Book - Financial Dynamics and Business Cycles

Fi n a n ci al D y n a mi c s
Willi S e m ml er
FI N A N CI A L D Y N A MI C S A N D B U SI N E S S C Y C L E S
a n d B u si n e s s C y cl e s
N e w P er s p e cti v e s
Willi S e m ml er
I S B N 9 7 8- 0- 8 7 3 3 2- 5 3 1- 8
,!7IA8H3-dcfdbi!
w w w.r o utl e d g e. c o m a n i nf or m a b u si n e s s
Rnancial Dynamics
and Business Cycles
Financial Dynamics
and Business Cycles
New Perspectives
Edited by
Willi Semmler
~l Routledge
m~ Taylor & Francis Group
LONDON AND NEW YORK

First published 1989 by M.E. Sharpe.Inc
Published 2016 by Routledge
2 Park Square, Milton Park, Abingdon, Oxon 0X14 4RN
711 Third Avenue, New York, NY 10017, USA
Routledge is an imprint of the Taylor & Francis Group, an informa business
Copyright © 1989 by Taylor & Francis. All rights reserved.
No part of this book may be reprinted or reproduced or utilised in any form or by

any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying and recording, or in any information storage or retrieval
system, without permission in writing from the publishers.
Notices
No responsibility is assumed by the publisher for any injury and/or damage to
persons or property as a matter of products liability, negligence or otherwise, or
from any use of operation of any methods, products, instructions or ideas contained
in the material herein.
Practitioners and researchers must always rely on their own experience and
knowledge in evaluating and using any information, methods, compounds, or
experiments described herein. In using such information or methods they should be
mindful of their own safety and the safety of others, including parties for whom they
have a professional responsibility.
Product or corporate names may be trademarks or registered trademarks, and are

used only for identification and explanation without intent to infringe.
Library of Congress Cataloging-in-Publication Data
Financial dynamics and business cycles : new perspectives / edited by

W illi Semmler.
p. cm.
Bibliography: p.
ISBN 0-87332-531-1
1. Business cycles—Mathematical models. 2. Economic stabiliza
tion—Mathematical models. 3. Debts, External—Mathematical models.
I. Semmler, W illi.
HB3722.F56 1989 88-30857

3 3 8 .5 '4 2 '0 7 2 4 -d c 19 CIP
ISBN 13: 978-0-87332-531-8 (hbk)

Contents
Foreword
Hyman P. Minsky vii
Introduction
Willi Semmler xi
I
Basic Models on Nonlinear Dynamics and Financial Instability
A Minsky Crisis
Lance Taylor and Stephen 0 'Connell 3
Finance, Instability, and Cycles

Michael Woodford 18
Debt-Financing of Firms, Stability, and Cycles in a

Dynamical Macroeconomic Growth Model
Reiner Franke and Willi Semmler 38
Accumulation, Finance, and Effective Demand

in Marx, Keynes, and Kalecki
Anwar Shaikh 65
The Real and Financial Determinants of Stability:

The Law of the Tendency Toward
Increasing Instability
Gerard Dumenil and Dominique Uvy 87
II
Stabilization Policy in Nonlinear Dynamical Models
with Money and Finance
Comparative Monetary and Fiscal Policy Dynamics

Richard H. Day 119
Monetary Stabilization Policy in a Keynes-Goodwin

Model of the Growth Cycle
Toichiro Asada 145
Qualitative Effects of Monetary Policy

in "Rich" Dynamic Systems
Peter S. Albin 168
Debt Commitments and Aggregate Demand:

A Critique of the Neoclassical Synthesis and Policy
Steven Fauari and John Caskey 188
III
Empirical Evidence on Debt and Financial Instability
The Cyclical Behavior of Corporate Financial Ratios

and Minsky's Financial Instability Hypothesis
Christopher J. Niggle 203
Theories of Financial Crises

Martin H. Wolfson 221
The Political Economy of the External Debt and Growth:

