exceed a limit value before price starts reacting. For example, existing
goods in firms’ inventories prevent a price rise before these goods are sold.
However, for simplicity we assume that prices do not have static friction.
The equations of motion for the aggregate flows of consumption and
production, and for price then become the following:
mkd q0kd (t) D hk .qkd (t); pk (t)/ pk (t);

mks q0ks (t) D pk (t) Ck0 .qks (t)/; (5.25)
1
The analogue between the price equation and a spring system in physics was suggested by Alia
Dannenberg (Spring 2016, PhD Physics).
1 0
p (t) D qkd (t) qks (t): (5.26)
kp k
The ‘free body’ diagram of the forces acting upon the consumption,
production, and price of good k is shown in Fig. 5.5, where the static
friction forces are omitted for simplicity. The direction of motion on
the right-hand side of the diagram—where the accumulated amounts of
consumption and production, Qd ; Qs , with unit kg increase—is defined
positive, and that on the left-hand side is negative. Similarly, positive
forces point to the right and negative to the left. The coordinate axis
was chosen so that the system is in a continuous motion on the right
(negative flows of consumption and production are impossible), and the
difference in velocities of the particles changes the length L of the spring,
L D Qkd Qks .
The length of the spring measures the difference in accumulated
amounts of consumption and production of good k in units kg, because
the unit of the coordinate axis in Fig. 5.5 is kg. The positive force
component acting upon consumption is hk (qkd ; pk ) and the negative one
is pk , and the positive force component acting upon production is pk and
the negative one is Ck0 (qks ).
The spring between the two ‘particles’ reflects the situation that an
equal mutual force pk is acting upon the two particles, and a difference in
velocities of the particles qkd qks changes their distance and, therefore,
also the length of the spring that changes the mutual force component pk .
Analogously with Hooke’s law in physics, price pk is the force by which
the spring acts upon the two particles.
§: In physics, Hooke’s law states that force F by which a spring acts
upon the particles tied in the spring is of form F D kX, where k is a
positive spring constant and X the deviation of the spring from its rest
length. Thus if X D 0, then F D 0. ˘
Force pk depends positively on the length of the spring L D Qkd Qks ,
and we assume the rest length of the spring to be zero. Thus when the
spring stretches, the price increases, which exerts a higher force on the
two particles and vice versa.
Now, from Eq. (5.26) we see that if qkd D qks , then p0k (t) D 0 that is
pk is constant, and if qkd < qks (qkd > qks ), the distance between particles
5 Goods Markets 221
mkd ; mks decreases (increases) in Fig. 5.5. According to Eq. (5.26), qkd <
qks (qkd > qks ) causes p0k (t) < 0 (p0k (t) > 0), and decreasing price increases
qkd and decreases qks and vice versa. Thus price with unit e/kg is a force
component that acts upon aggregate consumption and production, and
these interact back to price by the difference qkd qks . These interactions
keep the system in a relatively stable motion. Because economic units are
not actively changing price pk , here we do not treat price as a ‘particle’ in
the system but a variable that interacts with aggregate consumption and
production. In the equilibrium state, the system moves with equilibrium
speed qkd D qks and constant pk . If, however, hk (qkd ; pk ) < C0 (qks ),
the system stops because then pk ! 0. In this case, the firms producing
good k go into bankruptcy unless they also have other, more profitable,
products.
Let us rewrite Eq. (5.26) as
p0k (t) D kp (qkd qks ): (5.27)
By taking definite integrals from 0 ! t of both sides of Eq. (5.27), we get

the solution of the differential equation as
pk (t) pk (0) D kp .Qkd (t) Qks (t)/ kp .Qkd (0) Qks (0)/ ;
Rt
where Qki (t)Qki (0) D 0 qki (s)ds; i D d; s with unit kg are accumulated
amounts of consumption and production from the initial moment until
current time moment t (see Sect. 2.7.3). We can rewrite this equation as
pk (t) D kp .Qkd (t) Qks (t)/ C C0 ; C0 D pk (0) kp .Qkd (0) Qks (0)/ ;
(5.28)
where C0 is a constant with unit e/kg. At the beginning of the production
of good k, its consumption is zero, but some products have already
been produced and offered for sale at price pk (0). Thus Qd (0) D 0
and Qs (0); pk (0) > 0; these make C0 > 0. Thus in the beginning
of the production, ‘particle’ mks has been for some time on the right-
hand side of ‘particle’ mkd in Fig. 5.5. Once consumption then started,
the first sold goods were already produced and stocked, and no extra
production was needed. This explains how consumption could grew faster
than production in the beginning, and particle mkd —that demonstrates
accumulated consumption in Fig. 5.5—bypassed that of accumulated
production mks and consumption started to lead this production system
via the price system.
Equation (5.28) shows that the historical scarcity of the good measured
by its accumulated difference in consumption and production, Qd (t)
Qs (t), defines the current price level of the product, and positive constant
C0 explains that pk (t) 0 even if Qd (t) < Qs (t). As earlier, the shapes
of the ‘particles’ in Fig. 5.5 have no economic meaning, and actually the
particles should be drawn as points on the horizontal axis. However, the
box shapes of the variables better visualizes the analogy we make here with
Newtonian mechanics.
Assuming hk .qkd (t); pk (t)/ and Ck0 .qks (t)/ as in Eqs. (5.13), (5.14),
System (5.25) becomes:
b0 b1 b2
mkd q0kd (t) D C 2 qkd (t) C 2 pk (t) pk (t);
m m m
a0 a1
mks q0ks (t) D pk (t) 2 qks (t); (5.29)
n n
1 0
p (t) D qkd (t) qks (t):
kp k
With certain parameter values, System (5.29) is globally stable and will
converge with time toward its equilibrium state given in Eq. (5.17). We
can thus make the following definition.
§: In the equilibrium state of an industry, the forces acting upon the
aggregate flows of consumption, production, and price, vanish. ˘
The equilibrium state of the industry in Eq. (5.17) can be solved by
setting p0k (t) D q0kd (t) D q0ks (t) D 0 in System (5.29) and solving the
resulting three equations with respect to the three endogenous variables.
The speed of adjustment of the system toward the equilibrium depends on
the two inertial masses, the spring constant, and the values of the param-
eters in the model. An exact mathematical analysis of this adjustment is
given in Estola and Hokkanen (2008), and an interested reader can turn
to this work for more information. In this book we mainly concentrate
5 Goods Markets 223
on analyzing the equilibrium state, and in this it is essential to know that

the system is stable and will converge in the equilibrium state with time.
In Fig. 5.6 are annual industrial flows of production and prices in
Finnish industries: DD: Wood and wood products; DH: Rubber and
Fig. 5.6 Annual flows of production and prices in Finnish industries

plastic products; DI: Other non-metallic mineral products; DJ: Basic

metals and fabricated metal products. Prices are multiplied by 1000 to
get their numerical values close to those of the flows of production.
The corresponding figures for industries DA, DB+DC, DF, DG were
presented earlier in Fig. 4.13. Figure 5.6 show that an industrial price may
have a similar positive trend as in the annual flow of production, but, for
example, in industry DJ the flow of production increases while price stays
relatively constant.
Because the analytic solution of System (5.29) is rather complicated, we
demonstrate the solution with the following numerical values: m D n D
10, mkd D mks D kp D 1, a0 D 50; a1 D 10 and b0 D 200 C 1:5t, b1 D
50; b2 D 50. The time dependence in parameter b0 is assumed to get a
positive time trend in consumption, which causes via rising price a positive
time trend in production, similarly as in all industries in Fig. 5.6. The
time paths of the three variables qkd ; qks ; pk are shown in Fig. 5.7, where
on the horizontal axis is time and on the vertical axis are qkd ; qks , and pk .
Consumption is the thicker of the two upper curves, and price is the one
below the others. The three quantities are presented in the same figure to
demonstrate their connections, even though the measurement units of the
quantities differ. The figure shows that price increases (decreases) when
qkd > qks (qkd < qks ), and the system converges with time to achieve
stable growth in all three variables. More realistic time paths for variables
qkd ; qks ; pk than those in Fig. 5.7 can be obtained by assuming parameters
a0 ; a1 and b0 ; : : : ; b2 to be random quantities. The initial condition of
the solution is: qd (0) D qks (0) D 30 and pk (0) D 10.
As compared with the static neoclassical analysis, by System (5.25)
we can study, besides equilibrium, reasons for the growth of a perfectly
competed industry, by setting different kind of time dependencies in
Ck0 .qks (t)/ and hk .qkd (t); pk (t)/. On the other hand, the adjustment
process can be studied by solving System (5.29) with varying numerical
values for the parameters to reveal their effects. Concerning the speed of
adjustment, the static neoclassical framework is a special case of Eq. (5.25)
with an infinite speed of adjustment, that is, mkc D 0, c D d; s and
1=kp D 0.
5 Goods Markets 225
Fig. 5.7 Time paths of consumption, production and price
The advantages of the detailed analysis of dynamics in this section,

as compared with the pure price adjustment mechanism in Sect. 5.2.4,
are: (1) In the real world, the adjustment of firms and consumers takes
time, and we can estimate the inertial factors mkc , c D d; s and spring
constant k by observations from the real world. (2) The forces acting upon
quantities qkd ; qks explain the changes observed in these quantities, while
the pure price adjustment mechanism explains only changes in price. (3)
From the point of view of economic policy-making, if government likes to
promote the growth of this industry, the model shows that it can do this
by positively affecting hk .qkd (t); pk (t)/ and negatively affecting Ck0 .qks (t)/.
(4) Assuming different kinds of time-dependencies in marginal costs of
firms and in marginal willingness-to-pay of consumers, we can model the
time paths in production, consumption, and price by using System (5.25),
while Eq. (5.19) explains only price dynamics.
5.3 A Monopoly Firm in an Industry

The degree of monopolization of an industry, or the degree of firms’
abilities to affect the prices of their products, can be measured by the
share of 5 biggest firms, in terms of turnover, in the industry. Another
way to measure the degree of monopolization is to estimate the difference
between the price of the good and the average of marginal costs of firms
in the industry. The latter measuring is, though, more difficult. The
degree of monopolization in a few industries in Finland is demonstrated
in Table 5.1.
Next, we model the short-term behavior of a monopoly firm producing
good k, and we assume that the planning time horizon of the firm is one
week. Because a monopoly firm is the only producer in the industry, the
sales function of a monopoly firm is the market demand relation of
the good. Assuming that the monopoly firm knows its sales function, the
situation is identical as in Sect. 4.4.2. A reminder of this is illustrated in
Fig. 5.8.
In the profit maximizing situation of a monopoly firm, the firm’s
marginal costs equal with marginal revenues (no label exists for the latter),
and the optimal flow of production is qk and price is pk (see Fig. 5.8).
The weekly profit of the firm (=industry) with unit e=week is shown
as the area of the rectangle with sides pk Ck (qk )=qk and qk . The
values of all functions in the figure are measured in units e=kg, and the
flow of production in units kg=week. According to Fig. 5.8, the degree
of monopolization in the industry can be measured by the difference
Table 5.1 The share of five biggest firms in terms of turnover in Finnish industrya
Manufacture of food products, beverages, and tobacco 21.9%
Building of complete constructions or parts thereof and con- 26.5%
struction services
Manufacture of pulp, paper, and paper products 54%
Manufacture of basic metals and fabricated metal products 78.9%
Manufacture of refined petroleum products, coke, and nuclear 98.3%
fuel
a
Source: Statistic Centrum in Finland: ‘Finnish firms’, Helsinki 1994, pp. 40–41
5 Goods Markets 227
Fig. 5.8 Time paths of consumption, production and price
between pk and Ck0 (qk ), because in Sect. 5.2.3 we showed that in the
equilibrium state of a perfectly competed industry, price equals the average
of marginal costs of firms.
The weekly profit of the firm in Fig. 5.8 can be named as monopoly
profit the firm gains because it can set the price of the product in the
industry. A monopoly firm is a price-setter while firms in a perfectly
competed industry are price-takers (see Sect. 5.2.1). As a price-setter, a
monopoly firm is assumed to know the sales function of the good. The
flow of production of a monopoly firm equals the aggregate flow of
production in the industry, and the sales function of a monopoly firm
is the market demand relation of the good. The weekly profit of the firm
is then
…k (t) D pk (t)qk (t) Ck .qk (t)/ D Bk .qk (t)/qk (t) Ck .qk (t)/;
where the firm is assumed to know its sales and cost functions: pk D
Bk .qk /, B0k (qk ) < 0, and Ck .qk /, Ck0 (qk ) > 0. The time derivative of the
profit is:
d…k 0
…0k (t) D qk (t) D B0k .qk (t)/qk (t) C Bk .qk (t)/ Ck0 .qk (t)/ q0k (t):
dqk
Substituting the sales function pk (t) D Bk .qk (t)/ in the above expression,
the force acting upon the production of the monopoly firm becomes:
Fks D B0k .qk (t)/qk (t) C pk (t) Ck0 .qk (t)/: (5.30)
Suppose next that the costs of production of a monopoly firm equals

with the aggregate costs of all firms, if perfect competition prevails in the
industry. Then, by comparing the force in Eq. (5.30) with the resultant
force in a perfectly competed industry in Eq. (5.23) we see that in the
equilibrium state, the force in Eq. (5.30) is smaller by factor B0k .qk /qk <
0. In the consumers’ equilibrium holds pk D hk (qkd ; pk ), and in the
monopoly firm’s equilibrium holds pk D Ck0 (qk ) B0k (qk )qk > Ck0 (qk ).
The force components acting upon production thus depend on the degree
of competition in the industry so that increasing competition increases the
force. Because the market demand relation is the same in both cases, in the
equilibrium the aggregate flow of production is smaller and the price
is higher in a monopolized industry than in the same industry under
perfect competition.
We can now speculate that if a monopoly firm behaves in the optimal
way from its own point of view, by which kind of economic policy could
government make it produce more and decrease the price it charges? On
the basis of the force acting upon the production of the firm, all factors
that decrease the firm’s marginal costs, or increase the demand, positively
affect the production of the firm. In addition, factors that decrease the
absolute value of the negative quantity B0k (qk )qk , positively affect the
production. Factor dpk =dqk D B0k (qk ) measures the relationship between
the price of the good and the aggregate flow of consumption. Because
dpk =dqk D B0k (qk ) D 1=D0k (pk ) D 1=(dqk =dpk ), the more horizontal
the inverse demand function of the industry is, the more close to zero is
quantity B0k (qk ) because dqk =dpk D D0k (pk ) ! 1 implies dpk =dqk D
B0k (qk ) ! 0. Thus in the case of a horizontal inverse demand function, a
monopoly firm operates exactly as a firm under perfect competition. The
5 Goods Markets 229
government can try to change all three factors in the force acting upon
the production in the industry by using its policy variables.
5.4 A Firm in Monopolistic Competition

A firm in a monopolistically competed industry faces the sales function
of the product of the firm. The real or ostensible quality differences
in the products of the industry allow the firms to price discriminate
their products within certain limits. Monopolistic competition deviates
from oligopoly so that in the former the products are heterogeneous,
and no barriers exist for entry due to increasing returns to scale. In
a monopolistically competed industry, firms enter and exit from the
industry every now and then, which makes the industry more vulnerable.
The equilibrium state of a firm in a monopolistically competed indus-
try corresponds to that of a monopoly firm: the firm aims to produce at the
flow of production where its marginal revenues equal its marginal costs.
The difference between a monopoly firm and a firm in monopolistic com-
petition is that a monopoly firm faces the market demand of the only good
in the industry. A firm in monopolistic competition, on the other hand,
faces the sales function of its own product, which is one good that satisfies
a certain need of consumers for which non-perfect substitute goods exist.
The behavior of a firm in monopolistic competition exactly corresponds
to that we analyzed in Sect. 4.4.2, and we do not repeat it here.
In monopolistic competition, firms try to differentiate their products
from those of other firms to get more freedom for their pricing. From the
point of view of the society, the problem is that if increasing returns to
scale exist in this production, monopolistic competition does not take
advantage of this. Relatively small scale production of heterogeneous
goods by various firms keep the unit costs of the firms high. This is
seen in shops as a too-large product selection from the point of view
of society; for example, products such as washing powder, beverages
and candies. However, it is difficult to say how much consumers prefer
product selection and low prices. This will be solved with time according
to the success of firms with different strategies—large scale production
and low price, or small scale production of differentiated products at
higher price.
5.4.1 Firms’ Advertising
A characteristic feature in monopolistic and oligopolistic competition is

firms’ advertising. From the point of view of the welfare of society, all
other firms’ advertising, than the distribution of useful information,
is a waste of resources. Such advertising, where homogeneous goods are
seemingly differentiated by affecting the feelings of consumers, is not
socially useful. For the advertising firm, however, this may be profitable. It
is often difficult to say what kind of effect advertising has on consumers,
but there are laws and ethical principles that advertising should obey.
Advertising that affects the preferences of consumers can be considered
as harmful from the point of view of society if consumers cannot be shown
to benefit from the marketed goods. If, on the other hand, a good is
socially useful in that it, for example, increases the health of people or
promotes the functioning of society, a manipulative marketing of this
kind of good may be socially useful. Campaigns against smoking and
alcohol, or promoting sport, for example, belong in this category. On the
other hand, informative advertising is socially useful if it increases people’s
knowledge of useful products that reduce their work time or increase the
quality of various kinds of work.
As a method of competition between firms, marketing may sometimes
increase competition and in this way better the functioning of society.
For example, marketing may increase consumers’ information of product
prices which promotes firms’ price competition. This benefits society.
Marketing may, however, in some cases decrease the entry of new firms
into an industry, which decreases competition and is thus socially harmful.
5.5 Oligopolistic Competition in an Industry

In oligopolistic competition, the products of different firms are highly
homogeneous and there exists barriers for entry in the industry—like
increasing returns to scale until a certain scale of operation is reached—
effectively preventing smaller firms being able to engage in price com-
petition. For these reasons only a few, relatively large, firms operate in
5 Goods Markets 231
the industry and the number of firms stays fixed. A characteristic feature
in oligopolistic competition is the avoidance of price competition by
direct or indirect price agreements, and competition takes place by other
means.
Avoiding price competition is beneficial for all firms in an industry with
oligopolistic competition, because that would decrease the profitability of
the whole industry. Even though direct price contracts are usually illegal,
it is often difficult to prove their existence. A typical oligopoly situation
prevails, for example, in the gasoline station industry, in banking and
insurance, and in wholesale and retail industries.
5.5.1 Pricing in Oligopoly: Contract vs. Competition
In oligopolistic competition, the firms in the industry have a common

interest to set the price of the relatively homogeneous good in the industry
at the level that maximizes the profit of the industry. However, every firm
is still motivated to increase its share of the profit of the industry, which
it can do by slipping away from the price agreement by decreasing the
price of its product. These elements mean that we cannot present a unique
theory for the equilibrium state in oligopolistic competition as we defined
for the other studied market situations.
In oligopolistic competition, the firms can operate together with a
price agreement, similarly to how one monopoly firm would operate in
the industry. Firms can estimate the aggregate demand relation of the
relatively homogeneous good produced in the industry, and set the price
of the good at the level that maximizes the profit of the industry. Price
agreement requires a contract between firms concerning the segmentation
of the market so that every firm has a certain share of customers, or a
certain share of the aggregate flow of production. For example, during
the history of the union of oil-producing countries (OPEC) they have
frequently succeeded in negotiating production quotas for every member
country.
The first question in modeling the behavior of an industry with
oligopolistic competition is whether the firms do cooperate or not. This
situation can be analyzed by using game theory. In a game theoretic
analysis, defined for every player (firm) are the goals he/she/it aims to
achieve in the game, and the strategies he/she/it can use in attaining
these goals. After this, the next thing to look for is whether there exists
an aggregate flow of production/price combination in the industry that
is derived by firms’ cooperative or separate behavior, from which none of
the firms likes to deviate if other firms stay in it. This situation is called a
Nash equilibrium in the honor of Nobel laureate John Nash (1950), who
invented the concept. However, a game theoretic analysis of oligopolistic
competition would require the definition of various concepts applied in
game theory, and in this book we omit modeling oligopolistic competition
in game theoretic way. A reader interested in game theory can turn, for
example, to Gibbons (1992).
5.5.2 Price Contracts in Oligopoly
A price cartel is organized to prevent firms from engaging in price

competition. The aggregate profit of firms in a cartel is divided between
the firms according to some agreed principles. This may take place
by fixing the production quota, as with OPEC, where some firms or
countries may be paid even for not producing at all. In this way, the cartel
avoids the price decrease an excess supply would cause.
Price agreements may be direct or indirect, of which the former are
usually illegal according to the laws promoting competition. For this
reason, price agreements are usually made indirectly, by using a price
leader, for instance. A price change by a firm acting as the price leader is a
signal for other firms to change their prices accordingly. The price leader
may be the biggest firm in the industry or some other firm that more
sensitively than others observes changes in the demand of the product;
for example, by its international contacts.
The requirements for a price agreement are: (1) The entry of firms in
the industry can be controlled. This can be made, for example, by way
of issuing licenses. Opening a medical clinic or a law firm requires the
appropriate education and a license for the profession, to drive a taxicab
usually requires a license, and so on. (2) The products of the firms must
be relatively homogeneous. (3) The costs of firms must be similar enough
so that they can agree to a common price. The fewer firms in an industry,
the easier it is to arrange a price agreement.
5 Goods Markets 233
5.5.3 Non-cooperative Oligopoly
The less public information there is about prices, the more easy it is for
firms to slip out of a common price agreement. In this kind of situation,
firms base their operation on their assumptions of the strategies of other
firms.
1. Bertrand’s conjecture.2 Firms believe that other firms do not follow

their price changes. In this case, every firm is motivated to decrease its
product price to increase its market share. This leads to price competi-
tion that ends up over time with marginal cost pricing, similarly as in
a perfectly competed industry.
2. Cournot’s conjecture.3 Firms believe that other firms will produce at
the flow of production they have planned, independent of the price
level. In this, as in the former case, the individual behavior of firms
leads to a higher aggregate flow of production than a monopoly firm
would produce in the industry. This decreases the price level under
monopoly pricing, because every firm follows the price decreases of
other firms in order to get their production sold. The more firms there
are in the industry, the closer to marginal cost pricing the industry will
end up over time.
5.5.4 Cournot’s Model of a Duopoly
We assume two firms in an industry producing a homogeneous good,

and the flows of production of the firms with units kg=week are denoted
by q1 ; q2 , respectively. In a duopoly, the function that expresses the
maximum unit price p (e=kg) by which the two firms can sell the amount
q1 C q2 (kg=week) of the homogeneous product in a week—the inverse
demand function of the good—can be written as:
2
Joseph Louis Françis Bertrand (11 March 1822–5 April 1900) was a French mathematician that
worked in the fields of number theory, probability theory, thermodynamics and economics.
3
Antoine Augustin Cournot (28 August 1801–31 March 1877) was a French philosopher and
mathematician who also contributed to the development of economics theory.
p D f (z); f 0 (z) < 0; z D q1 C q2 :
The weekly cost function Ci .qi / (e/week) of firm i is assumed as
Ci .qi / D ci C gi .qi /qi ; i D 1; 2;
where by ci (e=week) is denoted fixed costs in a week and variable unit

costs gi (qi ) (e=kg) depend on the flow of production (see Sect. 4.3.2).
Both firms are assumed to know the demand relation and their own
cost function. The weekly profits of the firms …i (e/week), i D 1; 2, are
then:
…1 .q1 ; q2 / D f .q1 C q2 /q1 c1 g1 .q1 /q1 ;

…2 .q1 ; q2 / D f .q1 C q2 /q2 c2 g2 .q2 /q2 : (5.31)
Both firms wish to maximize their profit, and so in the equilibrium holds
@…1 @…2
D 0; D 0:
@q1 @q2
In order to demonstrate the equilibrium situation in the industry, we

assume the demand and cost functions as
f .q1 C q2 / D a b(q1 C q2 ); gi .qi / D di ; i D 1; 2; (5.32)
where positive constants a; b; di have units e/kg, (eweek)=kg2 and

e/kg, respectively. The profit functions are then
…1 .q1 ; q2 / D Œa b(q1 C q2 )q1 c1 d1 q1 ;

…2 .q1 ; q2 / D Œa b(q1 C q2 )q2 c2 d2 q2 : (5.33)
The equilibrium state in the industry can be solved from the following
system of equations
@…1
D 0 , a 2bq1 bq2 d1 D 0; (5.34)
@q1
5 Goods Markets 235
Fig. 5.9 Equilibrium state in Cournot’s duopoly
@…2
D 0 , a 2bq2 bq1 d2 D 0; (5.35)
@q2
as
a C d2 2d1 a C d1 2d2
q1 D ; q2 D : (5.36)
3b 3b
The flows of production q1 ; q2 maximize the firms’ profits. The
equilibrium state is shown in Fig. 5.9 as the crossing point of the lines
representing Eqs. (5.34), (5.35); the former of these is the less steep one.
5.5.5 Cournot’s Duopoly à la Newton*
Here we dynamize Cournot’s duopoly. We set the flows of production

of the two firms as functions of time t, and we assume demand and cost
functions as in Eq. (5.32). The time derivatives of the profit functions are:
@…1 0
…01 (t) D q (t) bq1 (t)q02 (t)
@q1 1
D Œa 2bq1 (t) bq2 (t) d1 q01 (t) bq1 (t)q02 (t); (5.37)
@…2 0
…02 (t) D q (t) bq2 (t)q01 (t)
@q2 2
D Œa 2bq2 (t) bq1 (t) d2 q02 (t) bq2 (t)q01 (t): (5.38)
Each firm’s leader aims to maximize the profit of their firm, and so they
change qi (t) towards increasing …i .q1 ; q2 ), i D 1; 2, with time as
@…i
q0i (t) > 0 if > 0;
@qi
@…i
q0i (t) < 0 if < 0; (5.39)
@qi
@…i
q0i (t) D 0 if D 0; i D 1; 2:
@qi
These rules increase the profits of the firms with time, and the last
rule implies that there is no reason to change the flow of produc-
tion if it is expected not to affect the profit. The time derivatives in
Eqs. (5.37), (5.38) show that a change in q2 affects the profit of Firm 1, but
these changes are outside control of the managers of Firm 1. Similarly, q1
affects the profit of Firm 2 but Firm 2 cannot affect it. Thus by adjusting
their flows of production as in Eq. (5.39) the firms increase their profits.
A relation that fulfills the rules in Eq. (5.39), is
@…i
q0i (t) D fi (Fi ); fi0 (Fi ) > 0; fi (0) D 0; Fi D ; i D 1; 2; (5.40)
@qi
where function fi obeys the above characteristics. In Eq. (5.40), Firm

i adjusts its flow of production according to the deviation between its
marginal revenues and costs. The first order Taylor series approximation
of function fi in Eq. (5.40) in the neighborhood of the optimum point
Fi D @…i =@qi D 0, is
fi (Fi ) D fi (0) C fi0 (0)(Fi 0) C i D fi0 (0) Fi C i ;

5 Goods Markets 237
where i is the error term. Assuming i D 0 we can write Eq. (5.40) as
@…i 1
mi q0i (t) D ; mi D 0 i D 1; 2: (5.41)
@qi fi (0)
Now, imitating Newtonian mechanics we identify @…i =@qi —the reason

for the acceleration of production of Firm i—as the ‘force acting upon
the flow of production of Firm i’. Positive constant mi with unit
(eweek2 )=kg2 is the ratio between force and acceleration, and its
magnitude measures the inertia in this adjustment. Following Newton,
we call mi the ‘inertial “mass” of flow of production of Firm i’ and
Eq. (5.41) ‘the Newtonian equation of production of Firm i’. We could
have added also static friction force in Eq. (5.41), but we omitted it for
simplicity.
The ‘free body’ diagram of the forces acting upon the production
flows of the two firms is shown in Fig. 5.10 where static friction forces
are omitted for simplicity. The forces are derived by using the profit
functions in Eq. (5.33). The ‘particle’ that is moving in the coordinate
system (q2 ; q1 ) is the vector q(t) D (q2 (t); q1 (t)) of simultaneous flows of
production of the two firms. As earlier, the shape of the ‘particle’ has no
Fig. 5.10 Forces acting upon the production system of two firms
economic meaning, and actually the particle should be drawn as a point

(vector) in the coordinate system. However, the box shape for the vector
function visualizes better the analogy we make here with Newtonian
mechanics.
Thus the point describing the value of the vector function moves with
time in space (q2 ; q1 ), and if the system is stable, the vector function will
converge to a fixed point with time. The positive force component acting
upon the flow of production of good 1 (pointing upward) is a, and the
negative force component acting upon the flow of production of good 1
(pointing downward) is b(2q1 C q2 ) C d1 . Similar force components are
acting upon the flow of production of good 2 so that the positive force
component is pointing to the right, and vice versa.
The force components show how the flow of production of one firm
affects negatively that of the other. These interdependencies between the
two firms arise from the common inverse demand function the firms have.
By using the profit functions in Eq. (5.33), we get the corresponding
pair of Newtonian equations as
m1 q01 (t) D a 2bq1 (t) bq2 (t) d1 ;

m2 q02 (t) D a 2bq2 (t) bq1 (t) d2 : (5.42)
In System (5.42), we assume that each firm can identify the flow of
production of the other, and that each firm can forecast its own costs
and the market price. System (5.42) has a unique solution with a fixed
initial condition. However, because this solution is rather complicated, we
simplify the analysis by assuming numerical values for the parameters. Let
a D 100, and b D di D mi D 1, i D 1; 2. The solution of System (5.42)
is then:
1 1
q1 (t) D 33 C (C1 C C2 )e3t C (C1 C2 )et ;
2 2
1 1
q2 (t) D 33 C (C1 C C2 )e3t C (C2 C1 )et ; (5.43)
2 2
where C1 ; C2 are the constants of integration (the initial conditions). With

t ! 1, the exponential functions vanish and System (5.43) converges
5 Goods Markets 239
Fig. 5.11 The force field in Cournot’s duopoly
to Cournot’s equilibrium: q1 D (a C d2 2d1 )=3b D 33 D q2 . The

graphical demonstration of this adjustment is shown in Figs. 5.11 and 5.12.
§: A curve in the state space of a dynamic system, where the direction
of motion changes, is called a demarcation curve. ˘
According to the above definition, lines q01 (t) D 0, q02 (t) D 0 in
Figs. 5.11 and 5.12 (that are the same lines as in Fig. 5.9) are demarcation
curves. The ‘force field’ denoted by the arrows shows the direction of
motion of the system in the four subspaces defined by the demarcation
lines. From Eq. (5.42) we can solve
a bq2 (t) d1
q01 (t) > 0 if q1 < ;
2b
and vice versa. Thus below demarcation line q01 (t) D 0 , q1 D (a
d1 bq2 )=2b holds q01 (t) > 0, and, similarly, above it holds q01 (t) < 0.
Below demarcation line q02 (t) D 0 holds q02 (t) > 0, and above it holds
q02 (t) < 0. For example, in the area below both demarcation lines holds
Fig. 5.12 The force field and demarcation lines in Cournot’s duopoly
q01 (t) > 0; q02 (t) > 0 (the direction of motion is thus upward and to
the right because q1 increases ‘up’ and q2 increases ‘right’). Above both
demarcation lines holds q01 (t) < 0; q02 (t) < 0, and so on. The defined
force field implies stability, and depending on the initial condition, the
adjustment may be smooth or oscillatory. Assuming time dependencies in
parameters a; b; di ; i D 1; 2, we can model the time path of production
in the industry by using System (5.42). This cannot be done by using
the static model described in Eqs. (5.34), (5.35), which demonstrates the
usefulness of the dynamic framework introduced in this section.
5.6 Price in Monopolistic Competition

§: By price inflation we understand the growth rate of the average price
level in an economy. ˘
Price inflation (hereafter abbreviated to ‘inflation’) is calculated at
macro level as a weighted average of growth rates of prices of goods in
all industries in an economy by using a weighting method; for example,
5 Goods Markets 241
the growth rate of a consumer or producer price index. Average inflation

in one industry, on the other hand, is calculated as a weighted average
of growth rates of unit prices of firms in the industry; for example, by
arithmetic average. Thus to understand inflation, we have to explain how
the prices of goods in various industries are determined. In Sect. 5.2.7
we studied price determination in a perfectly competed industry. In this
section, we analyze the determination of average price level in an industry
with monopolistic competition. In imperfect competition, firms set the
prices of their products as we studied in Sect. 4.7.1.
Factors that increase average price level and thus cause price inflation
in an industry have traditionally been divided between the factors causing
demand and those causing cost inflation. Demand inflation takes place
when the aggregate demand of products of the firms in an industry is
greater than the aggregate supply. This excess demand allows some firms
to increase their product prices so that the aggregate production of the
industry still gets sold; this increases the profits of these firms. Cost
inflation, on the other hand, takes place when the costs of the firms
increase in an industry. For the profitability of the firms to stay constant,
firms must raise their product prices accordingly.
Next we analyze the determination of the average price level in a
monopolistically competed industry that consists of n firms producing
somewhat heterogeneous goods. The weekly profit …i (e=week) of firm
i is
…i (t) D Ri (t) Ci (t) D pi (t)qi (t) Ci .qi (t)/;

qi (t) D Di .pi (t)/; Ci0 (qi ) > 0; D0i (pi ) < 0;
where Ri (e=week) and Ci (e=week) are the weekly revenues and costs of
firm i, pi (e=kg) the price of the product of the firm, qi (kg=week) the
flow of production (= flow of consumption of the products of the firm),
and qi D Di (pi ) the sales function of the firm. The derivative of the sales
function of the firm with respect to the price, dqi =dpi D D0i (pi ), measures
the change in weekly sales of the firm when a marginal change in price
takes place.
If the flow of production is replaced by the sales function of the firm,

the weekly profit of firm i can be expressed as a function of price only,

…i (t) D pi (t)Di .pi (t)/ Ci Di .pi (t)/ :
The time derivative of this profit function is
d…i 0 dRi 0 dCi 0

…0i (t) D pi (t) D pi (t) p (t) D .Di (pi ) C pi D0i (pi )/p0i (t)
dpi dpi dpi i

Ci0 (qi )D0i (pi )p0i (t) D Di (pi ) C (pi Ci0 (qi ))D0i (pi ) p0i (t);
where p0i (t) with unit (e=kg)=week is the flow of the price of the product
of firm i. Now, p0i (t) is the effect of firm i on the change in the average
price level in the industry, and so also on the economy level price inflation.
According to Sect. 4.7.1, we can name quantity
d…i dRi dCi
Fi D D D Di (pi ) C (pi Ci0 (qi ))D0i (pi )
dpi dpi dpi
with unit kg=week as the force firm i directs upon the average price level
in the industry. A profit-seeking firm changes the price of its product to
increase its weekly profit. The equation of motion for price pi is then
p0i (t) D fi (Fi ); fi0 (Fi ) > 0; fi (0) D 0; Fi D Di (pi ) C (pi Ci0 (qi ))D0i (pi );
(5.44)
where function fi obeys the defined characteristics. Average price level p
(e=kg) in the industry and its flow are then:
1X 1X 0
n n
p(t) D pi (t); p0 (t) D p (t):
n iD1 n iD1 i
Adding the adjustment equations in Eq. (5.44) of the n firms, we get:
X
n X
n
p0i (t) D fi (Fi ): (5.45)
iD1 iD1
5 Goods Markets 243
Dividing both sides of Eq. (5.45) by n we get:
1X 0 1X
n n
p0 (t) D pi (t) D fi Di (pi ) Ci0 (qi )D0i (pi ) pi D0i (pi ) : (5.46)
n iD1 n iD1
Equation (5.46) shows that the average price level in the industry increases
if the arithmetic average of changes in firms’ profits is positive due to a
price increase, and vice versa. The equilibrium price level in the industry
corresponds to situation Fi D 0; i D 1; : : : ; n. Then the firms cannot
increase their weekly profit by changing their product price. In the next
section we show that assuming fi0 (0) D f 0 (0) > 0 8i, we can make the
following approximation
1X
n z z1
fi Di (pi )Ci0 (qi )D0i (pi )pi D0i (pi ) f 0 (0)
0
C p(t) ; (5.47)
n iD1 n n
where z0 > 0; z1 < 0 are constants with units kg=week, kg2 =(e week),
respectively. The equilibrium price level can then be solved as
z0
p0 (t) D 0 , p(t) D > 0:
z1
The modeling in this section shows how we can model price inflation
in an industry with monopolistic competition. At the roughest level, the
empirical testing of this modeling requires data of average price level in
the industry. However, a more detailed data of the costs of the firms in
the industry would allow a more detailed testing of the theory by using
Eq. (5.44).
5.6.1 Price Dynamics in Monopolistic Competition*
The Taylor series approximation of function fi in Eq. (5.44) in the

neighborhood of the equilibrium state Fi D 0 is
fi (Fi ) D fi (0) C fi0 (0)(Fi 0) C i D fi0 (0) Fi C i : (5.48)

Assuming i D 0, we can approximate Eq. (5.44) as
p0i (t) D fi0 (0) Fi ; (5.49)
where fi0 (0) with unit e/kg2 is a positive constant. Adding the adjustment
equations in Eq. (5.49) of the n firms in the industry, we get:
X
n X
n
p0i (t) D fi0 (0) ŒDi (pi ) C (pi Ci0 (qi ))D0i (pi ): (5.50)
iD1 iD1
Then, dividing both sides of Eq. (5.50) by n we get:
1X 0 1X 0
n n
p0 (t) D pi (t) D f (0) ŒDi (pi ) C (pi Ci0 (qi ))D0i (pi ):
n iD1 n iD1 i
(5.51)
Next, we approximate the sum
X
n X
n
FD Fi D ŒDi (pi ) C pi D0i (pi ) Ci0 (qi )D0i (pi )
iD1 iD1
by the Taylor series in the neighborhood of the equilibrium point p0 D

(p10 ; : : : ; pn0 ) as
n h
X
F D F(p0 ) C 2D0i (pi0 ) C pi0 D00i (pi0 ) Ci00 (qi0 )ŒD0i (pi0 )2
iD1
i
Ci0 (qi0 )D00i (pi0 ) (pi pi0 ) C i
n h
X i
D F(p0 ) 2D0i (pi0 ) Ci00 (qi0 )ŒD0i (pi0 )2 C Œpi0 Ci0 (qi0 )D00i (pi0 ) pi0
iD1
n h
X i
C 2D0i (pi0 ) Ci00 (qi0 )ŒD0i (pi0 )2 C Œpi0 Ci0 (qi0 )D00i (pi0 ) pi C i ;
iD1
5 Goods Markets 245
where qi0 D Di (pi0 ) and F(p0 ) D 0. Then, assuming i D 08i and

denoting the constants as
n h
X i
z0 D F(p0 ) 2D0i (pi0 ) Ci00 (qi0 )ŒD0i (pi0 )2 C Œpi0 Ci0 (qi0 )D00i (pi0 ) pi0 ;
iD1
n h
X i
z1 D 2D0i (pi0 ) Ci00 (qi0 )ŒD0i (pi0 )2 C Œpi0 Ci0 (qi0 )D00i (pi0 ) ;
iD1
we can make the following approximation
z1 X
n
F z0 C pi ;
n iD1
where constants z0 ; z1 have units kg=week and kg2 =(e week), respec-
tively. Now, assuming fi0 (0) D f 0 (0) 8i, we can approximate Eq. (5.51) as:
!
z1 X
n
z0
p0 (t) D f 0 (0) C 2 pi (t) :
n n iD1
Finally, denoting f 0 (0) D 1=mp , where mp > 0 with unit kg2 =e

represents the inertial ‘mass’ of the average price level in the industry,
we can present the modified Newtonian equation for the average price
level as:
z0 z1
mp p0 (t) D C p(t): (5.52)
n n
Term ‘modified’ is used because p0 (t) is the velocity, and not the accel-
eration, of average price level. The force acting upon the average price
level in the industry is the arithmetic average of the firm-specific force
components causing demand and cost inflation. In the profit maximizing
situation of firm i holds: pi D Ci0 (qi ) > 0 and Ci00 (qi ) > 0. Assuming that
the firms operate near their profit maximizing situations, we can sign the
constants as z0 > 0; z1 < 0 because Fi (pi0 ) D 0 and D0i (pi ) < 0. These
are sufficient conditions for price stability. The higher the average price
level, the smaller its velocity and vice versa, and the equilibrium price level
corresponds to: p0 (t) D 0 ) p(t) D z0 =z1 > 0. The general solution
of Eq. (5.52) is:
z0 z1
t
p(t) D C H0 e mp n ; (5.53)
z1
where H0 (e=kg) is the constant of integration. If z0 > 0; z1 < 0, p(t)

converges with time into its equilibrium value: p(1) D z0 =z1 > 0.
The ‘free-body’ diagram of the forces acting upon the price level in the
industry does not essentially deviate from that made in Sect. 4.7.1, and
thus it can be omitted.

The first order Taylor series approximation of the marginal cost function
of firm i in the neighborhood of the equilibrium flow of production qi0 is:
Ci0 .qsi / D Ci0 .qsi0 / C Ci00 (qsi0 ).qsi qsi0 / C i ; i D 1; : : : ; n; (5.54)
where i is the residual term, see Sect. 10.10.1. Assuming i D 0 and

adding over i, we get4 :
X
n X
n
Ci0 .qsi / ŒCi0 (qsi0 ) Ci00 .qsi0 /qsi0 C Ci00 .qsi0 /qsi
iD1 iD1
a1
a0 C qs ;
n
Pn Pn Pn Pn
4
c q D c iD1 qsi C iD1 (ci c)qsi where c D (1=n) iD1 ci . The approximation
Pn iD1 i i Pn 00
iD1 ci qi c iD1 qsi is thus the more accurate the less ci D Ci (qsi ) or qsi vary, i D 1; : : : ; n.
5 Goods Markets 247
Pn
where qs D iD1 qsi and constants
X
n X
n
a0 D ŒCi0 .qsi0 / Ci00 .qsi0 / qsi0 ; a1 D Ci00 .qsi0 / ;
iD1 iD1
have units e=kg and (eweek)=kg2 , respectively. Because marginal costs

are positive at every qsi , then a0 > 0 (let qsi ! 0 and i ! 0 in Eq. (5.54)
8i). Increasing returns to scale in the industry Ci00 (qi ) < 08i correspond
to a1 < 0, and decreasing returns to scale correspond to a1 > 0.
We approximate the marginal willingness-to-pay function of consumer
j in the neighborhood of his equilibrium point zj0 D (qdj0 ; p0 ; Tj0 ; pG0 ) as
@hj @hj
hj (qdj ; p; pG ; Tj ) D hj (zj0 ) C (zj0 )(qdj qdj0 ) C (zj0 )(p p0 )
@qdj @p
@hj @hj
C (zj0 )(Tj Tj0 ) C (zj0 )(pG pG0 ) C j ; j D 1; : : : ; m:
@Tj @pG
(5.55)
Assuming j D 08j and adding over j, we get5 :
X
m Xm
@hj @hj
hj (qdj ; p; pG ; Tj ) D hj (zj0 ) (zj0 )qdj0 (zj0 )p0
jD1 jD1
@qdj @p
X m Xm
@hj @hj @hj @hj
(zj0 )Tj0 (zj0 )pG0 C (zj0 )qdj C (zj0 )p
@Tj @pG jD1
@qdj jD1
@p
Xm
@hj Xm
@hj
C (zj0 )Tj C (zj0 )pG
jD1
@Tj jD1
@pG
b1 b3
b0 C qd C b2 p C T C b4 pG ; (5.56)
m m
5
Similar approximations are made here as in footnote 2.
where thePaggregate flow of consumption of the m consumers Pmis denoted

by qd D m jD1 q dj , their aggregate budgeted funds by T D jD1 Tj , and
Xm h
@hj @hj @hj
b0 D hj (zj0 ) (zj0 )qdj0 (zj0 )p0 (zj0 )Tj0
jD1
@qdj @p @Tj
@hj i X @hj
m X @hjm
(zj0 )pG0 ; b1 D (zj0 ); b2 D (zj0 );
@pG jD1
@qdj jD1
@p
Xm
@hj Xm
@hj
b3 D (zj0 ); b4 D (zj0 ):
jD1
@Tj jD1
@pG
The units of the constants are: b0 W e=kg, b1 W (e week)=kg2 , b3 W

week=kg, and b2 ; b4 are dimensionless. Because the marginal willingness-
to-pay of every consumer is non-negative at every qdj ; Tj ; p; pG , we can
conclude that b0 0 (let qdj ; Tj ; p; pG ; j ! 0 8j in Eq. (5.55)). Because
@hj =@qdj < 0 and @hj =@p < 08j (Sect. 3.7.2), then b1 < 0; b2 < 0.
In the following we assume that pG D pG0 and Tj D Tj0 8j, which
eliminates quantities pG ; T from Eq. (5.55). An approximate average of
the consumers’ marginal willingness-to-pay is then:
1X
m
b0 b1 b2
h(qd ; p) hj (qdj ; p) C 2 qd C p; (5.57)
m jD1 m m m
where
m
X Xm
@hj @hj @hj
b0 D hj (zj0 ) (zj0 )qdj0 (zj0 )p0 ; b1 D (zj0 );
jD1
@qdj @p jD1
@qdj
X
m
@hj
b2 D (zj0 ):
jD1
@p
5 Goods Markets 249
An approximate average of the consumers’ marginal willingness-to-pay

for one kilogram of good k thus linearly depends on the aggregate flow of
consumption of the good and its price.
References
Estola, M., & Hokkanen, V.-M. (2008). Consumer, firm, and price dynamics:
An econophysics approach. An application of economic forces. Saarbrücken,
Germany: VDM Verlag Dr. Müller.
Gibbons, R. (1992). A primer in game theory. Bodmin: Hartnolls Ltd.
Samuelson, P. (1941). The stability of equilibrium: Comparative statics and
dynamics. Econometrica, 9(2), 97–120.
Samuelson, P. (1942). The stability of equilibrium: Linear and nonlinear systems.
Econometrica, 10(1), 1–25.
6
Labor as a Production Factor
§: By the derived demand of production factors we understand that the

demand of inputs of firms depend on the demand of their products. ˘
For example, the demand of labor of a firm producing good k depends
on the flow of production of good k of the firm, and the firm makes
its employment decision on this basis. The demand of labor of a firm
producing good k is thus derived from the demand of good k produced
by the firm.
6.1 A Firm’s Demand of Labor

In this section, we analyze the adjustment of the use of labor of a firm
in the short term. In the short term, firms’ machinery and production
methods are fixed and the only adjustable factor is their use of labor. For
simplicity, the labor input is assumed to consist of only one profession
and the hourly wage of the employees is assumed fixed. In the real world,
firms’ labor consists of workers of different professions and wages, but
here we keep the situation as simple as possible.

DOI 10.1007/978-3-319-46879-2_6
In Sect. 2.9 we defined the production function for a firm producing

good k as a relationship between the flow of production of the firm and its
use of labor. Here we assume that the planning time horizon of the firm is
one week, and we measure the firm’s use of labor in hours per week. The
production function of the firm producing good k is then
qk D fk (Lk ); fk0 (Lk ) > 0; fk00 (Lk ) 0; (6.1)
where the flow of production is denoted by qk (unit=week), the use of

labor of the firm by Lk (h=week), and the production function fk is
assumed continuous and differentiable. The assumptions of the produc-
tion function imply that the marginal productivity of labor is positive
(fk0 (Lk ) > 0), and the law of non-increasing marginal productivity holds
for labor (fk00 (Lk ) 0) (see Sect. 2.9). We denote the price of good k by
pk (e=unit) and the hourly wage by w(e=h). The weekly profit of the firm
is then
…k D pk qk C0 (1 C s)wLk ; qk D fk (Lk ); (6.2)
whereby C0 (e=week) is denoted fixed weekly costs and by 0 < s <

1 the dimensionless constant assumed social security rate of wage. Thus
weekly labor costs consist of wage costs wLk and social security payments
swLk . In Chap. 4 we studied the behavior of a firm adjusting its flow of
production. In profit function (6.2), however, we have taken one step
forward so that now the flow of production of the firm is no longer an
independent quantity, but it depends on the use of labor of the firm. Thus
the firm cannot any more choose its flow of production; only its use of
labor. However, the production function in Eq. (6.1) expresses a unique
relation between the flow of production and the firm’s use of labor. Thus
the firm decides its flow of production indirectly via deciding its use of
labor. In the real world, the relationship between the flow of production
of a firm and its use of labor is uncertain, but in the following we assume,
for simplicity, that no uncertainties exist in the production function of
the firm.
The (physical) marginal productivity of labor, dqk =dLk D fk0 (Lk )

(unit=h), measures the ratio of a change in the flow of production
and a change in the firm’s use of labor. Marginal productivity of labor
measures average productivity from a marginal increase in use of labor
(see Sect. 2.9). The value of marginal productivity of labor is obtained
by multiplying the (physical) marginal productivity by the price of the
product of the firm. Here the value of marginal productivity of labor is
pk fk0 (Lk )(e=h).
Example
Let the price of the product of a firm be 10 (e=unit), and let the marginal
productivity of labor be 100 (unit=h). What is, then, the value of marginal
productivity of labor from: (a) 1 hour, (b) 10 minutes, (c) 8 hours?
Answer.
(a) 10 (e=unit) 100 (unit=h) D 1000 (e=h):

(b) 1000 (e=h) D 1000 (e=60min) D 1000 (e=6 10min)
D 1000=6 (e=10min):
(c) 100 (unit=h) D (8=8) 100 (unit=h) D 800 (unit=8h)I
10 (e=unit) 800 (unit=8h) D 8000 (e=8h): ˘
The costs of the firm are assumed to consist of labor costs and fixed
costs, and the cost function is then Ck D C0 C (1 C s)wLk . Thus we
can no longer define the firm’s marginal costs as dCk =dqk like in Chap. 4,
because now the flow of production does not exist in the cost function.
On the other hand, we can analyze the dependence of weekly costs on the
firm’s use of labor.
§: By unit labor costs we understand the ratio between the costs of a
firm and its use of labor in a time unit. ˘
§: By marginal costs of labor we understand the ratio between a
change in the costs of a firm and a change in its use of labor in a time
unit. Marginal costs of labor measure unit labor costs from a marginal
increase in use of labor. ˘
If the cost function of the firm is continuous and differentiable, the

unit and the marginal costs of labor can be expressed as:
Ck Ck dCk
and lim D :
Lk Lk !0 Lk dLk
The unit and the marginal costs of labor of the firm studied in this section
are:
Ck C0 C (1 C s)wLk C0 dCk
D D C (1 C s)w and D (1 C s)w:
Lk Lk Lk dLk
We assume perfect competition in the industry in which the firm is

operating, and the wage the firm pays to its employees to be fixed.
Thus the firm cannot affect the price of its product or the wage that are
both determined in the corresponding markets. The firm’s use of labor is
optimal, if the firm cannot increase its weekly profit by changing its use
of labor. Mathematically, this can be expressed as:
d…k (1 C s)w
D 0 , pk fk0 (Lk ) D (1 C s)w , fk0 (Lk ) D : (6.3)
dLk pk
The firm’s use of labor is optimal if the value of marginal productivity of

labor is equal to the marginal costs of labor. This condition is presented
in Eq. (6.3) also with real quantities so that the (physical) marginal
productivity of labor is equal to the real hourly labor costs; that is, the
nominal hourly labor costs deflated by the price of the product of the
firm (see Sect. 2.3).
Next we study the force the firm directs upon the aggregate labor time
of the profession of skills the firm is employing. A profit-seeking firm
adjusts its use of labor with time so that this increases its weekly profit.
If the price of the product of the firm and the wage paid by the firm are
fixed, we can study the firm’s use of labor as a function of time Lk (t) as
follows. The time derivative of the weekly profit function of the firm in
Eq. (6.2) is:
@…k 0
…0k (t) D Lk (t) D pk fk0 (Lk .t)/ (1 C s)w Lk0 (t):
@Lk
The adjustment rules for the use of labor, that make the weekly profit of
the firm increasing with time (…0k (t) > 0), are:
Lk0 (t) > 0 if pk fk0 .Lk (t)/ (1 C s)w > 0;

Lk0 (t) < 0 if pk fk0 .Lk (t)/ (1 C s)w < 0;
Lk0 (t) D 0 if pk fk0 .Lk (t)/ (1 C s)w D 0:
Quantity pk fk0 .Lk (t)/(e=h) measures the value of the marginal pro-
ductivity of labor, and we can interpret such as this firm’s marginal
willingness-to-pay for one hour of labor time of this profession. Wage
and social security payments, on the other hand, are the costs of one
hour of work time for the firm. Similarly as a consumer compares his
marginal willingness-to-pay for a good and its price, a firm compares its
marginal willingness-to-pay for one hour of work time and its costs. A
profit-seeking firm increases its use of labor if its marginal willingness-to-
pay for one hour exceeds the hourly labor costs at the prevailing use of
labor, and vice versa.
Quantity Fd D pk fk0 .Lk (t)/ (1 C s)w (subindex d refers to demand)
can thus be named as the force this firm directs upon the weekly work
time of the profession. The explanation for this is that Fd causes the
acceleration of use of labor of the firm: Lk0 (t)(h=week2 ). Force Fd with unit
e=h consists of the revenues and costs the firm compares in its decision-
making concerning its use of labor. According to the law of non-increasing
marginal productivity, marginal productivity of labor is non-increasing
with increasing use of labor. This makes the force non-negative at small
and non-positive at large uses of labor. In the profit maximizing situation,
the force acting upon the use of labor vanishes:
pk fk0 .Lk (t)/

Fd D 0 , pk fk0 .Lk (t)/ D (1 C s)w , w D : (6.4)
1Cs
Equation (6.4) corresponds to the equilibrium (optimal) situation of the

firm.
Equation w D pk fk0 .Lk (t)/=(1 C s), that defines the equilibrium use
of labor of the firm, is called the demand relation of labor of this firm.
Fig. 6.1 The demand of labor of a firm
The term ‘relation’ is used because actually the last form of Eq. (6.4) is the
inverse demand function of labor. The corresponding demand function
would be of form Lk D f (w; s), which would require solving L from
Eq. (6.4) by taking the inverse function of fk (Lk ). This is not necessary. We
can generally call Eq. (6.4) labor demand relation, which term means that
the equation relates quantities Lk ; w together. It shows the firm’s optimal
use of labor at different wage levels: see Fig. 6.1. The slope of the labor
demand function in coordinates (Lk ; w),
dw pk fk00 (Lk )
D 0;
dLk 1Cs
depends on the non-increasing marginal productivity of labor. From the

demand relation of labor we see that an increase in price pk moves the
labor demand relation upward (or ‘right’) in coordinate system (Lk ; w).
Similarly, and increase in social security rate s moves the labor demand
relation downward (‘left’). The use of labor of the firm thus depends on
the price of the product of the firm, which reflects the ‘derived demand’
character of labor.
6.2 Newtonian Theory of Use of Labor*

Applying the force acting upon the use of labor of a firm defined in
the previous section, we can write an equation of motion for the labor
input as:
Lk0 (t) D g(Fd ); g0 (Fd ) > 0; g(0) D 0; Fd D pk fk0 .Lk (t)/ (1 C s)w;
(6.5)
where g is a function with the above characteristics. Function g expresses
the relationship between the acceleration of use of labor Lk0 (t)(h=week2 )
and the force acting upon the labor input of the firm. This relationship is
asymptotically stable (Sect. 3.8), if the following inequality holds strictly:
@Lk0 (t) @Fd pk f 00 (Lk )
D g0 (Fd ) D g0 (Fd ) k 0:
@Lk @Lk (1 C s)
Decreasing marginal productivity of labor, fk00 (Lk ) < 0, is thus a sufficient

condition for stability. Taking the Taylor series approximation of function
g(Fd ) in the neighborhood of point Fd D 0, assuming the error term zero
and denoting mLk D 1=g0 (0) > 0, Eq. (6.5) takes the form:
pk fk0 .Lk (t)/ (1 C s)w
mLk Lk0 (t) D pk fk0 .Lk (t)/ (1 C s)w , Lk0 (t) D :
mLk
(6.6)
This is the Newtonian equation of motion for the labor input of the firm.
According to Eq. (6.6), the firm increases its use of labor if the value of
marginal productivity of labor exceeds hourly labor costs and vice versa.
We could have also added static friction in Eq. (6.6) to explain that the
firm may not always change its use of labor when the force acting upon
it deviates from zero. This, however, is omitted as well as the finding of
solutions for the differential equation in Eq. (6.6). If an exact functional
form is assumed for function fk (Lk ), the solution of Eq. (6.6) defines the
time path for Lk (t).
If the equation of motion for the use of labor in Eq. (6.6) is solved and
substituted in the firm’s production function, the time path of production
can be determined. Even though the exact form of the production
function is not known, and the equation of motion for labor is not solved,
the acceleration of production can still be modeled by taking the time
derivative of the production function and substituting there Eq. (6.6) as:
Œpk fk0 .Lk (t)/ (1 C s)w
q0k (t) D fk0 .Lk (t)/Lk0 (t) D fk0 .Lk (t)/ :
mLk
The dynamics of production is thus determined by the firm’s use of labor.

6.3 Labor Supply of a Person

Here we analyze the labor supply of a person according to his choice
between work time and leisure time on the basis of the benefits and costs
of this decision. The decision-making situation is analogous to that of a
consumer in Chap. 3. A labor supplier makes his decision to supply work
time on the basis of his preferences concerning working and having leisure
time. The compensation from work (wage or salary) and the available time
restrict the choices of the labor supplier, similarly to how budget equation
limits the choices of a consumer.
In most countries, the laws concerning work time limit the daily and
weekly work time of an individual. For this reason, the planning time
horizon of a labor supplier is assumed long enough so that the number of
working hours can be considered as an adjustable quantity. Due to this,
the planning time horizon of a labor supplier is assumed to be one year.
The labor supplier ‘chooses’ the number of hours in a year he would be
willing to work at the hourly wage he believes to be receiving on average
during the year.
We denote the hourly wage as w(e=h), the constant assumed income
tax rate by dimensionless number (Greek letter tau), 0 < < 1, the
number of working hours in a year by L(h=y), and the annual after-tax
wage income by T(e=y). Then T D wL wL D (1 )wL, where wL
is annual gross wage income and wL annual income taxes. The maximal
possible number of working hours in a year is calculated as follows: work
5 days in a week for 8 hours per day, minus annual holiday of 4 weeks.
The maximal number of annual working days is then: 48 (week=y)
5 (d=week) D 240 (d=y) and the maximal number of annual working
hours is approximately 240 (d=y) 8 (h=d) D 1920(h=y). The labor
supplier has thus leisure time in a year H (h=y) D 1920 (h=y) L (h=y)
together with weekends, evenings, and nights.
If the labor supplier can work the amount of hours he likes in a year
at wage w, the possible choices for the labor supplier can be expressed by
the following equation between annual wage income and leisure time
T D (1 )wL D (1 )w(1920 H); (6.7)

Fig. 6.2 The possibilities of a labor supplier
where both sides of the equation are measured in units e=y; notice that
possible holiday time payments are omitted from this analysis.
The maximal annual after-tax wage income of the labor supplier is:
T D 1920(1 )w. Annual wage income is zero if L D 0, and then
H D 1920 (h=y). The possible choices of the labor supplier are shown in
Fig. 6.2. In coordinate system (H; T), the slope of the ‘budget line’ of the
labor supplier is dT=dH D (1 )w < 0; the greater the w, the steeper
the line.
A labor supplier enjoys income and leisure time, and the law of non-
increasing marginal utility is assumed to hold for both these ‘goods’. The
utility function of a labor supplier could be derived from a set of axioms
like that of a consumer. This is omitted, however, and we assume that the
utility of a labor supplier is measured by continuous and differentiable
function
@u(H; T) @u(H; T) @2 u(H; T)

u D u(H; T); > 0; > 0; 0;
@H @T @H 2
@2 u(H; T) @2 u(H; T) @2 u(H; T)
0; D ;
@T 2 @T@H @H@T
where the first order partials are the corresponding marginal utilities,
the non-positivity of the second order partials with respect to the same
quantity imply non-increasing marginal utility, and the second order cross
partials are equal due to the assumed continuity of the partial functions
(see Sect. 3.7).
The arguments of the utility function of a labor supplier are annual
after-tax wage income and leisure time. Thus this utility consists of
different factors than that of a consumer who gains utility from the
consumption of goods. The satisfaction from leisure time can, though,
be understood to be gained from the ‘consumption’ of leisure time, but
receiving wage income can hardly be understood as consumption. In this
decision-making, the labor supplier compares the satisfaction from leisure
time and its alternative cost, the lost income due to not working. In this
way the situation resembles that of a consumer. The measurement unit
of utility again does not have an essential role in the modeling. Utility is
an auxiliary quantity needed in defining the marginal willingness-to-pay
of the labor supplier for leisure time. Thus we set measurement unit ut=y
for the utility of a labor supplier.
The preferences of a labor supplier can be described by indifference
curves analogously with those of a consumer. One indifference curve
represents constant utility, and the slope of an indifference curve in
coordinates (H; T) can be derived as in Sect. 3.5:
@u(H;T)
@u @u dT @H
du D 0 , dH C dT D 0 ) D @u(H;T) < 0:
@H @T dH
@T
The law of non-increasing marginal utility defines the shapes of the

indifference Curves: see Fig. 6.3. The greater the annual leisure time of the
labor supplier, the smaller his marginal utility @u(H;T)
@H
and the less steep the
curve. The greater the annual after-tax wage income of the labor supplier,
the smaller his marginal utility @u(H;T)
@T
and the more steep the curve.
The equilibrium situation of the labor supplier is presented in Fig. 6.3.
The equilibrium defines the combination of annual after-tax income
and leisure time (H ; T ) that corresponds to the optimal annual work
time L D 1920 H at wage w. The slopes of the ‘budget’ line
and an indifference curve are as defined, and they both have unit e=h.
The former measures the objective exchange rate between annual wage
Fig. 6.3 The equilibrium state of a labor supplier
income and leisure time, and the latter the subjective rate at which the
labor supplier is willing to exchange these two quantities. In the optimal
situation, an indifference curve touches the ‘budget’ line and at this point
the slopes of the curves are equal.
The optimum condition can be expressed as a two equation system:
T D (1 )wL and
@u(H;T) @u(H;T)
@H @H
(1 )w D @u(H;T) , (1 )w D @u(H;T)
;
@T @T
which is analogous with the optimum condition for a consumer. The

after-tax hourly wage (1 )w is the objective exchange rate between
income and leisure time; it measures the alternative cost for one hour of
leisure time in the form of lost income. Quantity @u(H;T)
@H
= @u(H;T)
@T
with unit
e=h, on the other hand, measures the marginal willingness-to-pay of the
labor supplier for one hour of leisure time; that is, the amount of income
the labor supplier is willing to give up for one extra annual hour of leisure
time. The marginal willingness-to-pay of the labor supplier for one annual
hour of leisure time is the greater the higher his marginal utility of leisure
time, and the smaller his marginal utility of annual income at prevailing
levels of annual leisure time and income.
Fig. 6.4 The effect of wage change on labor supply
The reactions of the labor supplier to changes in wage are studied in

Fig. 6.4. An increase in wage makes the ‘budget’ line steeper so that it
twists with Hmax D 1920(h=y) as the fixed point. Suppose that the wage
increases first from w0 to w1 and then from w1 to w2 . The optimal choices
of the labor supplier at wages w0 ; w1 ; w2 are E0 ; E1 ; E2 , respectively. These
correspond to annual amounts of leisure time H0 , H1 and H2 , and annual
working times L0 D 1920H0 , L1 D 1920H1 and L2 D 1920H2 .
Thus the supply of annual work time increases with wage at low wage
levels, but after wage and annual income have increased enough, a wage
increase may lead to a decrease in the supply of labor time (see Fig. 6.4).
This results because then equal annual wage income is received at lower
work time.
We can also define the force acting upon the labor supply of the person.
We solve the ‘budget’ equation as T(t) D (1 )wL(t), use equation
H(t) D 1920 L(t), and substitute these in the utility function. The
annual work and leisure time as well as income are assumed to depend
on time t so that we can analyze their adjustment with time. The utility
function then becomes the following:
u D u.T(t); H(t)/ D u.(1 )wL(t); 1920 L(t)/;
where the ‘budget’ equation that restricts the choices of the labor supplier
is included in the utility function. At fixed wage and tax rates, annual
work time L(t) is the only quantity by which the labor supplier can affect
his utility. The time derivative of the utility function is

@u 0 @u @T @u @H @u @u
u0 (t) D L (t) D C L0 (t) D (1 )w L0 (t);
@L @T @L @H @L @T @H
where @T
@L
D (1 )w and @H @L
D 1 are obtained from the relations
substituted in the utility function. The changes in the annual work time,
that make the acceleration of utility, u0 (t)(ut=y2 ), positive, are:
@u
@u @u
L0 (t) > 0 if (1 )w > 0 , (1 )w @H
@u
> 0;
@T @H @T
@u
@u @u
L0 (t) < 0 if (1 )w < 0 , (1 )w @H
@u
< 0;
@T @H @T
@u
@u @u
L0 (t) D 0 if (1 )w D 0 , (1 )w @H
@u
D 0:
@T @H @T
As in Sect. 3.8, either of the quantities F1 or F2 ,
@u @u
F1 D (@u=@T)(1 )w @u=@H or F2 D (1 )w = ;
@H @T
can be used as the force this labor supplier directs upon the labor
supply of the profession he represents. Either of these forces cause
the acceleration of labor supply of this person: L0 (t)(h=y2 ). The latter
form of the force consists of the alternative cost of one hour of leisure
time, (1 )w(e=h), minus the marginal willingness-to-pay of the labor
@u @u
supplier for one hour of leisure time, @H = @T (e=h). The equilibrium state
@u @u
(1 )w D @H = @T corresponds to zero force. In the equilibrium, the
after-tax wage and the marginal willingness-to-pay of the labor supplier
for one hour of leisure time are equal. This relation is presented in Fig. 6.5
@u @u
in the form of w D (1=(1 )) @H = @T , and it can be interpreted as the
supply relation of labor of this person.
The labor supply relation (inverse supply function) in Fig. 6.5 defines
the optimal annual work time of the person at different wages. The choice
Fig. 6.5 The labor supply relation of a person
of the labor supplier is restricted by the equation that relates the annual
after-tax wage income and leisure time. This constraint may make the
labor supply relation ‘backward bending’ at high wage levels and annual
work time, which is denoted by the dotted part of the relation in Fig. 6.5.
The slope of the labor supply relation (1 )w @u=@T D @u=@H in
coordinate system (L; w) is:
@ u2 @ u 2
2 2@ u 2
dw 2(1 )w @H@T @H 2 (1 ) w @T 2
D @u @2 u @2 u
: (6.8)
dL (1 ) @T (1 )L @H@T C (1 )2 wL @T 2
The result in Eq. (6.8) can be derived by taking the total differential
(Sect. 10.10.2) so that other quantities in equation (1 )w @u=@T D
@u=@H except w and L are treated as constants. Equation (1 )w
@u=@T @u=@H D 0 is then of the general form f (w; L) D 0 from
which we get by totally differentiating:
@f
@f @f dw
dw C dL D 0 , D @L
@f
: (6.9)
@w @L dL
@w
Formula (6.8) deviates from that in Eq. (6.9) so that the partial derivatives
in Eq. (6.9) are written in a complete form in Eq. (6.8). The exact
derivation of the slopes of curves in graphical presentations hopefully
motivates a serious student of economics to learn the mathematical
techniques needed in this (see Sect. 10.13).
From formula (6.8)—the slope of the labor supply function in coordi-
nates (L; w)—we see that its numerator is positive if @2 u=@T@H is positive,
because the law of non-increasing marginal utility makes the two other
additive terms in the numerator non-negative. Quantity @2 u=@T@H can
be assumed to have a small absolute value because an increase in leisure
time may not have a great effect on the marginal utility of income, or an
increase in annual income may not have a great effect on the marginal
utility of leisure time. The sufficient condition for maximal utility is that
@2 u=@T@H > 0. The sign of the numerator is thus ambiguous but the
positive sign is more plausible.
The first additive term in the denominator of Eq. (6.8) is positive, the
second is ambiguous, and the last is non-positive. The greater the annual
income T, the smaller the marginal utility @u=@T. The denominator may
thus be negative; namely, the labor supply relation may be decreasing
in coordinate system (L; w) (the backward bending part of the relation
corresponds to dw=dL < 0). The probability of this is the higher the
greater is T.
In the following we assume that @2 u=@T@H > 0, which guarantees the
existence of an optimum for the labor supplier. The following results are
obtained from the marginal willingness-to-pay
@u @u of the labor supplier for
one hour of leisure time, g(L; w; ) @H = @T :
2 2
@ u @2 u @u @u @ u @2 u
@g @H 2 C @T@H (1 )w @T
@H @T 2
(1 )w @H@T
D @u 2 > 0;
@L
@T
2
@ u @u @u @ u2
@g (1 )L @T@H @T
@H @T 2
D > 0;
@w @u 2
@T
2
@ u @u @u @2 u
@g wL @T@H @T
@H @T 2
D @u 2 < 0:
@ (1 ) @T
These results imply that the labor supplier is the more willing-to-pay for
one hour of leisure time the higher are w and L, and the smaller is .
6.4 Newtonian Theory of Labor Supply*

According to the force a single labor supplier directs upon the labor
supply of his profession defined in the previous section, we can present
the following equation of motion for his labor supply:
@u
L0 (t) D f (Fs ); f 0 (Fs ) > 0; f (0) D 0; Fs D (1 )w @H
@u
; (6.10)
@T
where f is a function with the above properties. Function f expresses the

relation between the acceleration of labor supply L0 (t) and the force Fs
acting upon it (s refers to supply). This relation is asymptotically stable, if
2 3
@u @2 u @2 u @u @2 u @2 u
0
@L (t) @T @H 2
(1 )w @H@T C @H (1 )w @T 2 @T@H
D f 0 (Fs ) 4 @u 2 5
@L
@T
is negative. The sufficient condition for stability is thus @2 u=@H@T > 0,

which is the sufficient condition for maximal utility of the labor supplier.
Taking the first order Taylor series approximation of function f in
Eq. (6.10) in the neighborhood of point Fs D 0, assuming the error-
term zero and denoting the ‘inertial "mass" of labor supply’ by mL D
1=f 0 (0) > 0, we get:
!
@u @u
0 @H 0 1 @H
mL L (t) D (1 )w @u
, L (t) D (1 )w @u
: (6.11)
@T
mL @T
This is the Newtonian equation of motion for the labor supply of this
person. According to Eq. (6.11), the annual labor supply of this person
increases if the after-tax wage income from one work hour is greater than
the value of one hour of leisure time for the person, and vice versa. We
could have also added static friction in Eq. (6.11) to explain that the labor
supply of this person is not always changed when the force acting upon
it deviates from zero. This, however, is omitted as well as the finding of
possible solutions for Eq. (6.11).
6.5 Atomistic Labor Market

In Sect. 6.6 we will analyze the determination of wage and employment
of a profession in a region where all labor suppliers are members of one
trade union, and the union operates as the wage setter in the labor market.
Here we analyze, on the other hand, the determination of wage and
employment of a profession in a region where every person and firm
employing these workers behave independently. No trade union exists in
the labor market that would participate in the wage negotiation, and no
minimum wage exists for the profession. The differences between these
two labor market situations correspond to those between industries with
different kind of competition.
The labor suppliers of the profession we study here are assumed to work
at the same wage w, but differences may exist in personal exchange rates
between annual wage income and leisure time. Firms employing the labor
of this profession may deviate in their production methods expressed by
their production functions. If every labor supplier and demander operates
separately, and both types of partners are numerous, perfect competition
prevails in the labor market. One supplier (trade union) situation in a
labor market would correspond to a monopoly firm in an industry.
§: By the demand of labor of a profession in a region we understand
the aggregate uses of labor of this profession in a time unit with different
wages that correspond to the equilibrium states of every firm in the region
employing labor of this profession. ˘
§: By the supply of labor of a profession in a region we understand
the aggregate work time of laborers of this profession in a time unit at
different wages that correspond to the equilibrium states of every labor
supplier of this profession in the region. ˘
The above definitions assume that labor markets work locally; a limited
area exists around firms and homes of labor suppliers which define the
maximum distance for daily working. In the above definitions we talked
generally about labor and wage, because the use of labor can be measured
in different units. For example, if the use of labor is measured by the
number of employees working in a time unit, the unit price of labor is
the salary of one employee in the time unit. However, in the following
we measure firms’ use of labor by worked hours in a time unit; unit labor
costs are then hourly wage plus social security payments per one hour.
Our analysis here exactly corresponds to that in Sect. 5.2.7, and so
the presentation is a little shorter. Let perfect competition prevail in the
product market of every firm employing the studied type of labor as well
as in the labor market. Let there be n firms employing the type of labor we
study and m labor suppliers. The planning time horizon of every firm and
labor supplier is assumed to be one year. The modeling of firms’ behavior
is based on the assumption that firms try to increase their annual profit by
adjusting their use of labor. Labor suppliers are, similarly, assumed to seek
utility by changing their annual work time. A firm is assumed to increase
its annual use of labor if it believes that the revenues from production
during the extra work time exceed the extra labor costs, and vice versa.
Similarly, an individual labor supplier is assumed to increase his annual
work time, if the extra income received from this exceeds its alternative
cost, the decrease in annual leisure time and vice versa.
The profit and utility functions of firms and labor suppliers are
assumed continuous and differentiable, and both parties are assumed to
consider changes in their annual use of labor and work time. With these
assumptions, we can define the forces by which the firms and the labor
suppliers act upon the annual aggregate work time of this profession.
6.5.1 The Demand of Labor
Suppose the flows of production of firms are measured in units kg=y. The
profit of firm i producing a single good is then
…i (t) D pi fi .Ldi (t)/ Ci0 (1 C s)w(t)Ldi (t);

where fi is the production function of the firm, pi (e=kg) the price of the
product of the firm, Ci0 (e=y) the annual fixed costs of the firm, w(e=h)
the wage of the employed labor, 0 < s < 1 the dimensionless social
security rate of wage, and Ldi (h=y) the annual use of labor of the firm
(subindex d refers to ‘demand’). We assume that Ldi and w depend on
time t, because later we will analyze their adjustment with time.
The labor costs of the firm consist of the wage costs wLid and social
security payments swLid the firm pays to the state. The force by which
firm i acts upon the annual work time of the studied profession is:
@…i
D pi fi0 .Lid (t)/ (1 C s)w(t):
@Lid
This force measures the firm’s marginal profitability of its use of labor.
The rationale for the force interpretation is as before; the greater the above
quantity, the more profitable it is for the firm to increase its use of this
type of labor. The profit-seeking firm increases its use of this type of labor
if the above force is positive, and vice versa. For simplicity, firms’ use of
labor is assumed not to have a static friction, which assumption does not
hold in the real world.
In the following we analyze the forces directed by all firms using, and
by all labor suppliers supplying. the labor upon the annual work time of
the profession in a region. In practice, this can be assumed to take place
via the employment office in the region, where all labor suppliers and
demanders express their willingness to supply and demand work. The role
of the office is to guide potential workers to the firms interested in hiring
them, in which way employment offices work in real life.
We assume n firms employing the studied type of labor in the region.
At time moment t, the average force acting upon the firms’ use of labor is
1X
n
FdL D pi fi0 .Lid (t)/ (1 C s)w(t) D fd .Ld (t); p/ (1 C s)w(t);
n iD1
(6.12)
P
whereby fd .Ld (t); p/ (1=n) niD1 pi fi0 (Lid (t)) is denoted the average
value of marginal productivity of labor for the firms at their aggregate
P
annual use of the labor Ld D niD1 Lid . Suppose the production of every
firm is measured in units kg=y. The prices of the products then have unit
e=kg, and by p we denote the arithmetic average of these prices; see the
Appendix of this chapter.
6.5.2 Adjustment of Labor Demand
Firms employing the studied profession adjust the aggregate use of labor as
Ld0 (t) D Gd (FdL ); G0d (FdL ) > 0; Gd (0) D 0; FdL D fd .Ld (t); p/ (1Cs)w(t);
(6.13)
where Gd is a function with the above characteristics. From Eq. (6.13) we
get @Ld0 (t)=@Ld (t) D G0d (FdL ) @fd =@Ld , and according to the Appendix
of this chapter, this partial is non-positive. The adjustment of labor
demand is thus stable if @fd =@Ld < 0, see Fig. 3.9.
6.5.3 The Supply of Labor
The supply of labor of the studied profession can be analyzed by the

average willingness of the labor suppliers in the region to increase their
annual work time. The average force the m labor suppliers direct upon
their annual work time at time moment t is
0 @uj
1 @uj
1 Xm Xm
FsL D @(1 )w(t) @Hj
A D (1 )w(t) 1 @Hj
m jD1 @uj m jD1 @uj
@Tj @Tj
D (1 )w(t) fs .Ls (t); w(t); /; (6.14)
where the annual leisure time and wage income P of labor supplier j is
denoted by Hj and Tj , respectively, and Ls D m jD1 Ljs is the aggregate
annual work time of all labor suppliers (subindex s refers to ‘supply’).
By fs (Ls ; w; ) we denote the average marginal willingness-to-pay of the
m labor suppliers for one hour of leisure time; see the Appendix of this
chapter. According to Eq. (6.14), on average the labor suppliers are willing
to increase their annual work time, if the after-tax hourly wage exceeds
their average marginal willingness-to-pay for one hour of leisure time,
and vice versa.
6.5.4 Adjustment of Labor Supply
The labor suppliers are assumed to adjust the aggregate labor supply as
Ls0 (t) D Gs (FsL ); G0s (FsL ) > 0; Gs (0) D 0;

FsL D (1 )w(t) fs .Ls (t); w(t); /; (6.15)
where function Gs obeys the above characteristics. From Eq. (6.15) we get
@Ls0 (t)=@Ls (t) D G0s (FsL ) @fs =@Ls , and according to the Appendix of
this chapter, this partial is negative. The adjustment of labor supply is thus
stable, see Fig. 3.9.
6.5.5 The Force Acting upon Employment
The resultant force acting upon the annual work time of a profession is
defined as the sum of the force components created by labor demanders
and suppliers. Both force components in Eq. (6.12) and in Eq. (6.14) are
measured in units e=h, and so we can add them. The resultant force is
then:
FL D fd .Ld (t); p/ (1 C s)w(t) C (1 )w(t) fs .Ls (t); w(t); /

D fd .Ld (t); p/ ( C s)w(t) fs .Ls (t); w(t); /: (6.16)
The rationale for this resultant force is the same as earlier; the two force
components are caused by independent partners, and they both affect the
annual aggregate work time of the profession.
Let us assume, for a moment, that the wage tax and social security rates
of wage, and s, are zero. Then a positive force is acting upon the annual
work time of the profession, if the average marginal willingness-to-pay
of firms for one hour of annual work time exceeds the average marginal
willingness-to-pay of labor suppliers for one hour of annual leisure time,

and vice versa. Thus, if firms on average are willing to pay a wage that
exceeds the average value of leisure time for the labor suppliers, a positive
force is acting upon the aggregate annual work time of the profession, and
vice versa.
In most countries, and s are positive. The resultant force in Eq. (6.16)
then shows that and s negatively affect the force acting upon the
aggregate annual work time of the profession. The reason for this is that
a firm hiring labor pays a different cost of labor than employees receive
from their working. Quantity ( C s)w is called a tax wedge, because it is
a ‘wedge’ between the price paid by firms and received by employees for
one hour’s work. The resultant force shows that positive ; s decrease the
aggregate annual work time of the profession because they decrease firms’
marginal willingness to hire employees and workers’ marginal willingness
to work.
Note. The result that wage tax and social security rate decrease employ-
ment, is based on partial analysis. To find out their total effect on
employment, we should take account as to how public sector uses the
money it collects with these two forms of taxes. ˘
6.5.6 Wage Adjustment
In a perfectly competed labor market, wage adjusts with time according

to the excess demand or supply of labor, similarly as price adjusts in a
perfectly competed industry. The equation of motion for wage is:
w0 (t) D Gw (Ld (t) Ls (t)); G0w (Ld Ls ) > 0; Gw (0) D 0; (6.17)
where Gw is a function with the above characteristics. According to

Eq. (6.17), wage increases if annual aggregate demand of work time
exceeds annual supply, and vice versa. The explanation for Eq. (6.17) is the
following. If every labor supplier is working the annual amount of hours
he/she prefers, and firms like to increase their use of labor, the only way to
entice employees to increase their work time is to raise the wage offer. If,
on the other hand, labor supply is greater than demand at the prevailing
wage, and a labor supplier likes to increase his annual work time, he can
do this by decreasing his wage requirement because a decrease in wage
positively affects firms’ demand of labor.
Equation (6.17) corresponds to the principle of modeling in economics,
because in the excess demand situation of labor, firms can increase their
profit by increasing their use of labor, which takes place only by increasing
their wage offer. In a situation where there is an excess supply of labor,
labor suppliers can increase their utility by increasing their annual work
time, which takes place similarly by decreasing their wage requirement.
6.5.7 Equilibrium in Atomistic Labor Market
An equilibrium state in the atomistic labor market is shown in Fig. 6.6.

There w D fd (Ld ; p)=(1 C s) is the labor demand relation (inverse
demand) that corresponds to the equilibrium state of every firm employ-
ing this labor at different wages. The labor supply relation (inverse
supply) w D fs (Ls ; w; )=(1 ), on the other hand, expresses the
equilibrium state of every labor supplier of the profession in the region
at different wages. In Fig. 6.6 on the horizontal axis is measured Ld and
Ls in units h=y, and on the vertical axis is measured wage and quantities
fd (Ld ; p)=(1 C s) and fs (Ls ; w; )=(1 ) with unit e/h at different levels
of Ld ; Ls , respectively. In the equilibrium state holds:
w D fd (Ld ; p)=(1 C s) D fs (Ls ; w; )=(1 ) and Ld D Ls :
§: In the equilibrium state of an atomistic labor market, the forces

acting upon the demand, the supply, and the wage of the profession
vanish. ˘
Note. The labor demand and supply relations in Fig. 6.6 may not corre-
spond to the annual uses of labor of all firms and the annual amounts
of work time of all labor suppliers at a certain moment of time, but they
describe the aggregate amounts of annual work time in the region when
every firm and labor supplier are in their equilibrium state. The stability
analysis given earlier show, however, that both parties adjust their behavior
with time to reach their equilibrium states. ˘
Fig. 6.6 The labor supply relation of a person
The two non-equilibrium situations in Fig. 6.6 can be analyzed as in

Chap. 5. At wage w1 , excess demand Ld1 Ls1 > 0 of labor prevails, and
at wage w2 , excess supply of labor Ls2 Ld2 > 0 prevails. Equation (6.17)
shows that these excess demand and supply situations change the wage so
that the labor market settles into its equilibrium. Because wage adjustment
according to excess demand guarantees that the labor market reaches its
equilibrium state with time, the system is stable. The system of differential
equations describing an atomistic labor market exactly corresponds to that
in Sect. 5.2.8, and it is studied in this chapter in Sect. 6.5.9.
6.5.8 An Approximation of the Equilibrium
In this section, we separate supply and demand by subscripts s; d. Accord-

ing to previous sections, when every firm and labor supplier is in their
optimal situation, we have
@uj
@Hj
(1 C s)w D pi fi0 (Ldi ) and (1 )w D @uj
; (6.18)
@Tj
i D 1; : : : n and j D 1; : : : ; m. Adding the n and m equations in

Eq. (6.18), separately, and dividing the results by n and m, respectively,
we get
@uj
1X 0 1X
n m
@Hj
(1 C s)w D pi fi (Lid ); (1 )w D @uj
: (6.19)
n iD1 m jD1
@Tj
This corresponds to the neoclassical equilibrium in the labor market.