The Case of Peru
Felix Jimenez and Edward J. Nell 228
Foreword
Hyman P. Minsky
As the 55th anniversary of the bank holiday of March 1933 approached, financial
instability was a main topic in the financial press. Daily reports appeared of
international debt crises, of the covert bankruptcy of deposit insurance, and of the
near bankruptcy of one great financial institution after another.
The great stock market crash of October 19 and 20, 1987, demonstrated that
extreme instability can happen. It is generally asserted that the consequences of
October 19th and 20th would have been disastrous if the Federal Reserve and
Treasury interventions had not set things right. As a result of these interventions,
subsequent asset prices and financing terms were different from what uncon-
strained markets would have generated. However, the various and assorted state-
ments in the financial press, even by exalted economists, are not clear on what
caused the free fall of the 19th and 20th, and on how the interventions that took
place led to more normal behavior of the economy and various financial markets.
In 1933, financial markets in the United States and throughout the capitalist
world collapsed. The bank holiday of March 1933 remains an extreme example of
a policy intervention that dominates market processes. But in the nice, polite, but
largely apologetic economics of the time, breakdowns such as occurred during
the period from October 1929 to March 1933 were not supposed to happen. This
position of the orthodox theory of the early 1930s is quite surprising, as financial
breakdowns had occurred with regularity ever since modern capitalism emerged,
with its expensive, long-lasting capital assets, markets in which activity and
positions in capital are financed, and entrepreneurial banks riding high on a crest
of new technology.
In the light of historical experience, the past 55 years are the anomaly. During
this time span, no financial or economic collapse has taken place which comes
close to the magnitude of either the great collapse of 1929-1933 or even that of
1907. However, the orthodox economic theory of today, like the apologetic
economics of the 1920s, does not offer a lucid account of why the economy is now
vii
viii HYMAN P. MINSKY
so much more stable than in the past. Why are income, employment, and price
level repercussions of the so-evident financial instability now contained?
In the early 1930s a race was on among the eminent economists of the day to
explain the great collapse as a normal result of interactions among the monetary,
production, and portfolio management spheres of the economy. The announced
aim of Keynes's research program was to develop a theory of a monetary produc-
tion economy that was rich enough to encompass a smoothly expanding economy,
a constrained cyclical economy, and an economy that from time to time collapses
into depressions or explodes into accelerating inflation. In the verbal exposition
of The General Theory, endogenously determined changes in liquidity preference
could lead to rapid movements in the relative prices of different types of assets
and of assets and output. In turn, these changes can lead to rapid and large
movements in output and employment.
In this exposition, it was quite clear that the properties of a capitalist economy
with banks and bankers of various sorts and debts denominated in money were
being examined. If the structure of indebtedness was sufficiently binding in such
an economy, then normal market reactions to excess supply could increase, not
decrease, the excess supply. Specifically, in a financially complex economy, wage
flexibility in response to unemployment could make things worse.
The General Theory carried the day as an explanation of why a capitalist
economy can work as well as the American economy did in the 1920s and as
miserably as it did in 1929-1933. In winning the day, The General Theory beat
out serious efforts by von Hayek, Myrdal, and Schumpeter, who also attempted to
integrate money with the financing of production.
The popular semi-mathematical statements of The General Theory, most par-
ticularly the successful IS/LM version of Hicks, transformed what was essential-
ly a complex nonlinear system of the verbal exposition, which showed that the
economy had various modes of possible behavior, into a nice, polite interdepen-
dent equilibrium system. In the hands of Patinkin, the Hicks version of Keynes's
theory was transformed into a system that sought and sustained equilibrium:
Keynes's theory was stood on its head. The dynamics of the macroeconomics that
emerged out of the Hicks-Patinkin tradition of interpreting Keynes was retro-
grade when compared to the various operational equilibria (market, short-period,
and long-period) of the Marshallian tradition in which Keynes worked. The
reason for this retrogression is that the mathematical exposition of The General
Theory in terms of a multi-market equilibrium was cut to the mathematical
competence of the economists of the day, which, to put it kindly, was limited.
Poor and largely irrelevant economics does no harm when the economy be-
haves so as to generate a largely smooth trajectory. In the first decades after World
War II, the financial structure was so robust that, in truth, money and financial
relations mattered very little. Quite by accident, the fiscal structure of the United
States during this period was such that modest, largely built-in deficits and
surpluses were sufficient to stabilize the economy. In the era of financial tranquil-
FOREWORD ix
ity, the basic message of Keynes-that the proper subject of economics is a