In the Appendix of this chapter we approximate the average value of
firms’ marginal productivity of labor as
1X 0
n
a0 a1 a2
F(Ld ; p) pi fi (Lid ) C 2 Ld C p;
n iD1 n n n
X
n
1X
n
Ld D Lid ; pD pi ; (6.20)
iD1
n iD1
where constants a0 > 0; a1 < 0; a2 > 0 have units e=h, (e y)=h2 and
kg=h, respectively; see the Appendix of this chapter.
In the Appendix of this chapter, we approximate the labor suppliers’
average marginal willingness-to-pay for one hour of leisure time as
@uj
1X
m
@Hj b0 b1 b2 b3
G(Ls ; w; ) @uj
C 2 Ls C w C ;
m jD1 m m m m
@Tj
wherePthe aggregate labor supply of the m labor suppliers is denoted by

Ls D m jD1 Lsj , and constants b0 > 0; b1 > 0; b3 < 0 have units e=h,
(e y)=h2 and e=h, respectively; b2 > 0 is dimensionless.
The equilibrium state in the labor market can then be approximated as:
a0 a1 a2
Labor demand: (1 C s)w D C 2 Ld C p; (6.21)
n n n
b0 b1 b2 b3
Labor supply: (1 )w D C 2 Ls C w C : (6.22)
m m m m
Assuming pi D pi0 8i, we can eliminate p from System (6.21)–(6.22);

see the Appendix of this chapter. Then setting Ld D Ls , we can solve the
equilibrium state in the labor market (w ; Ld D Ls ) as:
a0 b1 n a1 m(b0 C b3 )
w D ;
a1 mŒb2 C ( 1)m C (1 C s)b1 n2
mnfa0 Œ(1 )m b2 (1 C s)(b0 C b3 )ng
Ld D Ls D : (6.23)
a1 mŒb2 C ( 1)m C (1 C s)b1 n2
The units of the constants can be used to check that the solutions are
dimensionally well-defined, namely the units of w , Ld are e=h, h=y,
respectively.
6.5.9 Labor Market Adjustment in Detail*
In Sects. 6.5.2–6.5.7 we presented the equations of motion for labor

demand, supply, and wage in a perfectly competed labor market. By taking
the Taylor series approximations of functions Gc in Eqs. (6.13), (6.15)
in the neighborhood of the equilibrium points Fc D 0, c D d; s,
and assuming the error terms zero, we can approximate the functions as
Gc (Fc ) D G0c (0)Fc , where G0c (0) are positive constants. Then, denoting
G0c (0) D 1=mLc , c D d; s, we can interpret these constants as ‘inverses
of the inertial "masses" mLc of the adjusting quantities’. These inertial
‘masses’ measure the factors resisting changes in the adjusting quantities,
and their measurement unit is e(y=h)2 when time t has unit y. Constant
G0w (0) in Eq. (6.17), on the hand, is denoted as G0w (0) D kw , where kw
with unit e=h2 is the ‘spring constant’ that measures the sensitivity of
wage to the excess demand of labor Ld Ls , see Sect. 5.2.7. These units
make the equations dimensionally homogeneous. The factors resisting
wage changes are the existing wage contracts, costs from renegotiation of
wage contracts, and so forth. The value of constant kw depends on these
factors.
fs(Ls, w, t) w w ¯
fd (Ld ,p)
mks mkd
tw SW
∫ Ld (t)dt, ∫ Ls (t)dt
Fig. 6.7 The forces acting upon wage and demand and supply of labor
The equations of motion for labor demand, supply and wage are then:
mLd Ld0 (t) D fd (Ld ; p) (1 C s)w;

mLs Ls0 (t) D (1 )w fs (Ls ; w; ); (6.24)
1 0
w (t) D Ld (t) Ls (t): (6.25)
kw
Notice that we could have also added static friction for wage in Eq. (6.25)
to explain the rigidities that may prevent wage adjustment in some
cases according to excess demand and supply. However, we assume, for
simplicity, that wages do not have static friction which assumption does
not hold in the real world.
The ‘free body’ diagram of the forces acting upon the demand and
supply of labor is in Fig. 6.7, where the static friction forces are omitted for
simplicity. The direction of motion to the right—where the accumulated
amounts of demanded and supplied labor time increase—is defined
positive, and to the left, negative. In Fig. 6.7, the system cannot move
to left direction, because the flows of demanded and supplied labor time
Ld ; Ls cannot be negative. The positive force component acting upon
the demand of labor is fd (Ld ; p), and the negative force consists of two
components; wage w and social security payments sw for one hour. Both
these cost components are paid by firms, and employees receive the wage
while government receives the social security payments. The positive force
component acting upon the supply of labor is wage w, and the negative
force components consist of average marginal willingness-to-pay of labor
suppliers for one hour of leisure time fs (Ls ; w; ) and income tax from one
hour’s wage, w. As earlier, the shapes of the ‘particles’ in Fig. 6.7 have no
economic meaning, and actually the particles should be drawn as a points
on the horizontal axis. However, the box shapes for the variables better
visualizes the analogy we make here with Newtonian mechanics.
In Fig. 6.7, the spring between the two ‘particles’ reflects the matter
that an equal mutual force w is acting upon the two particles, and the
difference in velocities of the particles Ld Ls changes their distance and
so also the length of the spring that changes the mutual force w. This is
seen in wage equation (6.25) that corresponds exactly to the price equation
given in Sect. 5.2.7. Thus wage with unit e/h is a force component that
acts upon the aggregate labor demand and supply, and these interact back
to wage by the difference Ld Ls . The diagram in Fig. 6.7 shows how
government as an outside actor can affect the system by changing its
policy variables s and . In the equilibrium state, the system moves with
the equilibrium velocity Ld D Ls in the direction of accumulated labor
demand and supply with wage w staying constant.
Assuming fd (Ld ; p) and fs (Ls ; w; ) as in Eqs. (6.21), (6.22) and
pi D pi0 8i to eliminate quantity p from System (6.24), we get:
a0 a1
mLd Ld0 (t) D C 2 Ld (t) (1 C s)w(t);
n n
b0 b1 b2 b3
mLs Ls0 (t) D (1 )w(t) Ls (t) w(t) ; (6.26)
m m2 m m
1 0
w (t) D Ld (t) Ls (t):
kw
With certain parameter values, System (6.26) is globally stable and will
converge with time to the equilibrium state given in Eq. (6.23). The
equilibrium state of the labor market in Eq. (6.23) can be solved by
assuming Ld0 (t) D Ls0 (t) D w0 (t) D 0 in System (6.26), and solving the
resulting system of three equations with respect to the three endogenous
variables. The speed of adjustment of the system toward the equilibrium
depends on the two inertial masses, the spring constant, and the values
of the parameters. In this book we concentrate on the equilibrium state,
however, and therefore it is necessary to know that the system is stable and
will converge into the equilibrium state with time.
Fig. 6.8 Annual working hours and earnings index in Finland
Fig. 6.8 displays the aggregate annual working hours in the Finnish
economy and an index of earnings level during 1975–2013. The figure
shows that the earnings level has been rising steadily while aggregate
working hours have been slightly decreasing during the time period. Thus
a steady rise in the standard of living of employees has occurred in Finland.
Because the analytic solution of System (6.26) is rather complicated, we
demonstrate the solution by the following numerical values: m D n D 10,
mLd D mLs D 2, kw D 1=2, s D D 0, a0 D 400 C 25:1t; a1 D 8,
and b0 D 10C24:4t; b1 D 10; b2 D 0:2. The time paths of quantities
Ld ; Ls ; w with these parameter values are shown in Fig. 6.9, where on the
horizontal axis is time and on the vertical axis are Ld ; Ls , and w. The
three quantities are presented in the same figure to demonstrate their
connections, even though the measurement units of the quantities differ.
The time path of wage is the increasing one, and that of Ld is the thicker
one of those remaining. The figure shows that wage increases faster if Ld >
Ls , and Ld ; Ls stay relatively constant while wage increases continuously.
The initial condition of the solution is: Ld (0) D 400; Ls (0) D 300 and
w(0) D 5. A positive time trend in labor demand (a0 D 400 C 25:1t)
and a negative time trend in labor supply (b0 D 10C24:4t) is assumed.
These trends cause scarcity of labor that explains the continuous increase
in wage shown in Figs. 6.8 and 6.9.
Fig. 6.9 Labor demand, supply, and wage level
As compared with the static neoclassical analysis, by System (6.24)

we can study the dynamics of the labor market by assuming time
dependencies in quantities fd (Ld ; p) and fs (Ls ; w; ). On the other hand,
the adjustment process can be studied by solving System (6.26) with
different numerical values for the parameters in the model. Concerning
the speed of adjustment, the static neoclassical framework is a special case
of System (6.24) with an infinite speed of adjustment, that is, mLc D 0,
c D d; s and 1=kw D 0.
6.6 Trade Unions in the Labor Market

In the previous section we analyzed the determination of employment
and wage of a profession in a perfectly competed labor market. There
the labor suppliers were not members of a trade union, and the firms
were not members of an employer union. However, in most countries the
labor suppliers with certain education and work experience belong in a
trade union, firms belong in an employer union, and wage negotiation
takes place between these unions. In spite of a wage contract between
these unions, every firm and labor supplier makes their decision to hire
a person or accept a work offer independently. These matters make the
functioning of the labor market more complicated, because many of the
work contracts are not made at the wage agreed by the unions. It is a
commonly occurrence that contracts of work are agreed at a higher wage
than unions have agreed. The difference between a contract wage and the
actual wage is called a wage slide.
In this section, we analyze the behavior of a unionized labor market
by one simplified model called Monopoly Union model (see Dunlop
(1944)). This model assumes that a trade union makes the choice on
behalf of its members with respect to their work time and wage. Let the
time unit be one year. Analogously with a monopoly firm, a trade union
is supposed to know the demand relation of its ‘product’ (labor), and the
union sets the wage like a monopoly firm sets the price of its product. Like
a monopoly firm, a monopoly union chooses the optimal (annual use of
labor, wage) combination from the annual aggregate demand relation of
its members’ work time.
The demand of the work time of union members is assumed to be a
unique relation between the use of labor time of union members and wage.
The utility of the union is assumed to depend on the after-tax wage and
the aggregate annual work time of union members. The target function
of the monopoly union is then:
@u @u @2 u @2 u
u D u.wN ; L/; > 0; > 0; 0; 0;
@wN @L @wN 2 @L2
@2 u @2 u
D ; wN D (1 )w;
@L@wN @wN @L
where wN (e=h) is the after-tax hourly wage of union members, 0 < <
1 the wage tax rate, w(e=h) the hourly wage, and L(h=y) the annual
aggregate work time of union members. The marginal utilities of after-tax
wage and employment of union members, @u=@wN , @u=@L, are positive,
the second order partials of the utility function with respect to the same
quantity are non-positive due to non-increasing marginal utility, and the
second order cross partials are equal due to the continuity of the partial
functions.
Note 1. Talking about a utility function in connection with a trade union

is somewhat absurd because a union cannot feel pleasure or satisfaction. A
trade union has goals it aims to reach, and the target (utility) function of a
union consists of some measures for these goals. Thus when we talk about
the utility function of a trade union, we actually refer to the target function
that represents the goals of the union. The existence of a continuous target
function for a union should be derived, starting from a set of axioms, as
we did in the case of a consumer. This derivation is, however, omitted,
and we assume that the ‘utility’ of a trade union can be measured by a
continuous and differentiable function with unit ut=y. ˘
Note 2. Analogously to consumer theory, we can define a family of indif-

ference curves for a monopoly union where one curve represents constant
utility. The difference between the utility function of an individual laborer
and that of a trade union is that a trade union benefits from every rented
hour of its members’ work time, while a decrease in leisure time decreases
the utility of that laborer whose work time increases. A trade union is
thus a macro unit, the goals of which cannot be directly added from
those of its members. Union members gain if their leisure time increases,
all other things being equal, while a union cannot enjoy leisure time. A
union gains from every rented hour of its members’ work time in the
form of the payments it receives from its members. Unemployed union
members, on the other hand, cause expenditures for the union in the form
of unemployment benefits. The higher the wage the union can rent out
the work time of its members, and the higher the share of employed union
members, the more the union receives in the way of payments and the
more certain union leaders can be of their re-election. ˘
Of course, there exists models of trade union behavior where the target
function of the union is defined by aggregating over utility functions of
individual union members; see, for example McDonald and Solow (1981).
In this book we omit these models, however, because this section is only
a minor part of this book.
The slope of an indifference curve of the trade union in coordinates
(L; w) can be derived as follows. The total differential of the utility
function is:

@u @u 1 @u @u
du D dwN C dL D (1 )dw C dL; (6.27)
@wN @L 1 @w @L
where @u=@wN was obtained as follows
@u @u @wN @u @u 1 @u
D D (1 ) , D ;
@w @wN @w @wN @wN (1 ) @w
and dwN D (1 )dw was obtained from wN D (1 )w. If we now set

du D 0 in Eq. (6.27), we get the slope of an indifference curve as:
@u
dw @L
D @u < 0:
dL @w
Because the marginal utilities of wage and employment are positive, an

indifference curve is decreasing in coordinate system (L; w).
The existence of a trade union does not affect the demand of labor, and
we assume the labor demand relation for union members as in Sect. 6.5.1,
(1 C s)w D fd (L; p); (6.28)
where the subindex d in the use of labor is omitted because here we

do not need to separate labor demand and supply. The labor demand
relation in Eq. (6.28) expresses a unique relation between wage and
annual equilibrium use of labor of every firm employing this labor at
different wages. A trade union in a monopoly position sets the wage for
its members, and firms decide the annual work time they will employ at
this wage.
Similarly, as a monopoly firm, a monopoly union has only one quantity
by which it can affect its utility; either wage or aggregate annual work
time. If the union desires to get a certain annual work time rented, the
demand relation of the work time of union members shows at which wage
this takes place. If, however, the union likes to get a certain wage for its
members, the demand relation of the work time of union members shows
the annual work time that can be rented at this wage.
Fig. 6.10 The equilibrium of a monopoly union
The optimal situation of the monopoly union (L ; w ) is presented

in Fig. 6.10. In the optimum, the slope of an indifference curve is equal
to that of the annual demand relation of work time of union members:
@u
1 @fd (L; p)
@L
@u
D :
@w
(1 C s) @L
This condition—together with the demand relation for annual work time
of union members—uniquely defines the optimal wage and annual work
time of union members.
The force the monopoly union directs upon the annual work time of
its members can be derived as follows. We substitute wage from the utility
function of the union by using the labor demand relation as

(1 )
u D u(wN (t); L(t)) D u fd .L(t); p/; L(t) ;
(1 C s)
where the aggregate annual work time of union members is set to depend
on time t. The union can now affect its utility only by the annual work
time L(t). The time derivative of the utility function is:

@u (1 ) @fd .L(t); p/ @u
u0 (t) D C L0 (t):
@wN (1 C s) @L @L
Changes in the aggregate annual work time, that make the acceleration of
utility u0 (t)(ut=y2 ) positive, are:
@u
0 (1 ) @fd .L(t); p/ @L
L (t) > 0 if C @u
> 0;
(1 C s) @L @wN
@u
(1 ) @fd .L(t); p/
L0 (t) < 0 if C @L
@u
< 0;
(1 C s) @L @wN
@u
(1 ) @fd .L(t); p/
L0 (t) D 0 if C @L
@u
D 0:
(1 C s) @L @wN
) @fd .L(t);p/ @u @u
As before, quantity Fu D (1(1Cs) @L
C @L = @wN can be named as the
force the union directs upon the annual work time of the members
of the union in the region. The first negative component in the force
measures the decrease in utility due to a decrease in wage required for an
increase in employment. The second positive component measures the
increase in utility due to an increase in annual work time. The union
compares the gain of extra work time and its cost, the decrease in wage,
and decides the optimal (annual work time, wage) combination on this
basis. The zero force situation corresponds to the optimal state of the
union.
In the above defined force, the measurement unit of utility cancels
out. Thus the force is measurable in units (e=h)=(h=y) D e y=h2 ,
when the benefits and losses for the union from this decision can be
quantified. Another way to model the decision-making of the trade union
is to substitute the aggregate annual employment from the utility function
by using the labor demand relation, and analyze the utility as a function of
wage. This way we could define the force the union directs upon the wage
of union members. This, however, is omitted because in solving the labor
demand relation with respect to Ld we should take the inverse function
of fd that would complicate the modeling.
6.6.1 Dynamic Trade Union Behavior*
The analysis in the previous section can be expressed mathematically as

@u
(1 ) @fd .L(t); p/
L0 (t) D f (Fu ); f 0 (Fu ) > 0; f (0) D 0; Fu D C @L
@u
;
(1 C s) @L @wN
where the union is assumed to know the ‘force’ Fu acting upon the
employment of union members in the region. The union adjusts the
employment according to this force with the aim of increasing its utility
with time. Assuming a specific utility function for the union we could
study the stability of the model. However, for shortness we omit these
analyses.

The first order Taylor series approximation of the value of marginal pro-
ductivity of labor of firm i in the neighborhood of the point (Lid0 ; pi0 ) is
pi fi0 .Ldi / D pi0 fi0 .Ldi0 / C pi0 fi00 (Ldi0 ).Ldi Ldi0 / C fi0 (Ldi0 ).pi pi0 / C i ;
(6.29)
where i is the residual term. Assuming i D 0 and summing over i, we get
X
n X
n

pi fi0 .Ldi / pi0 fi0 .Ldi0 / pi0 fi00 (Ldi0 )Ldi0 fi0 (Ldi0 )pi0
iD1 iD1
X
n X
n
a1
C pi0 fi00 (Ldi0 )Ldi C fi0 (Ldi0 )pi a0 C Ld C a2 p;
iD1 iD1
n
Pn Pn
where Ld D iD1 Ldi , p D (1=n) iD1 pi and1
X
n

a0 D pi0 fi0 .Ldi0 / pi0 fi00 (Ldi0 )Ldi0 fi0 (Ldi0 )pi0
iD1
X
n X
n X
n
D pi0 fi00 (Ldi0 )Ldi0 ; a1 D pi0 fi00 (Ldi0 ); a2 D fi0 (Ldi0 ):
iD1 iD1 iD1
The units of a0 ; a1 ; a2 are e=h, (e y)=h2 , and kg=h, respectively, and

our assumptions of firms’ marginal productivity make a0 0; a1 0,
and a2 > 0. .
@u @u
The Taylor series approximation of gj (Lsj ; w; ) @Hjj @Tjj in the
neighborhood of the equilibrium point zj0 D (Lsj0 ; w0 ; 0 ) is:
@gj (zj0 ) @gj (zj0 )

gj (Lsj ; w; ) D gj (zj0 ) C (Lsj Lsj0 ) C (w w0 )
@Lsj @w
@gj (zj0 )
C ( 0 ) C j ; j D 1; : : : ; m: (6.30)
@
Assuming j D 08j and summing over j, we get2
X
m
gj (Lsj ; w; )
jD1
m
X
@gj (zj0 ) @gj (zj0 ) @gj (zj0 )
D gj (zj0 ) Lsj0 w0 0
jD1
@Lsj @w @
X
m
@gj (zj0 ) X
m
@gj (zj0 ) X
m
@gj (zj0 )
C Lsj C w C
jD1
@Lsj jD1
@w jD1
@
b1
b0 C Ls C b2 w C b3 ;
m
Pn Pn Pn Pn
1
ci xi D c iD1 xi C iD1 (ci c)xi where c D (1=n) iD1 ci . The approximation
PniD1 Pn
iD1 ci xi c iD1 xi is the more accurate the less ci or xi vary, i D 1; : : : ; n.
2
See footnote 1.
Pm
where Ls D jD1 Lsj and
Xm
@gj (zj0 ) @gj (zj0 ) @gj (zj0 )
b0 D gj (zj0 ) Lsj0 w0 0 ;
jD1
@Lsj @w @
X
m
@gj (zj0 ) X
m
@gj (zj0 ) X
m
@gj (zj0 )
b1 D ; b2 D ; b3 D :
jD1
@Lsj jD1
@w jD1
@
The units of b0 ; b1 ; b3 are: e=h, (ey)=h2 , and e=h, respectively, and b2

is dimensionless. Because gj (Lsj ; w; ) is positive at every Lsj ; w; , then
b0 > 0 (let Lsj , w, , j ! 0 8j in (6.30)). In Sect. 6.3 we showed that
@gj =@Lsj > 0, @gj =@w > 0, and @gj =@ < 0; thus b1 > 0, b2 > 0 and
b3 < 0.
References
Dunlop, J. T. (1944). Wage determination under trade unionism. New York:
Macmillan.
McDonald, I., & Solow, R. (1981). Wage bargaining and employment. American
Economic Review, 71, 896–908.
7
Capital Goods as Firms’ Inputs
We repeat here the definition given in Sect. 4.3 for firms’ capital goods.
§: The plant, machinery, and equipment owned by a firm, that create
revenues during several fiscal periods, are called the physical capital of
the firm. ˘
The raw materials used by firms are not counted as capital goods
because they wear out in production. The term ’physical capital’ makes a
distinction from the financial or monetary capital of a firm, and name
capital refers to a stock concepts as distinct from flow concepts. The
amounts of different kinds of capital goods are measured, for example,
in units kg, unit, and so on. Flow quantities, on the other hand, are
time-related quantities and their measurement units are, for example,
unit=week, kg=h (see Sect. 2.7.1).
All stocks are accumulated from some kind of a flow process, and
a stock measures the sum of a flow at a certain moment of time (see
Sect. 2.7.3). For example, the inventory of a firm’s produced but unsold
products, the number of production factors acquired and not scrapped
during the history of the firm, the common value of all production factors
of a firm at a certain time moment, are all stocks. Pieces of land measured

DOI 10.1007/978-3-319-46879-2_7
in units acre; hectare, and so forth, can also be included in a firm’s

physical capital, because land gains revenues for its owner during various
fiscal periods in the form of crops, rents, or savings; the owner of a land
does not need to pay rent for its use.
When we analyze a firm’s use of capital goods in its production
processes, it is essential to make a distinction between stock and flow
concepts. Rent is the compensation for the services of a capital good in
a time unit; for example, the rent 20 (e/h) for using a drilling machine.
The measurement units of rents can be (e=unit)=mn, (e=unit)=h, and
so on. The price of a capital good with unit e=unit, on the other hand,
is the compensation for the services given by the capital good during
its time of use; that is, the compensation with regard to the stock of
services from the capital good during its time of use. Because prices have
dimension (money=real) and monetary values have dimension (money),
in the following we treat the values of capital goods. Thus we should talk
about the value of a capital good rather than its price. However, in buying
a capital good it is more easily understood that the buyer pays the price,
and not the value of the good, which is the reason we talk about the price
of a capital good rather than its value. For example, price 1000 (e=unit)
corresponds to the value 1000 (e) D 1000 (e=unit) 1 (unit) of one
capital good. This distinction between stocks and flows was not needed
in connection with firms’ use of labor, because labor can be bought only
in slavery societies; usually it can only be rented.
7.1 Renting Capital Goods

If a firm rents capital goods—and does not buy them—the firm’s use of
capital goods exactly corresponds to its use of labor; see Sect. 6.1. In that
case a firm compares the hourly rent of a capital good and the value of
its marginal productivity from one hour, and attempts to choose on this
basis its optimal use of capital goods for its planning time period. Suppose
the weekly production of a firm depends on its use of capital goods with
unit h=week. If the rent of the capital goods is fixed, and their physical
productivity is non-increasing with increasing use, we can show that there
exists a unique optimal use of these capital goods, as in the case of labor.
This analysis is carried on here shortly because it exactly corresponds to
that of the use of labor.
Suppose the weekly production of a firm producing good k depends
on its use of one kind of labor Lk (h=week) and one kind of capital
goods Bk (h=week) as qk D f (Lk ; Bk ), where qk (kg=week) is the flow
of production of the firm. Production function f obeys the following
characteristics:
@qk @qk @2 qk @2 qk @2 qk @2 qk
> 0; > 0; 0; 0; and D D 0:
@Lk @Bk @Lk2 @B2k @Lk @Bk @Bk @Lk
The marginal productivity of both factors is positive, non-increasing

marginal productivity holds for both factors, the partial functions of
the production function are continuous, and the two production factors
are assumed to be independent in the production process. The last
assumption is made because with it we can analyze the use of both factors
independently.
We denote the hourly wage of labor by w (e=h) and the rent of capital
goods by z (e=h). The weekly profit of the firm is then
…k (t) D pk qk (t) C0 wLk (t) zBk (t); qk (t) D f .Lk (t); Bk (t)/;
where C0 (e=week) are fixed costs, all the three prices are assumed
constant, and the firm’s use of both production factors is set to depend on
time t. The firm is assumed to adjust its use of labor and capital to increase
its weekly profit with time. The time derivative of the profit function is
(Sect. 10.9.4):

@qk @qk
…0k (t) D pk w Lk0 (t) C pk z B0k (t):
@Lk @Bk
The firm can now affect its weekly profit by adjusting its use of labor
and capital independently of each other. In Sect. 6.1 we analyzed a firm’s
adjustment of its use of labor, and thus we can omit it here. The
adjustment of capital goods is made accordingly. The adjustment rules

for renting capital goods, that increase the weekly profit of the firm with
time, are:
@qk
B0k (t) > 0 if pk z > 0;
@Bk
@qk
B0k (t) < 0 if pk z < 0;
@Bk
@qk
B0k (t) D 0 if pk z D 0:
@Bk
These adjustment rules show that the firm increases its use of capital goods
if the value of marginal productivity of capital from one hour is greater
than hourly rent, and vice versa. In the optimal situation, the value of
marginal productivity of capital from one hour is equal to hourly rent.
7.1.1 A Dynamic Theory of Renting Capital Goods*
The analysis in the previous section can be presented mathematically as

@qk
B0k (t) D g(F); g0 (F) > 0; g(0) D 0; F D pk z; (7.1)
@Bk
where g is a function with the above characteristics. Because B0k (t)

(h=week2 ) is the acceleration of use of capital goods caused by quantity
@qk
F D pk @B k
z, we can interpret F as the force acting upon the use of
capital goods of the firm. Equation (7.1) is the equation of motion for
the use of the capital goods of the firm, and its linear form corresponds
to the Newtonian equation. Because
@B0k (t) @2 qk
D g0 (F)pk 2 0;
@Bk @Bk
the equation of motion in Eq. (7.1) is stable if @B0k (t)=@Bk < 0. This
analysis exactly corresponds to that of a firm’s use of labor, and we omit
all details here.
7.2 Investing in Capital Goods

According to the definition, a capital good produces services during
various fiscal periods. The value of one capital good of a firm can be
calculated as the difference between the present value of the services
and costs it creates during its time of use. Another way to analyze the
profitability of an investment is to compare the present values of the costs
of buying a capital good and renting the corresponding services from other
firms. The revenues and costs at different time periods are not directly
comparable, however, and we have to transform these money flows at
different time units comparable with each other. For this analysis, we have
to define a method of present value calculation, which is introduced next.
7.3 Interest Calculation and Discounting

In economic analysis, interest and discount1 calculation have been defined
in discrete and in continuous time. We start with the discrete one.
7.3.1 Discounting in Discrete Time
Suppose that at time moment t0 , the amount of money x(t0 ) (e) is

deposited in a bank account with a fixed interest rate. In order to separate
discrete and continuous time calculations, in the following discrete time
interest rate is denoted by rd and continuous time interest rate by r.
Interest rate rd with unit 1=t is the following growth rate of the
deposited capital:
Œx(t0 C t) x(t0 )=t

D rd , x(t0 C t) x(t0 ) D rd tx(t0 )
x(t0 )
, x(t0 C t) D x(t0 ) C rd tx(t0 ) D (1 C rd t)x(t0 ):
1
The term ‘discount’ comes from the reduction in future debt payments a debtor receives if he repays
his debt before the terminal date.
Table 7.1 Discrete time interest calculation

Time moment Money on a bank account
t0 x(t0 )
t0 C t x(t0 C t) D (1 C rd t)x(t0 )
t0 C 2t x(t0 C 2t) D (1 C rd t)2 x(t0 )
:: ::
: :
t0 C nt x(t0 C nt) D (1 C rd t)n x(t0 )
The last form of the equation shows that the money on the bank account
at moment t0 C t, x(t0 C t), can be expressed by the interest rate rd .
Assuming rd to be fixed, we can calculate how a monetary capital increases
during n time intervals when interest revenues are added in the capital
at the last moment of every time unit. We call this compound interest
calculation.
§: By compound interest calculation we understand the calculation
of interest returns in the way that interest revenues from one time unit are
added in the invested capital after every time unit. ˘
The money on the bank account at the last moment of every time
interval is shown in Table 7.1. The deposit at moment t0 , x(t0 ), grows
during time unit t so that at the end of the time unit, the money in
the bank account is x(t0 C t) D x(t0 ) C rd tx(t0 ) D (1 C rd t)x(t0 ),
where x(t0 ) is the invested capital and rd tx(t0 ) the interest revenues from
the time unit. The dimensionless term 1 C rd t, where the measurement
units of rd and t cancel each other out, is called the interest factor for
one time unit. At time moment t0 C 2t the capital is, similarly,
x(t0 C2t) D (1Crd t)x(t0 )Crd t(1Crd t)x(t0 ) D (1Crd t)2 x(t0 );
where (1 C rd t)x(t0 ) is the capital at the beginning of the time unit and
rd t(1 C rd t)x(t0 ) the interest revenues from the time unit.
Next we define the amount of money at time moment t0 , x(t0 ), that
corresponds to the money x(t0 C t) at moment t0 C t, as

1
(1 C rd t)x(t0 ) D x(t0 C t) , x(t0 ) D x(t0 C t):
1 C rd t
We call x(t0 ) the present value of x(t0 C t) and 1=(1 C rd t) the
discount factor between time moments t0 and t0 C t. The discount
factor is a dimensionless quantity like the interest factor. In general, the
discount factor between time moments t0 and t0 C nt is 1=(1 C rd t)n .
If the interest rates of time units differ, that is rd1 ¤ rd2 ¤ where rdj
is the interest rate of the jth time unit, the interest and discount factors
between t0 and t0 C nt are:
1
(1 C rd1 t) (1 C rdn t) and : (7.2)
(1 C rd1 t) (1 C rdn t)
The reader can check this by constructing a similar table to Table 7.1
varying the interest rates of the time units.
Example
Let interest rate be 10 (%=y) = 0.1 (1=y) and t D 1 (y). Then, setting x(t0 C
t) D x(t0 C 2t) D x(t0 C 3t) D 1 (e) we get the corresponding present
values as: 1=(1 C 0:1 1) D 0:91 (e), 1=(1 C 0:1 1)2 D 0:83 (e), and 1=(1 C
0:1 1)3 D 0:75 (e). If interest rate is 5 (%=y) and t D 1 (y), the present
values are: 0.95 (e), 0.91 (e), and 0.86 (e), respectively. Thus an increase in
interest rate decreases present values. ˘
Dimensionally, monetary quantities at every time unit belong in the

dimension of money, and are thus additive. A positive interest rate means,
however, that the value of one euro at a future time moment is smaller than
that at the current moment. The reason for this is the interest revenues
that are obtained for deposited money. Current euro corresponds in the
future to one euro plus interest revenues for one euro.
§: Interest and discount factors are transformation rules between
monetary quantities at different time moments. They work identically
as the transformation equations between measurement units belonging in
one dimension. ˘
For example, the transformation equation between euros at moments

t0 C 4t and t0 C t is
x(t0 C 4t) .1 C rd t/4 x(t0 ) x(t0 C 4t)

D , D .1 C rd t/3 ,
x(t0 C t) .1 C rd t/ x(t0 ) x(t0 C t)
x(t0 C 4t) D .1 C rd t/3 x(t0 C t) etc.:
Interest and discount calculation can be defined for real values too. Sup-
pose x (e) is deflated by an average price level p (e/kg) of the economy.
The growth rate of the real quantity x(t)=p(t) (kg) then corresponds to
the real interest rate rR :

x(t0 Ct)
p(t0 Ct)
x(t0 )
p(t0 )
=t x(t0 C t) x(t0 )
D rR , D (1 C rR t): (7.3)
x(t0 )
p(t0 )
p(t0 C t) p(t0 )
This can be seen by comparing Eqs. (7.3) and (2.12) and remembering
that t1 D t0 C t or t D t1 t0 . The latter form of Eq. (7.3) defines the
transformation equation between values of real quantity x=p at different
time units. Quantity x(t0 Ct)=p(t0 Ct) measures the purchasing power
of x(t0 C t) at time moment t0 C t, that is, the amount of goods in
the economy that can be obtained by x (e) at time moment t0 C t.
It is a matter of preference as to whether discounting is made in nominal
or in real terms. It is essential that if we operate with nominal (real) terms,
the interest rate to be used in discounting is the nominal (real) one.
7.3.2 Interest Calculation in Continuous Time*
Analogously with discrete time, continuous time interest rate r(t) is

defined as the instantaneous growth rate of a monetary quantity at time
moment t, r(t) D x0 (t)=x(t). The solution of this differential equation is
(Sect. 10.15)
Rt
x0 (t) D r(t)x(t) ) x(t) D Ae t0 r(s)ds
; (7.4)
where A (e) is the integration constant, by s is denoted running time

during time interval (t0 ; t), and e is the base of the natural logarithm. The
reader can check the above result by taking the time derivative of the latter
equation, substituting in this the solved function for x(t), and verifying
that the latter equation equals the former. Setting t D t0 in Eq. (7.4),
we get
Rt
r(s)ds
x(t0 ) D A ) x(t) D x(t0 )e t0 : (7.5)
Equation (7.5) defines the money on the bank account at time moment
t when x(t0 ) was deposited at moment t0 with continuous time interest
rate r that may vary during (t0 ; t). In a continuous time compound interest
calculation, interest revenues are added in the capital after every instant of
time, which creates the exponential growth. If t D t0 , then x(t) D x(t0 )
in Eq. (7.5).
Using the definition r(s) D x0 (s)=x(s), we get:
Z Z ˇt
t t
x0 (s) ˇ x(t)
r(s)ds D ˇ
ds D ˇ ln(x(s)) D ln :
t0 t0 x(s) t0 x(t0 )
We can thus express Eq. (7.5) as

x(t)
ln
x(t) D x(t0 )e x(t0 )
;
which is trivially true because the inverse operations—exp and ln—cancel

each other out. The present value of x(t) at moment t0 is obtained by
solving Eq. (7.5) with respect to x(t0 ):
Rt
x(t0 ) D x(t)e t0 r(s)ds
:
If the interest rate is constant during time interval (t0 ; t), then:
x(t0 ) D x(t)er(tt0 ) , x(t) D x(t0 )er(tt0 ) : (7.6)
The first form of Eq. (7.6) expresses the present value of x(t) at time
moment t0 , and the latter is the corresponding compound interest for-
mula. The continuous time correspondent for the discrete time discount
factor (1 C rd t)n with constant interest rate is thus er(tt0 ) , and

the corresponding interest factor is er(tt0 ) . These interest and discount
factors are dimensionless quantities because the unit of r is 1=t and time
unit t t0 is measured in units t.
Note. One essential difference exists between the two discount factors. In
the continuous time discount factor, the length of time interval t t0 is
measured in time units. In the discrete time discount factor, on the other
hand, exponent n is a pure number that represents the order of the time
interval. If time is measured in years in discrete analysis, then n D 3
implies that the discount factor is that between the initial and the third
year, and so forth. ˘
7.3.3 Parities Between Interest Rates
Which is the correct way to calculate interest revenues? There is no

unique answer to this question because banks calculate interest on their
deposits in different ways. It is common is to take the end-of-the-day
or the smallest amount of money on a bank account in a month as
the capital for which the interest is calculated. Discrete time interest
returns are calculated for deposited capital either on daily or monthly
basis, and interest revenues are added in the capital usually at the end
of the year. Some banks also apply a continuous time interest calculation.
Thus the interest calculation principles vary between banks and between
different accounts in the same bank. For this reason, we present in the next
conversation rules that make the interest rates comparable when different
compound interest calculation methods are applied.
Discrete and Continuous Time Interest Rates*
According to the previous sections, with compound interest calculation

the discrete and continuous time calculated values of capitals at time
moment tn , deposited in a bank account at moment t0 , are:
R tn
r(t)dt
x(tn ) D x(t0 )(1 C rd t)n and x(tn ) D x(t0 )e t0 :
Solving x(tn )=x(t0 ) from both equations and setting them equal, we get:
R tn
r(t)dt
(1 C rd t)n D e t0 : (7.7)
Thus both methods give identical results if the two interest (and discount)
factors are equal. Suppose then that the two interest rates with unit 1=t
are constant. The amount of money on the bank account after time
interval t, calculated by both methods, is then:
x(t1 ) D x(t0 )(1 C rd t) and x(t1 ) D x(t0 )er(t1 t0 ) :
Setting these equal and dividing by x(t0 ) we get:
ln(1 C rd t) D r (t1 t0 ) or rd t D ert 1: (7.8)
By using the transformation Eq. (7.8), continuous and discrete time

compound interest calculation gives identical results (notice that t1 t0 D
t and rd t is a dimensionless quantity the numerical value of which is
equal to rd ). We can thus define the continuous time interest rate rc for
time unit t, conformal with the corresponding discrete time rate, as:

1 x(t1 ) ln(1 C rd t)
rc D ln D : (7.9)
t1 t0 x(t0 ) t
Notice that in Eq. (7.9), x(t1 ) is calculated by using the corresponding

discrete time interest rate and discrete time method of calculation.
Example
We calculate the capital on a bank account by discrete and continuous
time methods when 100 (e) is deposited at moment t0 D 0 for time unit
tn t0 D 20 (y), and rd D r D 0:1 (1=y). The time paths of the capitals are in
Fig. 7.1a. The continuous time capital (the curve) somewhat overestimates
the discrete time one (the dots). However, using rc D ln(1 C rd t)=t D
0:0953 (1=y) in continuous time, the two time paths coincide, see Fig. 7.1b.
Hull (2000, pp. 51–52) shows that even without the adjustment r D rc , the
continuous time interest calculation gives almost identical results as the
daily discrete time analysis with ‘normal’ levels of interest rates. ˘
Fig. 7.1 (a) Discrete and continuous time interest rates. (b) Two conformal
interest rates
Discrete Time Interest Rates of Varying Time Units
Another parity can be made between discrete time interest rates of varying
time units. Let us denote the discrete time interest rate for time unit k
by rdk , and that for time unit z by rdz , where s D k=z , k D sz
is the transformation equation between the time units. For example,
1 (y) D 12 (mn) where k D y; z D mn, and s D 12. In both cases,
interest returns are added in the deposited capital at the ending moment
of the corresponding time unit. Let s > 1, which implies k > z; thus
rdz represents a more dense splitting of time, that is, the interest revenues
are added in the capital more frequently. In monthly (annual) compound
interest calculation, interest revenues are added in the capital after every
month (year). The deposited capital x(t0 ) (e) increases in both cases
during time unit k as:
x(tk ) D x(t0 )(1 C rdk k) and x(tsz ) D x(t0 )(1 C rdz z)s :
Setting these two capitals equal and dividing by x(t0 ), gives:
rdc;k k D (1 C rdz z)s 1 , rdc;z z D (1 C rdk k)1=s 1: (7.10)

Fig. 7.2 (a) Capitals with two discrete time interest rates. (b) Capitals with
conformal discrete time interest rates
These transformation rules make any two discrete time interest rates
comparable with each other, because relation k D sz holds for every two
time units k; z. Thus Eq. (7.10) defines two discrete time interest rates
conformal with each other in compound interest calculation.
Example 1
Suppose rdy D 0:1 (1=y), rdm D 0:01 (1=m) and 1 (y) D 12 (m). Using
the rules in Eq. (7.10), the corresponding conformal monthly and annual
interest rates are rdc;m D 0:00797 (1=m) and rdc;y D 0:1268 (1=y), respectively.
These somewhat differ from the approximate ones we get by transforming
simply by using time units: rOdm D 1=12 (y=m) 0:1 (1=y) D 0:0083 (1=m) and
rOdy D 12 (m=y) 0:01 (1=m) D 0:12 (1=y). ˘
Example 2
Figure 7.2a shows the differences in capitals during 20 years when 100 (e)
is deposited at time moment 0 and interest rates rdy D 0:1 (1=y), rOdm D
0:0083 (1=m) are applied (the curve refers to monthly and dots to annual
analysis). This difference disappears when the conformal monthly interest
rate rdc;m D 0:00797 (1=m) is applied in the annual analysis, see Fig. 7.2b. ˘
7.4 Present Values of Money Flows

7.4.1 Present Values in Discrete Time
In calculating present values of money flows, we transform monetary

quantities at various time moments comparable with current money, see
Sect. 7.3.1. Let us suppose money flow N(t0 C jt) (e/t), j D 1; 2; : : : ,
where the money is received at the ending moment of every time unit t
with a fixed interest rate rd (1=t). In discrete time, the present value of
an n period flow at time moment t0 is
N(t0 C t)t N(t0 C 2t)t N(t0 C nt)t

P(t0 ) D C C C
1 C rd t (1 C rd t)2 (1 C rd t)n
Xn j
1
D N(t0 C jt)t
jD1
1 C rd t
X
n
D N(t0 C jt)t .1 C rd t/j ; (7.11)
jD1
where N(t0 C jt)t has unit e 8j.

To compare the discrete and continuous time present values, we assume
that the money flow is fixed, that is, N(t0 C jt) D N 8j. We can then
take the common factor Nt of every term in front of the sum (7.11) and
study the obtained sum of positive terms aj ,
j
1
aj D ; j D 1; 2; : : : ; n; 0 < a < 1 because rd > 0:
1 C rd t
This sum is called a geometric series with positive terms. Let us denote
the sum of the series of n terms Sn as:
X
n
Sn D a1 C a2 C a3 C C an D aj :
jD1
Multiplying this sum (every term of the sum) by a, we get: aSn D a2 C

a3 C a4 C C anC1 . Subtracting these two series from each other, we
can solve the sum of the geometric series of n terms as:
a(1 an )
Sn aSn D a anC1 , (1 a)Sn D a(1 an ) , Sn D :
1a
(7.12)
The above geometric series Sn with positive terms converges if 0 a < 1.
The sum of a convergent geometric series with an infinite number of terms
is: limn!1 Sn D a=(1 a) because limn!1 an D 0 if 0 a < 1. We
can then express the present value P(t0 ) with n time units as
a(1 an ) N
P(t0 ) D Nt D Œ1 (1 C rd t)n ; (7.13)
1a rd
where

a(1 an )
1
1Crd t
1 1
(1Crd t)n
1
Œ1 (1 C rd t)n
1Crd t
D D 1Crd t1
1a 1 1
1Crd t 1Crd t
1 (1 C rd t)n
D :
rd t
A quick test for the correctness of the last form of Eq. (7.13) can be made
by dimensional analysis. The measurement unit of the left hand side is e,
that of N=rd is e, and the term in parenthesis is dimensionless. At least
the measurement units match; hopefully also the form of the equation.
Equation (7.13) is a complicated connection of four quantities: P(t0 ),
N, rd and n. There exists computer programs that solve rd from Eq. (7.13)
in a numerical way to see what is the yield (the internal rate of return
per annum) of an asset with initial price P(t0 ) D 10;000 (e) that pays
Nt D 1300 (e) at the end of every year during n D 10 years. This yield
can be compared with the prevailing risk-free interest rate in the economy
to see whether this asset is competitive in creating revenues. Another way
to use Eq. (7.13) is to calculate the necessary annual payment N related to
an n period fixed flow with initial price P(t0 ) and discount rate rd as:
rd P(t0 )
ND : (7.14)
1 (1 C rd t)n
One clear result can still be obtained from Eq. (7.13): the higher the
interest rate rd , the smaller the present value of a fixed payment instru-
ment. Although quantities N, n and P(t0 ) are known at the moment of
the investment decision, interest rate rd during the term to maturity of
the asset is not known, and so there exists an interest rate risk in this
investment decision.
Assuming an infinite time horizon (limn!1 ) in Eq. (7.13), we get
N
P(t0 ) D : (7.15)
rd
Notice that the measurement unit of the interest rate transforms the
measurement unit of the present value as: (e/t)=(1=t) D e.
Note. In calculating present values of money flows, the interest rate

applied in the discount factor represents the alternative rate of return that
could be obtained for the money if invested elsewhere. Many times the
interest rates offered by government bonds (see Sect. 8.2.3) are considered
as a proper measure for the risk-free interest rate that can be used in
discounting money flows of other financial instruments. ˘
We calculate the present value of a money flow where 4 (e) is received

after four weeks. Thus the flow is 4 (e=4week) even though the money is
not received until the interval ends. The interest rate is assumed to be 10
(%=y) = 4/520 (1=4week), where the interest rate is transformed simply
by using time units. The present value of this flow is then
4 (e=t)t 4 (e=4week) (4week) 4 (e )

D 1 D D 3:969 (e):
1 C rt 1 C 520 4week (4week)
4
1 C 520
4
Next we calculate the present value of a money flow, where one euro is
received at the end of every week in a time unit of four weeks. Notice
that the total of the flow is the same as above: 4 (e=4week). Again,
transforming the interest rate simply by time units the weekly interest
rate is 10 (%=y) = 1=520 (1=week), and the present value is:
1 (e=t)t 1 (e=t)t 1 (e=t)t 1 (e=t)t

C C C
1 C rt (1 C rt)2 (1 C rt)3 (1 C rt)4
1 (e=week) (1week) 1 (e=week) (1week)
D 1 C 1 2
1 C 520 week (1week)
1
1 C 520
1
(1week)
week
1 (e=week) (1week) 1 (e=week) (1week)
C 3
C 1 4
1 C 520
1 1
week
(1week) 1 C 520
1
week
(1week)
D 3:981 (e):
Thus the more frequently the payments take place in a fixed flow, the
greater is the present value of a positive money flow. The four week interest
rate rd (1=4week), that gives the same present value as the above when
calculated on the four week basis, can be obtained as follows:
4 (e=4week) (4week)
3:981 (e) D )
1 C rd (1=4week) (4week)

4 3:981 1
rd D D 0:00477 :
3:981 4week
This is clearly smaller than the four-week interest rate we used without
making the compounding correction: 4=520 D 0:00769 (1=4week).
7.4.2 Present Values in Continuous Time*
Here we analyze present values of money flows in continuous time.

Discrete time is transformed to continuous by letting t ! 0. This is
done by dividing time unit t in k equal subintervals and letting k ! 1.
At the time interval t0 Cjt=k, the interest rate is rd (t0 Cjt=k) with unit
(1=(t=k)), and the money flow is N(t0 C jt=k) with unit (e/(t=k)),
j D 1; 2; : : : ; nk. With fixed k, the present value of this flow during nk
time periods is:
X
t t j
nk
t t
P(t0 ) D N t0 C j 1 C rd t0 C j :
jD1
k k k k
If k ! 1, then t=k ! dt, rd (t0 C jt=k) ! r(t0 C jdt), and N(t0 C

jt=k) approaches the instantaneous flow at time moment t0 C jdt.
The above given discount factor can be modified as
00 1 .
jrd t0 Cj t
k /
1 k=t k=t
.
rd t0 Cj t /
t t j B 1 k
C
1 C rd t0 C j D @@1 C k=t
A A ;
k k
rd .t0 Cj t
k /
see Chiang (1984, p. 293). Now we know that

z
1
lim 1C D e;
z!1 z
and because limk!1 k=(rd (t0 C jt=k)t) D 1, we can simplify the

above formula by the definition of the number e. The limiting process
transforms the exponent of e to r(t) (t t0 ). This occurs because
with k ! 1, jt=k ! jdt and rd (t0 C jt=k)jt=k ! r(t0 C jdt)jdt.
The following definition t D t0 C jdt, j D 1; 2; : : : for continuous time
completes the proof.
Taking the limit k ! 1 transforms the sum to an integral with
integration limits t0 and t0 C nt D tn . We can thus write
X
t t j t
nk
t
lim N t0 C j 1 C rd t0 C j
k!1
jD1
k k k k
Z tn
D N(t)er(t)(tt0 ) dt
t0
where the instantaneous flow at moment t D t0 C jdt is denoted by N(t).

Assuming r to be constant, the present value of an infinite fixed flow
in continuous time becomes the following:
Z ˇ1
1 ˇ N N 1 N
Ner(tt0 )
dt D ˇˇ er(tt0 ) D e e0 D :
t0 t0 r r r
Thus if rd D r, we get equal present values for the same fixed infinite
money flow by discrete and continuous time discounting.
Next we calculate the present value of flow 4 (e=4week) in continuous
time. Time is first measured in units 4week and then in units week. The
present value of flow 4 (e=4week) is in the first case:
Z ˇtDt0 C1
tDt0 C1 ˇ 4 4
4er(tt0 ) dt D ˇˇ er(tt0 ) D er e0
tDt0 tDt0 r r
4

D 520 1 e 520 D 3:985 (e):
Notice that the marginal change in time, dt, is measured above in

units 4week. The complete form of the integrated factor is then 4
(e=4week) dt (4week) er(tt0 ) = 4 dt er(tt0 ) (e), where er(tt0 )
is dimensionless.
Next time is measured in units week. Flow 4 (e=4week) corresponds
to 4/4 (e=week) = 1 (e=week). The present value of this flow is
Z ˇtDt0 C4
tDt0 C4 ˇ 1 1
1er(tt0 ) dt D ˇˇ er(tt0 ) D e4r e0
tDt0 tDt0 r r
4

D 520 1 e 520 D 3:985 (e);
where dt is measured in units week. Thus, independent of the applied

time unit, we get an equal present value for the same money flow in
continuous time discounting. It is essential that the limits of integration
and the money flow are measured in the same time unit. Present values
of money flows in continuous time are a bit higher than the corre-
sponding ones in discrete time. The more dense the splitting of time in
discrete analysis, the more close the present value is to that in continuous
time.
7.5 Investment Decision of a Firm

We assume that a firm is planning to increase its stock of capital goods,
because the firm is operating at full capacity and excess demand exists for
its products. Let us first study the situation where the firm cannot rent
the required capital goods from other companies. Thus the firm considers
buying a capital good (later a machine) that produces the required services.
The decision whether to buy a machine or not is based on the present
values of the revenues and costs one machine causes during its time of
use. The costs consist of the price of the machine and its user costs.
We analyze the investment decision isolated from the other manage-
ment of the firm at time moment t0 . Time is divided in units t:
t0 ; t0 Ct; t0 C2t; t0 C3t; : : : , and we assume that the machine is used
for n time units. We denote the price (value) of one machine as C0 (e)
at moment t0 , and its user costs and revenues from time unit t0 C it,
i D 1; : : : ; n are denoted as C(t0 Cit) (e=t) and R(t0 Cit) (e=t),
respectively. The present value of revenues from one machine during n
time units is
R(t0 C t)t R(t0 C 2t)t
Rpv D C C
1 C r1 t (1 C r1 t)(1 C r2 t)
R(t0 C nt)t
C ;
(1 C r1 t) (1 C rn t)
where the money flows R(t0 C it) (e=t) are assumed to be received at
the ending moment of every time unit, and the discrete time interest rates
are denoted by ri (1=t), i D 1; : : : ; n. The deviation of the interest rates
at different time units makes the discount factors as products of different
discount factors, and not as powers of factor 1=(1 C rt) (see Sect. 7.3.1).
A firm can finance its investment by its cash reserves, by issuing
equities, or by taking a loan. If a firm finances its investment by a loan, we
assume that the firm pays back the loan with a fixed payment in a time
unit that contains interest cost and part payment. The present value of
costs of one machine from n time units, that contains the possible costs
of a loan too, is
C(t0 C t)t C(t0 C 2t)t

Cpv D C0 C C C
1 C r1 t (1 C r1 t)(1 C r2 t)
C(t0 C nt)t
C :
(1 C r1 t) (1 C rn t)
Investing in a machine is profitable, if the present value of revenues from

using the machine exceeds the present value of its costs, that is, Rpv >
Cpv . By subtracting the costs from the revenues at every time unit, the
condition for the profitability of the investment can be presented as:
.R(t0 C t) C(t0 C t)/t

Rpv Cpv > 0 ,
1 C r1 t
.R(t0 C 2t) C(t0 C 2t)/t
C C
(1 C r1 t)(1 C r2 t)
.R(t0 C nt) C(t0 C nt)/t
C > C0 :
(1 C r1 t) (1 C rn t)
Thus the investment is profitable, if the net present value of revenues

from one machine during its time of use exceeds the price (value) of one
machine. If Rpv > Cpv , the firm should pay for one machine at most the
present value of net revenues from using the machine; this is the value of
one machine for the firm. Essential in the above analysis is that the future
net revenues are not directly compared with the price of the machine, but
the present value of future net revenues is compared with the price paid at
current moment. The future revenues have the smaller present value the
further away the revenues are, and the greater is the interest rate.
In firms’ investment decisions, firms’ managers do not know the future
interest rate and the revenues and costs from different machines for
certain, and they have to estimate these quantities. Suppose the managers
of a firm can estimate the average revenues and user costs of one machine
at one time unit, which, for simplicity, are assumed constant during the
use of the machine. We assume that the managers of the firm estimate the
profitability of the investment by using one interest rate that represents
the alternative rate of return the managers believe they could get for the
invested money elsewhere. We denote the constant net revenues from one
machine during time unit t0 Cit as N(t0 Cit) D R(t0 Cit)C(t0 C

it) D N (e=t), 8i D 0; 1; : : : ; n. The managers’ estimate of present
value of net revenues from using one machine is then
Nt Nt Nt Nt

Npv D C C C C
1 C rt (1 C rt) 2 (1 C rt) 3 (1 C rt)n

1 1 1 1
D Nt C C C C
1 C rt (1 C rt)2 (1 C rt)3 (1 C rt)n
X n i Xn
1 1
D Nt D Nt ai ; a D ;
iD1
1 C rt iD1
1 C rt
where r (1=t) is the discount rate. In Sect. 7.4.1 we derived the sum of
the above geometric series with positive terms 0 < a D 1=(1Crt) < 1.
In the case of an infinite money flow, the present value Npv is:
a Nt Nt
Nt N
Npv D Nt D 1Crt1 D 1Crt
D D : (7.16)
1a 1 1Crt 1Crt1
1Crt
rt r
Notice that the measurement unit of the interest rate transforms the
measurement unit of the present value as: (e=t)=(1=t) D e.
Next we assume, for simplicity, that the machine lasts forever with its
maintenance costs that are included in user costs. We can then use the
sum of an infinite geometric series in the calculations, which simplifies
the analysis. The condition for profitability of the investment is then:
N
Npv > C0 , > C0 : (7.17)
r
The last form of this inequality shows how interest rate r affects the
investment decision. If N > 0, r ! 0 makes the investment surely
profitable because then the left hand side of the inequality increases
without limit while C0 is limited. If N < 0, the investment is never
profitable. The higher the interest rate, the smaller the left hand side of
the inequality (when N > 0), and thus the smaller the probability that
the investment is profitable.
We can also present the last inequality in Eq. (7.17) as:
N
r< or N > rC0 :
C0
The former of these inequalities shows that the interest rate used in
discounting must be smaller than the rate of return of the investment—
the fixed net revenues from one time unit N (e=t) divided by the
invested capital C0 (e). The latter form of the inequality shows that the
net revenues at one time unit from one machine must exceed the interest
revenues that could be obtained for the invested capital C0 . In Sect. 7.4.2
we showed that the present value of a fixed infinite money flow is the same
in continuous and in discrete time discounting; thus these results hold in
continuous time too.
Example
A firm is planning to buy a machine that is estimated to be used 3 years
producing services of net value 4000 (e=y), and that can be sold after
three years at 10,000 (e). Suppose the firm uses interest rate 10 (%=y) in
its discounting. How much should the firm pay for this machine?
Answer. The discount factors for the three years were calculated in
Sect. 7.3.1. Assuming that the revenues are obtained at the ending moment
of every year, the present value of revenues from the machine is
Npv D 0:91 4000 C 0:83 4000 C 0:75 4000 D 9960 (e);
and the present value of reselling the machine is 0:75 10000 D 7500 (e).
The firm should thus pay at most 9960 C 7500 D 17;460 (e) for the machine.
˘
Suppose a firm has two possible ways to increase its use of capital goods
(machines) at time moment t0 : (1) rent the machines at a fixed hourly rent,
or (2) buy the machines. The need of machine hours per time unit t is
assumed to be fixed B (h=t). The firm can rent these machines at z (e=h)
or buy them at price (value) C0 (e). We assume that the firm cannot
operate without these machines, and so renting or buying the machines is
compulsory. The reselling price of one machine is assumed zero, and the
user plus maintenance costs of one machine at time unit t are denoted
by a (e=t) (< zB (e=t)). With these costs, the rented and bought
machines can be used forever.
The managers of the firm are assumed to use a fixed interest rate
r (1=t) in their discounting. The renting costs of B (h=t) machine
hours from time unit t are zBt (e). By using the sum of an infinite
geometric series, the present value Vpv (e) of renting costs from an infinite
future is:
zBt zB
Vpv D D :
rt r
The present value of costs from buying one machine that produces the
required machine hours and using it an infinite time, Cpv (e), is similarly:
at a
Cpv D C0 C D C0 C :
rt r
The condition for profitability of the investment is then:
a zB zB a
Cpv < Vpv , C0 C < , C0 < , rC0 < zB a:
r r r
The profitability of this investment is based on the difference between

present value of cost savings created by one machine during its time of
use and its price. The higher the interest rate, the more probable the
investment is not profitable because zB a > 0, and the right-hand
side of the third form of the inequality decreases when r increases. The
more expensive a machine is, and the less cost savings it creates, the more
probable the investment is not profitable. Now, zB a (e=t) is the
cost savings of one bought machine at time unit t, and rC0 (e=t)
the interest revenues to be obtained for capital C0 during t. The last
form of the inequality shows that the investment is profitable if the cost
savings from buying a machine are greater than the interest revenues to
be obtained for capital C0 during t.
§: We can name quantities Nr C0 and zBa
r
C0 as forces acting upon
the capital stock of the firm in these two situations. If these forces are
positive, it is profitable for the firm to invest in these capital goods in the
case the firm plans to operate long enough. ˘
The capital stock of a firm is measured by the accumulation function
of investments of the firm, because investments increase the firm’s capital
stock. The accumulation of the capital stock of a firm can be analyzed as a
continuous process, even though in the real world additions in the capital
stock are made in a discrete way. This will be studied in the next section.
Note. Usually firms use their capital goods a finite time. In that case, a
similar analysis of investment decisions can be made as we have made here
by calculating present values of net revenues from a finite time. Examples
of this will be given in Sect. 8.2.6 and here we omit these cases. ˘
7.5.1 Accumulation of Firms’ Capital Stocks
The physical capital of a firm at a certain time moment is a stock quantity

that has accumulated via the firm’s investments. Now, adding the amounts
of different capital goods (computers, trucks, and so on) is difficult due
to the different measurement units for amounts of these goods, and so the
stock of physical capital of a firm is usually measured in monetary terms.
If the stock of physical capital of a firm is expressed in monetary units,
the investments in capital goods are money flows.
Suppose the production of a firm began at time moment t0 , and let us
denote the firm’s investment at moment t0 C it by I(t0 C it) (e=t),
i D 1; 2; : : : . In order to simplify the situation, we assume that the capital
stock does not depreciate due to wearing out or aging. At time moment
t D t0 Cnt, the value of the capital stock of the firm, K(t) (e) (notation
K comes from ‘das Kapital’ used by Karl Marx), is
X
n
K(t) D I(t0 )t C I(t0 C t)t C C I(t0 C nt)t D I(t0 C it)t;
iD0
which is the accumulated sum of investments of the firm during its

history. Notice that we do not use discounting in this; that is, we value
past investments of the firm in terms of current money. The reason for this
is that because these capital goods have been bought and paid for, there
is no reason to compare the past and present money by taking account of
possible interest returns that could have been obtained for the money that
has been used in buying these capital goods. On the other hand, we could
adjust the value of the capital stock of a firm by correcting the values of
old capital goods in terms of their wearing out and aging, or according to
their resale value. However, to simplify the analysis we do not make these
corrections, and we value the capital stock of a firm by the sum of its past
investments.
Next we transform time continuous by letting t ! 0. The value of
the capital stock of the firm at moment t D t0 Cnt can then be expressed
by the following definite integral (Sect. 10.14.2):
X
n Z t
K(t) D lim I(t0 C it)t D I(s)ds; (7.18)
t!0 t0
iD0
where running time during interval (t0 ; t) is denoted by s. According to

Eq. (7.18), the firm’s investment equals the instantaneous flow of its capital
stock,
dK
D K 0 (t) D I(t):
dt
Notice that the derivation here is made with respect to the upper limit of
the integral; see Sect. 10.14.2.
Assuming time continuous does not make an essential difference in
the analysis. Even though the investments are made so that during the
first half of a year one machine with a value of 1000 (e) is purchased, and
during the latter half of the year two similar machines with a value of 1000
(e) are purchased, the annual value of investments is 3000 (e=y). This
corresponds to the average weekly investment flow 3000/52 (e=week),
or average hourly flow 3000=(52 7 24) (e=h). Continuing in this
way we can define the average investment flow from as long a time unit
we like. The change in the capital stock during one year is then
Z ˇ1
1 ˇ
K(0; 1) D 3000 ds D ˇˇ 3000 s D 3000 (1 0) D 3000 (e);
0 0
where running time ds during the year is measured in units y. The same
result can be obtained by using the weekly flow of investment as:
Z ˇ52
52
3000 ˇ 3000 3000
K(0; 52) D ds D ˇˇ sD (52 0) D 3000 (e):
0 52 0 52 52
We can thus analyze the investments of a firm as a continuous process in

time even though the investments are made, for example, once a year.
7.5.2 Investment Decisions in Continuous Time*
Next we assume time continuous and measure it in units y. We analyze

a firm’s investment decision by defining the force acting upon the firm’s
capital stock. Investments are acquirements of capital goods of firms that
are executed when this is profitable. The capital stock of a firm is assumed
to consist of m types of capital goods of the firm. At time moment t, the
capital stock of a firm consists of all capital goods acquired in the firm
during its history starting at moment t0 . Because the values of different
kind of capital goods can be added in monetary units, the capital stock is
measured in units e. We denote the value of the capital stock of the firm
K(t) (e) at moment t as
X
m Z t
dKi
K(t) D Ki (t); Ki (t) D Ii (s)ds; D Ii (t);
iD1 t0 dt
where Ki (t) is the value of the capital stock of the firm of goods of type i
at moment t, and Ii (t) (e=y) the investment of the firm in type i capital
goods at moment t. If Ii (t) < 0, the firm decreases its capital stock of type
i goods at moment t by selling or scrapping the P machines. Investment
Pm
0
means an increase in capital stock, and K 0 (t) D m K
iD1 i (t) D iD1 Ii (t)
with unit e=y is the flow of the value of the capital stock, or the net
investment flow of the firm at time moment t. The capital stock of the
firm can be modeled as:
Ki0 (t) D fi (Fi ); fi0 (Fi ) > 0; fi (0) D 0; Fi D Npvi C0i ; i D 1; : : : ; m;

(7.19)
where functions fi , i D 1; : : : ; m obey the described characteristics. Npvi

is the present value of cost savings of one capital good of type i for the firm
in the case these goods can be bought and rented, and the present value of
net revenues of one capital good in the case capital goods of type i cannot
be rented. C0i (e) is the price (value) of capital good i at moment t. The
force acting upon the capital stock of the firm of type i goods, Fi , is
the difference between present values of cost savings or net revenues of
one capital good of type i and its price (value). The higher this difference
is, the greater is the force and vice versa. If one of the force components
in Fi is negative, it is profitable for the firm to sell these capital goods by
assuming that these can be sold at value C0i . In the zero force situation,
the capital stock of the firm stays constant.
7.5.3 A Dynamic Theory of Investment*
We analyze the investment decision of a firm in certain type of machines at

time moment t. Time is assumed continuous and to be measured in units
y. For simplicity, the interest rate r (1=y) used in discounting is assumed
fixed. We denote by N (e=y) the estimated annual cost savings from one
bought machine in the case that the firm can buy and rent these machines.
The maintenance costs of machines are included in the costs so that with
these costs the machines last forever. In a case where the firm cannot rent
the machines but only buy them, we denote by N (e=y) the annual net
revenues from one machine. For simplicity, N is assumed fixed. If the firm
takes a loan for its investment, the interest cost and part payment of the
loan are included in the annual costs of one machine.
In Eq. (7.19) we claim that the capital stock of a firm increases, K 0 (t) >
0, if F D Npv C0 > 0, Npv D N=r, where N (e/y) is the net income
flow from investing in capital goods and r(1=y) the discount rate of the
firm. In order to get this modeling comparable with that to be done in
Chap. 9, we transform the force Fi acting upon the capital stock of the
firm as:
N N
K 0 (t) > 0 if F > 0 , > C0 , > r; and vice versa:
r C0
Thus the capital stock of the firm increases if the internal rate of return of
investing in capital goods is greater than discount rate r.
We linearize function fi in Eq. (7.19) in the neighborhood of point Fi D
0 by the Taylor series approximation, assume the error term zero, and omit
the subindex for shortness. The investment decision of the firm at time
moment t can then be modeled as

0 0 N 1
K (t) D f (0) r C FKS ; f 0 (0) D > 0; I(t) D K 0 (t); (7.20)
C0 mK
where FKS (1=y) is the static friction force of the capital stock, and
f 0 (0) D 1=mK a positive constant where mK with unit 1/e represents
the ‘inertial "mass" of the capital stock of the firm’. Static friction is
needed in the model to explain that the capital stock of the firm may not
always be changed when a non-zero force is acting upon it. The static
friction force contains all factors resisting changes in the capital stock not
included in the revenues and the costs of the machines. These factors are
added up to constant FKS the numerical value of which can be estimated
on the basis of observations.
The ‘free body’ diagram of the forces acting upon the capital stock
K is in Fig. 7.3, where the static friction force is omitted for simplicity.
The direction of motion on the right is defined positive, and on the left,
negative. The positive force component acting upon the capital stock
is N=C0 , and the negative force component is r. Quantity K(t) on the
horizontal axis measures the capital stock, and quantity mK with unit
y resists changes in the motion of the ‘particle’. Similarly, as described
earlier, the shape of the ‘particle’ has no economic meaning, and the
particle should be drawn as a point on coordinate axis K(t). However,
the box shape for the variable the capital stock of the firm visualizes better
the analogy we make here with Newtonian mechanics.
Fig. 7.3 Free body diagram of the forces acting upon the capital stock
The force acting upon the capital stock of the firm, FK D N=C0
r C FKS , has unit 1=y. Notice that Eq. (7.20) does not exactly correspond
to Newton’s equation F D ma, because K 0 (t) is velocity, and not
acceleration, of the capital stock. According to Eq. (7.20), the capital
stock increases (decreases) with time if force FK is positive (negative). An
increase in N positively, and increases in C0 and r negatively affect the
force. This investment behavior corresponds to the principle of modeling
in economics (see Sect. 1.2.4). In the case the active force component
N=C0 r exceeds the static friction force FKS , the following dynamic
equation and its solution result:

0 N 1 N
mK K (t) D r ) K(t) D K0 C r t; (7.21)
C0 mK C0
where K0 with unit e is the constant of integration. Thus the capital stock
increases linearly with time if CN0 > r, and the greater the mK the slower
this adjustment.
In Fig. 7.4 is annual capital formation in current prices of four macro
sectors in Finland: non-financial firms, financial firms, public sector, and
households. Non-financial firms have been the greatest investor of these
30000
25000
20000
15000
10000
5000
0
75
77
79
81
83
85
87
89
91
93
95
97
99
01
03
05
07
09
11
13
19
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
–5000
Non-financial firms Financial firms Public sector Households
Fig. 7.4 Gross fixed capital formation in Finland

sectors in Finland, and financial firms the least. A positive time trend exists
in investment in all other sectors except financial firms, and a relatively
great variability exists in every sector. Thus to get the model for firms’
investment behavior in Eq. (7.21) to correspond more accurately with
observed cyclical behavior as in Fig. 7.4, a more detailed force should be
created acting upon the capital input of a firm. The observed fluctuations
in investments originate either from changes in interest rate or from
changes in expected net revenues from firms’ investments. These elements
can be added in Eq. (7.21) by modeling the expected net revenues in a
more detailed way.
References
Chiang, A. C. (1984). Fundamental methods of mathematical economics (3rd ed.)
Singapore: McGraw-Hill International Editions.
Hull, J. C. (2000). Options, futures, & other derivatives (4th ed.). USA: Prentice-
Hall International, Inc.
8
Money and Financial Markets
8.1 A Short History of Money

There have been various periods in the history of money. Before money
was invented, the relative values of goods were expressed in the units of the
goods that were bartered with each other. For example, a certain amount
of cotton was exchanged for a negotiated amount of salt. Gradually,
certain goods began to be used as measurement units of values of other
goods. In this way the first goods money was invented. Corn, cattle, fur,
cocoa and precious metals are among the examples of such goods money.
In the Finnish language, the original meaning of the term ’money’ was
squirrel fur. The benefits of precious metals as compared with other
goods money were:
1. Transportability: Great intrinsic value relative to the size and weight.

2. Divisibility: ‘Money’ could be used in small exchanges too.
3. Homogeneity: The ‘quality of money’ need not be checked.
4. Stability of value : In some transactions, the medium of exchange
needed to be stored for some time during which its value had to stay
constant.

DOI 10.1007/978-3-319-46879-2_8
Precious metals fulfilled these criteria better than, for instance, furs
and corn, and this explains their gained popularity with time. The
transportation, storing, and division of precious metals was not easy,
however. Gradually people realized that they did not have to carry precious
metals (money) with them. Rather, they could store (deposit) the metals
in the vaults of goldsmiths, for example, who had safe places for their
raw materials. Goldsmiths gave receipts for these deposits, in the form of
documents stating the amount of precious metals a person had stored in
the vault, and people started to use these receipts, which were guaranteed
by the goldsmiths, as means of payments. These were the predecessors of
modern paper money, which gradually started to take over from precious
metals as the medium of exchange.
When the exchanges of goods increased in Europe in the beginning of
seventeenth century, the payment system was centered in trading houses
called Giro banks. Giro banks accepted precious metals as deposits, coins
were struck from these metals, and receipts for these ‘deposits’ were given
to the ‘depositors’. These receipts–the first banknotes– were then used as
means of payments of goods. Giro banks were predecessors of the modern
central banks that have the exclusive right to issue paper money. By the
eighteenth century, in some countries alongside precious metals money,
banknotes were also in circulation with a limited use; they were accepted
only locally. Gradually, public authorities gave itself the uniform right to
issue money, and this way money started to lose its meaning as a good. In
Finland, the Bank of Finland has had the exclusive right to issue money
since 1886.
In a precious metals-based money system, banknotes are so-called
representative money. The issuer of the banknote (has written the receipt
for the depositor of gold or silver), has promised to give the precious metal
to the holder of the receipt (note) at full value. In this way banknotes
represented the amount of gold or silver stored in an organization, which
thereafter began to be called banks. As mentioned earlier, the first such
organizations were goldsmiths who needed the metals as raw material for
their products. Later on some goldsmiths started to concentrate on the
deposit business, and they were the predecessors of modern banks. The
advantage of banknotes as compared with precious metals was that the
‘deposited’ gold or silver did not need to be transported from one person
to another; only the owner of the receipt changed in the exchange, and
precious metals stayed safely on the ‘bank account’.
The next innovation in the means of payments came in the form of a
check.
§: A check is a request to pay its holder a written amount of money
from the checking account of the writer of the check. ˘
The advantage of checks as compared with banknotes is that there is less
need to hold cash money. This decreases the risk of losing money or being
robbed. The newest forms of medium of exchange are bank and credit
cards that work like checks. Their advantages as compared with checks are
lower transaction costs and more smooth functioning. The disadvantage
of bank and credit cards is the uncertainty about the payer’s ability to pay.
Various banks have guaranteed the payments by their bank cards up to a
limit because banks can judge the credibility of their clients.
The last innovation in the means of payments is a computer-based
payment system that charges the payer’s bank account by the amount of
the payment, and transfers the money to the receiver’s account. No cash is
needed, and time will tell as to how much cash money will eventually be
needed in the future, after most people begin to use electronic methods
of making payments.
8.1.1 Money Systems in Brief
In a gold standard paper money system, the issuer of the paper money
is responsible for holding the required gold reserves that secure the gold
value of the paper money. In the beginning of twentieth century, the
world’s monetary system was secured by the gold reserves of central banks.
The gold reserves in central banks equated to the total value of banknotes
(paper money) in circulation. Gradually, central bankers realized that
they could issue more banknotes than they had gold reserves, because
the probability that all people would simultaneously come to redeem
their gold was low. In this way, the fractional reserve banking principle
was developed, where banks keep only a fraction of their deposits in
reserve.
In the early stages of central banks, alongside the issuing of paper

money they took part also in the lending business along with commercial
banks. Thus central banks were motivated to issue more banknotes than
they had gold reserves. A central bank that issues more notes than it
has gold reserves applies the fractional reserve principle in its issuing of
money. Commercial banks, on the other hand, apply the fractional reserve
banking principle in their lending. Commercial banks hold a part of their
deposits in reserve and issue most of their deposits forward in the form of
loans. The fractional reserve principle makes the banking business risky,
but it is the only way banks can make profit because those banks that do
not issue loans of their deposits make losses due to the interest payments
on their deposits.
In the beginning of 1930s, a worldwide depression occurred. This
depression collapsed stock prices, which decreased people’s trust in securi-
ties, and this distrust extended to paper money too. The distrust of paper
money encouraged people to exchange their paper money for gold, which
was believed to better keep its value during the depression. Consequently,
central banks started to lose their gold reserves, and soon they realized
that they could not maintain the fixed exchange rate between banknotes
and gold due to the excess amount of notes issued. For example, the gold
reserves in the Bank of England diminished so much during the 1920s
that England had to give up the gold standard in 1931. This devalued the
pound sterling, which had a serious effect on the world’s payment system.
The international banking and payment system is so highly integrated
that a panic in a major banking center affects the payment system of
the whole world. Due to the integration of the international payment
system, in the 1930s various European countries also had to give up the
gold standard. It was thought then that giving up the gold standard was
a temporary event, but it has turned out to be permanent. The current
monetary system in almost every Western economy is the so-called free
money system or a paper money system.
In a paper money system, notes are not redeemed by gold or silver.
Modern notes are pure paper (fiat) money, and their value is based
on people’s trust on the purchasing power of the money in the future.
Central banks hold gold, financial assets, and other currencies in reserve
so that people can exchange their cash money to other currencies if

they are not confident about the value of the home currency. The
exchange rates of currencies change with time via their demand and
supply, so that when people become skeptical of a certain currency, they
start exchanging this to other currencies which decreases the value of
the currency. Flexible exchange rates guarantee that the central banks’
reserves of foreign currencies do not get exhausted. All exchange rate
collapses, of which we have evidence, have occurred in situations
where a central bank has applied a fixed exchange rate policy in
the case where domestic currency has been overvalued. Once people
realize this situation, they start exchanging the domestic currency to
other currencies, and eventually the reserves of foreign currencies of
the central bank get exhausted and it must devalue the domestic cur-
rency.
A change from a gold standard paper money system to a pure paper
money system is made so that public authorities declare central ban-
knotes as the legal medium of exchange. This way money loses its value
as a good, and a fixed transformation rule between money and gold ceases
to exist. If the central bank can then freely change the exchange rates of
the domestic currency, the system is called a free paper money standard.
If, on the other hand, the exchange rates of the home currency are kept
fixed with respect to other currencies, for instance, via agreements in
international trade or policy, the system is called a binding paper money
standard.
A paper money system requires something corresponding to the rep-
resentative money. In a gold standard paper money system, the amount
of gold in the vaults of the central bank was equal to the value of all
paper money in circulation. However, according to the fractional reserve
banking principle, a 100% gold coverage is not necessary. We can then
ask, how much cash money can be issued so that it does not threaten
the behavior of the monetary system? Nobel laureate Milton Friedman
(1968) suggested that the amount of money in circulation can increase at
a growth rate equal to that of real production. Nowadays, this monetarist
view is generally accepted by central banks so that in modern paper money
systems, changes in the aggregate money in circulation reflect changes in
the aggregate flow of production in the monetary area; a country or a
monetary union. The more goods are produced, the more money can be
issued to buy these goods. An excess supply of money over real production
causes excess demand for goods, which causes either price inflation or a
stock market bubble.
We can remark here that before money was invented, goods were
exchanged with each other, and the ‘amount of money in circulation’
was equal to the amount of exchangeable goods. A modern paper money
system more closely resembles the exchange system of goods where every
produced good increases the monetary base (see the next section) than
the gold-based paper money system. This is reasonable because a limited
amount of gold exists in the Earth, and the aggregate flow of production
of goods in the world has no upper limit. Thus, there is no sense in tying
the monetary base of a growing economic system to a fixed quantity.
8.1.2 The Functions of Money
In the 1870s, William Stanley Jevons (1970) presented three main func-
tions of money as: (1) A means of making payments, (2) A measure of
monetary values of goods, and (3) A liquid form of wealth. By ’liquid
wealth’ we understand such forms of wealth that can easily and quickly be
transformed to cash money. In addition to these functions, Jevons defined
another: A measure for the amount of savings and debts. Later, this fourth
function of money was seen as a part of a measure of values. We can thus
state the second function as: A measure of the values of goods, future
payments, and savings and debts. However, we can still define the fourth
function of money as: (4) The final amortization of a debt. People can
amortize their debts by other debts, but only by paying with cash money
can a loan finally be amortized.
The concepts of money and a monetary economy can be defined on
the basis of the functions of money.
§: We call money, or legal tender, the legal medium of exchange that
every participant in the economy is obligated to accept. ˘
§: Essentially, two forms of legal tender exists: Principal money (notes)
and exchange money (coins). Everyone has to accept principal money as
the medium of exchange, but exchange money needs to be accepted only
up to a small amount. However, the official issuer of legal tender has to

accept exchange money without limit. ˘
Example 1
It is common for customers to request shopkeepers’ permission to pay for
their purchases with large amounts of coins, because legally shopkeepers
do not have to accept this. However, those shopkeepers who like to keep
their customers usually agree to this request. ˘
§: A monetary economy is one where a common good called money

(or currency) has been defined with the earlier given functions. ˘
§: A monetary system covers the laws and regulations concerning legal
means of payments. ˘
Besides notes and coins—together called cash—banking and credit
organizations have also created bank money that has a limited use in
the medium of exchange; organizations and people do not have to accept
bank money as a means of payment. For instance, bank cards, credit cards
and checks are bank money.
Example 2
Bank and credit cards are not accepted by all shops, and so they have a
limited use in the medium of exchange ˘
National means of payments consist of the legal means of payments

together with bank money. When bank money is added into legal means
of payments, we get a money aggregate of the economy. A money
aggregate of an economy can be defined in various ways depending on
the forms of money included in the concept; that is, how liquid are
the means of payments that are included in it. The money aggregates
are defined in somewhat different ways in various countries, and the
following definitions are applied by the European Central Bank. M1
is the narrowest money aggregate that includes only the most liquid
means of payments: currency in circulation plus overnight deposits.

M2 comprises M1 plus deposits with an agreed maturity of up to and
including three months. The broadest money aggregate M3 comprises
M2 plus repurchase agreements, money market fund shares, and units as
well as debt securities with a maturity of up to and including two years;
see http://www.ecb.int/pub/.
Besides money aggregates, we need to define the monetary base.
§: The monetary base of an economy (monetary area) consists of
the liabilities of the central bank of the economy (monetary area). The
liabilities of a central bank consist of the currency in circulation (notes
and coins), and the deposits of banks in the central bank called bank
reserves. ˘
The monetary base is also called high-powered money, because the
monetary base affects the amount of bank money, but bank money does
not affect the monetary base. Banks create bank money from the amount
of currency in circulation, and so the monetary base is the key factor that
affects the money aggregates. Central banks control the money aggregates
by the monetary base, which we call monetary policy. These matters are
studied further in textbooks on macroeconomics.
8.2 Fundamentals of Financial Markets

§: We call financial markets the markets where the surplus funds
(savings) of economic units– firms, households, government, local gov-
ernments, investors, and so on–are channeled to those economic units
that have a shortage of funds; that is, to those who wish to borrow
funds. ˘
Why is this channeling of funds from savers to borrowers important
for an economy? The answer is that those who save are frequently not
the same economic units that have profitable investment opportunities.
Suppose you have saved 1000 (e), you have no use for that money, and
financial markets do not exist in the economy. Then you will keep the
money in your pocket, no one can use your money for investment, and
you do not get interest earnings on it. Suppose then that a firm with a
profitable investment opportunity is willing to borrow your money and

pay it back after a year with an annual interest of 5 (%=y). In this case
you would get compensation for not using your money, and a profitable
investment gets financed in the economy. The introduction of financial
markets, where the demand of loans and supply of savings meet, is thus
a Pareto improvement for an economy as compared with the situation
that such markets would not exist.
§: According to Italian economist Wilfred Pareto (1848–1923), we call
Pareto improvement a change that improves the welfare of some people
without decreasing that of others. ˘
Another example of the importance of financial markets is housing
loans. A person who does not own wealth, but has a regular salary, can get
a mortgage loan for buying a house (see Sect. 8.2.3). Without financial
markets, only a few of us could buy a house, and so lively housing markets
would not exist in economies. This shows also the importance of collateral
in the lending business.
§: A collateral is a general term for various kinds of property that
borrowers offer to lenders to guarantee their repayment of their debt. ˘
8.2.1 Direct Finance
We call direct finance borrowers’ direct borrowing of funds from lenders

by issuing them securities (or assets or financial instruments). Securities
are claims on the borrowers’ future income and assets. While securities are
assets for their holders, they are liabilities for their issuers. For example,
if General Motors aims to build a factory, it can raise funds from savers
by issuing them the shares or the bonds of GM; see Sect. 8.2.3.
§: The collection of the securities owned by an economic unit is called
the portfolio of his assets. ˘
The reason that securities are financial instruments is that various
securities can be used as ‘instruments’ through which investors can affect
the risk, return (yield), and liquidity of their portfolio of assets. Because
every asset has a specific combination of yield, risk, and liquidity, a
change in the content of a portfolio affects its aggregate yield, risk, and
liquidity.
8.2.2 Indirect Finance
Funds from lenders to borrowers can move directly or indirectly. We call

the latter indirect finance and it needs a middleman, called a financial
intermediary.
§: A financial intermediary issues liabilities to lenders–savers and
then, in turn, issues loans to borrowers–spenders. ˘
Different kinds of financial intermediaries exist in various countries.
For example, the following financial intermediaries exist: (1) Depository
institutions, (2) Contractual savings institutions, and (3) Investment
intermediaries. Depository institutions are: (1) Commercial banks, (2)
Savings and loan associations, (3) Mutual savings banks, and (4) Credit
unions. Contractual savings institutions are: (1) Life insurance companies,
(2) Pension funds, and (3) Fire and casualty insurance companies. Invest-
ment intermediaries are: (1) Mutual funds, (2) Money market mutual
funds, and (3) Finance companies (Fig. 8.1).
Financial intermediaries collect money from financial markets by issu-
ing securities, and earn money by making loans of these funds or by
investing them, and by taking care of the risks of economic units. In the
following, we do not study the behavior of different kinds of financial
intermediaries in a detailed way. We identify all financial intermediaries as
banks, and we study the behavior of a typical bank that acquires funds by
DIRECT FINANCE
Lenders/savers Financial markets Borrowers/spenders

-Houesholds Funds -Money market Funds -Houesholds
-Firms -Capital market -Firms
-Government -Government
-Non-residents -Non-residents
Funds
Financial intermediaries
Funds Funds
-Credit institutions
-Other financial institutions
INDIRECT FINANCE
Fig. 8.1 Direct and indirect finance

issuing liabilities in the form of various types of deposits, and earns money
by making loans of these funds or by investing them. For example, a bank
can give loans to firms and households or buy government or corporate
bonds (see Sect. 8.2.3). This way funds are transferred from lenders–
savers to borrower-spenders with the help of a financial intermediary (the
bank). By charging a higher interest rate for its loans than it pays for its
deposits (the interest differential), a financial intermediary earns money.
Financial intermediaries improve the efficiency of financial markets
in the following ways:
1. Small savings can gain a higher interest rate as a part of a large loan.
2. Small firms, for which it would be impossible to get funds in the form
of direct finance, can get relatively large loans from banks.
3. Financial intermediaries reduce information costs. It would be very
expensive for lenders to identify all potential borrowers, and for bor-
rowers to identify all potential lenders.
4. Once a lender finds a potential borrower, he/she/it has to consider the
probability that the borrower will default (see the definition later).
Financial intermediaries, on the other hand, have regular information
on the financial situation of their clients through their bank accounts.
This gives them superior information of the risks of their clients as
compared with other lenders.
5. Financial intermediaries reduce transaction costs as compared with the
situation that every lender and borrower writes a loan contract, or pays
a brokerage commission for the transaction. Smaller transaction costs
related to one large loan as compared with many small loans creates
economies of scale in the lending business (lower unit costs at a larger
scale of operation).
6. Financial intermediaries can create scale and maturity transformations
between financial agreements. For example, from a continuous inflow
of small short-term deposits from various sources with different interest
rates, a bank can issue large long-term loans with a fixed interest rate.
7. The expertise and education of the personnel in banks allows them to
make better investment decisions as compared with many small savers
with less information. Investing large sums of money, though, may
create large losses in a case where banks make bad investment decisions.
8. If a bank has enough independent depositors and borrowers, the risks

related to one client do not threaten the existence of the bank, which
might happen in the case of a small financial unit.
§: By default we understand the situation that an economic unit is

unable to fulfill its financial obligations. ˘
8.2.3 Financial Market Instruments
The distinction between money and capital markets is made on the basis
of the maturity of securities traded in the markets. In practice, specific
market places do not exist for these markets, and the securities traded in
both markets are often sold by the same financial intermediaries.
§: The maturity of a debt instrument is the length of the time (term)
to the instrument’s expiration date. ˘
§: The money market is the financial market where short-term debt
instruments (maturity less than one year) are traded. The capital market
is the market where long-term debt and equity instruments (maturity over
a year) are traded. ˘
§: An equity is a financial instrument that gives its holder a specified
ownership concerning the underlying matter. For example, shares of
common stocks of corporations are equities. ˘
Money Market Instruments
The following assets are traded in the money market:
1. A Negotiable bank certificate of deposit (CD) is a debt instrument

issued by banks to gather deposits. A CD pays fixed annual interest and
pays back the principal at the date of maturity.
2. A treasury bill is a short-term debt instrument issued by governments
usually in three-, six-, or twelve-month maturity. T-bills pay the
principal (nominal, face) value at maturity and do not pay interest.
However, T-bills effectively pay interest by initially being sold at a
discount, that is, at a lower price than the principal value. T-bills are the
most liquid asset in the money market with the minimum possible risk
of default, because they are issued by governments and actively traded
every weekday.
3. A commercial paper is a short-term debt instrument issued by large
banks and well-known corporations.
4. A banker’s acceptance is a bank draft (a promise of payment) issued
by a firm payable at some future date, and guaranteed by the bank who
stamps it as ‘accepted’. The firm issuing a banker’s acceptance deposits
the required funds in its bank account at the date of maturity to cover
the draft. If the firm fails to do this, the bank’s guarantee obligates
the bank to cover the draft. These ‘accepted’ drafts can be resold in
secondary markets (see the next section) at discount similarly to T-bills.
5. A Repurchase agreement (RP) is an agreement to buy securities on the
understanding that they will be repurchased at some specified time and
price in the future. In this deal, the seller corresponds to the borrower
and the buyer the lender.
6. An Interbank deposit is a loan between banks ranging from one day
to one year. This market is highly sensitive to the credit needs of banks,
and the interest rate on these loans is a closely watched barometer of
the tightness of the credit market. In the euro area, this interest rate is
called the euribor and in the USA the federal funds rate.
Capital Market Instruments
The following assets are traded in the capital market:
1. Shares of the common stock of a corporation are equity claims on

the net income and assets of the corporation. Shares usually give annual
dividend payments, and capital gains to shareholders in the event the
share price increases.
2. Mortgages are loans given to individuals or firms to purchase land,
housing, or other real structures where the structure or land serves as
the collateral for the loan.
3. Corporate bonds are long-term debt instruments issued by corpo-

rations with strong credit ratings. A typical corporate bond pays the
holder a fixed interest (coupon) payment once or twice a year, and
pays back the face value at the date of maturity.
4. Government bonds are long-term bonds issued by governments.
5. Municipal bonds are long-term bonds issued by local governments.
6. Consumer and firm loans are consumers’ and firms’ liabilities issued
by financial intermediaries.
Usually, no secondary market (see the next section) exists for loans, and
so they are the least liquid of all capital market instruments. However, in
some countries banks are allowed to securitize the loans they have issued
by grouping them in a pool, and selling equities as shares of ownership of
these pooled loans in secondary markets. Thus bank loans can be resold
in secondary markets after being securitized in similar debt instruments
to bonds.
8.2.4 Primary and Secondary Markets
We studied direct and indirect finance in Sects. 8.2.1 and 8.2.2. In direct
finance, organizations raise funds from financial markets by issuing bonds
or equities that are claims on the net income and assets of the issuers.
If you own one share of the common stock of a corporation that has
issued a hundred shares, you have a claim on 1% of future net income
and assets of the firm. Equities give their holders annual payments called
dividends, while bonds give their holders annual or semi-annual (twice-a-
year) coupon payments and pay back the principal at the date of maturity.
Thus bonds have a fixed maturity while equities have no maturity. Equities
can, though, be resold in secondary markets.
An organization can raise either own (capital with ownership rights) or
foreign capital (debts) from financial markets. For example, equities are
‘own’ and bonds and loans are ‘foreign’ capital. The main disadvantage of
holding a corporation’s equities rather than its debt instruments is that an
equity holder is a residual claimant; that is, the corporation must fulfill its
obligations to its debt holders before its equity holders. On the other hand,
the advantage of holding equities is that equity holders benefit from the
increased profitability of a corporation and its asset value, since equities
confer ownership rights to their holders. The debt holders of a firm do not
benefit from firm’s increased profitability, because the payments on loans
are not affected by the firm’s profitability. However, the risks of holding
equities are greater than those of debt instruments issued by a firm. These
questions are studied further in Sect. 8.2.6.
§: The Primary market is the market where the issuing of new
securities, such as bonds and equities, takes place. ˘
§: The Secondary market is the market where previously issued
securities are resold. ˘
The primary markets of securities are not well known to the public
because the issuing of new securities takes place behind closed doors.
The issuing of new securities usually takes place as follows. A financial
firm buys a given amount of new securities (shares, bonds, etc.) from
the issuing organization in the primary market at an agreed price. In this
case, we call the financial company a dealer because just like with playing
cards, someone must deal the cards (securities) to the players before the
game can start. The dealer resells the securities in the secondary market
at the price determined there by the demand and supply of the securities.
Investors buy and sell securities in the secondary market like card players
who try to change ‘bad’ cards to ‘better’ ones in order to win the game. The
dealer must thus estimate the price at which the securities can be resold
in secondary markets, which secondary market will be applied and when
the emission is profitable to be done. If the average price index of a stock
market is decreasing, and no demand exists for new securities, there is no
sense in issuing new securities in that market, and vice versa. Nowadays,
it is common for big corporations to issue their equities and bonds in
secondary markets in various countries, depending on the situation in
these markets.
When someone buys a security in a secondary market, the person who
sells the security receives the money in exchange for the security, but
the corporation that has originally issued the security receives no funds.
The issuing corporation receives funds only when it issues new securities
in a primary market. The New York and Helsinki Stock Exchanges are
examples of secondary markets. Other kinds of secondary markets are the

foreign exchange, futures, and options markets. In this book we do not
treat financial derivatives like futures and options, however, and a reader
interested in these matters can turn, for example, to Mishkin (2001) or
Hull (2000).
Secondary markets serve three functions. First, they allow the reselling
of financial instruments, which improves their liquidity. The increased
liquidity of financial instruments makes them more desirable, which helps
in issuing new securities in primary markets. Second, the higher the price
of a security in a secondary market, the higher the price for which the
issuing organization can sell similar new securities in a primary market.
This connection between asset pricing in secondary and primary markets
has focused the research of asset price determination in secondary and
not in primary markets. Third, secondary markets open a market for risk.
Those who consider the assets they hold too risky for them can sell them
in secondary markets, and vice versa. This third function of secondary
markets opens a possibility to share the risks of economic units at a
competitive price. The operation of insurance companies, for example,
is based on this function.
8.2.5 Exchanges and OTC Markets
Secondary markets can be organized basically in two ways. One way is to

organize the exchanges so that the buyers and sellers of securities (or their
agents called brokers) meet in a central location to conduct the trades.
The New York Stock Exchange for shares and the Chicago Board of Trade
for commodities (wheat, silver, etc.) are examples of organized exchanges.
Another way to organize a secondary market is the Over-The-Counter
(OTC) market, where the dealers at different locations with inventories
of securities buy and sell securities ‘over-the-counter’. Because OTC
dealers are in contact with each other via computers and they know the
prices set by each of them, the OTC-market is very competitive and does
not much differ from an organized exchange.
8.2.6 Pricing Financial Instruments
§: We call risk-free arbitrage a situation where an economic unit buys a

security or a good from one market, and sells it immediately in another
market at a profitable price. ˘
Risk-free arbitrage gives the arbitrator a risk-free profit. If such sit-
uation occurs in any two markets, we can believe that someone will
utilize this opportunity because various professionals search for profitable
opportunities in financial markets at every minute. Due to this, most
pricing assumptions in financial markets are based on the idea that risk-
free arbitrage is impossible. This is the idea behind efficient market
hypothesis; see, for example, Fama (1970).
§: In an efficient market, asset prices are adjusted so that risk-free
arbitrage is impossible. ˘
In Sect. 8.2.8 we will study asset price dynamics further, but already
now we can notice that if a possibility exists for a risk-free arbitrage, this
will increase the demand of these assets (goods) in the low price market
and increase their supply in the high price market. The adjustment of the
two prices via excess demand and supply then eliminates this possibility
with time.
Consols
§: A financial asset of infinite horizon with a fixed payment in a time unit,

and no repayment of the principal, is called a consol. ˘
According to Sect. 7.4.1, the present value P(t0 ) (e) of a constant
money flow N (e/t) with discount rate rd (1=t) at time moment t0 is:
N
P(t0 ) D : (8.1)
rd
The pricing formula for a consol, or its equilibrium price (value), is given
in Eq. (8.1). The price of a consol is determined in the market according
to the fixed flow of payments N (e/t) it offers and the prevailing
interest rate; the greater the former and the lower the latter, the greater the
equilibrium price of a consol. Consols were sold for the time by British
Treasury during the Napoleonic Wars, and are still traded today.
Equation (8.1) can also be written as
N
rd D or N D rd P(t0 ):
P(t0 )
The former of these equations shows that the discount rate applied equals
the rate of return of the consol—the fixed money flow N (e=t) divided
by the invested capital P(t0 ) (e). The latter equation shows that the
revenues from a consol are equal to the interest revenues to be obtained
for capital P(t0 ) (e) at time unit t. At the equilibrium price of a consol
P(t0 ) (e), investing in consols is as profitable as investing with a risk-free
interest rate; thus at price P(t0 ), risk-free arbitrage is impossible.
Loans, Bonds, and T-Bills
The calculation of present values of money flows is important in valuing

financial instruments. For example, a bond creates a discrete income flow
during its maturity, and the asset is valued (priced) in a secondary market
according to the present value of this money flow and the present value
of repayment of the face value at the date of maturity.
The value of a loan for the lender can be calculated as the present
value of its future interest and part payments. Suppose someone wants
to borrow from you 100 (e) and promises to pay back 110 (e) after one
year. What is the rate of return of this financial instrument, if no default
risk exists in the loan? The rate of return rd of the loan is
Œ110 (e) 100 (e)=t

rd D D 0:1 (1=t); (8.2)
100 (e)
and because t D 1 (y), the rate of return is 10 (%=y). We call this

the yield to maturity or the internal rate of return of this financial
instrument. The term ‘to maturity’ stresses that the yield is calculated from
the whole time the security exists. This shows the principle of calculating
the internal rate of return of a treasury bill that does not explicitly pay
interest returns. Due to this, T-bills are also called zero-coupon bonds.
Now, the present value of payment 110 (e) after one year with interest
rate 10 (%=y) is:
110 (e) 110 (e) 110 (e)

D D D 100 (e):
1 C rd t 1 C 0:1 1 1:1
If, however, interest rate increases from 10 (%=y) to 12 (%=y), the present
value of the repayment of the loan decreases to
110 (e) 110 (e)

D D 98:21 (e);
1 C 0:12 1 1:12
and if the interest rate decreases to 5 (%=y), the present value increases to
110 (e) 110 (e)

D D 104:76 (e):
1 C 0:05 1 1:05
These examples show that the present value of a fixed payment asset
decreases when the interest rate in the economy increases, and vice versa.
Thus an interest rate risk exists in a fixed payment asset although no
risk exists in monetary payments of such assets. This holds for all fixed
payment securities like T-bills, bonds, and consols.
Note. If we know the current price and the future payments of a security,
we can calculate its rate of return. We call this the internal rate of return
of the financial asset. On the other hand, if we know the future payments
of an asset, and we have an estimate of the future interest rate in the
economy, we can calculate the expected present value of this asset. Thus
we call the interest rate that is used in discounting either the expected
future interest rate in the economy—when we calculate present values—
or the internal rate of return of a security—when we solve the unknown
rate of return of a security. ˘
§: The yield or the rate of return of a security is calculated by taking

account all forms of income the security gives to its holder. ˘
This definition means that in calculating the yield of buying one share
of the common stock of a corporation, together with dividend income we
must take account of the capital gains or losses that are caused by changes
in the share price.
The present value of a fixed payment N (e=t) asset with n years can
be calculated by using formula (7.13) in Sect. 7.4.1 as
n
X j
1 N
P(t0 ) D Nt D Œ1 (1 C rd t)n : (8.3)
jD1
1 C rd t rd
Together with annual or semiannual payments, many debt instruments

(like bonds) repay the principal at maturity. The current value (price) of
a bond with term to maturity of n time units is then:
N Z
P(t0 ) D Œ1 (1 C rd t)n C : (8.4)
rd (1 C rd t)n
The pricing formula in Eq. (8.4) is obtained from Eq. (8.3) by adding
to it the present value of repayment of the principal denoted by Z (e).
Similarly, as in the previous case, an interest rate risk exists in this asset.
Next we test numerically how the pricing formula in Eq. (8.4) works.
Example 1
Suppose the principal is Z D 100 (e), the bond offers 5 (%/y) coupon rate,
that is, N D 0:05 (1=y) 100 (e) D 5 (e=y), and the interest rate in the
economy is rd D 0:05 (1=y). Then the present value P(t0 ) with different values
of n, n D 0; 1; 2; 3; : : : , is always P(t0 ) D 100 (e). ˘
Example 1 shows that if the coupon rate of a bond equals the prevailing
interest rate in the economy, the price of the bond, that eliminates the
risk-free arbitrage possibility, is equal to its principal value. This result
does not depend on the term to maturity of the bond.
Example 2
Suppose Z D 100 (e), N D 0:1 (1=y) 100 (e) D 10 (e=y) and rd D 0:05 (1=y).
Then if n D 1, P(t0 ) D 104:8 (e), and if n D 10, P(t0 ) D 138:6 (e). ˘
Example 3
Suppose Z D 100 (e), N D 5 (e=y) and rd D 0:1 (1=y). Then if n D 1, P(t0 ) D
95:5 (e), and if n D 10, P(t0 ) D 69:3 (e). ˘
Examples 2 and 3 show that if the coupon rate of a bond is greater

than the prevailing (or expected) interest rate in the economy, the price
of the bond is greater than its principal value, and vice versa. The greater
the term to maturity of a bond, the higher is its price if it offers a higher
coupon rate than the interest rate in the economy. The longer you get
higher earnings than on the average in the economy, the greater is the
price of such asset.
Shares of Common Stocks of Corporations
The pricing formula for one share of the common stock of a corporation
equals with that in Eq. (8.4), but in this case n denotes the investor’s
investment horizon (the length of time the investor plans to hold the
share), Z the price (value) at which the investor believes he can sell the
share at the ending moment of his investment period, and N the investor’s
estimate of dividends for one share at every time unit t during the time
the investor holds the share. This current value is based on the investor’s
expectation of future share price, dividends, and interest rate. Thus there
exists a risk in quantities N, rd , and Z, but P(t0 ) and n are known by the
investor at the initial moment. The risks in holding an equity then consist
of different factors than those in holding a bond.
An investor makes his decision whether to invest in the shares of a
corporation on the basis of the internal rate of return rd he expects to
gain from one share. He can calculate this by putting in formula (8.4)
his expectations of quantities N and Z, the known values of P(t0 ) and n,
and solving the equation numerically with respect to rd . In the decision-
making, the investor must still consider the risks in this decision. This
can be made, for example, by calculating the internal rate of return with
different estimates for the unknown quantities.
8.2.7 Dynamics of Asset Prices*
The Demand of Different Assets
In Sect. 8.2.6 we studied the equilibrium prices (values) of bonds and

shares of common stocks of corporations. At time moment t, the value
of these assets with term to maturity of n time units and discount rate
rd (1=t) is:
N Z
P(t) D Œ1 (1 C rd t)n C : (8.5)
rd (1 C rd t)n
Formula (8.5) consists of the present value of money flow N (e=t)

(the coupon payments of a bond or the dividends of a share), and the
repayment of the principal of a bond or the expected future selling price
of a share; these both are denoted by Z (e).
We can now ask what makes this equilibrium price hold, that is, what
causes the dynamics of asset prices? An investor observes the prevailing
price of a security, makes his estimate of the present value of future
revenues of this asset, and buys one if he expects to gain from this and
sells these assets in the opposite case. If the investor does not have the
shares he would like to sell, he can often short them.
§: By shorting we understand a situation where an investor borrows an
asset he does not own, sells the asset at current spot price, buys an identical
asset from the market at a later time, and returns the asset to its owner. ˘
As the above definition describes, shorting is profitable if current asset
price is high, and the investor who shorts gets the asset later at a lower
price than current one. Thus shorting has a risk, and it creates losses if the
investor wrongly forecasts future asset prices.
In the following, we analyze the price dynamics of the share of the

common stock of a corporation, and later we show that the model can be
applied to other securities too. The net demand of investor i of the shares
of common stock of a corporation at time moment t depends on quantity
FiS (e):
Ni Zi
FiS (t) D Œ1 (1 C rdi t)ni C P(t); (8.6)
rdi (1 C rdi t)ni
where Ni ; rdi ; Zi are the expectations of investor i of these quantities, and

ni is the length of investment horizon of the investor. We can interpret
the expected net present value of future payments of the asset in Eq. (8.6)
as the marginal willingness-to-pay of investor i of this asset, and P(t) is
its price. Thus the decision-making of investor i is analogous to that of a
consumer; buy an asset if you are willing to pay its price—that is, if you
value the asset higher than its price—and sell it in the opposite case if you
have it, or you can short it. We can interpret quantity FiS as the force by
which investor i is acts upon the demand of the asset.
The net demand Di (t) (unit=t) of the shares of investor i is can thus
be formulated as
Di (t) D fi (FiS (t)); fi0 (FiS ) > 0; fi (0) D 0: (8.7)
The following results can be obtained from Eq. (8.7):
@Di (t) f 0 (FiS )

D i Œ1 (1 C rdi t)ni > 0;
@Ni rdi
@Di (t)
D fi0 (FiS ) < 0;
@P
@Di (t) fi0 (FiS )
D > 0;
@Zi (1 C rdi t)ni

@Di (t) 0 (Ni rdi Zi )ni t Ni Œ1 (1 C rdi t)ni
D fi (FiS ) ;
@rdi rdi (1 C rdi t)ni C1 2
rdi

@Di (t) 0 (Ni rdi Zi )ln(1 C rdi t)
D fi (FiS ) :
@ni rdi (1 C rdi t)ni
These results show that increases in Ni and Zi increase the demand. On

the other hand, result @Di (t)=@P < 0 implies the stability of asset prices
because an increase in asset price decreases its demand and vice versa. This
explains the observed fluctuation in asset prices: once an asset price has
increased enough, it starts to decrease and vice versa. Asset prices change
according to their net demand, and when the majority of investors believe
that an asset has lost its growth potential, they start selling these assets
because they like to invest in assets with the greatest growth potential. An
increase in asset price decreases its internal rate of return, and once an asset
price has increased enough with respect to its expected future value Z, it
has lost its growth potential and investors start selling it. This explains the
similarity between investors and card players; both exchange bad cards for
better ones.
Now, if Ni < rdi Zi then @Di (t)=@rdi < 0; otherwise this result is
ambiguous. If Ni < rdi Zi holds, then @Di (t)=@ni < 0 and vice versa. Thus
the longer the investment horizon of investor i, the smaller his demand
of these assets if Ni < rdi Zi , and vice versa.
We can now model the demand of bonds, consols, and T-bills of
investor i as special cases of Eq. (8.7). For bonds, Ni D N; Zi D Z; ni D n
for all investors in Eq. (8.6), for consols, Ni D N, Zi D 0, ni D n D 1,
and for T-bills, Ni D N D 0; Zi D Z, and ni D n D 1. These change FiS
as follows:
N Z
FiB (t) D Œ1 (1 C rdi t)n C P(t);
rdi (1 C rdi t)n
N
FiC (t) D P(t);
rdi
Z
FiT (t) D P(t);
1 C rdi t
where subscripts B; C; T refer to Bond, Consol, and T-bill, respectively.

The net demand of investor i of these assets is then:
Dij (t) D fij (Fij (t)) fij0 (Fij ) > 0; fij0 (0) D 0; j D S; B; C; T: (8.8)
In Fij , j D C; T, result @Dij =@rdi < 0 holds certainly.

Let m investors exist in the market. In the equilibrium state in the

market of shares for the common stock of a corporation, the following
holds:
Ni Zi
P(t) D Œ1 (1 C rdi t)ni C ; i D 1; : : : ; m: (8.9)
rdi (1 C rdi t)ni
The equilibrium states of different assets are obtained as special cases of

Eq. (8.9). Adding the m equations in Eq. (8.9) and dividing by m, we get:
m
1 X Ni ni Zi
P(t) D Œ1 (1 C rdi t) C : (8.10)
m iD1 rdi (1 C rdi t)ni
Thus in the equilibrium, asset price equals the arithmetic average of

investors’ estimates of present values of net revenues from this asset.
8.2.8 Aggregate Investor Behavior*
Aggregate Behavior of Asset Buyers
We create here a simple model for the dynamics of an asset price. For this,
all investors are separated in three classes: those who are willing to buy
the studied asset, those who are willing to sell them, and those who are
not interested in this asset. The third class of investors does not affect the
price of the asset, and thus their behavior is not modeled. In this one-
period setting, one investor can belong in only one of the three classes. In
the real life, these one-period settings repeat, and investors may change in
between these classes. We assume that the asset is the share of the common
stock of a corporation, but a similar model can be created for all other
assets too. For all investors i D 1; : : : ; m willing to buy these shares, the
following holds:
N Zi
FiS (t) D Œ1 (1 C rdi t)ni C P(t) > 0:
rdi (1 C rdi t)ni
Thus every investor willing to buy these shares must value the share higher
than its current price. Adding the terms FiS over the m investors and
dividing by m, we get
m
1 X Ni ni Zi
Fd (t) D Œ1 (1 C rdi t) C P(t): (8.11)
For shortness, we denote the average marginal willingness-to-pay of the

m investors for one share by Hd ,
m
1 X Ni Zi
Hd (Ni ; Zi ; rdi ; ni ) Œ1 (1 C rdi t)ni C :
(8.12)
The investors’ aggregate demand D(t) (unit=t) of these shares for time
unit t at time moment t can then be expressed as
D(t) D fd (Fd (t)); Fd D Hd P(t); fd0 (Fd ) > 0; fd (0) D 0: (8.13)
The demand of the shares is thus an increasing function of quantity Fd ,

and D(t) D 0 if Fd D 0. By the demand of an asset we understand an
official buying offer of certain amount of the assets at a fixed price in a
specified time window. Thus the greater is Fd , the more buying offers the
investors issue at price P(t).
Aggregate Behavior of Asset Sellers
For all investors j D 1; : : : ; n willing to sell these shares, the following

holds:
N Zj
FjS (t) D P(t) Œ1 (1 C rdj t)nj C > 0:
rdj (1 C rdj t)nj
Thus every investor willing to sell these shares must value the shares lower
than their current price. Adding the terms FjS over the n investors and
dividing by n, we get
n
1 X Nj nj Zj
Fs (t) D P(t) Œ1 (1 C rdj t) C : (8.14)
n jD1 rdj (1 C rdj t)nj
For shortness, we denote the average marginal willingness-to-pay of the n

investors for one share by Hs ,
n
1 X Nj Zj
Hs (Nj ; Zj ; rdj ; nj ) Œ1 (1 C rdj t)nj C :
n jD1 rdj (1 C rdj t)nj
(8.15)
The supply S(t) (unit=t) of the shares for time unit t at time moment
t can then be expressed as
S(t) D fs (Fs (t)); Fs D P(t) Hs ; fs0 (Fs ) > 0; fs (0) D 0: (8.16)
The supply of the shares is thus an increasing function of quantity Fs , and

S(t) D 0 if Fs D 0. By the supply of an asset we understand an official
selling offer of a certain amount of the assets at a fixed price in a specified
time window. Thus the greater is Fs , the more selling offers the investors
issue at price P(t).
Price Determination
In a perfectly competed asset market, asset prices change according to their

excess demand or supply. Trades are conducted so that when a buying and
a selling offer meet, the trade is confirmed at the common price of the
offers. In buying offers, price is usually fixed from above so that smaller
prices are preferred and trades are conducted at the lowest possible price
below the fixed level. Similarly, in selling offers price is usually fixed from
below so that higher prices are preferred and trades are conducted at the
highest possible price above the fixed level. If more buying than selling
offers exist at current price, those who really like to buy the assets must
accept selling offers at a higher price, and this way the asset price increases.
Similarly, if more selling than buying offers exist at current price, those
who really like to sell these assets must accept buying offers at a lower
price, and thus asset price decreases. The equation of motion for the asset
price is then
P0 (t) D fP .D S/; fP0 (D S) > 0; fP (0) D 0: (8.17)
8.2.9 Explicit Adjustment in Share Price*
We simplify the studied model by linearizing the system in the neighbor-

hood of its equilibrium point. Thus we take Taylor series approximations
of functions fd ; fs in Eqs. (8.13), (8.16) in the neighborhood of points
Fc D 0, c D d; s by assuming the error terms zero. Then we
denote fc0 (0) D 1=mc , c D d; s, where mc , with unit et=unit are
positive inertial ‘masses’ of quantities D(t); S(t), respectively. However,
in Eq. (8.17) we denote fP0 (0) D kP , where kP > 0 with unit e/unit
is the ‘spring constant’ that measures the sensitivity of share price P on
difference D S, see Fig. 8.2. Time t is measured in units t. The system
for share price behavior is then:
md D(t) D Hd (Ni ; Zi ; rdi ; ni ) P(t); (8.18)

ms S(t) D P(t) Hs (Nj ; Zj ; rdj ; nj ); (8.19)
1 0
P (t) D D(t) S(t): (8.20)
kP
The ‘free body’ diagram of the forces acting upon the demand and
the supply of the shares is in Fig. 8.2. The direction of motion on the
Fig. 8.2 Forces acting upon the demand, supply, and price of an asset
30
25
20
P(t)
15
10
0 2 4 6 8 10 12 14
time
Fig. 8.3 Two time paths for share price
right-hand side is defined positive, and on left-hand side, negative. The

‘particles’ denoted by md ; ms represent aggregate accumulated buying
and selling offers at prevailing prices during the history. The positive
force component acting upon the demand of shares is Hd (Ni ; Zi ; rdi ; ni ),
and the negative component is P(t). The positive force component
acting upon the supply of shares is P(t), and the negative component is
Hs (Nj ; Zj ; rdj ; nj ). As earlier, the shapes of the ‘particles’ in Fig. 8.3 have no
economic meaning, and actually the particles should be drawn as points
on the two horizontal axes. However, the box shapes for the variables
better visualizes the analogy we make here with Newtonian mechanics.
The spring between the two ‘particles’ reflects the matter that an equal
mutual force P(t) is acting upon the two particles, and nonzero difference
D(t) S(t) changes the length of the spring that changes the mutual
force
R component
R P(t), see Eq. (8.20). The length of the spring is LP D
D(t)dt S(t)dt, and the force by which it acts upon the two particles
is P D kP LP . By taking the time derivative of this equation we get
Eq. (8.20). Thus share price P(t) with unit e reflects the historical scarcity
of the asset, and it changes according to difference D(t) S(t). These
interactions keep the system in a relatively stable motion.
Solving Eqs. (8.18) and (8.19) with respect to D(t) and S(t), respectively,
and substituting these in Eq. (8.20), we get it in the following form:
1 0 1 1
P (t) D .Hd P(t)/ .P(t) Hs / ; (8.21)
kP md ms
where P0 (t) (e=t)) is the flow of the share price. The equilibrium share
price can be solved from Eq. (8.21) by setting P0 (t) D 0:
ms Hd C md Hs
P D : (8.22)
md C ms
Because Hd ; Hs ; md ; ms 0, P is nonnegative and it depends positively

on both marginal willingness-to-pay values Hd ; Hs . An increase in Hd
increases D(t) which stretches the spring from its right end, and an
increase in Hs decreases S(t) which stretches the spring from its left end;
see Fig. 8.2. The resulting change in difference D(t)S(t) is transmitted to
price P(t) by spring constant kP ; see Eq. (8.20). This explains why Hd ; Hs
both affect price P(t) positively. One special equilibrium situation in the
share price is that where both types of investors have identical expectations
of future revenues of the asset, and current price equals these expectations.
Thus Hd D Hs D P(t), which makes P0 (t) D 0 in Eq. (8.21). In that case,
none of investors is interested in buying or selling these assets, and thus
the asset price stays fixed.
Increasing price increases the mutual force between the particles, which
guarantees the stability of the system. The price stability can be proved by
the negativity of the partial

@P0 (t) 1 1
D kP C < 0I
@P(t) md ms
see Fig. 3.9. In the Appendix of this chapter, Hd (Ni ; Zi ; rdi ; ni ) is approx-
imated as:
a0 a1 a2 a3 a4
Hd (N d ; Z d ; rdd ; nd ) C N d C Z d C rdd C nd ; (8.23)
m m m m m
whereby a bar above a variable is denoted arithmetic average. The units

of a0 ; a1 ; a3 ; a4 are: e, t, e t, and e, respectively, and a2 is
dimensionless; a1 > 0, a2 > 0, and the signs of a3 ; a4 are ambiguous,
see the Appendix of this chapter. Hs (Nj ; Zj ; rdj ; nj ) can be approximated
similarly:
b0 b1 b2 b3 b4
Hs (N s ; Z s ; rds ; ns ) C N s C Z s C rds C ns ; (8.24)
n n n n n
where the average variables N s ; Z s ; rds ; ns as well as constants b1 ; : : : ; b4

deviate from those of asset buyers.
If we now substitute Eqs. (8.23) and (8.24) in Eq. (8.21), we get a
solvable model for asset price dynamics. However, due to the several
variables in Eq. (8.21), its solution is not very informative. The aim of this
modeling is to explain in a detailed way how the expectations of investors
of different factors affect their valuations of assets, and how asset prices
react to differences in these expectations. In order to demonstrate the
behavior of the model, we apply Eq. (8.21) by assuming first (a) Hd D 25,
Hs D 10, and then (b) Hd D 10, Hs D 25. For both cases the following
values for the constants md D 1; ms D 2; kP D 1=2 are assumed. The
reason to assume ms > md is that asset holders may have a higher inertia
in their decisions, but this assumption does not affect the results in a
noticeable way. The corresponding solutions of Eq. (8.21) are
3 3
Pa (t) D 20 C Ca e 4 t ; Pb (t) D 15 C Cb e 4 t ; (8.25)
where Ci , i D a; b with unit e are the constants of integration. The

difference in the two equilibrium prices, 20 and 15, depends on the
assumed difference in inertial factors: md D 1 < ms D 2. However,
if md D ms holds, we get the solution for the equilibrium price as:
P D (Hd C Hs )=2, see Eq. (8.22).
Thus the system is stable and will converge into the equilibrium state
independent on whether buyers or sellers valuate the asset more. In
Fig. 8.3 is graphed two solutions of Pb in Eq. (8.25) with different initial
conditions: Cb D 50 and Cb D 10. The figure shows the stability of
the model independent of the initial condition of the solution.
3500
3250
3000
2750
2500
2250
2000
1750 © Kauppalehti
7/11 1/12 7/12 1/13 7/13 1/14 7/14 1/15 7/15 1/16
OMX Helsinki 25
Fig. 8.4 The time path of OMX stock price index
Fig. 8.4 presents the OMX Helsinki 25 stock price index that measures
the average stock price of 25 Finnish corporations. The index behaves
like all stock prices; there is sometimes positive and sometimes negative
trend, and the index fluctuates around its trend. Thus to get model
(8.21) accurately mimicking observed stock price behavior, quantities
Hd ; Hs should be modeled more exactly. Model (8.21) explains how the
asset price reaches its equilibrium value with fixed Hd ; Hs . In real life,
however, investors receive every day and every hour new information
about firms’ future revenues and costs, and this information changes their
expectations. These changes in expectations affect the investors’ buying
and selling behavior, as Eqs. (8.18), (8.19) describe.
A commonly used stock price model is the random walk (RW) model,
see, for example, Fama (1965). An example of an RW process is presented
in Fig. 8.5 where the graph is created as
P(t) D P(t 1) C (t); P(0) D 100; t D 1; : : : ; 150;
where is a random quantity with unit e that has uniform distribution

between (4; 4). Uniform distribution means that every number between
(4; 4) has equal probability. Thus the RW model for share price P(t) is
Fig. 8.5 The time path of a random walk process for 150 time units
formulated by adding to P(t 1) a random scalar between (4; 4) so

that the initial asset price is P(0) D 100. This process is repeated so that
after every time unit t the share price has equal probability of going
up or down. Another realization of the same random walk process as in
Fig. 8.5 is in Fig. 8.6. In the RW model, every trial gives a completely
different kind or realization for the 150 time units depending on what
values random term (t) will get. The same holds for stock prices, which
means that the long run predictability of both variables is difficult. These
are the reasons that the RW model has been commonly used in modeling
stock market behavior, even though various similar models have been
applied too.
The behavior of the graphs in Figs. 8.4, 8.5, and 8.6 is remark-
ably similar. In Fig. 8.4, new information changes investors marginal
willingness-to-pay for shares at every time unit, and the probability of
getting good or bad information is almost equal. This new information
changes investors’ buying and selling of assets, and thus stock prices have
almost equal probability of going up or down. Here we do not study other
140
120
100
80
60
0 20 40 60 80 100 120 140
Fig. 8.6 The time path of a random walk process for 150 time units
specific stock price models in a detailed way, but we show that the RW
model can be obtained as a special case of model (8.21).
The complete model in Eq. (8.21) with Hd ; Hs as in Eqs. (8.23), (8.24),
respectively, shows how different factors affect the share price. In order to
show how the RW process is obtained as a special case of Eq. (8.21), we
make the following assumptions. First, the average time unit of investors
is assumed to be very short, because only investors that are active in stock
market actually affect stock prices. Usually corporations pay dividends
annually, and thus short term investors—that plan to hold the shares they
buy for a few days or hours—do not earn dividends. Thus we can set Nk D
0 8k, k D i; j, and we can also omit the discount factor 1=(1 C rdk t)nk ,
k D i; j from Eqs. (8.12) and (8.15) when nk is less than a week. Then we
get from Eqs. (8.12) and (8.15):
1X 1X
m n
Hd Zi D Z d ; Hs Zj D Z s :
m iD1 n jD1
Next we assume that investors willing to buy the shares expect the share
price to increase from that of current price, and those willing to sell these
shares expect the share price to decrease from that of current price. Thus
we set Hd D P(t) C d and Hs D P(t) s , where k , k D d; s are
random variables with unit e with positive mean and constant variance.
In the RW process time is discrete, and assuming this we transform the
velocity of share price as
P P(t C 1) P(t) P(t C 1) P(t)

P0 (t) D D ;
t t (t 1) 1
where symbol 1 in the denominator is not pure number but one unit of
time, and P is the change in the price during the time unit. In order
to avoid errors in measurement units of the quantities, we keep in the
following t instead of 1 in denoting the time unit. Then we get the
discrete time version of Eq. (8.21) as follows:
1 P 1 1
D ŒHd P(t) ŒP(t) Hs ,
kP t md ms
1 P(t C 1) P(t) 1 1
D ŒP(t) C d P(t) ŒP(t) (P(t) s ) ,
kP t md ms
1 P(t C 1) P(t) d s
D ,
kP t md ms

d s
P(t C 1) P(t) D tkP ,
md ms

d s
P(t C 1) D P(t) C P ; P D tkP ;
md ms
where P is a random quantity with unit e. If the expectations of the

two types of investors deviate in opposite directions by almost equal
amounts—which is a plausible assumption—and the inertial factors of
investors are roughly equal, namely, md ms , then random quantity P
has besides constant variance also zero mean. Thus if investors plan to hold
the shares they buy only a very short time, and the described assumptions
hold for investors, then the RW model is a special case of Eq. (8.21).

We approximate quantity Hi ,
Ni Zi
Hi (Ni ; Zi ; rdi ; ni ) Œ1 (1 C rdi t)ni C (8.26)
rdi (1 C rdi t)ni
in the neighborhood of point xi0 D (Ni0 ; Zi0 ; rdi0 ; ni0 ) by taking the
Taylor series approximation as:
@Hi (xi0 ) @Hi (xi0 )

Hi (Ni ; Zi ; rdi ; ni ) D Hi (xi0 ) C (Ni Ni0 ) C (Zi Zi0 )
@Ni @Zi
@Hi (xi0 ) @Hi
C (rdi rdi0 ) C (ni ni0 ) C i : (8.27)
@rdi @ni (xi0 )
Adding the approximations of Hi in Eq. (8.27) by assuming i D 0 8i,

we get
X
m
Hi (Ni ; Zi ; rdi ; ni ) a0 C a1 N C a2 Z C a3 rd C a4 n; (8.28)
iD1
Pm
where x D (1=m) iD1 xi is the arithmetic average of x D N; Z; rd ; n,
and:
m
X @Hi (xi0 ) @Hi (xi0 ) @Hi (xi0 )
a0 D Hi (xi0 ) Ni0 Zi0 rdi0
iD1
@Ni @Zi @rdi
Xm Xm
@Hi (xi0 ) @Hi (xi0 ) @Hi (xi0 )
ni0 ; a1 D ; a2 D ;
@ni iD1
@Ni iD1
@Zi
X
m
@Hi (xi0 ) X
m
@Hi (xi0 )
a3 D ; a4 D :
iD1
@rdi iD1
@ni
The units of a0 ; a1 ; a3 ; a4 are: e, t, e t, and e, respectively, and

a2 is dimensionless. Now, Hi (xi0 ) is positive at every Ni ; Zi ; rdi ; ni because
every asset gives some revenues in the future and thus has a positive present
value. Thus a0 > 0 (let Ni ; Zi ; rdi ; ni ; i ! 0 8i in Eq. (8.27)). In
Sect. 8.2.7 we showed that @Hi =@Ni > 0, @Hi =@Zi > 0, and @Hi =@rdi ,
@Hi =@ni are ambiguous; thus a1 > 0, a2 > 0 and the signs of a3 ; a4 are
ambiguous.
Dividing both sides of Eq. (8.28) by m, we get an approximation for
the average value of Hi , i D 1; : : : ; m as:
1X
m
a0 a1 a2 a3 a4
Hi (Ni ; Zi ; rdi ; ni ) C N C Z C rd C n: (8.29)
m iD1 m m m m m
By taking the same steps we can derive the following approximation for
the average value of Hj , j D 1; : : : ; n as:
1X
n
b0 b1 b2 b3 b4
Hj (Nj ; Zj ; rdj ; nj ) C N C Z C rd C n: (8.30)
n jD1 n n n n n
References
Fama, E. (1970). Efficient capital markets: A review of theory and empirical work.
Journal of Finance, 25, 383–417.
Fama, E. F. (1965). Random walks in stock market prices. Financial Analysts
Journal, September–October, pp. 55–59.
Friedman, M. (1968). The role of monetary policy. American Economic Review,
March, pp. 1–17.
Hull, J. C. (2000). Options, futures, and other derivatives (4th ed.). Upper Saddle
River, NJ: Prentice-Hall International, Inc.
Jevons, W. S. (1970). Theory of political economy (2nd ed.), Original publication
at 1879. Harmondsworth: Penguin.
Mishkin, F. S. (2001). The economics of money, banking, and financial markets
(6th ed.) Reading, MA: Addison Wesley.
9
Saving, Borrowing, and Interest Rates
In this chapter, we model the determination of interest rates by the

demand for loans and the supply of savings in a perfectly competed
loan market. We assume a two-period setting where households1 have a
fixed income at both periods. Households consume all their income and
possible interest returns at the two time units during the time units. At
the latter period saving households consume their income at that period
and their savings and interest returns from the first period, and borrowing
households pay back their loans and interest costs from the first period.
The two periods represent current time and future, and in real life these
may correspond to, for example, two adjacent months or years, the youth
and the old age of households and so on.
We assume that the amounts of all goods in the economy are measured
in units kg, and the weighted averages of prices of all goods at periods
0 and 1 with unit e/kg are denoted by p0 and p1 . This simplification is
made because here we are not interested in the composition of consumers’
consumption during one time unit as in Chap. 3, but only of its timing.
1
By a household we understand a decision-making unit that may consist of a family with a unique
utility function, or of only one person.

DOI 10.1007/978-3-319-46879-2_9
The average price levels are assumed fixed and known by every household
at the first period when they make their consumption/saving decision. At
the first period, households may consume their income exactly, or more or
less, depending on whether they act as borrowers or depositors in the loan
market. Every household belongs in one of the following three groups: net
savers, net borrowers, or those whose net saving is zero. The behavior of
the last group is not modeled, however, because these households do not
affect the interest rate. The saving and borrowing households are grouped
into two groups that are studied separately. In this two-period setting,
one household can belong in only one of these three groups. In the real
life, these two-period settings repeat and households may change between
these groups.
9.1 Saving Households

Saving household i has budget equation
Ti1 p1 Xi1 p1 Xi1 Ti1

Ti0 C D p0 Xi0 C , Ti0 p0 Xi0 D (9.1)
1 C rs t 1 C rs t 1 C rs t
for the two time units. The income and consumption of the household
with units e=t and kg=t, respectively, are denoted by Tic ; Xic at period
c D 0; 1. In the following we assume t D 1 (y), and then the deposit
rate denoted by rs has unit 1=y. On the left-hand side of the first form
of the equation are the present values of income at the two time units
discounted by the savings rate rs , and on the right hand side are the present
values of consumption expenditures at both periods. The explanation for
the discount rate is that a saving household compares present and future
consumption on the basis of the deposit rate that defines the interest
returns from savings. On the left-hand side of the second form of the
equation are the savings from the first period Si D Ti0 p0 Xi0 > 0,
and on the right hand side the present value of excess consumption over
income at the latter period, p1 Xi1 Ti1 > 0. Subindex 0 is omitted from
savings because households can save only in the first period.
Equation (9.1) can also be written as
(1 C rs t)(Ti0 p0 Xi0 ) D p1 Xi1 Ti1 ; (9.2)
where on the left-hand side are the savings and interest returns on these
savings from the first period, and on the right-hand side is the excess
consumption expenditures in the latter period. All these three forms of
the equation express the same thing. In a two-period setting with fixed
incomes, prices, and deposit rate, the consumption decision at the first
period Xi0 exactly defines the savings Si and the consumption at the latter
period Xi1 . We can thus define the consumptions at both periods by
savings Si . Defining the savings of household i as we did above, we can
write:
Ti0 Si
Si D Ti0 p0 Xi0 , Xi0 D :
p0
By using the budget equation, the consumption at the latter period can
be expressed in terms of savings as
(1 C rs t)Si C Ti1
(1 C rs t)Si D p1 Xi1 Ti1 , Xi1 D :
p1
The satisfaction household i gains during the two time units is measured
by utility function ui with unit ut=(2y). Function ui is assumed continu-
ous and differentiable:
@ui @ui @2 ui @2 ui
ui D ui .Xi0 ; Xi1 /; > 0; > 0; 0; 0;
@Xi0 @Xi1 @Xi0
2
@Xi1
2
@2 ui @2 ui
D > 0:
@Xi1 @Xi0 @Xi0 @Xi1
These assumptions of the utility function guarantee the existence of a

unique maximum for the household’s utility maximization problem; see
Sect. 3.7.
Fig. 9.1 The budget equation of saving household i
The budget of household i defines the possible consumption bundles

for the household at the two periods; see Fig. 9.1. As in Chap. 3, we can
define an indifference curve for a saving household that represents the
points of constant utility.
The shape of the indifference curve results from the assumptions of the
utility function. In coordinate system (Xi0 ; Xi1 ), the slopes of the budget
line and an indifference curve are:
@ui
dXi1 p0 dXi1 @X
D (1 C rs t) < 0 and D @ui0 < 0: (9.3)
dXi0 p1 dXi0 @X
i
i1
The slope of an indifference curve is derived as that for a consumer in

Sect. 3.5, and the slope of the budget line is obtained from Eq. (9.2). Thus
both are decreasing in coordinate system (Xi0 ; Xi1 ), and the higher the rs

the steeper the budget line. The optimal consumption bundle (Xi0 ; Xi1 )

and optimal savings Si D Ti0 p0 Xi0 for household i are shown in
Fig. 9.2.
At point (Ti0 =p0 ; Ti1 =p1 ), the household does not save or borrow, and

in the optimum household i consumes by (Ti0 =p0 ) Xi0 less than it could
afford by its first period income. The monetary value of this bypassed
Fig. 9.2 The equilibrium state of a saving household i
Fig. 9.3 Two non-optimal points of a saving household i

consumption p0 Œ(Ti0 =p0 ) Xi0 equals with optimal savings Si . In the
optimum, the slopes of the budget equation and indifference curve are
equal,
@ui
p0 @Xi0
(1 C rs t) D @ui
; (9.4)
p1 @Xi1
where the minus sign is cancelled from both sides of the equation.
If household i is consuming too much at current period, the situation
corresponds to point (Xi0A ; Xi1A ) in Fig. 9.3, where the slope of the budget
equation is steeper than that of the indifference curve. Thus it holds
@ui
p0 @Xi0 @ui 1 @ui 1
(1 C rs t) > @ui
, (1 C rs t) >
p1 @Xi1
@Xi1 p1 @Xi0 p0
@ui 1 @ui 1
!
@Xi0 p0 1 @Xi0 p0
, 1 C rs t > @ui
, rs > @ui 1
1 : (9.5)
@Xi1
1
p1
t @X p i1 1
We can understand the last form of Eq. (9.5) as follows. Quantities
@ui 1 @ui 1
; ;
@Xi0 p0 @Xi1 p1
with unit ut/e measure the household’s marginal utilities from consump-
tion at the two time units by one euro. If the household decides to save
one euro, the gain from this is the interest returns to be obtained for one
euro by which the household can increase its future consumption. The
loss from this is the lost consumption at current period by one euro. Thus
on the left-hand side of the second form of Eq. (9.5) is the received gain
in future utility by increased consumption with one euro plus interest
returns rs t on one euro, and on the right-hand side is the required loss in
current utility due to decreasing current consumption by one euro. Thus
in situation (Xi0A ; Xi1A ), the household would gain from saving more and
changing current consumption to future. The rate of loss in utility due to
passing consumption to future is
!
@ui 1 @ui 1
@Xi0 p0
@Xi1 p1
=t 1
@ui 1
@Xi0 p0
1 @ui
D @ui
1 :
p1 @Xi1
t @Xi1
1
p1
According to the last form of Eq. (9.5), utility-seeking household i

increases its saving if interest rate rs is greater than the ‘rate of household’s
loss in utility due to passing consumption to future’. For shortness, we
denote the latter quantity with unit 1=t by zi ,
@ui 1
!
1 @Xi0 p0
zi D @ui
1 ;
t @Xi1
1
p1
and we call it the subjective rate of time preference of the household.

The greater zi is, the more the household prefers current consumption
as compared with future consumption, and thus the less willing the
household is to save. Notice that zi > 0, if marginal utility of one euro
from current consumption is higher than that of future consumption,
which we can assume to hold for most people. Thus if rs > zi , utility-
seeking household i increases its saving and vice versa. In the event that
the household is consuming too little at current period, the situation
corresponds to point (Xi0B ; Xi1B ) in Fig. 9.3, where the slope of the
indifference curve is steeper than that of the budget equation. This
situation corresponds to rs < zi , which the reader can prove similarly
as we made for the opposite case.
The slope of the equilibrium supply function of savings of household
i in coordinate system (Si ; rs ), obtained from equation rs D zi , is:
@ ui p1
2
@ ui 2
@ ui 2
drs @X 2 p C 2(1 C rs t) @X @X (1 C rs t)2 pp01 @X 2

D i0 0
:
i1 i0 i1
dSi @ui p 0 @2 u i @2 u i
p t C S t (1 C r t)
@Xi1 0 i s p1 @Xi1
2 @Xi1 @Xi0
The numerator of this partial is positive and the sign of the denominator is
ambiguous. However, assuming that the savings of household i are small,
that is, Si 0, then the denominator and also the derivative are positive.
Thus with small savings the supply function of savings of household i,
that corresponds to the equilibrium state of the household, is increasing
in the coordinate system (Si ; rs ); see Fig. 9.4. In the dotted part of the
curve, drs =dSi < 0.
In order to model the dynamics of savings of household i, we assume
that savings Si (t) depend on time t with unit y. This makes the household’s
consumption at the two time units time-dependent too, Xi0 (t); Xi1 (t), and
the other defined quantities are assumed fixed. We can then write the
utility function of household i as a function of its savings Si (t) (e=y) as:

Ti0 Si (t) (1 C rs t)Si (t) C Ti1
ui (t) D ui .Xi0 (t); Xi1 (t)/ D ui ; :
p0 p1
(9.6)
Fig. 9.4 The supply function of savings of saving household i
According to Eq. (9.6), household i has only one variable by which it can
affect its utility, namely Si (t). The time derivative of the utility function,
or the acceleration of utility u0i (t) with unit ut=(4y2 ), is then2 :

@ui 1 @ui (1 C rs t) 0
u0i (t) D C Si (t): (9.7)
@Xi0 p0 @Xi1 p1
For the utility to increase with time, both factors in the product on the
right-hand side of Eq. (9.7) must be of equal sign. Household i likes to
increase its utility with time, namely, to get u0i (t) > 0, and so it changes
its saving to:
@ui 1 @ui (1 C rs t)

Si0 (t) > 0 if C > 0;
@Xi0 p0 @Xi1 p1
Si0 (t) < 0 if C < 0; and
@Xi0 p0 @Xi1 p1
2
Unit ut=4y2 D ut=(2y 2y) results from dividing the measurement unit of the flow of utility
ut=2y by time unit 2y.

Si0 (t) D 0 if C D 0;
@Xi0 p0 @Xi1 p1
where situation Si0 (t) D 0 is the equilibrium state of the household. To

express these adjustment rules in an understandable form, we transform
the above inequalities according to Eq. (9.5) as

Si0 (t) > 0 if C > 0;
@Xi0 p0 @Xi1 p1
@ui 1 @ui 1
Si0 (t) > 0 if (1 C rs t) > ;
@Xi1 p1 @Xi0 p0
@ui 1
!
@Xi0 p0
Si0 (t) > 0 if 1 C rs t > @ui 1
;
@Xi1 p1
@ui 1
!
1 @Xi0
Si0 (t)
p0
> 0 if rs > @ui
1 ; and vice versa.
t @Xi1
1
p1
The interpretation is thus the same as earlier: household i increases its

saving if rs > zi , and vice versa. Identical transformations can be made
for cases S0 (t) < 0 and S0 (t) D 0.
Now, Si0 (t) (e=y2 ) is the acceleration of savings of household i, and
imitating Newton, we name the quantity
@ui 1
!
1 @Xi0 p0
Fi D rs zi ; zi D @ui
1
t @Xi1
1
p1
with unit 1=t as the ‘force acting upon saving of household i’, because
it is the cause for the household’s acceleration of saving. Household i
decreases its consumption at first period if rs > zi , and vice versa, and
the utility maximizing situation corresponds to zero force: rs D zi .
The following results can be derived from zi :

(Ti0 Si ) @2 ui @ui @2 ui @ui @ui @ui
@zi p0 @Xi0 @Xi1 @Xi0
2 @Xi1
@Xi0
@Xi0 @xi1
D 2 ;
@p0 p20 t @ui
p1 @Xi1

Œ(1Crs t)Si CTi1 @2 ui @ui @2 ui @ui @ui @ui
@zi p1 2 @Xi0
@Xi1
@Xi1 @Xi0 @Xi1
C @Xi0 @xi1
D 2 ;
@p1 @ui
p0 t @Xi1
@2 ui @ui
@X@i0 @X @ui
ui 2
@zi 2 @Xi1
@Xi0 i1 @Xi0
D 2 < 0;
@Ti0 p20 t @ui
p1 @Xi1
@2 ui @ui @2 ui @ui
@zi @Xi1 @Xi0 @Xi1
2 @Xi0
@Xi1
D 2 > 0;
@Ti1 @ui
p0 t @Xi1
2
@ ui @ui @ ui @ui 2
@zi
1
p0 @Xi0 @Xi1 @Xi0
@X 2 @Xi1
D 2
i0
@Si p0 t @ui
p1 @Xi1

(1Crs t) @2 ui @ui @2 ui @ui
p1 @Xi1 @Xi0 @Xi1
2 @Xi0
@Xi1
C 2 > 0;
p0 t @ui
p1 @Xi1
@2 ui @ui @2 ui @ui
@zi @Xi1 @Xi0 @Xi1
2 @Xi0
@Xi1
D 2 > 0;
@rs p0 @ui
Si @Xi1
These results are based on the utility function in Eq. (9.6), and we can
interpret them as follows. An increase in zi means that the household
prefers current consumption more, which decreases its saving. Increases in
p0 ; p1 change the relative prices of current and future consumption, and
their effects on saving are ambiguous. An increase in Ti0 decreases zi and
thus increases the saving of household i, and an increase in Ti1 decreases
the saving of household i. Result @zi =@Si > 0 implies that the adjustment
is stable; the higher the savings of household i, the higher is zi which

decreases the household’s saving. Sign @zi =@rs > 0 implies that increasing
interest rate increases the household’s rate of time preference, which results
because with higher interest rate saving household can decrease its saving
and still get the same interest revenues.
9.1.1 Dynamics of Savings*
The reasoning in the previous section can be modeled as
Si0 (t) D G(Fi ); G0 (Fi ) > 0; G(0) D 0; Fi D rs zi ; (9.8)
where Fi is the force acting upon the saving of household i. The acceler-
ation of saving Si0 (t) is positive if Fi is positive, and vice versa. Taking the
first order Taylor series approximation of function G in Eq. (9.8) in the
neighborhood of point Fi D 0, and assuming the error term zero, we get:
Si0 (t) D G0 (0) (rs zi ): (9.9)
Now, denoting 1=G0 (0) D mi , we can call Eq. (9.9) the ‘Newtonian
equation of savings of household i’. The name comes from the analo-
gous form of Eq. (9.9) with Newton’s equation ma D F, and constant
mi > 0 with unit y/e represents the inertial factors (‘mass’) in this
adjustment.
The ‘free body’ diagram of the forces acting upon the saving of
household i is in Fig. 9.5. The positive force component acting upon
saving is rs and the negative force component is zi .
Fig. 9.5 The forces acting upon the saving of household i

9.2 Borrowing Households

Borrowing households face a fixed lending rate rb (> rs ) with unit 1=y.
The budget equation of borrowing household j for the two time units is
(1 C rb t)(p0 Xj0 Tj0 ) D Tj1 p1 Xj1 ; (9.10)
where the income and consumption of the household at the two periods
with units e=t and kg=t, respectively, are denoted by Tjc , Xjc , c D
0; 1, and t D 1 (y). On the left-hand side of Eq. (9.10) is the loan the
household takes at the first period Bj D p0 Xj0 Tj0 > 0 and its interest
costs, and this equals with the savings of the household at the latter period
Tj1 p1 Xj1 > 0. Equation (9.10) can also be written as
p1 Xj1 Tj1
p0 Xj0 C D Tj0 C ;
1 C rb t 1 C rb t
where on the left-hand side is the household’s present value of consump-

tion and on the right-hand side the present value of income discounted
by rb . A borrowing household compares present and future consumption
on the basis of the lending rate that defines the costs of borrowing.
The first period consumption of a borrowing household uniquely
defines its loans at the first period and consumption at the latter period.
From the given definition we get:
Bj C Tj0
Bj D p0 Xj0 Tj0 , Xj0 D ; (9.11)
p0
and from Eq. (9.10) we get:
Tj1 (1 C rb t)Bj
(1 C rb t)Bj D Tj1 p1 Xj1 , Xj1 D : (9.12)
p1
Household j gains utility uj (ut=(2y)) from its consumption as:
@uj @uj @2 uj @2 uj
uj D uj .Xj0 ; Xj1 /; > 0; > 0; 0; 0;
@Xj0 @Xj1 @Xj0
2
@Xj1
2
@2 uj @2 uj
D > 0:
@Xj1 @Xj0 @Xj0 @Xj1
The assumptions of the utility function are equal as for saving households.
The budget equation of household j in Fig. 9.6 shows the consumption
bundles the household can choose at the two periods. An indifference
curve for a borrowing household is defined analogously as for a saving
household. The shape of the indifference curve results from our assump-
tions of the utility function. In the coordinate system (Xj0 ; Xj1 ), the slopes
of the budget line and an indifference curve are:
@uj
dXj1 p0 dXj1 @Xj0
D (1 C rb t) < 0 and D @u < 0:
dXj0 p1 dXj0 j
@Xj1
Thus both are decreasing, and the higher the rb the steeper the budget
line.
kg
y
(1 + rb Dt )T j 0 + T j1
p1
(1 + rb Dt )( p0 X j 0 – T j 0 ) = T j1 – p1 X j1
æ T j1 æ 1
çT j 0 + ç kg y
è 1 + rb Dt è p0
Fig. 9.6 The budget equation of borrowing household j

kg
y
(1 + rb Δt )( p 0 X j 0 – T j 0 ) = T j1 – p1 X j1
T j1
p1
X *j1 u j ( X j 0 , X j1 ) = u j 0
T j 0 / p0 X *j 0 kg y
Fig. 9.7 The equilibrium state of borrowing household j

The optimal consumption bundle (Xj0 ; Xj1 ) and optimal borrowing
Bj of household j are shown in Fig. 9.7. At point (Tj0 =p0 ; Tj1 =p1 ), the
household does not save or borrow, and in the optimum the household

consumes by Xj0 (Tj0 =p0 ) > 0 more than it could by its first period

income. The monetary value of this excess consumption p0 Xj0 Tj0 equals

with Bj .
Substituting Eqs. (9.11) and (9.12) in the utility function and assuming
the loans taken to depend on time t, Bj (t)—the consumption of house-
hold j at both periods then depends on time too, Xj0 (t); Xj1 (t)—the utility
function becomes as:

Bj (t) C Tj0 Tj1 (1 C rb t)Bj (t)
uj (t) D uj .Xj0 (t); Xj1 (t)/ D uj ; :
p0 p1
(9.13)
Household j has now only one quantity by which it can affect its utility,
namely Bj (t) (e=y). The acceleration of utility u0j (t) with unit ut=(4y2 ) is:

@uj 1 @uj (1 C rb t)
u0j (t) D B0j (t):
@Xj0 p0 @Xj1 p1
The rules to adjust borrowing that make the utility of household j increase
with time, are:

B0j (t) > 0 if > 0;
@Xj0 p0 @Xj1 p1
B0j (t) < 0 if < 0; and
@Xj0 p0 @Xj1 p1
B0j (t) D 0 if D 0:
@Xj0 p0 @Xj1 p1
To express these rules in an understandable form, we transform the above

inequalities as:

B0j (t) > 0 if > 0;
@Xj0 p0 @Xj1 p1
@uj 1 @uj 1
B0j (t) > 0 if > (1 C rb t) ;
@Xj0 p0 @Xj1 p1
@uj 1
0 @u 1
j 1
@Xj0 p0 1 @ @Xj0 p0 1A > rb :
B0j (t) > 0 if @uj 1
> 1 C rb t ,
t @uj 1
@Xj1 p1 @Xj1 p1
Quantity
0 @uj
1
1
1 @ @Xj0
1A
p0
zj D @uj
t 1
@Xj1 p1
with unit 1=t measures the household’s subjective rate of time preference
between current and future consumption. Using the form in Eq. (9.13) of
the utility function, we can derive the following results from zj :

(Bj CTj0 )p1 @uj @2 uj @2 uj @uj @u @u
@Xj0 @Xj0 @Xj1
2 @Xj1 p1 @Xj0j @Xj1j
@zj p0 @Xj0
D 2 ;
@p0 @uj
tp20 @Xj1

ŒTj1 (1Crb t)Bj @uj @2 uj @2 uj @uj @uj @uj
@Xj0 @Xj1
@Xj1 @Xj0 @Xj1
C @Xj0 @Xj1
@zj p1 2
D 2 ;
@p1 @uj
tp0 @Xj1

@uj @2 uj @2 uj @uj
p1 @Xj1 @Xj0
@Xj0 @Xj1 @Xj0
@zj 2
D 2 < 0;
@Tj0 @uj
tp20 @Xj1
@uj @2 uj @2 uj @uj

@zj @Xj1 @Xj1 @Xj0 2 @Xj0
@Xj1
D 2 > 0; (9.14)
@Tj1 @uj
tp0 @Xj1

@uj @2 uj @2 uj @uj
(1 C rb t) @Xj0 @Xj1
@Xj1 @Xj0 @Xj1
@zj 2
D 2
@Bj @uj
tp0 @Xj1

p1 @uj @2 uj @2 uj @uj
p0 @Xj1 @Xj0
2 @Xj0 @Xj1 @Xj0
C 2 < 0;
@uj
tp0 @Xj1

@uj @2 uj @2 uj @uj
Bj @Xj0 @Xj1
@Xj1 @Xj0 @Xj1
@zj 2
D 2 < 0: (9.15)
@rb @uj
p0 @Xj1
We can interpret these results as follows. An increase in zj means that

the household prefers more current consumption, which increases its
borrowing. Increases in p0 ; p1 change the relative prices of current and
future consumption, and their effects on borrowing are ambiguous. An
increase in Tj0 decreases zj and thus decreases the borrowing of household
j, and an increase in Tj1 increases the borrowing of household j. Result
@zj =@Bj < 0 implies that the adjustment is stable; the higher the
borrowing of household j, the smaller is zj which decreases the household’s
borrowing. Sign @zj =@rb < 0 implies that increasing interest rate decreases
the household’s time rate of preference.
Fig. 9.8 The demand function of loans of household j
The slope of the equilibrium demand function of loans of household j

in coordinate system (Bj ; rb ), zj D rb , is (see Fig. 9.8):
p 1 @2 u j p0 @2 u @2 u
drb p0 @Xj02 C p1
(1 C rb t)2 @X 2j 2(1 C rb t) @Xj0 @Xj j1
D < 0:
j1
dBj @u (1Crb t)p0 Bj t @2 uj @2 u

p0 t @Xj1j p1 @Xj1
2 C Bj t @Xj1 @Xj j0
Because dBj =drb D 1=(drb =dBj ), we can interpret this result so that an
increase in rb decreases the optimal borrowing of household j.
According to the above, we can name quantity Fj ,
Fj D zj rb ;
as the ‘force acting upon borrowing of household j’ because it is the

cause for the acceleration of borrowing B0j (t) (e=y2 ) of the household.
The force with unit 1=t shows that household j increases its borrowing
if its subjective time rate of preference is higher than lending rate, and vice
versa.
Fig. 9.9 The forces acting upon the borrowing of household j
9.2.1 Dynamics of Borrowing*
The reasoning in the previous section can be summarized as
B0j (t) D H(Fj ); H 0 (Fj ) > 0; H(0) D 0; Fj D zj rb ; (9.16)
where Fj is the force acting upon borrowing of household j. The acceler-

ation of borrowing of household j, B0j (t), is positive if Fj > 0, and vice
versa. Taking the first order Taylor series approximation of function H in
Eq. (9.16) in the neighborhood of point Fj D 0 and assuming the error
term zero, we get:
B0j (t) D H 0 (0) (zj rb ): (9.17)
Now, denoting 1=H 0 (0) D mj we can name Eq. (9.17) as the ‘Newtonian
equation of borrowing of household j’, where constant mj > 0 with
unit y/e represents the inertia (‘mass’) in this adjustment.
The ‘free body’ diagram of the forces acting upon the borrowing of
household j is in Fig. 9.9. The positive force component acting upon
borrowing is zj and the negative component is rb .
9.3 Loan Market Behavior

We assume that saving households deposit their savings in banks, and
borrowing households borrow these funds from banks. Banks benefit from
this in the form of the interest differential rd they charge, rd D rb rs >
0, which is assumed fixed for simplicity. Earlier on we showed that an
increase in interest rate decreases the borrowing of borrowing households,
and either increases or decreases the saving of saving households. However,

in the following we assume that the savings of saving households are small
enough so that an increase in interest rate increases their savings.
For banks to have enough funds for their lending in the case of excess
demand of loans, they can raise the deposit and lending rates (the interest
differential is assumed fixed) to get more deposits and to decrease the
demand of loans. In the case of excess supply of deposits, banks can do the
opposite. This behavior maximizes the interest earnings of banks because
then maximal amount of deposits is lent forward with a fixed interest
differential. This means that the lending rate is determined according to
the excess demand of loans over savings, and the savings rate follows the
lending rate. Individual households take the two interest rates as given in
the perfectly competed loan market. For simplicity, banks are assumed to
hold enough cash reserves to prevent bank runs.
§: A bank run is a situation where large numbers of depositors
simultaneously make withdrawals from deposit accounts at a bank, and
the bank’s cash reserves run out and it must close its doors. ˘
The ultimate reason for bank runs is the fractional reserve banking
principle explained in Sect. 8.1.1. However, in order to earn money, banks
must issue loans from their deposits. A well-managed bank keeps the
necessary cash reserves, and banks may also borrow from other banks
or from the central bank if they have a shortage of liquid funds. Thus
a well-managed bank has no fear of a bank run. A bank run occurs when
a bank makes losses in its lending or investment policy, and its depositors
get information about this and start to fear about the bankruptcy of the
bank.
9.4 Aggregate Analysis

According to Sects. 9.1 and 9.2, when every saving and borrowing house-
hold have adjusted optimally, we have
! 0 @uj
1
@ui 1 1
1 @Xi0 1 @ @Xj0
1A ;
p0 p0
rs D @ui
1 and rb D @uj
(9.18)
t @Xi1
1 t 1
p1 @Xj1 p1
i D 1; : : : n; j D 1; : : : m. Adding the n and the m equations in Eq. (9.18)

and dividing the results by n and m, respectively, we get:
! 0 1
@ui @uj 1
1X 1 X
n 1 m
@Xi0 1 1 @ @Xj0 p0
1A :
p0
rs D @ui
1 ; rb D
n iD1 t @Xi1
1 m jD1 t @uj 1
p1 @Xj1 p1
The average force saving households direct upon the aggregate saving in
the economy is
@ui
!
1X 1
n 1
@Xi0 p0
FS D rs @ui
1 D rs zs (S; rs ; p0 ; p1 ; Ts0 ; Ts1 );
n iD1 t @Xi1
1
p1
whereby zs is denoted the average rate of time P preference of saving

n
households
Pn at prevailing
Pn aggregate saving, S D iD1 Si , and Ts0 D
T
iD1 i0 , T s1 D T
iD1 i1 , respectively, are the aggregate incomes of
saving households at the two time units; see the Appendix of this chapter.
The average force borrowing households direct upon the aggregate
borrowing in the economy is
0 1
@uj 1
1 Xm
1 @ @Xj0 p0
FB D rb C 1A D rb Czb (B; rb ; p0 ; p1 ; Tb0 ; Tb1 );
m jD1 t @uj 1
@Xj1 p1
whereby zb is denoted the average rate of time preference

Pm of borrowing
households at prevailing aggregate borrowing, B D jD1 Bj , and Tb0 D
Pm Pm
jD1 Tj0 , Tb1 D jD1 Tj1 , respectively, are the aggregate incomes of
borrowing households at the two time units; see the Appendix of this
chapter.
The resultant force Fres acting upon borrowing in the economy is
then:
Fres D FS C FB D rs zs rb C zb D zb zs rd ;
where in the last form rs was substituted by equation rs D rb rd . Fres is

positive, if the rate of time preference for borrowing households exceeds
that for saving households more than the interest differential, and vice
versa. Thus in the economy borrowing increases if borrowing households
prefer current consumption more than saving households. The aggregate
savings at time unit 0 are
X
n X
n X
n X
n
SD Si D (Ti0 p0 Xi0 ) D Ti0 p0 Xi0 D Ts0 p0 Xs0 ;
iD1 iD1 iD1 iD1
where Xs0 is the aggregate consumption of saving households at time unit

0. The aggregate loans at time unit 0 are
X
m X
m X
m X
m
BD Bj D (p0 Xj0 Tj0 ) D p0 Xj0 Tj0 D p0 Xb0 Tb0 ;
jD1 jD1 jD1 jD1
where Xb0 is the aggregate consumption of borrowing households at time

unit 0.
9.4.1 Adjustment in Aggregate Saving
According to Sects. 9.1.1, 9.4, and the Appendix of this chapter, the
equation of motion for aggregate saving in the economy is
S0 (t) D Rs (FS ); R0s (FS ) > 0; Rs (0) D 0; FS D rs zs (S; rs ; p0 ; p1 ; Ts0 ; Ts1 );

(9.19)
where function Rs obeys the defined characteristics. Thus aggregate saving
increases with time if rs > zs (S; rs ; p0 ; p1 ; Ts0 ; Ts1 ), and vice versa.
9.4.2 Adjustment in Aggregate Borrowing
According to Sects. 9.2.1, 9.4, and the Appendix of this chapter, the
equation of motion for aggregate borrowing in the economy is
B0 (t) D Rb (FB ); R0b (FB ) > 0; Rb (0) D 0; FB D rb Czb (B; rb ; p0 ; p1 ; Tb0 ; Tb1 );
(9.20)
where function Rb obeys the defined characteristics. Thus aggregate

borrowing increases with time if zb (B; rb ; p0 ; p1 ; Tb0 ; Tb1 ) > rb , and vice
versa.
9.4.3 Adjustment in Interest Rate
In the case of a fixed interest rate differential, either of the two interest
rates rs , rb can be considered as the adjusting quantity in the loan market,
and the other is determined by equation rb D rs C rd . Here we choose
lending rate rb to be the adjusting quantity. The lending rate is assumed
to be determined according to the excess demand of loans as:
rb0 (t) D Rr (B S); R0r (B S) > 0; Rr (0) D 0;

B S D p0 (Xs0 C Xb0 ) (Tb0 C Ts0 ): (9.21)
The velocity of lending rate rb0 (t) with unit 1=y2 depends positively on
net borrowing at time unit 0, because p0 (Xs0 C Xb0 ) (e=y) is the value
of aggregate consumption and Ts0 C Tb0 the aggregate income in the
economy at time unit 0.
9.4.4 Equilibrium State in the Loan Market
In the Appendix of this chapter, the average rate of time preference of

saving households is approximated as
@ui
!
1X 1
n 1
@Xi0 p0
zs (S; rs ; p0 ; p1 ; Ts0 ; Ts1 ) @ui
1
n iD1 t @Xi1
1
p1
a0 a1 a2 a3 a4 a5 a6
C 2 S C rs C p0 C p1 C Ts0 C Ts1 ; (9.22)
n n n n n n n
where constant a0 > 0 has unit 1=y, a2 > 0 is dimensionless, the signs
of a3 ; a4 with unit kg/(ye) are ambiguous, and a1 > 0; a5 < 0; a6 > 0
all have unit 1/e. In the Appendix of this chapter, the average rate of time
preference of borrowing households is approximated as
0 @u 1
1 X 1 @ @Xj0 p0
m j 1
zb (B; rb ; p0 ; p1 ; Tb0 ; Tb1 ) 1A

m jD1 t @uj 1
@Xj1 p1
b0 b1 b2 b3 b4 b5 b6
C 2 B C rb C p0 C p1 C Tb0 C Tb1 ; (9.23)
m m m m m m m
where constant b0 > 0 has unit 1=y, b2 < 0 is dimensionless, the signs
of b3 ; b4 with unit kg/(ye) are ambiguous, and b1 < 0; b5 < 0; b6 > 0
all have unit 1/e.
Assuming pc D pc0 and Thc D Th0c0 , h D s; b and c D 1; 2, we can
eliminate these quantities from Eqs. (9.22), (9.23); see the Appendix of
this chapter. Substituting rs from Eq. (9.22) by equation rs D rb rd , we
can approximate the equilibrium state in the loan market as:
a0 a1 a2
Supply function of savings: rb rd D C 2 S C (rb rd ); (9.24)
n n n
b0 b1 b2
Demand function of loans: rb D C 2 B C rb : (9.25)
m m m
Then, setting S D B we can solve the above system for (rb ; S D B ) as
ma1 b0 nb1 (a0 C rd (n a2 ))

rb D ;
ma1 (m b2 ) C nb1 (a2 n)
mnŒa0 (b2 m) a2 (b0 C rd (b2 m)) C n(b0 C b2 rd mrd )
S D B D :
ma1 (m b2 ) C nb1 (a2 n)
(9.26)
The units of the constants can be used to check that the solutions in
Eq. (9.26) are dimensionally well-defined, that is, the units of rb , S are
1=y and e=y, respectively.
The equilibrium state in the loan market, where the forces acting upon
saving, borrowing, and interest rate vanish, is shown in Fig. 9.10, and
where the demand and supply functions are presented in reduced forms
of Eqs. (9.24), (9.25), with n > a2 . This assumption holds when there
are enough saving households, i.e. n is large. On the horizontal axis are
Fig. 9.10 Equilibrium state in the loan market
the aggregate saving and borrowing at time unit 0, and on the vertical axis
is the lending rate. In the equilibrium, aggregate saving equals aggregate
borrowing S D B and this takes place at lending rate rb D rs C rd .
9.4.5 The Explicit Adjustment Process*
In Sects. 9.4.1–9.4.3 we presented the equations of motion for aggregate

saving, borrowing, and interest rate in a perfectly competed loan market.
Taking the Taylor series approximations of functions Rc in Eqs. (9.19)–
(9.20) in the neighborhood of equilibrium points Fc D 0, c D S; B
and assuming the error terms zero, we can approximate functions Rc as
Rc (Fc ) D R0c (0) Fc , where R0c (0) > 0 are constants. Denoting these
constants as R0c (0) D 1=mc , c D s; b, we can interpret them as ‘inverses
of inertial "masses" mc of the adjusting quantities’. These inertial
‘masses’ measure factors resisting changes in the adjusting quantities,
and their measurement unit is y2 =kg when time t has unit y. Constant
R0r (0), however, is denoted by R0r (0) D kr because kr with unit 1=(ye)
corresponds to spring constant in physics; see Fig. 9.11. These units make
the equations dimensionally homogeneous. The factors resisting changes
in interest rate that affect constant kr are existing bank loan contracts with
a fixed interest rate, costs from changing the interest rate, and so on.
Fig. 9.11 The forces acting upon saving, borrowing, and interest rate
The equations of motion for aggregate saving, borrowing, and interest

rate are then:
ms S0 (t) D rb rd zs (S(t); rb (t); rd ; p0 ; p1 ; Ts0 ; Ts1 );

mb B0 (t) D rb C zb (B(t); rb (t); p0 ; p1 ; Tb0 ; Tb1 ); (9.27)
1 0
r (t) D B(t) S(t): (9.28)
kr b
The ‘free body’ diagram of the forces acting upon saving, borrowing,
and interest rate is in Fig. 9.11. The positive force component acting upon
saving is rb and the negative components are zs and rd . The positive force
component acting upon borrowing is zb and the negative one is rb . Notice
that interest differential rd affects negatively the motion of the system.
The horizontal coordinate axis is chosen so that in the equilibrium state
of the system, it moves with constant velocity B D S in the coordinate
system of accumulated saving and borrowing with interest rate staying
constant. A difference in velocities of the particles changes the interest
rate, which balances the difference. The connection between accumulated
and velocity of saving is:
Z Z
d
S(t)dt D S(t); where S(t)dt has unit e and that of S(t) is e=y;
dt
and the same holds for borrowing. As earlier, the box shapes of the
‘particles’ in Fig. 9.11 have no economic meaning, and they are assumed
only to improve Rthe clarityRof the figure. The length of the spring Lr in
Fig. 9.11 is Lr D B(t)dt S(t)dt, and the force by which the spring acts
upon the two particles is rb D kr Lr . By taking the time derivative of this
equation we get Eq. (9.28). The spring between the two particles reflects
the matter that an equal mutual force rb is acting upon the two particles,
and the difference in velocities of the particles B(t) S(t) stretches the
spring that increases the mutual force rb .
Assuming functions zs ; zb as in Eqs. (9.24), (9.25), we can write System
(9.27) as:
a2 a a0 a1
ms S0 (t) D 1
2
rb (t) C 1 rd 2 S(t);
n n n n

b0 b1 b2
mb B0 (t) D C 2 B(t) C 1 rb (t); (9.29)
m m m
1 0
r (t) D B(t) S(t);
kr b
where we used rs D rb rd in substituting rs from the system. Notice that

in System (9.29), the coefficient of rb in the savings equation is positive
with n great enough, and the coefficient of rb in the borrowing equation
is negative with m great enough. The equilibrium state in the loan market
in Eq. (9.26) is obtained by setting S0 (t) D B0 (t) D rb0 (t) D 0 in System
(9.29), and solving the resulting system of three equations with respect to
the three endogenous variables. The speed of adjustment of the system
toward the equilibrium state depends on the two inertial masses, the
spring constant, and the values of the parameters. In this book we mainly
concentrate on the equilibrium state, however, which is meaningful only
if the system is stable and will converge to the equilibrium state with
time. With certain values for the constants in the model, System (9.29)
converges with time to the equilibrium state given in Eq. (9.26).
In Fig. 9.12 is presented the aggregate deposits received and the aggre-
gate loans issued by Finnish banks in units 109 e together with the
EONIA rate of interest. EONIA (Euro OverNight Index Average) is the
one-day interbank interest rate for the eurozone at which banks provide
loans to each other within the duration of one day. Therefore EONIA
can be called the one-day Euribor (though expressed in units 1=year).
The figure shows that a continuous increase has taken place in aggregate
savings and loans, while the trend in the EONIA rate is either constant
or decreasing. Because the issued loans by Finnish banks have exceeded
the deposits they have received during the whole time period, banks have
also acquired funds by issuing certificates of deposits to their clients.
Because Finnish financial markets are a relatively small part of European
financial markets, Finnish bank loans and deposits do not much affect
the EONIA rate. On the other hand, changes in the EONIA rate do
affect Finnish saving and borrowing, but because the changes have been
relatively modest during time period 2006–2013 these effects have been
minor.
Because the general solution of System (9.29) is rather complicated, we
demonstrate it by using the following values for the parameters: rm D 0:5,
mc D 1, c D s; b, kr D 1=5, a0 D 100; a1 D 2000; a2 D 0:02,
b0 D 200 C 0:4t; b1 D 5; b2 D 2, m D 100; n D 200. The interest
rate is expressed in percent and the loans and savings are expressed in
units 109 e. A positive time trend is assumed in zb , b0 D 200 C 0:4t,
to get model (9.29) to resemble the observed behavior in Fig. 9.12. With
these parameter values, the time paths of variables S; B; rb are presented
in Fig. 9.13 where on the horizontal axis is time and on the vertical axis
are S; B, and rb . The three quantities are presented in the same figure
to demonstrate their connections, even though their measurement units
Fig. 9.12 Deposits and bank loans in Finland together with Eonia rate
Fig. 9.13 Time paths of aggregate saving, borrowing and interest rate
differ. The time path of rb is the one below those of S; B, and that of S is
the thicker one. The figure shows that rb increases if B > S, and vice versa.
As compared with the static neoclassical analysis, by System (9.27) we
can study the reasons for the dynamics of a perfectly competed loan mar-
ket by assuming time dependencies in zs (S(t); rb (t); rd ; p0 ; p1 ; Ts0 ; Ts1 )
and in zb (B(t); rb (t); p0 ; p1 ; Tb0 ; Tb1 ). On the other hand, the adjustment
process in the loan market can be studied by solving System (9.29) with
different numerical values for the parameters in the model. Concerning
the speed of adjustment, the neoclassical framework is a special case of
System (9.27) with an infinite speed of adjustment, that is, mc D 0,
c D s; b and 1=kr D 0.
9.4.6 Adding Investment in the Model*
In Chap. 7 we modeled firms’ investment behavior. Firms may finance

their investments by their cash reserves, by issuing equities, or by taking
loans. If firms take investment loans, they act as borrowers in the loan
market and we have to add their effect on aggregate borrowing. On the
other hand, we do not model firms’ depositing their cash reserves in banks
because we assume that firms mainly act as borrowers in the loan market.
The equation of motion for a firm’s investment was given in Sect. 7.5.3.
For simplicity, we assume that every firm applies loan financing in only
one kind of capital goods, and we denote by Npvi (e) the present value of
net revenues (or cost savings) one capital good (machine) causes for firm
i, and by Ci (e) the price (value) of one machine. These net revenues also
include the costs of loans.
We assume v firms in the economy and denote by Ii (t) (e=t) the
loan financed investments of firm i during time unit t D y. According
to Sect. 7.5.3, the following model describes the investments of firm i:
Npvi
mi Ii (t) D Fi ; Fi D rb ; Ii (t) D Ki0 (t); (9.30)
Ci
where Ni (e=y) is the net revenues (or cost savings) one capital good
causes for firm i during time unit y, and rb (1=y) is the interest rate on
loans. A firm financing its investments by a loan applies the loan rate in its
discounting due to the same reason as borrowing households. Constant
mi > 0 with unit 1/e represents the inertial factors in the adjustment of
the capital stock of firm i.
In the Appendix of this chapter, we show that we can approximate
P the
dynamics of loan financed investments of all firms, I(t) D viD1 Ii (t), in
the neighborhood of point xi0 D (Ni0 ; rb0 ; Ci0 ), i D 1; : : : ; v, as:
z0 z1 z2
mI I(t) D C 2 N C 2 C rb ;
v v v
P
where N D viD1 Ni is the aggregate netPrevenues one machine causes for
all firms during time unit y, and C D viD1 Ci is the aggregate of prices
(values) of the investment goods. Denoting the aggregate investment loans
by I(t), System (9.29) becomes the following:
a a0 a1 a2
ms S0 (t) D
2
1 rd 2 S(t) 1 rb (t);
n n n n

b0 b1 b2
mb B0 (t) D C 2 B(t) C 1 rb (t);
m m m
z0 z1 z2
mI I(t) D C 2 N C 2 C rb (t);
v v v
1 0
r (t) D B(t) C I(t) S(t): (9.31)
kr b
Then, comparing Systems (9.29) and (9.31) we see that now the equilib-
rium state in the loan market, S0 (t) D B0 (t) D rb0 (t) D 0, is:

b1 mI n .a0 C (n a2 )rd / v 2 C a1 b1 (vz0 C Nz1 C Cz2 ) b0 mmI v 2
rb D ;
.a1 .b1 C (b2 m)mmI / C b1 mI (a2 n)n/ v 2
na0 .b1 C (b2 m)mmI / v 2

S D
(a1 (b1 C (b2 m)mmI ) C b1 mI (a2 n)n)v 2

n(a2 n) b0 mmI v 2 Cm(m b2 )mI rd v 2 b1 (rd v 2 Cvz0 CNz1 CCz2 )
;
(a1 (b1 C(b2 m)mmI )Cb1 mI (a2 n)n)v 2
m .mI n ..a0 (m b2 )Cn(b0 b2 rd Cmrd ) a2 .b0 C (m b2 )rd //// v 2

B D
2
ma1 b0 v C (b2 m)(vz0 C Nz1 C Cz2 )
;

a1 m b0 v 2 C (b2 m)(vz0 C Nz1 C Cz2 )
I D
b1 n(a0 v 2 (a2 n)(rd v 2 C vz0 C Nz1 C Cz2 ))
: (9.32)
(a1 (b1 C (b2 m)mmI ) C b1 mI (a2 n)n)v 2
The equilibrium state in the loan market thus changes somewhat. Notice
that I ¤ 0, that is, in the equilibrium the aggregate loan-financed
investments change the equilibrium aggregate saving, borrowing, and
interest rate in the economy. If we had assumed that in the equilibrium
state holds I(t) D K 0 (t) D 0, this would mean that the capital stock
remains constant, and then investments would not affect the loan market.
We could compare the solutions in Eqs. (9.26) and (9.32) by assuming
numerical values for the exogenous variables and the parameters in the
model. For shortness, however, we omit this.

We take the Taylor series approximation of function
@ui 1
!
1 @Xi0 p0
zsi (Si ; rs ; p0 ; p1 ; Ti0 ; Ti1 ) @ui
1
t @Xi1
1
p1
in the neighborhood of point xi0 D (Si0 ; rs0 ; p00 ; p10 ; Ti00 ; Ti10 ) as
@zsi (xi0 ) @zsi (xi0 )
zsi D zsi (xi0 ) C (Si Si0 ) C (rs rs0 )
@Si @rs
@zsi (xi0 ) @zsi (xi0 ) @zsi (xi0 )
C (p0 p00 ) C (p1 p10 ) C (Ti0 Ti00 )
@p0 @p1 @Ti0
@zsi (xi0 )
C (Ti1 Ti10 ) C i : (9.33)
@Ti1
Now, assuming i D 0 8i and summing over i, we get3

X
n
zsi (Si ; rs ; p0 ; p1 ; Ti0 ; Ti1 )
iD1
Xn
@zsi (xi0 ) @zsi (xi0 )
D zsi (xi0 ) Si0 rs0
iD1
@Si @rs

@zsi (xi0 ) @zsi (xi0 ) @zsi (xi0 ) @zsi (xi0 )
p00 p10 Ti00 Ti10
@p0 @p1 @Ti0 @Ti1
X
n
@zsi (xi0 ) X
n
@zsi (xi0 ) X
n
@zsi (xi0 ) X
n
@zsi (xi0 )
C Si C rs C p0 C p1
iD1
@Si iD1
@rs iD1
@p0 iD1
@p1
X
n
@zsi (xi0 ) X
n
@zsi (xi0 )
C Ti0 C Ti1
iD1
@Ti0 iD1
@Ti1
a1 a5 a6
a0 C S C a2 rs C a3 p0 C a4 p1 C Ts0 C Ts1 ;
n n n
Pn Pn Pn Pn
3
Because iD1 ci xi D c iD1 xi C iD1 (ci c)xi where c D (1=n) iD1 ci , the approximation
is the more accurate the less ci and xi vary, i D 1; : : : ; n.
Pn Pn Pn
where S D iD1 Si , Ts0 D iD1 Ti0 , Ts1 D iD1 Ti1 , and
n
X @zsi (xi0 ) @zsi (xi0 ) @zsi (xi0 )
a0 D zsi (xi0 ) Si0 rs0 p00
iD1
@Si @rs @p0

@zsi (xi0 ) @zsi (xi0 ) @zsi (xi0 )
p10 Ti00 Ti10 ;
@p1 @Ti0 @Ti1
X
n
@zsi (xi0 ) X
n
@zsi (xi0 ) X
n
@zsi (xi0 )
a1 D ; a2 D ; a3 D ;
iD1
@Si iD1
@rs iD1
@p0
X
n
@zsi (xi0 ) X
n
@zsi (xi0 ) X
n
@zsi (xi0 )
a4 D ; a5 D ; a6 D :
iD1
@p1 iD1
@Ti0 iD1
@Ti1
The unit of a0 is 1=y, a2 is dimensionless, those of a3 ; a4 are kg/(ye),

and a1 ; a5 ; a6 all have unit 1/e. Because the rate of time preference
of the saving household i is positive at every Si ; rs ; : : :, then a0 > 0
(let Si ; rs ; : : : ; i ! 0 8i in Eq. (9.33)). In Sect. 9.1 we showed that
@zsi =@Si > 0, @zsi =@Ti0 < 0, @zsi =@rs > 0, and @zsi =@Ti1 > 0. Thus
a1 > 0, a2 > 0, a5 < 0, a6 > 0, and the signs of a3 ; a4 are ambiguous.
We take the Taylor series approximation of function
0 @uj
1
1
1 @ @Xj0
1A
p0
zbj (Bj ; rb ; p0 ; p1 ; Tj0 ; Tj1 ) @uj
t 1
@Xj1 p1
in the neighborhood of point xj0 D (Bj0 ; rb0 ; p00 ; p10 ; Tj00 ; Tj10 ) as
@zbj (xj0 ) @zbj (xj0 )

zbj D zbj (xj0 ) C (Bj Bj0 ) C (rb rb0 )
@Bj @rb
@zbj (xj0 ) @zbj (xj0 ) @zbj (xj0 )
C (p0 p00 ) C (p1 p10 ) C (Tj0 Tj00 )
@p0 @p1 @Tj0
@zbj (xj0 )
C (Tj1 Tj10 ) C j : (9.34)
@Tj1
Then, assuming j D 0 8j and summing over j, we get4
X
m Xm
@zbj (xj0 )
zbj (Bj ; rb ; p0 ; p1 ; Tj0 ; Tj1 ) D zbj (xj0 ) Bj0
jD1 jD1
@Bj
@zbj (xj0 ) @zbj (xj0 ) @zbj (xj0 ) @zbj (xj0 )

rb0 p00 p10 Tj00
@rb @p0 @p1 @Tj0
X m Xm
Tj10 C Bj C rb
@Tj1 jD1
@Bj jD1
@rb
X
m
@zbj (xj0 ) X
m
@zbj (xj0 ) X
m
@zbj (xj0 )
Cp0 C p1 C Tj0
jD1
@p0 jD1
@p1 jD1
@Tj0
X
m
@zbj (xj0 ) b1 b5 b6
C Tj1 b0 C B C b2 rb C b3 p0 C b4 p1 C Tb0 C Tb1 ;
jD1
@Tj1 m m m
where
m
X @zbj (xj0 ) @zbj (xj0 ) @zbj (xj0 )
b0 D zbj (xj0 ) Bj0 rb0 p00
jD1
@Bj @rb @p0

p10 Tj00 Tj10 ;
@p1 @Tj0 @Tj1
X
m
@zbj (xj0 ) X
m
@zbj (xj0 ) X
m
@zbj (xj0 )
b1 D ; b2 D ; b3 D ;
jD1
@Bj jD1
@rb jD1
@p0
X
m
@zbj (xj0 ) X
m
@zbj (xj0 ) X
m
@zbj (xj0 )
b4 D ; b5 D ; b6 D :
jD1
@p1 jD1
@Tj0 jD1
@Tj1
The unit of b0 is 1=y, b2 is dimensionless, those of b3 ; b4 are kg/(ye),

and b1 ; b5 ; b6 all have unit 1/e. Because the rate of time preference
of borrowing household j is positive at every Bj ; rv ; : : :, then b0 > 0
4
See footnote 3.
(let Bj ; rb ; : : : ; j ! 0 8j in Eq. (9.34)). In Sect. 9.2 we showed that

@zbj =@Bj < 0, @zbj =@Tj0 < 0, @zbj =@Tj1 > 0, and @zbj =@rb < 0. Thus
b1 < 0, b2 < 0, b5 < 0, b6 > 0, and the signs of b3 ; b4 are ambiguous.
Next we approximate the force acting upon the value of the stock of
physical capital of firm i, Fi D Ni =Ci rb , by taking the Taylor series
approximation in the neighborhood of point Fi (xi0 ), xi0 D (Ni0 ; rb0 ; Ci0 )
as
@Fi (xi0 ) @Fi (xi0 ) @Fi (xi0 )
Fi D Fi (xi0 )C (Ni Ni0 )C (Ci Ci0 )C (rb rb0 )Ci ;
@Ni @Ci0 @rb
(9.35)
@Fi @Fi
where @N i
D 1=Ci > 0, @C i
D (Ni =Ci2 ) < 0, @F @rb
i
D 1 < 0, and i
is the error term. Assuming i D 0 8i and adding the forces acting upon
the values of stocks of physical capital of every firm, we get:
v
X Xv
@Fi (xi0 ) @Fi (xi0 ) @Fi (xi0 )
Fi D Fi (xi0 ) Ni0 Ci0 rb0
iD1 iD1
@Ni @Ci @rb
v
X v
X v
X
@Fi (xi0 ) @Fi (xi0 ) @Fi (xi0 )
C Ni C Ci C rb
iD1
@Ni iD1
@Ci iD1
@rb
z1 z2
z0 C N C C vrb ;
v v
Pv Pv
where N D iD1 Ni , C D iD1 Ci , and5
Xv
@Fi (xi0 ) @Fi (xi0 ) @Fi (xi0 )
z0 D Fi (xi0 ) Ni0 Ci0 rb0
iD1
@Ni @Ci @rb
v
X v
X v
X
@Fi (xi0 ) @Fi (xi0 ) @Fi (xi0 )
z1 D ; z2 D ; z3 D D 1 v:
iD1
@Ni iD1
@Ci iD1
@rb
5
See footnote 3.
Taking the average of Fi over the v firms we get

v
1X z0 z1 z2
Fi D C 2 N C 2 C rb
v iD1 v v v
Because the investment of every firm may be positive or negative, z0 is of

ambiguous sign. On the other hand, @Fi =@Ni > 0, @Fi =@Ci < 0, and
@Fi =@rb D 1, 8i, so 1 > 0; z2 < 0.
Pwe can sign the other constants as: zP
Approximating as viD1 mi Ii (t) mI I(t), where I(t) D viD1 Ii (t) and
v
1X
mI D mi > 0;
v iD1
we can model the aggregate investment behavior of all firms as

z0 z1 z2
mI I(t) D C 2 N C 2 C rb :
v v v
10
Mathematical Appendix
10.1 Introduction
In this chapter we present the mathematical concepts and rules of calcu-
lation necessary to understand the main parts of the book.
In general, quantities can be divided into scalar and vector quantities.
Scalars have a certain magnitude and vectors have, besides magnitude,
also a direction, which means that vectors have several dimensions (see
Section 10.4). All real numbers and dimensional quantities are scalars,
and vectors may have real numbers or scalars as their components. In this
book, complex numbers or vectors with complex number components are
not treated.
Note. We use the term dimension in two ways. By the dimension of

a vector we understand the number of its components, so that vector
r1 D (a1 ; a2 ) has two dimensions, r2 D (a1 ; a2 ; a3 ) has three dimensions,
and so forth. On the other hand, in dimensional analysis, by the
dimension of a quantity we understand a set of additive quantities in the

DOI 10.1007/978-3-319-46879-2_10
measurement system of a real science. For example, dimension Œlength in

physics defines the set of additive quantities measuring length, and so on.
˘
Now, mathematics operates with dimensionless quantities while real

sciences, like physics and economics, operate with dimensional quantities,
that is, quantities with a measurement unit. Mathematical operations with
quantities having a measurement unit set certain rules for calculation;
only quantities with equal dimension (measurement unit) can be added
together, and so on. In this appendix, mathematical concepts are mostly
presented without taking account of dimensional analysis. In the exam-
ples, however, the measurement units of applied quantities are exactly
specified.
10.2 Straight Line

Plane curves can be expressed as equations of two variables in the
coordinate system of the plane. Here we introduce more closely the most
familiar plane curve, namely, the straight line.
The equation of a straight line is y D kx C b, where constants k and
b are real numbers or scalars. This line cuts the x-axis at point (x; y) D
(b=k; 0), the y-axis at point (x; y) D (0; b), and the direction of the line
is defined by its slope k; see Fig. 10.1. A straight line, that has the same
direction as y-axis and cuts x-axis at point (a; 0), has equation x D a.
Expression y D kx C b is called the solved form of the equation of
a straight line. The equation can also be expressed in the general form
Ax C By C C D 0, where A ¤ 0 or B ¤ 0:
A straight line is uniquely determined if two points of it are known.
Suppose a line goes through points (x1 ; y1 ) and (x2 ; y2 ) 2 R2 , whereby
R is denoted the set of real numbers. Its slope is
y2 y1
kD ; x1 ¤ x2 ;
x2 x1
Fig. 10.1 Graphical presentation of the equation of a straight line
and the equation of the line is
y y1 D k(x x1 ) , y D (y1 kx1 ) C kx;
where the slope k has been defined earlier; see Fig. 10.1. The latter form
of the equation shows that this equation can be presented analogously as
in the solved form by denoting the constant as: b D y1 kx1 .
Example
Suppose it is known that the costs of production of a firm from a good
depend linearly on the amount of production. It is also known that when 12
products were produced, the costs were 2520 (e), and 22 products created
costs 3620 (e). With this knowledge, we can formulate the equation that
describes the relationship between costs and amount of production.
We denote by x (unit) the amount of production and by y (e) the
costs of production. The line goes through points (12 (unit); 2520 (e)) and
(22 (unit); 3620 (e)). The equation of the line is then:
3620 (e) 2520 (e)

y (e) 2520 (e) D (x (unit) 12 (unit))
22 (unit) 12 (unit)
1100 e
, y (e) 2520 (e) D (x (unit) 12 (unit))
10 unit
e
, y (e) 2520 (e) D 110 x (unit) 1320 (e)
unit
e
, y (e) D 110 x (unit) C 1200 (e);
unit
where symbol is used for multiplication when it improves readability;

usually it is omitted. The equation of the line describing the relationship
between costs y and amount of production x is thus:
e
y (e) D 110 x (unit) C 1200 (e): ˘ (10.1)
unit
One way to define correct measurement units for dimensional con-

stants in equations and functions—that is common in physics—is to
define the corresponding equation for measurement units. The corre-
sponding equation for measurement units with Eq. (10.1) is
e D A unit C B; (10.2)
where on the left-hand side is the unit of y, by A is denoted the unknown

measurement unit of constant 110, unit is the measurement unit of x,
and by B is denoted the unknown unit of constant 1200. For Eq. (10.2)
to be dimensionally well-defined, that is, well-defined with respect to
measurement units, both sides of the equation must have an equal unit,
and all terms added together must have equal units. Writing Eq. (10.2) in
the form
e B D A unit
we see that B D e must hold that the adding on left-hand side of the
equation is dimensionally well-defined. On the other hand, the unit of
the right-hand side of the equation A unit must be e for both sides to
have equal unit. Thus we can write equation
e
e D A unit , A D
unit
from which we solved the unknown measurement unit A for constant 110.
10.3 Closed and Complete Sets

A set consists of elements. Sets are denoted by capital letters like A, B, and
so forth. If element x belongs in set A, we denote this: x 2 A. If element
x does not belong in A, we denote: x … A. The set of real numbers
R contains all positive and negative numbers together with number 0.
The elements of R2 are points of the plane (x; y), where x; y 2 R is
an ordered pair of real numbers. However, if the coordinates of the plane
have a measurement unit, then (x; y) is an ordered pair of scalars. A special
situation is the so-called empty set ; that contains no elements.
§: Set A
R2 is closed, if it contains all its border points. ˘
Example
The circular disc of a plane, B((x0 ; y0 )I r), is not a closed set because it does
not contain its border points. On the other hand, a set that contains points
with distance from (x0 ; y0 ) less than or equal with r, is a closed set. This set
contains its border points the distance of which from point (x0 ; y0 ) is exactly
r, too. ˘
§: In mathematical analysis, metric space M is called complete (or a

Cauchy space), if every Cauchy sequence of points in M has a limit point
in M or, alternatively, if every Cauchy sequence in M converges in M. ˘
We do not have space here to study the above definition exactly, but
every Cauchy sequence is a convergent sequence, and thus in a complete
metric space M, every convergent sequence has a limit point in M.
Intuitively this means that a space is complete if there are no ‘missing
points’ in it (inside the space or at its boundary).
10.4 Vectors
Quantities that have an exactly defined direction in addition to magnitude
are called vectors. Vectors are usually denoted by small letters with a little
arrow above the variable like vE, uE, or by bold small letters like v, u. We
use the latter notation.
Fig. 10.2 Graphical presentation of a vector
We can graph vector v 2 R2 with the starting point in the origin (0; 0)
and ending point at (x0 ; y0 ). The coordinates of the end pointq
define the
direction of vector v exactly, and the length of vector v is jvj D x02 C y20 ;
see Fig. 10.2.
10.4.1 Calculation Rules for Vectors
Here we give the calculation rules for vectors of R2 .
Adding and Subtracting Vectors
Let a D (x1 ; y1 ) and b D (x2 ; y2 ). Then their sum is
a C b D (x1 C x2 ; y1 C y2 );
and their difference is
a b D (x1 x2 ; y1 y2 ):
Multiplying Vectors by Real Numbers
Let r 2 R. The product of vector a D (x; y) and real number r is
ra D (rx; ry):
The length of vector ra is
jraj D jrjjaj;
and its direction is the same as that of a if r > 0, and opposite to a if

r < 0. The calculation rules for vectors of multidimensional spaces, Rn ,
n D 3; 4; : : : are defined analogously.
10.5 Functions
10.5.1 The Definition of a Function
§: Function f from set A to set B, denoted by f W A ! B, is a rule that

connects to every element a 2 A exactly one element b 2 B. ˘
Set A is called the domain and set B the range of function f . If f (a) D
b, we say that b is the value of function f at point a.
Example 1
Function
f W RC ! R; f (x) D 214x C 515;
is a one-variable real valued function. ˘
Example 2
Function
f W R2 ! R; f (x; y) D 24xy;
is a real-valued function of two variables x and y. ˘

When functions are used in modeling phenomena in real sciences, the

values of the functions usually have a measurement unit. These are called
scalar-valued functions. We call a function vector-valued, if the range
of the function is a set of vectors or points in a multidimensional space.
Example 3
Function
f W R2 ! R2 ; f (x; y) D (x2 ; xy)
is a vector-valued function of two variables x and y. ˘
10.5.2 The Graph of a Function
§: The graph of a one-variable function f W A ! B is the set of points

(x; f (x)), where x 2 A and f (x) 2 B. The graph of a one-variable function
f W R ! R is the curve y D f (x) in plane R2 ; see Fig. 10.3. ˘
§: The graph of function f W A ! R, A
R2 consists of points
(x; y; f (x; y)) in the three-dimensional space. The values of function
f (x; y) are measured on the vertical axis, and the arguments of the function
(x; y) 2 A, A
R2 , are measured on the horizontal plane R2 . ˘
The graph of function z D f (x; y), f W R2 ! R is a two-dimensional
surface in a three-dimensional space; see Fig. 10.4a. The values of the
function are measured on the vertical axis z, and every point (x; y) on
the horizontal plane belongs in the domain of the function. The value
of function z D f (x0 ; y0 ) at point (x0 ; y0 ) is found on the ‘hill-shaped’
Fig. 10.3 The graph of a one-variable real valued function

Fig. 10.4 (a) The surface of z D f (x; y): (b) The level curves of z D f (x; y)
surface above the point; see Fig. 10.4a. The values of the function are
constant at constant height, which is demonstrated in the figure by
horizontal plane at z D c, c constant. The cutting curve of plane z D c
and the surface of function values is a curve in the plane, where the value
of the function is constant, z D c.
The surface of function z D f (x; y) in Fig. 10.4a can be described
in a two-dimensional plane, similar to a two-dimensional topographical
map being used to describe the three-dimensional shape of the Earth; see
Fig. 10.4b. In a topographical map, the points at the same height from
the level of sea are connected by a level curve. These level or indifference
curves are drawn on the horizontal ‘bottom’ plane of the map. The level or
contour curves of function f (x; y) similarly consist of points (x; y) 2 R2
in which the value of the function has constant value: f (x; y) D c,
c 2 R, constant. In Fig. 10.4b is shown two contour curves of function
z D f (x; y); the further away from origin the level curve is, the smaller
function value it represents.
10.5.3 Composite Function
In certain situations, one quantity depends on another, and in turn

that depends on a third. By connecting the functions describing these
relationships, we can express the initial quantity as a function of the third
one; see Fig. 10.5.
Fig. 10.5 Graphical presentation of function g ı f W A ! C
§: Let f W A ! B and g W B ! C be functions. A composite function

g ı f is a function from set A to set C so that
(g ı f )(x) D g(f (x)) for all x 2 A: ˘
Example
Suppose the daily costs of production of a good C (e/day) follow the
function
C(x) D a0 x2 C a1 x C 900;
where the flow of production is denoted by x (unit=day). Here a0 and a1 are

dimensional constants. For the function to be dimensionally well-defined,
the measurement units of a0 and a1 must be, respectively:
e day e
a0 W and a1 W :
unit2 unit
The unit of a0 x2 and a1 x is then e/day, which is also the unit of constant 900.
Suppose the flow of production at a normal day is 25 (unit=h), and we
want to express the costs of production as a function of daily work time
L (h=day). This is
unit h unit
x(L) D 25 L D 25L :
h day day
The daily costs can then be expressed as the composite function:
C(x(L)) D a0 .x(L)/2 C a1 x(L) C 900

D a0 (25L)2 C 25a1 L C 900
D 625a0 L2 C 25a1 L C 900:
Next we show that this result is dimensionally well-defined:

unit 2 e day h 2
C(x(L)) D 252 a0 L2
h unit2 day
unit e h e
C25 a1 L C 900
h unit day day
e e e
D 625a0 L2 C 25a1 L C 900 :˘
day day day
In this example, constants a0 , a1 , 25, and 900 have certain measure-

ment units. In theoretical analysis, the measurement units of constants
are defined so that the applied functions and equations are dimensionally
well-defined. These units help in interpreting the constants, and their
numerical values can be defined in empirical studies.
In this chapter, in clear cases, we do not treat the measurement units
of dimensional constants explicitly. The units of constants are also left
out from equations, if they make the equations unclear. The reader can,
however, always define the measurement units of the elements in the
domain and range of a function and check that they are dimensionally
well-defined.
10.5.4 Inverse Function
The definition of an inverse function is:

§: Let f W A ! B be a function. The inverse function of f is function
f 1 W B ! A that defines for all elements b 2 B an element a 2 A for
which holds f (a) D b. ˘
Note. Not all functions have an inverse function. An inverse function can
be defined only for functions f W A ! B for which for every element
in the range b 2 B corresponds to one and only one element in the
domain a 2 A. This kind of function is called a bijection. Function
f W A ! B is a bijection, if it obeys the following characteristics:
• f gets its all values only once, that is f (x1 ) D f (x2 ) , x1 D x2 , and
• f gets all values in its range. ˘
Fig. 10.6 (a) A bijective function. (b) A non-bijective function
For example, if function f W A ! B is strictly monotonic (strictly

increasing or decreasing), it is a bijection; see Fig. 10.6.
If function f W A ! B is a bijection, it has an inverse function
f 1 W B ! A. Function f and its inverse function f 1 have the following
connection:
y D f (x) , f 1 (y) D f 1 (f (x)) D x:
Example
Let the weekly cost function of a firm C Œ0; 1Œ! Œ45; 1Œ be C(x) D 15x C 45,
where constants 15 and 45 have units e/unit and e=week, respectively. The
inverse function of C expresses the flow of production that can be produced
at costs y (e/week). Function C is strictly increasing and its range is Œ45; 1Œ.
Thus C is a bijection and it has an inverse function. The inverse function is
obtained as follows:
C(x) D y; y 45
e unit
e

e

15 x C 45 Dy ; x0
unit week week week
e e
unit y week 45 week
x D e
week 15 ( unit )

unit y unit unit
x D 3 :
week 15 week week
y
Thus by costs y (e/week) the firm can produce x D 15 3 (unit=week). The
inverse function of C is C1 W Œ45; 1Œ! Œ0; 1Œ, C1 (y) D 15 3. ˘
y
10.6 Limit Value

Limit value expresses the value of function f (x) when x ! x0 .
§: A one-variable function f has a limit value a at x D x0 denoted by
lim f (x) D a;
x!x0
if f (x) approaches a when the values of x are close enough to x0 . The limit
value can also be denoted as: f (x) ! a when x ! x0 . ˘
The limit value of a two-variable real-valued function is defined as:
§: The limit value of a two-variable real-valued function f W A ! R,
A
R2 at point (x0 ; y0 ) 2 A, is
lim f (x; y) D a;
(x;y)!(x0 ;y0 )
if the values of f approach a when point (x; y) approaches (x0 ; y0 ). ˘
Note. If function f W A ! R, A
R2 has limit value a at point (x0 ; y0 ),
the values of the function must approach a independently of the direction
from where (x; y) ! (x0 ; y0 ). ˘
10.7 Derivative
10.7.1 The Definition of Derivative
In this section we study one-variable functions. The starting point is

the problem: What is the rate of change (or velocity) of a one-variable
function f at point x0 ? Let the values of function f (x) be f (x0 ) and f (x1 ) at
points x0 and x1 , respectively. We can then make the following definition:
§: The average rate of change of function f with respect to variable x,
when x changes from x0 to x1 , is
f (x1 ) f (x0 ) f
D ;
x1 x0 x
where a change in a quantity is denoted by Greek letter Delta, . ˘

Fig. 10.7 Graphical presentation of the derivative of function f (x) at x0
§: The instantaneous rate of change of function f with respect to x at

point x0 is the limit value:
f (x1 ) f (x0 ) f
lim D lim :˘
x1 !x0 x1 x0 x!0 x
Geometrically, the instantaneous rate of change of function f at point

x0 is the slope of the tangent of the graph of function f at (x0 ; f (x0 )). We
see this by drawing line S0 going through points (x0 ; f (x0 )) and (x1 ; f (x1 ));
see Fig. 10.7. The slope k of line S0 is:
f (x1 ) f (x0 ) f
kD D :
x1 x0 x
Slope k measures the average rate of change of function f when x changes

from x0 to x1 . When x1 ! x0 , line S0 turns to tangent S1 for the graph
of the function at x0 , see Fig. 10.7. The slope of the tangent of the graph
of function f at x0 is the limit value of difference quotient f =x when
x1 ! x0 . This limit value denoted by f 0 (x0 ) is called the derivative of
function f at x0 .
§: The derivative of function f at point x0 , f 0 (x0 ), is
f (x1 ) f (x0 )
f 0 (x0 ) D lim :˘
x1 !x0 x1 x0
We say that function f is differentiable at point x0 , if the limit value

of difference quotient f 0 (x0 ) exists. The derivative of function f also has
other notations than f 0 (x). These are
df
Df or :
dx
If the derivative function f 0 W A ! R of function f W A ! R is
differentiable, we can define the derivative function of function f 0 , that is,
the second order derivative function of f , f 00 . This can also be denoted as

d df d2 f
D :
dx dx dx2
In this notation, dd is denoted by d2 and (dx)2 shortly by dx2 .
10.7.2 Some Rules of Derivation
Derivative of a Product
Let f (x) D g(x) h(x). The derivative of a product of two functions is:
f 0 (x) D g0 (x)h(x) C h0 (x)g(x):
Derivative of a Quotient
g(x)
Let f (x) D h(x)
. Then f is differentiable if h(x) ¤ 0, and its derivative is:
g0 (x)h(x) h0 (x)g(x)
f 0 (x) D ; h(x) ¤ 0:
(h(x))2
Derivative of a Composite Function
Let f be a combined function: f (x) D g(h(x)). The derivative of f is:
f 0 (x) D g0 (h(x)) h0 (x):

This derivative can also be denoted as

dg dh
f 0 (x) D :
dh dx
This is known as the chain rule of differentiation, and it shows explicitly

with respect to which quantity every derivation is made.
Derivative of an Inverse Function
Let f W R ! R, y D f (x) be invertible so that x D f 1 (y) holds 8 x; y 2

R; 8 is an abbreviation for ‘for all’. Then we can write:
dy dx 1 1
D f 0 (x) , D (f 1 )0 (y) D 0 D dy :
dx dy f (x) dx
Derivatives of Logarithmic and Exponential Functions
In this chapter we do not treat logarithmic functions other than the

natural logarithm with base e. The calculation rules for other logarithms
with bases 2, 10, : : : are identical. Let y D ln(x) where ln is the
abbreviation for the natural logarithm function with e D 2:718 : : : as its
base number. The inverse function of the natural logarithmic function is
the exponential function x D exp(y) D ey , where exp is its abbreviation.
Then the following holds:
y D ln(x) , x D exp(y) D ey : (10.3)
Thus in solving variable x from the first equation in Eq. (10.3) we apply
the exponential function on both sides of Eq. (10.3) as
ey D eln(x) D x: (10.4)
For the logarithmic and exponential functions hold:
ln(x y) D ln(x) C ln(y); exCy D ex ey :
The derivatives of logarithmic and exponential functions are:
d 1 d x
ln(x) D ; e D ex :
dx x dx
Let us have composite functions y D ln(f (x)) and z D exp(g(x)). By

applying the derivative rule for a composite function, the derivatives of
these functions are, respectively:
dy f 0 (x) dz
D and D g0 (x)eg(x) :
dx f (x) dx
10.7.3 Examples of Derivatives
Example 1
Let y D 2 ln(x). Dividing by 2 and applying the exponential function on both
sides of the equation, we get the inverse function as:
y
D ln(x) , ey=2 D eln(x) , ey=2 D x:
2
The derivatives of the original and inverse function are:
dy 2 dx 1
D and D ey=2 :
dx x dy 2
Now, substituting formula x D ey=2 in dy=dx above, we get:
dy 2
D D 2ey=2 :
dx x
Comparing this and the formula for dx=dy we see that dy=dx D 1=(dx=dy). ˘
Example 2
Let L(t) describe the labor input of a firm as a function of time, and let q(L(t))
be the production function of the firm. The instantaneous rate of change
q(L(t))
of productivity of labor of the firm, L(t) , with respect to time, is then:
q(L(t))
d( L(t)
) q0 (L(t)) L0 (t) L(t) L0 (t) q(L(t))
D
dt ŒL(t)2
L0 (t) 0 L0 (t) q(L(t))
D q (L(t))
L(t) L(t) L(t)
0

L (t) 0 q(L(t))
D q (L(t)) :˘
L(t) L(t)
Example 3
Let the weekly profit …k of a firm be
…k D Bk (qk (t)) qk (t) Ck (qk (t));
where qk (t) is the flow of production of good k at time moment t, Bk (qk (t))
the inverse demand function of the good, and Ck (qk (t)) the weekly cost func-
tion of the firm. Then, according to the rule of derivation of a composite
function, the time derivative of the profit function is:
…0k (t) D B0k (qk (t)) q0k (t) qk (t) C q0k (t) Bk (qk (t)) Ck0 (qk (t)) q0k (t)

D B0k (qk (t)) qk (t) C Bk (qk (t)) Ck0 (qk (t)) q0k (t): ˘
10.8 Applications of Derivatives

10.8.1 Increasing and Decreasing Functions
The direction of change of a function can be studied according to the

sign of its derivative function. Let function f be continuous on interval
Œa; b and differentiable on a; bŒ. Then the following holds on a; bŒ: if
f 0 (x) > 0, f is strictly increasing, if f 0 (x) < 0, f is strictly decreasing, and

if f 0 (x) D 0, f is constant on interval Œa; b.
10.8.2 Extremum Values
§: Function f W A ! R, A 2 R has a local minimum (maximum)

at point x0 , if in the neighborhood of x0 at every point x holds f (x)
f (x0 ) (f (x) f (x0 )). Local minimums and maximums of a function are
called its extremum values. ˘
If function f W A ! R, A 2 R has a local extremum at an inner point
x0 2 A, and f is differentiable at x0 , then
f 0 (x0 ) D 0:
The assumption of an inner point is essential because on a border point

x1 2 A, the function may have a local extremum also in situations f 0 (x1 ) <
0 and f 0 (x1 ) > 0; see Fig. 10.8. Suppose that function f in Fig. 10.8 is
defined on set x 0. Then in Fig. 10.8a f has a local maximum at point
x D 0 and a local minimum at point x0 . Similarly, in Fig. 10.8b f has a
local minimum at point x D 0 and a local maximum at point x0 .
If f 0 W A ! R, A 2 R is differentiable, the type of the extremum
point of function f W A ! R, A 2 R can be studied by the second order
derivative of f . Let f 0 (x0 ) D 0 so that x0 is an inner point in A. Then
• If f 00 (x0 ) > 0, f has a local minimum at point x0 (Fig. 10.9a).
Fig. 10.8 (a) Two local extremums. (b) Two local extremums
Fig. 10.9 (a) Local minimum, (b) maximum, and (c) turning point of f (x)
• If f 00 (x0 ) < 0, f has a local maximum at point x0 (Fig. 10.9b).

• If f 00 (x0 ) D 0, unique result is not obtained, and x0 may be a minimum,
a maximum, or not at all an extremum point (Fig. 10.9c).
Second derivative f 00 (x) measures the rate of change of f 0 (x). In the case
of local minimum, f 0 (x) increases when x increases in the neighborhood
of x0 , that is, f 00 (x) > 0; x 2 Œx1 < x0 < x2 , see Fig. 10.9a. Similarly,
in the case of local maximum, f 0 (x) decreases when x increases in the
neighborhood of x0 , that is, f 00 (x) < 0; x 2 Œx1 < x0 < x2 , see Fig. 10.9b.
Point x0 is a turning point of function f , if f 0 (x0 ) D 0 and f 00 (x) changes
its sign at point x0 . In Fig. 10.9c, f 00 (x) 0; x x0 and f 00 (x) > 0; x > x0 .
Example
Let the flow of production of a firm be q (unit=week), q > 0, and its product
price p (e/unit). The sales function of the firm is p D 4000 33q, where the
units of constants 4000 and 33 are e=unit and eweek=unit2 , respectively. The
revenues of the firm R (e=week) are then
R(q) D pq D (4000 33q)q D 4000q 33q2 :
Let the costs C (e=week) be
C(q) D 2q3 3q2 C 400q C 5000;
where the constants have units:

e week2 e week e e
2W ; 3W ; 400 W ; and 5000 W :
unit3 unit2 unit week
Next we calculate the flow of production that maximizes the weekly profit
of the firm. The weekly profit function of the firm, …, is the following:
…(q) D R(q) C(q) D 4000q 33q2 (2q3 3q2 C 400q C 5000)

D 2q3 30q2 C 3600q 5000:
As a polynomial function, the profit function is differentiable and it can

have a maximum only at a point where its derivative function vanishes:
…0 (q) D 6q2 60q C 3600 D 6(q2 C 10q 600) D 0 ,

p
10 ˙ 100 C 2400 10 ˙ 50
qD D , q D 20 or (q D 30 < 0):
2 2
Let us continue to study the type of the extremum at the zero point of the
derivative, q D 20 (unit=week). The second order derivative of function … is
…00 (q) D 12q 60:
Now …00 (20) D 12 20 60 D 300 < 0, and thus function … has a local
maximum at q D 20. Because this is the only possible (positive) extremum
point, the profit of the firm is maximized at the flow of production 20
(unit=week). The maximal value of weekly profit is:

e
…(20) D 2 203 30 202 C 3600 20 5000 D 39;000 :
week
Next we show that the derivative of the profit function is well-defined with
respect to measurement units. The unit of derivative d…=dq is the same as
that of …=q,
e
… e
unit D
W week :
q week
unit
The measurement unit of function …0 (q) D 6q2 60q C 3600 is:

2
e week2 unit e week unit
…0 (q) D 6 q 60 q
unit3 week unit2 week

e e e e
C3600 D 6q2 60q C 3600 :
unit unit unit unit
Thus both sides of the derivative equation have equal unit. ˘

10.9 Partial Derivatives

10.9.1 Partial Functions
In this section, we study two-variable real-valued functions. Let f W A !

R, A
R2 , and let (x0 ; y0 ) 2 A be a fixed point. Then the partial
functions of f at point (x0 ; y0 ) are
ˆ(x) D f (x; y0 ); x 2 R and ‰(y) D f (x0 ; y); y 2 R:
Partial functions ˆ(x) and ‰(y) describe the behavior of function f when
one of its arguments changes and the other stays fixed.
The geometrical interpretation of the graph of function ˆ(x) is the
cutting curve of the surface of function f and plane y D y0 ; see Fig. 10.10.
The geometrical interpretation of function ‰(y) is the cutting curve of the
surface of f and plane x D x0 . However, plane x D x0 is not presented in
Fig. 10.10 to keep the figure clear.
Fig. 10.10 The surface of z D f (x; y) and partial functions ˆ(x); ‰(y)
10.9.2 Partial Derivatives
§: The partial derivatives of function f W A ! R, A

R2 , at point
(x0 ; y0 ) 2 A, are:
@f (x0 ; y0 ) f (x; y0 ) f (x0 ; y0 ) f (x0 C h; y0 ) f (x0 ; y0 )

D lim D lim ;
@x x!x0 x x0 h!0 h
@f (x0 ; y0 ) f (x0 ; y) f (x0 ; y0 ) f (x0 ; y0 C h) f (x0 ; y0 )
D lim D lim ;
@y y!y0 y y0 h!0 h
when these limit values exist; x x0 D h; y y0 D h. ˘

Partial derivatives are also commonly denoted as:
@f @f
D fx D Dx f and D fy D Dy f :
@x @y
Note. In the above notation, the arguments of the partial derivative

functions are omitted. Thus in the following we denote partial deriva-
tives abbreviated like @f =@x, and not by using the complete notation
@f (x; y)=@x, for notational clarity. ˘
Geometrically, the partial derivatives of function f are the slopes of the

tangent lines of the surface of f directed along the x- and y-coordinate axes
at point (x0 ; y0 ); see Fig. 10.11. Partial derivative @f =@x expresses the rate
of change of function f in the direction of x-axis, and @f =@y expresses
the rate of change of function f in the y-direction. Practically, partial
derivative @f =@x (analogously @f =@y) is calculated by taking the derivative
of function f with respect to x (with respect to y) by keeping the other
variable constant.
§: Function f W A ! R, A
R2 is differentiable at point (x0 ; y0 ) 2
A, if it has both partial derivatives @f =@x and @f =@y at (x0 ; y0 ). Function
f W A ! R is differentiable in A
R2 if it is differentiable at every point
in A. ˘
Fig. 10.11 Graphical presentation of partial derivatives of z D f (x; y)
If the partial derivative functions of function f W A ! R, A

R2 ,
@f =@x, @f =@y, are differentiable at point (x; y) 2 A, we can define the
@ @f @2 f
second order partial derivatives for function f as: @x @x D @x 2,

@ @f @ f
2
@ @f @ f
2
@ @f @ f
2
@y @x
D @y@x , @x @y
D @x@y , and @y @y
D @y 2.
Note. If @2 f =@x@y and @2 f =@y@x are continuous at (x0 ; y0 ), then

@2 f =@x@y D @2 f =@y@x holds; see Apostol (1967, p. 278). ˘
10.9.3 Rules of Partial Differentiation
The rules of partial differentiation are analogous to those for one-variable

functions. Partial differentiation is taken with respect to one of the
arguments of a function by keeping the other variables fixed.
Partial Derivatives of a Product
Let f (x; y) D g(x; y) h(x; y). The partial derivatives of f are:
@f @g(x; y) @h(x; y)
D h(x; y) C g(x; y);
@x @x @x
@f @g(x; y) @h(x; y)
D h(x; y) C g(x; y):
@y @y @y
Partial Derivatives of a Quotient
Let f (x; y) D g(x; y)=h(x; y), g; h W A ! R, A

R2 , h(x; y) ¤ 0. The
partial derivatives of f are:
@g(x;y)
@f @x
h(x; y) @h(x;y)
@x
g(x; y)
D ;
@x Œh(x; y)2
@g(x;y) @h(x;y)
@f @y
h(x; y) @y
g(x; y)
D :
@y Œh(x; y)2
10.9.4 Chain Rule of Partial Differentiation
Let function z D f (x; y) describe the dependency of quantity z on quan-

tities x and y. Suppose then that x and y depend of a third quantity, for
example time t. Then function f depends on time t as z(t) D f (x(t); y(t)).
The instantaneous rate of change of quantity z(t) D f (x(t); y(t)) with
respect to time, or its time derivative, is
dz @f dx @f dy
D C :
dt @x dt @y dt
This rule is known as the chain rule of partial differentiation.

Example 1
Suppose a firm producing good k has the following profit function:
…k .pk (t); qk (t)/ D pk (t) qk (t) Ck .qk (t)/ ;
where qk (t) is the flow of production of the firm, pk (t) the price of good k,
and Ck .qk (t)/ the cost function of the firm. The time derivative of the profit
function is
@…k dpk @…k dqk

…0k (t) D C D qk (t) p0k (t) C Œpk (t) Ck0 (qk (t))q0k (t);
@pk dt @qk dt
where
@…k @…k
D qk (t) and D pk (t) Ck0 (qk (t)): ˘
@pk @qk
Example 2
Let
z D f .x(t); y(t)/ D x(t) y(t):
The time derivative of function z is then
dz @f dx @f dy dx dy
D C D y(t) C x(t) D x0 (t) y(t) C y0 (t) x(t):
dt @x dt @y dt dt dt
This result corresponds to the derivative of a product of functions. ˘
Example 3
Let z D f .x(t); y(t)/ D x(t)
y(t)
, y(t) ¤ 0. The partial derivatives of this function
are:
@f 1 @f x(t)
D and D :
@x y(t) @y (y(t))2
The time derivative of function z is then

dz @f dx @f dy 1 dx x(t) dy
D C D
dt @x dt @y dt y(t) dt (y(t))2 dt
x0 (t) y0 (t)x(t) x0 (t) y(t) y0 (t) x(t)
D 2
D :
y(t) (y(t)) (y(t))2
This result corresponds to the derivative of a quotient of functions. ˘
10.10 Approximation
The tangent line of function f W A ! R, A
R at point (x0 ; f (x0 ))
approximates the curve of the function near this point. We can thus use
the tangent line in approximating changes in the function value with small
changes in the argument close to x0 ; see Fig. 10.12.
Change x D x x0 causes the change in function value
f (x0 ) D f (x0 C x) f (x0 );
which is denoted in Fig. 10.12 by y. The derivative of f in x0 is
y
lim D f 0 (x0 );
x!0 x
and thus we can write
Fig. 10.12 The tangent of function y D f (x) in approximation

y
D f 0 (x0 ) C (x0 ; x); where (x0 ; x) ! 0 with x ! 0:
x
Because y D f (x0 C x) f (x0 ), we can write
f (x0 C x) f (x0 ) D f 0 (x0 )x C (x0 ; x)x: (10.5)
The equation of the tangent line in Fig. 10.12 at x0 is
y D f 0 (x0 )x;
and the tangent line approximates the change in y with change in x,

x, by line segment CA. Equation (10.5) shows that the error in this
linear approximation is (x0 ; x)x, which is denoted in Fig. 10.12 by
line segment AB.
According to Fig. 10.12, the change in the function value at x0 can be
calculated by adding CA and AB D (x0 ; x)x. Giving smaller values
for x the error in the approximation AB gets smaller, and letting x !
0 we get
dy D f 0 (x0 )dx; (10.6)
where limx!0 D dx, limy!0 D dy. Equation (10.6) is called the

differential of function f at point x0 , and it shows the effect of a marginal
change in x, dx, on quantity y. Equation (10.6) can also be derived from
the definition of derivative, dy=dx D f 0 (x), at x0 .
Now, if function f is differentiable at x0 we can write
f (x0 ) D f 0 (x0 )x C (x0 ; x)x;
where (x0 ; x) ! 0 with x ! 0. With a small change in x, x dx,

we can approximate the change in the value of function f at x0 as:
dy D df (x0 ) f 0 (x0 )dx:

10.10.1 Taylor Series in Approximation
§: Let f W R ! R be a function with derivatives of order n at point x0 .

Then there exists one and only one polynomial P of degree n, called
Taylor polynomial that satisfies the following n C 1 conditions:
P(x0 ) D f (x0 ); P0 (x0 ) D f 0 (x0 ); :::; Pn (x0 ) D f n (x0 ):
The Taylor polynomial is given by formula
f 0 (x0 ) f 00 (x0 ) f 000 (x0 )

P(x) D f (x0 ) C (x x0 ) C (x x0 )2 C (x x0 )3
1Š 2Š 3Š
f n (x0 )
C C (x x0 )n ;
nŠ
where 1Š D 1, 2Š D 2 1, 3Š D 3 2 1, and so on. ˘

Polynomial P(x) approximates function f in the neighborhood of point
x0 as closely as necessary by adding the required number of power terms in
the polynomial. However, the first or the second order Taylor polynomials
are commonly used in applications in real sciences.
Example
Let us have function y D f (x) D 2ln(x), where ln is the natural logarithmic
function. We approximate this function by the second order Taylor polyno-
mial P2 (x) in the neighborhood of point x D 1:
2 2
P2 (x) D 2 ln(1) C (x 1) (x 1)2
1 21
D 2 C 2x (x 1)2 D 3 C 4x x2 ;
where ln(1) D 0. From P2 (x) we get the first order polynomial as: P1 (x) D
2 C 2x. We can then use the first and the second order Taylor polynomials,
P1 (x), P2 (x), in approximating the function values near point x D 1. The
corresponding values of the function and the polynomials at points x D
1; 1:2; 1:5; 2 are the following:
f (1) D 0 P1 (1) D 0 P2 (1) D 0

f (1:2) D 0:36 P1 (1:2) D 0:4 P2 (1:2) D 0:36
f (1:5) D 0:81 P1 (1:5) D 1:0 P2 (1:5) D 0:75
f (2) D 1:39 P1 (2) D 2:0 P2 (2) D 1:0
The above table shows that both polynomials are more accurate the
closer to the fixed point the approximation is made, and the second order
polynomial is more accurate if the distance is increased.
10.10.2 Total Differential
Let us have function f W R ! R, y D f (x). In the previous section, we

showed that if x changes by x D x x0 , the corresponding change in
the function value at point x0 , f (x0 ), can be approximated as
f (x0 ) f 0 (x0 )x:
Suppose now that quantity z depends on quantities x and y as z D f (x; y),

f W R2 ! R. Let x then change by x and y by y. The corresponding
change in the value of function f at point (x0 ; y0 ) is
z D f (x0 C x; y0 C y) f (x0 ; y0 ):
Let function f have continuous first order partial derivative functions.

Then, denoting by x; y marginal changes in x; y, respectively, we can
write:
@f (x0 ; y0 ) @f (x0 ; y0 )
z D x C y C 1 (x; y)x C 2 (x; y)y;
@x @y
where the error terms 1 ; 2 ! 0 with x; y ! 0, respectively. With

small changes in x; y, we can then approximate the change in z as
@f @f
z x C y;
@x @y
and denoting limx!0 D dx, limy!0 D dy, and limz!0 D dz, we

can write
@f @f
dz D dx C dy:
@x @y
This is called the total differential of function f ; it approximates the

change in the function value in the neighborhood of a fixed point.
Example 1
Let f (x; y) D c, c 2 R constant, be the level curve of a two-variable function.
We like to define the slope of the tangent of this level curve dy=dx in
coordinate system (x; y). Suppose function f is differentiable. On a level
curve, the change in the function value is zero, that is, dc D 0, and so
@f @f
dx C dy D dc; dc D 0:
@x @y
From this we get
@f
@f @f dy
dy D dx; which can be solved as D @x
@f
: (10.7)
@y @x dx
@y
Notice that in solving Eq. (10.7), we can multiply and divide by real-valued
quantities dx; dy; @f =@x; @f =@y ¤ 0 like with real numbers, if they deviate from
zero. The expression we solved for dy=dx is the slope of the tangent of level
curve f (x; y) D c in coordinate system (x; y). ˘
Note. If we can solve y as a function of x from the equation of the level

curve f (x; y) D c of a two-variable function, we can derive the slope of
the tangent by taking the derivative of this function with respect to x. ˘
Example 2
We define the slope of the tangent of a level curve of a two-variable
function in two ways. Let y 3x2 C 4x D 5 be the equation of the level
curve. The slope of the tangent of the curve is then:
@f
dy 6x C 4
D @x
@f
D D 6x 4:
dx 1
@y
In this case, we can solve y from the equation of the level curve as: y D
3x2 4x C 5. The slope of the tangent of the curve can then be obtained as:
dy
D 6x 4: ˘
dx
10.10.3 Taylor Formula for Scalar Fields
Here we present a generalization for the Taylor polynomial approximating

function values near a fixed point. The Taylor polynomial can be extended
to nth power in approximating m-variable scalar-valued functions, but
here we restrict to the second order polynomial P2 (x; y) in a two-variable
case in approximating f W R2 ! R.
§: Let f be a scalar valued function f W A ! R, A
R2 with
continuous second-order partial derivatives in A. Then 8(x; y); (x0 ; y0 ) 2
A we have:
@f (x0 ; y0 ) @f (x0 ; y0 )
P2 (x; y) D f (x0 ; y0 ) C (x x0 ) C (y y0 )
@x @y
1 @2 f (x0 ; y0 ) 1 @2 f (x0 ; y0 )
C (x x0 )2
C (x x0 )(y y0 )
2Š @x2 2Š @x@y
1 @2 f (x0 ; y0 ) 1 @2 f (x0 ; y0 )
C (x x0 )(y y0 ) C (y y0 )2 : ˘
2Š @y@x 2Š @y2
Polynomial P2 (x; y) is a generalization for the one-variable Taylor polyno-

mial presented in Sect. 10.10.1, and it contains the first order polynomial
P1 (x; y) as a special case. Polynomial Pi , i D 1; 2 is the more accurate
the closer to point (x0 ; y0 ) the approximation is made, and adding
higher order terms in the polynomial makes the approximation more
accurate. However, we omit all examples here, and the reader can refer
to Sect. 10.10.1 in understanding the principle in this approximation.
10.11 Extremum Values

§: Function f W A ! R, A
R2 has a local maximum at point (x0 ; y0 ) 2
A, if a disc of radius r exists with center at (x0 ; y0 ), B((x0 ; y0 )I r), so that
for all (x; y) 2 B((x0 ; y0 )I r) \ A holds f (x; y) f (x0 ; y0 ). Similarly,
function f has a local minimum at point (x0 ; y0 ), if there exists a disc
of radius r > 0, B((x0 ; y0 )I r), so that f (x; y) f (x0 ; y0 ) for all (x; y) 2
B((x0 ; y0 )I r)\A. Local maximums and minimums of function f (x; y) are
called its local extremum values. ˘
Suppose function f W A ! R, A
R2 has a local extremum at point
(x0 ; y0 ) 2 A. Then point (x0 ; y0 ) is an extremum point when the point
is crossed over in every possible direction, especially in the directions of
x and y-axes. Partial functions ˆ(x) D f (x; y0 ) and ‰(y) D f (x0 ; y) thus
have local extremum values at points x0 and y0 .
If function f W A ! R, A
R2 has a local extremum at an inner point
(x0 ; y0 ) 2 A, and both its first order partial derivatives exist at (x0 ; y0 ),
then
@f (x0 ; y0 ) @f (x0 ; y0 )
ˆ0 (x0 ) D D 0 and ‰ 0 (y0 ) D D 0:
@x @y
Thus in the extremum point of a differentiable two-variable function,

both first order partial derivatives are zero, if the extremum point is not
located on the border of the domain. The point of a function, where
both its first order partial derivatives vanish, is called a critical point.
The tangent plane that touches the surface in a critical point must thus
be horizontal.
In Fig. 10.13 is the minimum point of a bowl-shaped surface and its
tangent plane in the optimum point. In Fig. 10.14a is the surface of
function f (x; y) D 1 x2 y2 in the neighborhood of its maximum
point (x; y) D (0; 0) together with the tangent plane f (x; y) D 1 at
(x; y) D (0; 0). In Fig. 10.14b is the surface f (x; y) D 2x x2 C 3y C y2
and its tangent plane f (x; y) D 1:25 around the critical point (x; y) D
(1; 1:5). The critical point (x D 1; y D 1:5) of function f (x; y) D
2x x2 C 3y C y2 can be solved from equations @f =@x D 2 2x D
0; @f =@y D 3 C 2y D 0. The saddle-shaped surface has a local maximum
Fig. 10.13 The minimum point (x0 ; y0 ) of function z D f (x; y)
Fig. 10.14 Functions (a) f (x; y) D 1 x2 y2 . (b) f (x; y) D 2x x2 C 3y C y2
in the direction of x and a local minimum in the direction of y at point

(x; y) D (1; 1:5), and thus the critical point is not an optimum point
for the function.
In a critical point, a function may have a local minimum or maximum,
or otherwise the critical point is not an extremum point. A critical point,
that is not an extremum point, is called a saddle point. The name saddle
comes from the shape of the surface of a function with a critical point that
is not an optimum; see Fig. 10.14b.
The type of the critical point of a two-variable function can be studied
by the sign of the second order differential of the function. This topic
is, however, omitted here; see e.g. Chiang (1984, pp. 315–319).
10.12 Constrained Extremum Values

In various practical optimization problems of two-variable functions,
some constraints exist for the two variables. These kinds of optimization
problems are called constrained extremum value problems.
Fig. 10.15 presents two examples of how a constraint restricts the
optimization of a function. In the figure on the left, the constraint is of
the form (0 x x0 ; 0 y y0 ), and in the figure on the right, a linear
equation between points (x0 ; 0) and (0; y0 ) restricts the points (x; y) on
the horizontal plane. These constraints restrict the area of the surface of
the optimized function from which the optimum can be found.
A constrained two-variable extremum value problem may sometimes be
expressed in the form of an unconstrained one-variable extremum value
problem. Let the optimized function be f (x; y) and the values of variables
x and y to be constrained by equation g(x; y) D 0. Suppose equation
g(x; y) D 0 can be solved with respect to one of the variables as x D h(y)
or y D k(x). Then the optimized function can be expressed as a one-
variable function:
f (x; k(x)) or f (h(y); y):
In this way, a constrained two-variable optimization problem can be

transformed to an unconstrained one-variable optimization problem.
Fig. 10.15 Constrained parts of surface z D f (x; y)

10.12.1 The Method of Lagrange
Let f (x; y) W A ! R, A
R2 be a differentiable function, and g(x; y) D
k, k constant, a condition that restricts variables x and y. The following
so called Lagrange’s function can then be defined:
F(x; y; z) D f (x; y) C z(k g(x; y));
where dimensional constant z is called Lagrange’s multiplier.

Now, constraint g(x; y) D k can be expressed as k g(x; y) D 0.
Thus when the constraint equation holds, the value of Lagrange’s function
equals with that of function f (x; y), that is, F(x; y; z) D f (x; y). The
possible constrained extremum points of F(x; y; z) are those (x; y; z) that
fulfill the following group of equations:
@F
D 0;
@x
@F
D 0;
@y
@F
D 0: (10.8)
@z
Note. Because @F=@z D k g(x; y), Eq. (10.8) guarantees that the
constraint holds in the possible extremum point. ˘
The analysis of the type of extremum point is omitted in this book

because that would require the definition of various mathematical con-
cepts. These matters are studied, for example in Chiang (1984, pp. 379–
387).
10.12.2 Uniqueness of Marginal Willingness-to-Pay
Let the weekly utility function and budget equation of a consumer be

u D u(q1 ; q2 ) and T D p1 q1 C p2 q2 , respectively, where T is budgeted
funds, qi , the flow of consumption, and pi the price of good i, i D 1; 2;
see Sect. 3.7. Substituting q2 from the utility function by the budget
equation, we get:

1
u D u q1 ; (T p1 q1 ) :
p2
The necessary condition for maximal utility is then:

@u @u
du @u p1 @u @q1 @q2
D0 , D 0 , p1 D @u
p2 , p2 D @u
p1 : (10.9)
dq1 @q1 p2 @q2 @q2 @q1
Next we derive the same result by using Lagrange’s function
F D u(q1 ; q2 ) C z1 (T p1 q1 p2 q2 ):
The necessary conditions for maximum are:
@F @u
D z1 p1 D 0;
@q1 @q1
@F @u
D z1 p2 D 0;
@q2 @q2
@F
D T p1 q1 p2 q2 D 0:
@z1
From these we can solve:

1 @u
p1 D ;
z1 @q1
1 @u
p2 D ;
z1 @q2
T D p1 q1 C p2 q2 : (10.10)
If we now solve z1 from the first of these equations,
1 @u
z1 D ;
p1 @q1
and substitute this in the middle one, we get:

@u
p1 @u @q2
p2 D @u
D @u
p1 :
@q1
@q2 @q1
This is the same result as in Eq. (10.9) and it holds in the optimum
together with the budget equation in Eq. (10.10).
Next we transform the utility function by function f (u), f 0 (u) > 0.
The new Lagrange’s function is F D f (u(q1 ; q2 )) C z2 (T p1 q1 p2 q2 ),
and in the optimum holds:
@F @u
D f 0 (u) z2 p1 D 0;
@q1 @q1
@F @u
D f 0 (u) z2 p2 D 0;
@q2 @q2
@F
D T p1 q1 p2 q2 D 0:
@z2
From these we can solve:

f 0 (u) @u
p1 D ;
z2 @q1
f 0 (u) @u
p2 D ;
z2 @q2
T D p1 q1 C p2 q2 : (10.11)
Then, solving z2 from the first equation, we get

@u
f 0 (u) @q
z2 D 1
;
p1
and substituting this in the middle one gives:

@u
@q2
p2 D @u
p1 :
@q1
This equation together with Eq. (10.11) corresponds to that in Eq. (10.9).
Because p1 , p2 are the same in both cases, in the optimum holds:
f 0 (u) @u 1 @u f 0 (u) @u 1 @u
D and D :
z2 @q1 z1 @q1 z2 @q2 z1 @q2
Thus in the optimum, the consumer’s marginal willingness-to-pay values

are independent of the chosen utility function. This occurs because
marginal utilities and the Lagrange’s multiplier change with the transfor-
mation f (u) so that in the optimum their ratio stays fixed.
10.12.3 The Interpretation of Lagrange’s Multiplier
Here we show that in a consumer’s optimum, the value of Lagrange’s

multiplier equals with the consumer’s marginal utility of income. Let
the weekly utility function and budget equation of a consumer be u D
u(q1 ; q2 ) and T D p1 q1 C p2 q2 , respectively. Suppose then that the
consumer is in his optimum. Let q1 and q2 be the optimal consumption
flows and z the value of Lagrange’s multiplier in the optimum. These
depend on budgeted funds T, and so we can write (notice that qi , z
depend on pi too, i D 1; 2, but here we omit these relationships by
assuming pi to stay fixed):
q1 D q1 (T); q2 D q2 (T); and z D z (T):
In the optimum, Lagrange’s function is then:
F(q1 ; q2 ; z ) D u(q1 ; q2 ) C z (T p1 q1 p2 q2 ):

The partial derivative of Lagrange’s function with respect to T in the

optimum point is:
dF @u dq @u dq dz dq dq

D 1 C 2 C(T p1 q1 p2 q2 ) Cz 1p1 1 p2 2
dT @q1 dT @q2 dT dT dT dT
@u dq @u dq dz

D z p1
1
C z p2
2
C(T p1 q1 p2 q2 ) Cz :
@q1 dT @q2 dT dT
Because in the optimum holds
@u
z p1 D 0;
@q1
@u
z p2 D 0;
@q2
T p1 q1 p2 q2 D 0;
the partial derivative of Lagrange’s function with respect to T is: dF

dT
D
z . Because the budget equation holds in the optimum, the value of
Lagrange’s function is the same as that of the utility function, that is,
F(q1 ; q2 ; z ) D u(q1 ; q2 ). We can then write:
dF du
z D D :
dT dT
Thus in the optimum, the value of Lagrange’s multiplier z equals with

the consumer’s marginal utility of income.
10.13 Implicit Differentiation

So far we have presented two-variable functions in the solved or explicit
form as
z D f (x; y):
This relationship between the three variables can also be expressed as
F(x; y; z) D 0;
which is called the general or implicit form of the equation. Under

certain conditions, this equation defines the function z D f (x; y).
Suppose now that the relationship between quantities x, y, and z
described by function F(x; y; z) D 0 contains function z D f (x; y). Next
we want to define the first order partial derivatives @f =@x and @f =@y of
this implicit function. This can be done by taking the partial derivatives of
equation F(x; y; z) D 0 with respect to x and y, and solving @f =@x, @f =@y
from the obtained equation as a function of x, y, and z. This method is
called implicit differentiation.
Example 1
Let 2z D 4x C 6xy. We can solve this equation with respect to z to get the
explicit form of the function as
z D 2x C 3xy: (10.12)
We can then calculate the partial derivatives as:
@z @z
D 2 C 3y; D 3x:
@x @y
However, in implicit differentiation we assume z D f (x; y), and we apply this

assumption in differentiating 2z D 4x C 6xy with respect to x:
@z @z
2 D 4 C 6y ) D 2 C 3y:
@x @x
Similarly, by differentiating both sides of equation 2z D 4xC6xy with respect

to y, we get:
@z @z
2 D 6x ) D 3x:
@y @x
Thus implicit differentiation gives the same results as direct derivation. ˘

Example 2
Let us study the labor supply of a person. We denote hourly wage by w (e/h),
income tax rate by , and annual working hours by L (h=y). The annual after-
tax wage income T (e/y) of the person is then
T D (1 )wL: (10.13)
Let us denote the annual leisure time by H (h=y). We can then write
H D 1920 L; (10.14)
where 1920 (h=y) is the maximum annual work time; see Sect. 6.3. Let
u D u(H; T) with unit ut=y be the utility function of the labor supplier. Then
equation
@u @u
(1 )w D
@T @H
describes the equilibrium state of the labor supplier. Assuming to be

constant, the above equation can be presented in the general form w D
f (T; H). Because annual labor income T and annual leisure time H depend
on the annual number of working hours L, we can consider w as a function
of only one variable L as w D f (T; H) D f (T(L); H(L)). Then, according to the
chain rule of partial differentiation, we can write:
dw @w dT @w dH
D C : (10.15)
dL @T dL @H dL
Next we define this derivative by solving the partial derivatives, @w=@T and
@w=@H, in it by implicit differentiation. The partial derivative of equation
@u @u
(1 )w D (10.16)
@T @H
with respect to T is
@w @u @2 u @2 u
(1 ) Cw 2 D :
@T @T @T @T@H
From this we can solve @w=@T as:
@2 u @2 u
@w
1
1 @T@H
w @T 2
D @u
:
@T @T
Next we repeat this with respect to H. Differentiating Eq. (10.16) with

respect to H, we get:
@w @u @2 u @2 u
(1 ) Cw D :
@H @T @H@T @H 2
From this we can solve @w=@H as:
1 @2 u @2 u
@w 1 @H 2
w @H@T
D @u
:
@H @T
Then, substituting these results in Eq. (10.15), we get:
dw @w dT @w dH
D C
dL @T dL @H dL
@2 u @2 u
1
1 @T@H
w @T 2 dw
D @u
(1 ) LCw
@T
dL
@2 u @2 u
1
@H 2
w @H@T
C 1 @u
(1); (10.17)
@T
where dT=dL D (1 )((dw=dL)L C w) and dH=dL D 1 are obtained from

Eqs. (10.13) and (10.14), respectively, by assuming w(L). Next we take dw=dL
as the common factor in Eq. (10.17):
" !# !
@2 u @2 u @2 u @2 u
@T@H
(1 )w @T 2 dw @T@H
(1 )w @T 2
1L @u
Dw @u
@T
dL @T
!
1 @2 u @2 u
1 @H 2
w @H@T
@u
:
@T
Dividing both sides of this equation by the coefficient of dw=dL, we get:

2

@ u @2 u 1 @2 u @ u2
w @T@H (1)w2 1 w @H@T
@T 2 @H 2
@u
dw @T
D @u 2
@ u 2u :
dL @T L @T@H C(1)wL @
@T 2
@u
@T
This can be simplified as follows:

@2 u @ u 2 1 @2 u @2 u
dw w @T@H
(1 )w2 @T 2 1 @H 2
w @H@T
D @u @2 u @2 u
:
dL @T
L @T@H
C (1 )w L @T 2
If we now combine the second order partial derivatives by setting

@2 u=(@T@H) D @2 u=(@H@T) and expand the result by (1 ), we get (Sect. 6.3):
@ u2 @ u2
2 2@ u2
dw 2(1 )w @T@H @H 2 (1 ) w @T 2
D @u @ u
2 @2 u
:˘
dL (1 ) @T (1 )L @T@H C (1 )2 w L @T 2
10.14 Integral Calculus

10.14.1 Integral Function
Integration is the inverse operation for derivation. The starting point is

that we know the derivative (the velocity) of a function, and we want
to solve the original function. For example, an economist knows the
instantaneous flow of production of a firm, and he wants to forecast the
accumulated production of the firm for a certain time unit.
The operation, where we know the derivative of a function and derive
the original function, is called integration.
§: Function F is the integral function of function f W A ! R, if
F 0 (x) D f (x)
holds for all x 2 A. We denote:

Z
F(x) D f (x)dx:
The symbol x after the d-letter is the variable with respect to which the
integration is made, and function f is called the integrand. ˘
Note. If a function has one integral function, it has several other integral
functions. For example, one of the integral functions of function f (x) D
3x2 is F(x) D x3 , because F 0 (x) D 3x2 . However, all functions G(x) D
x3 CC, where C 2 R is a constant, are also integral functions of f , because
G0 (x) D 3x2 D f (x). Constant C is called the constant of integration.
If F is an integral function of function f , then all integral functions of f

are of form F(x) C C, where C 2 R is a constant. ˘
10.14.2 Definite Integral
Suppose f is a continuous function on interval Œa; b and let interval Œa; b

be divided in n subintervals with the division points a D x1 , x2 , x3 , : : : ,
xnC1 D b. Let the lengths of the subintervals be x1 , x2 , : : : , xn . We
can now define the area restricted by the graph of the function and the
horizontal axis between points Œa; b as follows:
X
n
In D f (x1 )x1 C f (x2 )x2 C C f (xn )xn D f (xj )xj I
jD1
see Fig. 10.16. Now, if n ! 1, the number of the division points increases
without limit, and xj ! 0 8 j D 1; 2; : : : ; n. Then In approaches a
limit value which we call the definite integral of function f from a to b;
see Fig. 10.16.
§: The definite integral of function f from a to b is the limit value
Z b X
n
f (x)dx D lim f (xj )xj
a n!1
jD1
Rb
Fig. 10.16 The principle of calculation of definite integral a f (x)dx
when this limit value exists. Numbers a and b are called the lower and
the upper limit of the integration, respectively. ˘
Rb
The geometric interpretation of the absolute value of a f (x) dx is the
area between the graph of y D f (x) and x-axis on interval Œa; b. We can
understand this as follows. If function f is continuous and positive on
interval Œa; b, f (xj )xj measures the area of the rectangle having xj as
the basis and f (xj ) as the height. Then the sum
X
n
In D f (x1 )x1 C f (x2 )x2 C C f (xn )xn D f (xj )xj
jD1
is an approximation for the area between the graph of function f and the
x-axis on the interval Œa; b; see Fig. 10.16. If n ! 1, then xj ! 0 8 j D
1; 2; : : : ; n, and the approximation becomes more accurate. The limit
value of In —given in the definition of the definite integral—measures
the area between y D f (x) and the x-axis on interval Œa; b.
In the graphical demonstration in Fig. 10.16, we assumed that f (x) >
Rb
0 8 x 2 Œa; b. However, if f (x) < 0 8 x 2 Œa; b, then a f (x)dx < 0,
and this definite integral has an equal absolute value as the area limited
by function f (x) > 0, the x-axis, and the lines x D a, x D b.
The first fundamental theorem of calculus: Let function f be contin-
uous on interval Œa; b and x 2 a; bŒ. Then function
Z x
F(x) D f (t)dt
a
is differentiable, and F 0 (x) D f (x). ˘

The second fundamental theorem of calculus: Let function f be
continuous on interval Œa; b and let F be one of its integral functions.
Then 8 x 2 Œa; b:
Z x
F(x) D F(a) C f (t)dt: ˘
a
Corollary. Let function f be continuous on interval Œa; b. Then, according

to the second fundamental theorem of calculus, we can write
Z b
f (x)dx D F(b) F(a);
a
where F is an integral function of f . ˘
Additivity with respect to the interval of integration: Let f W R ! R

be continuous on interval Œa; c so that a < b < c. Then we can write
Z c Z b Z c
f (x)dx D f (x)dx C f (x)dx:
a a b
This theorem reflects the additive property of areas illustrated in

Fig. 10.16. If interval Œa; c is decomposed in separate intervals Œa; b
and Œb; c, the sum of the two areas equals with the whole area. ˘
10.15 Differential Equations

In modeling real-world phenomena, we sometimes need to describe the
relationships between quantities where the equations contain an unknown
function and its derivatives. These equations are called differential equa-
tions.
Example 1
Suppose the velocity of the price of a product p0 (t) is linearly related to the
difference between its demand D(p(t)) and supply S(p(t)) that both depend
on price p(t). This relationship can be expressed as:
p0 (t) D a(D(p(t)) S(p(t))); (10.18)
where the magnitude of dimensional constant a > 0 defines the strength

of the relation between the velocity of price p0 (t) and excess demand D S.
If demand is greater than supply, D S > 0, then, according to Eq. (10.18),
price increases with time, p0 (t) > 0, and vice versa. ˘
The general form of a differential equation is
F(x; f (x); f 0 (x); f 00 (x); : : : ; f (n) (x)) D 0 or

df d2 f dn f
F x; f ; ; 2 ; : : : ; n D 0; (10.19)
dx dx dx
where f (x) is an unknown function that fulfills the differential equation in

(10.19). A differential equation defines the conditions function y D f (x)
must fulfill for Eq. (10.19) to be true. Differential equations are expressed
by using the argument x, function f (x), and its derivatives: f 0 (x); f 00 (x),
and so on.
The order of a differential equation is the order of the highest
derivative in the equation. For example,
y000 (x) 5y0 (x) D 7x C 3
is a third order differential equation. In an ordinary differential equation,

only one independent variable exists, for example, x or time t. If several
independent variables exist in a differential equation together with partial
derivatives of an unknown function, the equation is called a partial
differential equation.
The solution of a differential equation is function y D f (x) that
fulfills the differential equation.
Example 2
We show here that function
2apf
T t
qf (t) D C C 0 e pv m f (10.20)
2pf
is the general solution of differential equation
a
mf q0f (t) D (T 2pf qf (t))I (10.21)
pv
see the Newtonian theory of consumer behavior in Sect. 3.8.

We take the time derivative of the solution in Eq. (10.20) as:

2apf
2apf t
q0f (t) D C 0 e pv m f ; (10.22)
pv mf
and multiply both sides of Eq. (10.22) by mf . This gives:
2apf
2apf t
mf q0f (t) D C 0 e pv m f : (10.23)
pv
Next we substitute function qf (t) in Eq. (10.20) on the right-hand side of

Eq. (10.21). This gives
a a T

2apf
t
(T 2pf qf (t)) D T 2pf C C 0 e pv m f
pv pv 2pf
a 2pf T
2apf
t
D T 2pf C0 e pv mf
pv 2pf
2apf
2apf t
D C 0 e pv m f : (10.24)
pv
From Eqs. (10.23) and (10.24) we see that both sides of the differential
equation in Eq. (10.21) are equal, and so function qf (t) in Eq. (10.20) fulfills
the differential equation and is thus its solution. Dimensional constant C0
in the solution of qf (t) is the constant of integration with unit kg=week.
Function qf (t) in Eq. (10.20) is called the general solution of the differential
equation in Eq. (10.21), and all solutions of Eq. (10.21) with different values
of C0 are its special solutions. ˘
Note. In Example 2 we did not explain how we found the solution

of the differential equation. There exists specific solution methods for
different kind of differential equations, and the reader can study these
from textbooks on mathematics, for example, Apostol (1969, pp. 142–
188). Example 2 gives the reader a method to check that the solutions of
differential equations given in this book actually are the functions that
fulfill the differential equations. ˘
10.16 Scales of Measurement

In statistical analysis, the studied phenomena are expressed in a numerical
form.1 In the following we study research objects ai , i D 1; 2; : : : , that
in statistical analysis are called statistical units. These objects belong in
set E. Objects ai have characteristics x, y, z, : : :, that are called statistical
variables. Set E could be, for example, the inhabitants of a city, and the
characteristics of the objects could be, for instance, age, gender, annual
salary, and so on. To be able to present the studied characteristics in
numerical form, we have to measure the characteristics of the research
objects.
§: By the measurement of characteristic x we understand a rule that
attaches to research object ai a number or a symbol of measurement that
measures characteristic x in the object. ˘
Mathematically, measurement can be interpreted as a scalar-valued
function f defined in the domain E of the research objects. Every
characteristic has an own measurement or indicator function. In the
following we denote the indicator function of characteristic x by fx , and
the number of measurement of characteristic x from research object ai by
fx (ai ).
Note. Notation fx should not be confused with the partial derivative of

function f with respect to quantity x. In the following, we denote by fx
the indicator function of characteristic x. ˘
Usually indicator function fx can be defined in numerous equivalent

ways. However, there are certain rules that function fx must fulfill. Func-
tion fx must be chosen so that for the empirical relations characterizing
the statistical units, exactly equal mathematical relations characterize their
numbers of measurement.
In the following, we introduce different scales of measurement and
study the possible ways of choosing indicator function fx in every case.
The principles of solving this problem are:
1
This section is based on Stevens (1946).
• Characteristic x defines the properties indicator function fx must fulfill.

• Suppose indicator functions f1x and f2x fulfill the required properties.
Then we can study how the values of functions f1x and f2x depend
on each other. The idea is to study whether we can find function f W
R ! R that gives the number of measurement f2x when result f1x
is known. Mathematically, we are searching for function f for which
holds f2x (ai ) D f (f1x (ai )) 8 ai 2 E. Function f2x can then be expressed
as the following composite function: f2x D f ı f1x .
Measurements made at different scales are classified according to the

transfer function f defined above.
10.16.1 Nominal or Classification Scale
When statistical units are classified in different classes according to a

characteristic, we say that this characteristic is measured on a classification
(nominal) scale. The classification scale is used when two statistical units
ai and aj are compared so that we can conclude whether they are identical
(equivalent) with respect to the studied characteristic or not. If statistical
units ai and aj are equivalent with respect to characteristic x, we denote
this as:
ai ,x aj :
Relation ,x is called equivalence relation with respect to characteristic

x, and it has the following properties:
• Reflexivity: ai ,x ai 8 ai 2 E.
• Symmetry: If ai ,x aj then aj ,x ai .
• Transitivity: If ai ,x aj and aj ,x ak , then ai ,x ak .
§: If we can decide 8 ai ; aj 2 E whether relation
ai ,x aj
holds or not, characteristic x is measured on a classification scale. ˘

Usually, relation ,x defines a limited number of equivalence classes

among the statistical units ai 2 E. Every equivalence class consists of
identical statistical units with respect to characteristic x. For example, if E
is the set of all inhabitants in a city, then relation ‘gender’ is an equivalence
relation that contains two equivalence classes: male and female.
Let characteristic x be measured on a classification scale. Then it is
natural to require of the indicator function fx the following condition:
fx (ai ) D fx (aj ) if and only if ai ,x aj :
An infinite number of indicator functions fx W E ! R exist that fulfill

this condition. If f1x is an indicator function on a classification scale, then
function
f2x D f (f1x )
is also a proper indicator function on a classification scale, if transfer

function f is a bijection (see Function).
Example 1
For the set E of all inhabitants in a city, we can construct the following
indicator function f1x :
8
ˆ
ˆ 0; if ai is unmarried
ˆ
ˆ
<1; if ai is married
f1x (ai ) D
ˆ2;
ˆ if ai is divorced
ˆ
:̂
3; if ai is widow:
Let function f be: f W f0; 1; 2; 3g ! f27; 8; 1; 0g, f (x) D x3 . Then the
composite function f2x D f (f1x ) is an indicator function on the classification
scale (f is a bijection), and its values are:
8
ˆ
ˆ 0; if ai is unmarried
ˆ
ˆ
<1; if ai is married
f2x (ai ) D
ˆ8;
ˆ if ai is divorced
ˆ
:̂
27; if ai is widow: ˘
The numbers of measurement in the classification scale do not tell anything

else about the statistical units except that in which equivalence class they
belong. Thus arithmetic operations with these numbers of measurement
do not have a meaningful interpretation. In the place of the numbers of
measurement, we could just as well set any non-equal symbols like A, B, C,
:::
10.16.2 Ordinal Scale
Characteristic x is measured in an order scale, if the members of different

equivalence classes are not only different with each other with respect to
characteristic x, but there also exists an order relation x between the
equivalence classes. Order relation x has the following properties:
• ai x ai 8 ai 2 E,
• if ai x aj and aj x ak , then ai x ak ,
• if ai x aj and aj x ai , then ai ,x aj .
If 8 ai ; aj 2 E it holds ai x aj or aj x ai , we say that the order relation

is complete.
§: Characteristic x is measured on ordinal scale, if 8 ai 2 E there
exists an equivalence relation and a complete order relation with respect
to characteristic x so that
ai x aj and aj x ai if and only if ai ,x aj : ˘
For indicator function fx on ordinal scale, the order of the numbers of

measurement must be the same as that between statistical units, that is,
fx (ai ) fx (aj ) if and only if ai x aj : ˘
Note. Order relation x is different that ‘smaller or equal than’ relation

. For example, an order relation does not exist for vectors; that is, we
cannot say, for example which one of vectors x Dp(2; 4) or y D (1; 6) is
‘bigger’. We can though calculate their lengths as 22 C 42 D 4:47 and
p
22 C 42 D 6:08 and decide, which one of them is ‘longer’. However,
if these vectors represent bundles of consumption flows of a consumer,

a consumer may be able to define his/her preference order for these two
consumption bundles as x y, if the consumer prefers good 2 more than
good 1; see Sect. 3.3.
Example 2
An example of an indicator function fx on ordinal scale in the set E of
military staff in the USA is:
8
ˆ
ˆ 1; if ai is Private
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ 2; if ai is Corporal
<
fx (ai ) D 3; if ai is Sergeant
ˆ
ˆ
ˆ
ˆ 4; if ai is Staff Sergeant
ˆ
ˆ
ˆ: ::
:̂:
: : ˘
If f1x W E ! R is an indicator function on ordinal scale, then function
f2x D f (f1x )
is also an indicator function on ordinal scale, if the transfer function f is

strictly increasing.
Note. Every utility function of a consumer that represents the same

preference order must fulfill the above requirement, because consumer
preferences are measured on ordinal scale; see Sects. 3.3–3.5. ˘
10.16.3 Interval Scale
Let fx be an indicator function of characteristic x, and let us denote the

number of measurement of indicator function fx for statistical unit ai as
fx (ai ) D xi :
An interval scale is the first scale of measurement where adding and sub-
tracting the numbers of measurement can be interpreted in a meaningful
way. The difference in numbers of measurement, xj xi (xi xj ),
measures the change in characteristic x between statistical units ai and aj .
We denote the increment in characteristic x from statistical unit ai to unit
aj (ai x aj ) by (ai ; aj ), whereby x is denoted relation ‘smaller or equal
than’ with respect to characteristic x. If we measure characteristic x on the
interval scale, increments (ai ; aj ) and (am ; an ) can be compared, that
is, we can determine which of the following alternatives holds:
• (ai ; aj ) x (am ; an ) or (am ; an ) x (ai ; aj ).
If both these alternatives hold, we denote this as:
(ai ; aj ) ,x (am ; an ):
Using an interval scale, we can order the increments in characteristic x in

a unique way. For example, the measurement of temperature on the basis
of the expansion of volume of material due to increasing temperature is
made on the interval scale. An analogous change in the volume of the
material takes place after a fixed change in temperature.
If characteristic x is measured on the interval scale, indicator function
fx must also fulfill, together with the requirements for ordinal scale,
xj xi D xn xm if and only if (ai ; aj ) ,x (am ; an ):
Let f1x be an indicator function on the interval scale. Then function
f2x D f (f1x )
is also an indicator function on the interval scale, if function f has the

form
f (x) D ax C b;
where a and b are dimensional constants; a > 0 and b may take any value.
The connection between indicator functions f1x and f2x on the interval
scales 1 and 2 is then:
f2x D af1x C b:
Let us denote by f1x a change in measurement by using scale 1. The

corresponding change in measurement by using scale 2 is then:
f2x D af1x :
Now, if f1x D 1 then f2x D a, that is, one unit change in scale 1
corresponds to a units change in scale 2. If f1x D 0, then f2x D b, that is,
the zero point in scale 1 corresponds to number b in scale 2.
For example, when measuring temperature, we use two indicator
functions with different scales: Celsius and Fahrenheit scales. If we denote
the values of the indicator function on the Celsius scale by f1x and at
Fahrenheit scale by f2x , the following relation holds between the measured
values:
f2x D 1; 8f1x C 32:
10.16.4 Ratio Scale
If we use a ratio scale in measuring a characteristic, the earlier described

degrees of freedom in using an interval scale disappear. In a ratio scale,
the measured characteristic has a unique zero point. If the measured
characteristic fulfills the requirements for an interval scale, and we can
also talk about a unique zero point with respect to the characteristic, the
characteristic is measurable on a ratio scale.
Suppose the requirements for an interval scale are fulfilled and there also
exists an equivalence class A0 for statistical units ai for which characteristic
x is at its minimum possible level. Then for indicator function fx on a ratio
scale, we require that the statistical units in equivalence class A0 have the
number of measurement 0:
fx (ai ) D 0 if and only if ai 2 A0 :

Proposition. Let f1x be an indicator function on a ratio scale. Then
f2x D f (f1x )
is also an indicator function of characteristic x on a ratio scale, if the transfer

function f is of the form
f (x) D ax;
where a is a positive constant.
Proof. If f2x is an indicator function on the ratio scale, it fulfills the

requirements for an indicator function on the interval scale. Thus it holds
for f2x :
f2x D af1x C b; a > 0; b 2 R:
Besides this, it is required that f2x (ai ) D 0 8 ai 2 A0 . Because f2x D

af1x C b and f1x (ai ) D 0 8 ai 2 A0 , this requirement is met if b D 0. ˘
The term ‘ratio scale’ reflects that at this scale of measurement, the ratio
of numbers of measurement xi and xj of statistical units ai and aj , xi =xj , is
independent of the indicator function. Most of the traditional measures
for quantities, like length, weight, income, time interval, and so on, are
measurable on a ratio scale.
10.16.5 Absolute Scale
If characteristic x can be measured only by a unique indicator function,

we say that this measurement is made on an absolute scale. Two indicator
functions f1x and f2x on a ratio scale are connected by the relation f2x D
af1x , a > 0. If the characteristic is measured on an absolute scale, then
a D 1 must hold in the above equation, that is, the indicator function is
unique. For example, shares like the share of tax of gross wage, or the share
of unemployed people of a labor force, are measured on an absolute scale.

The number of objects in a box is also a unique number of measurement
on an absolute scale.
References
Apostol, T. M. (1967). Calculus (2nd ed., Vol. I). New York: Wiley International
Edition.
Apostol, T. M. (1969). Calculus (2nd ed., Vol. II). New York: Wiley International
Edition.
Chiang, A. C. (1984). Fundamental methods of mathematical economics (3rd ed.).
New York: McGraw-Hill International Editions.
Stevens, S. S. (1946). On the theory of scales of measurement. Science, 103(2684),
677–680.
Author Index
Note: Page number followed by ‘n’ refers to footnotes.
A D
Allen, Roy George Douglas, 36, 47 Dannenberg, Alia, 14, 219n1
Apostol, Tom, 122, 443 Darwin, Charles, 6
Ausloos, Marcel, 22 Debreu, Gérard, 100, 107, 110
de Jong, Fritz, 41, 47, 48
Drãgulescu, Adrian, 23
B Dunlop, John, T., 281
Bachelier, Louis, 20
Becker, Gary S., 7, 25
Bentham, Jeremy, 7, 14, 25 E
Bridgman, Percy Williams, 39–40 Einstein, Albert, 20
Brown, Robert, 20 Estola, Matti, 14, 19, 24, 83, 88,
185, 222
C
Canard, Nicolas-Francois, 19 F
Chen, Shu-Heng, 20 Fama, Eugene, 337, 352
Chiang, Alpha Chung-i, 428, 430 Fisher, Irving, 21, 62
Comte, Auguste, 7 Friedman, Milton, 18, 325

DOI 10.1007/978-3-319-46879-2
454 Author Index
G Mariani, Maria C., 23

Galam, Serge, 23 Markowitz, Harry, 22
Gibbons, Robert, 232 Marshall, Alfred, 16
Giffen, Robert, 197 Marx, Karl, 7, 16, 17, 313
Mas-Colell, Andrew, 21
McCabe, George P., 14
H McCauley, Joseph, 23
Hawking, Stephen, 6 McDonald, Ian M., 282
Hokkanen, Veli-Matti, 222 Mirowski, Philip, 20, 23
Hull, John C., 299, 336 Mishkin, Frederic, 336
Moore, David, 14
I
Ising, Ernst, 22 N
Nash, John, 232
Newton, Isaac, 4, 68
J
Nicolescu, Basarab, 21
Jevons, William Stanley, 9, 16,
20, 326
Jovanovic, Franck, 22
O
Ohanian, Hans C., 68, 80, 218
K
Keynes, John Maynard, 18
Kondor, Imre, 19 P
Kusmartsev, Feodor, 23 Pareto, Wilfred, 329
L Q
Leibniz, Gottfried Wilhelm, 68 Quatelet, Adolphe, 20
Lenin, Vladimir Iljits, 17
Li, Sai-Ping, 20
Liu, Y., 23 S
Lucas, Robert, 24 Samuelson, Paul, 213
Lux, Thomas, 23 Schinckus, Cristophe, 22
Smith, Adam, 9, 19
Solow, Robert, 282
M Sornette, Didier, 22
Mandelbrot, Benoit, 20 Stanley, Eugene, 19, 22
Mantegna, Rosario, 23 Stevens, Stanley Smith, 444
Author Index 455
T W
Takayasu, Hideki, 22 Walras, Leon, 9, 16, 20
V
Vihriälä, Vesa, 151 Y
von Wright, Georg Henrik, 12 Yakovenko, Viktor M., 23
Subject Index
A approximation, 421–6
absolute change, 59 average acceleration of production,
absolute scale, 451–2 80–1
acceleration of borrowing, 376 average fixed unit costs, 160, 163
acceleration of consumption, 135 average marginal costs of firms, 217
acceleration of labor supply, 266 average marginal willingness-to-pay,
acceleration of production, 28, 204 217
acceleration of savings, 367, 369 average productivity, 92
acceleration of use of labor, 255 average revenues from a product, 168
acceleration of utility, 121, 263, average unit costs, 160, 163, 164
366, 372 average utility, 112
accumulated amount of production, average variable unit costs, 160, 163,
69, 161 164
accumulation function of production, average velocity, 60
76, 82 average velocity of production, 71,
agent-base modeling, 20 75, 161
aggregate demand relation, 145–7, axioms of consumer behaviour,
166 99–100
alternative cost, 44, 103 axioms of economics, 5–14

DOI 10.1007/978-3-319-46879-2
458 Subject Index
B consumption flow, 101

balance sheet, 159 consumption space, 104, 107
bank card, 323 continuous time interest rate, 296–8
bankers’ acceptance, 333 contour plot, 110
bank money, 327 corporate bond, 157, 334
bank reserves, 328 corporation, 156
bank run, 377 cost-based price effect, 189
Bertrand’s conjecture, 233 cost function, 162, 175, 178
bijection, 405 cost inflation, 241
bond, 329, 331 coupon payment, 157, 334
Bridgman’s axion, 40 Cournot’s conjecture, 233
broker, 336 credit card, 323
Brownian motion, 20
budget equation, 100–3, 360, 370
D
dealer, 335, 336
C debt finance, 157
capital gain, 333 decision-making steelyard, 26–9
capitalism, 16 decreasing returns to scale, 162, 247
capital market, 333 default, 331
chain rule of differentiation, 410 definite integral, 439–41
chain rule of partial differentiation, degree of monopolization, 226
419–21 demand, 128, 133
check, 323 demand-based price effect, 189
collateral, 329 demand function (relation), 144, 150
commercial paper, 333 demand inflation, 241
competitive market system, 8 demand of assets, 342–5
complement good, 48 demand of labor, 267
composite function, 403–5 demand relation of labor, 255
compound interest calculation, 294, demand relation of loans of a
296–8 household, 375
conformal interest rate, 299, 301 demarcation curve, 239
consol, 337–8 depreciation, 158
constrained extremum value, 429–34 derivative, 407–12
consumer, 97 derived demand of production
consumer preferences, 103–8 factors, 251
consumer price index (CPI), 91 descriptive analysis, 4
consumption, 2 differential equation, 441–3
Subject Index 459
dimension, 395 equation of motion for wage,

dimensional analysis, 36–41, 395 272, 277
dimensional constant, 40 equilibrium state in Cournot’s
dimensionally homogeneous, 41, 51 duopoly, 235
dimensionless quantity, 39 equilibrium state in labor market,
dimension of monetary values, 46, 273, 276
295 equilibrium state in loan market, 382
dimension of satisfaction, 48 equilibrium state in the market of
dimension of the volume of shares of a common stock, 345
goods, 42 equilibrium state of a borrowing
dimension of time differences, 42 household, 372
direct finance, 329 equilibrium state of a consumer,
discount factor, 295, 298 119–21
discount rate, 360 equilibrium state of a labor
discrete quantity, 57–9 supplier, 261
dividend, 156, 334 equilibrium state of an industry, 222
durable good, 2 equilibrium state of a saving
dynamic consumer behavior, 121–34 household, 363, 367
equity, 333
equity finance, 157
E equivalence class, 446
economic force, 24, 25, 29 equivalence relation, 445
economic kinematics, 68–88 euribor, 333, 384
economic unit, 2 evolutionary economy, 25
economies of scale, 331 exchange money, 326
economy, 1 exchange rate, 56, 57
econophysics, 19–24, 29 expected value, 89
efficient consumption, 114 exponential function, 410–11
efficient market hypothesis, 337 export price index, 91
emission, 335 extremum values, 413–15
empirical research, 4
EONIA rate, 384
equation of motion for asset F
price, 382 face value, 334
equation of motion for borrowing, federal funds rate, 333
379, 383 financial intermediary, 330
equation of motion for price, 242 financial market, 328
equation of motion for saving, first fundamental theorem of
379, 383 calculus, 440
460 Subject Index
fiscal period, 157 G

Fisher equation, 62 game theory, 231
flow of consumption, 172 geometric series, 302
flow of production, 163, 172, 177 Giffen good, 149, 197
force acting upon asset demand, gold standard, 323
348 good, 1
force acting upon asset price, 348 goods money, 321
force acting upon asset supply, government bond, 334
348 gross domestic product (GDP), 88
force acting upon borrowing, growth rate, 60
375, 376
force acting upon capital stock, 292,
312, 315, 318 H
force acting upon consumption, high powered money, 328
125, 135
force acting upon labor input,
256, 263 I
force acting upon price, 189, 242, idealized particle, 68
245 imperfect competition, 200
force acting upon production, 180, implicit differentiation, 434–8
183, 204, 228, 237 Income Statement, 159
force acting upon saving, 367, increasing returns to scale, 162, 200,
369, 378 247
force acting upon work time, 255, index number, 88–91
269, 270, 284 indifference curve, 106, 115–18, 282,
force field, 239 363, 403
foreign capital, 334 indirect finance, 330–2
foundation for modeling economic industry, 196
phenomena, 26 inertial factor, 25
fractional reserve banking principle, inertial mass of asset demand, 347
324, 377 inertial mass of asset supply, 347
free body diagram, 140, 182, 192, inertial mass of borrowing, 382
220, 246, 277, 317, 348, 376, inertial mass of capital stock, 317, 387
383 inertial mass of consumption, 135,
free money system, 324 219
full-capacity production, 163 inertial mass of labor demand, 276
function, 401–6 inertial mass of labor supply, 266,
functions of money, 326–8 277
Subject Index 461
inertial mass of price, 190, 245 Lagrangian coefficient, 126, 430, 433
inertial mass of production, 181, 183, Lagrangian function, 125, 361
219, 237 law of large numbers, 14
inertial mass of saving, 382 law of non-increasing marginal
inferior good, 149 productivity, 94
inflation, 54, 63 law of non-increasing marginal utility,
instantaneous acceleration of 113
production, 81–2, 180, 183 legal tender, 326
instantaneous acceleration of liability, 329
utility, 121 limit value, 407
instantaneous velocity of production, linear approximation of nonlinear
73–4, 76 relation, 148
instantaneous velocity vector, 86 logarithmic function, 410–11
integral calculus, 438–41
integral function, 438–9
interbank deposit, 333 M
interest differential, 331, 376 macroeconomic thinking, 18
interest factor, 294, 295 marginal costs, 160, 163, 174, 176,
interest rate, 65–8 207
interest rate risk, 304, 339 marginal costs of labor, 253
internal rate of return, 303, 339 marginal productivity, 92, 275
interval scale, 40, 109, 448–50 marginal productivity of capital, 292
inventory, 158 marginal productivity of labor, 252,
inverse demand function, 144, 150 269, 286
inverse function, 405–6 marginal revenue, 169, 171, 172,
inverse supply function, 198 174, 175
investment, 158, 313 marginal utility, 112, 116, 131
irrational behaviour, 15 marginal utility of budgeted funds,
124, 126
marginal willingness-to-pay, 44, 111,
K 124, 126–9, 248, 255, 261, 265,
kinematics, 68 275, 343, 346, 350, 430–3
kinematics of production, 162 market supply relation, 198
market value of a good, 45
marxian political economics, 5
L maturity, 332
labor, 2 maturity transformation, 331
labor demand relation, 273, 283 maximum, 413
labor supply relation, 273 measurement, 444
462 Subject Index
measurement function, 444 Newtonian equation of motion for

measurement system for economics, labor, 257, 266, 277
41–51 Newtonian equation of motion for
measurement unit for utility, 111 price, 190
metatheoretical isomorphism, 21 Newtonian equation of production,
methodological basis for economics, 181, 183, 237
30 Newtonian equation of savings, 369
methodological monism, 3 Newtonian theory of a consumer,
minimum, 413, 428 134–42
modified Newtonian equation for nominal quantity, 52–7
price, 245 nominal value, 332
monetarism, 18 non-increasing marginal utility, 118
monetary base, 326, 328 norm, 86
monetary economy, 328 normal good, 149, 197
monetary policy, 328 normative economics, 3
monetary system, 327 normative scientific methodology, 3,
money market, 332 18
money-voting mechanism, 15
monopolistic competition, 200
monopoly, 199 O
monopoly profit, 227 official profit, 157
monopoly union, 281 oligopoly, 200
monopsony, 199 order relation, 447
mortgage loan, 329 order scale, 109, 447–8
municipal bond, 334 OTC-market, 336
outstanding bill, 158
own capital, 159, 334
N
Nash equilibrium, 232
national means of payments, 327 P
natural monopoly, 200 Pareto improvement, 329
neoclassical economics, 16–19, 151 partial derivative, 112, 416–21
neoclassical equilibrium, 210 partial differentiation, 418
net worth, 159 partnership, 155
Newtonian equation of borrowing, per cent, 61
376 perfect competition, 199, 202–25
Newtonian equation of consumption, physical capital, 157, 289
138, 217 point particle, 83
Subject Index 463
portfolio, 329 real quantity, 53

positive economics, 3 real world of a science, 4
positivism, 3, 18 relative change, 60
power-law scaling, 20 rent, 290
practical syllogism, 12 representative money, 322
preference relation, 104 repurchase agreement, 333
present value, 293, 295, 297, 302–5, resource, 2
308, 360 resultant force, 25, 378
price cartel, 202 returns to scale in production, 199
price inflation, 241 revenues, 169, 172, 173
primary dimension, 38 risk-free arbitrage, 337
primary dimensions in economics,
52–4
primary market, 334–6 S
principal money, 326 saddle point, 428
principal value, 332 sales function, 167, 168, 171, 184
principle of modeling in economics, saving, 360
13–14, 153, 180, 182 scalar, 40, 395
production, 2 scale of measurement, 444–52
production function, 91–5, 252 secondary dimension, 38, 49
production method, 154 secondary market, 333–6
Profit and Loss Account, 159 second fundamental theorem of
public good, 200 calculus, 440
purchasing power of money, 45, security, 329
47, 52 set, 399
pure number, 39 share of a stock, 329, 333, 341
shorting, 342
slope, 396
Q socialists, 16, 17
quantity, 36 sociology, 17
sociophysics, 23
sole proprietorship, 155
R spring constant, 219, 276, 348, 382
random walk, 352 spring equation, 219
rate of time preference, 365, 373 static friction, 137, 181, 219, 317
rational behaviour, 6–9 statistical unit, 444
ratio scale, 40, 450–1 statistical variable, 444
real number, 396 stockholder, 156
464 Subject Index
straight line, 396–8 utility, 49, 108, 115–18

substitutability, 195 utility function of a labor supplier,
substitute good, 48, 166 259
superposition principle, 218
supply function, 198
supply of labor, 267 V
supply relation of a firm, 203, 206 variable, 36
supply relation of labor of a person, variable costs, 160
263 vector, 399–401
supply relation of savings, 366, 381 vector function, 83, 395
system of measurement units for velocity of lending rate, 380
economics, 36 velocity of mass, 28
velocity of price, 188
velocity of production, 69–72, 77
T
tax wedge, 272
Taylor series approximation, 148, W
246, 286, 376, 382, 423–4, wage slide, 281
426 weighted average, 89
testing of hypothesis, 4 worldline, 69, 71
theoretical research, 4
total differential, 264, 424–6
treasury bill, 332 Y
yield, 329, 338, 340
U
unit costs, 160 Z
unit labor costs, 253 zero-coupon bond, 339

Newtonian Microeconomics PDF

Uploaded by

Copyright:

Available Formats

Newtonian Microeconomics PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Newtonian Microeconomics PDF

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What are some of the main weaknesses of neoclassical microeconomics according to the author?

What are some of the main weaknesses of neoclassical microeconomics according to the author?

NEWTONIAN

ISBN 978-3-319-46878-5 ISBN 978-3-319-46879-2 (eBook)

Library of Congress Control Number: 2016961252

© The Editor(s) (if applicable) and The Author(s) 2017

Printed on acid-free paper

This Palgrave Macmillan imprint is published by Springer Nature

Motivation for the Book

We change the assumed optimal behavior in the neoclassical framework

Innovations in the Book

testing of economic theories becomes rigorous. (3) The economic ‘forces’

About the Audience

Joensuu, Finland Matti Estola

Estola, M. (2015). Neoclassical and newtonian theory of production: An empir-

3.5 Utility Function and Indifference Curves 115

4 The Behavior of Firms 151

5 Goods Markets 195

5.2 Perfect Competition in an Industry 202

6 Labor as a Production Factor 251

6.5.5 The Force Acting upon Employment 271

7 Capital Goods as Firms’ Inputs 289

8 Money and Financial Markets 321

8.2 Fundamentals of Financial Markets 328

9 Saving, Borrowing, and Interest Rates 359

10 Mathematical Appendix 395

10.5 Functions 401

10.16 Scales of Measurement 444

Author Index 453

Subject Index 457

© The Author(s) 2017 1

§: By consumption we understand the eating or using up of something

1.1 Classification of Economics

Another classification in scientific analysis can be made between the-

Like other sciences, economics can be divided into various subfields,

1.2 The Axioms of Economics

There are certain regularities in human sciences: people read newspapers,

The needs of human beings in the twenty-first century have not

Axiom 1: The Regular Needs of Human Beings. People like to satisfy

1.2.2 Business Competition and Rational Behavior

In a market economy, people and organizations have to compete with

Axiom 2: Competitive Market System. A market system is based

Corollary to Axioms 1 and 2. People’s and firms’ willingness to manage

The observed regularities in economies can be demonstrated by the

1.2.3 The Birth of Organizations

all creative innovation is usually hard competition to achieve success, that

The examples above demonstrate that we can understand the behavior

Axiom 3: The Birth of Organizations. People build organizations to

The goals of an organization may conflict with those of its members,

Axiom 4: The Goals of Organizations. Organizations have goals they

1.2.4 The Principle of Modeling in Economics

Economic phenomena originate from the decisions of economic units:

Practical Syllogism. Person A intends to bring about B. A considers that he

In economics, however, the two scientific methodologies, Galilean

Axiom 5: The Principle of Modeling in Economics. Economic unit

The extension we make to PS takes care of the following matters: (1)

intentions; rather he does a benefit-harm analysis for his intentions before