monetary production economy and that once in a while the normal functioning of
such an economy leads to deep and serious depressions-was lost.
In the 1980s, financial turbulence reappeared. The Federal Reserve no longer
could assume that playing around with the money supply was sufficient to stabi-
lize the economy. The lender oflast resort function became the dominant determi-
nant of the time and extent of Federal Reserve intervention. In the turbulent world
of the 1980s, the lack of relevance of the formal models, such as rational expecta-
tions that had achieved prominence in the 1970s, became evident. Fanciful
schemes to develop business cycle models on the basis of either no or poorly
conceived monetary relations emerged in the 1980s. The irrelevance of these
models became evident as financial turbulence and intervention to constrain the
effects of this turbulence became common.
Even as financial instability emerged as a dominant observation, the combina-
tion of the ability to simulate complex dynamical behavior and greater mathemat-
ical sophistication enabled economists, who were not wearing the blinders of
neoclassical orthodoxy, to develop dynamic models that were capable of generat-
ing the complex paths through time that the verbal exposition of The General
Theory encompassed. Mathematical formulations have finally caught up with the
theoretical results based upon a combination of close verbal reasoning of the
repercussions of profit maximization in a monetary production economy and a
close understanding of how capitalist economies behave.
The papers collected in this volume come from various backgrounds and
research paradigms. A common theme runs through these papers that makes the
collection both interesting and important: The authors take seriously the obvious
evidence that capitalist economies progress through time by lurching. Whether a
particular study starts from household utility maximization or from the processes
by which productive structures are reproduced and expanded, the authors are
united in accepting the evidence that financial instability is a significant charac-
teristic of modern capitalism.
If a theory holds that a capitalist economy is best thought of as a complex
interdependent system whose trajectory depends on initial conditions and an
internal dynamics that is capable of endogenously generating periods of incoher-
ence, then the question "Why is incoherence so rare?" moves onto the research
agenda. The obvious answer is that the impact of the institutional structure and
the system of interventions that result from this structure dominates the econo-
my's endogenous dynamics. This dominance can be interpreted as the intermit-
tent imposition of new "initial conditions." Coherence is observed not because
markets yield order, but because institutions and policy interventions dominate
markets. In the view that follows from this argument, though not all chapters in
the volume will agree on it, markets are social devices to adjust details of the
economy once the structures of intervention and regulation are in place that can
abort the full development of chaos.
x HYMAN P. MINSKY
The main message of the combination of close observation of the economy and
the ability to handle complex nonlinear mathematical systems that is evident in
these studies is that capitalism is too complex for laissez-faire to be an acceptable
policy rule. The subtext is that because "apt government" is necessary for
capitalist economies to realize their potential, economic theory will be of use as it
provides insights that enable us to design "apt government. "
Introduction,
Willi Semmler
The General Theme

The chapters in this volume were contributions to a conference on "Economic
Dynamics and Financial Instability,'' held at the Economics Department of the
New School for Social Research in April 1986. This conference focused on the
role of money, finance, and debt in micro and macro dynamical models as well as
on an evaluation of stabilization policy in the context of a dynamical theory of a
monetary market economy. Another aim of the conference was to give an account
of the state of empirical research relating to the evidence of debt and financial
instability, especially in the U.S. economy. This general perspective attempted to
connect the analysis of real economic activities (production, investment, and
consumption) with financing activities in market economies. This view is shared,
by and large, hy the contributors in this volume.
The writings of Hyman Minsky, in particular, have stimulated the revival of
the type of research which also considers the financial side of the economy. As
can be seen from the chapters, major topics of the entire conference were initiated
by Hyman Minsky's work on the financial fragility of market economies. Other
writers, recently, also have pointed out that extensive external financing of eco-
nomic units (firms, but also households and financial institutions) and the feed-
backs from asset and financial markets to economic activities make market
economies very fragile. As Tobin puts it in his Nobel Prize Lecture:
In short, the financial and capital markets are at best highly imperfect coordi-
nators of savings and investment, an inadequacy which I suspect cannot be
remedied by rational expectation. The failure of coordination is a fundamental
source of macroeconomic instability. . . . (Tobin 1982, p. 179)
Starting with Fisher ( 1933) and then based on the frameworks of Keynes and
Kalecki, an elaborate theory of the role of money and finance in the macroeco-
xi
xii WILLI SEMMLER
nomic process has been put forward. By referring to both Keynes and Kalecki,
Minsky has elaborated a theory of financial instability in which the role of the
degree of indebtedness and cash payment commitments of firms, as well as the
revaluation of capital assets due to financial perturbations, are considered central
(Minsky 1975, 1982, 1986). Minsky specifically stresses that the increase of
firms' debt and their extensive borrowing lead to a dangerous liability structure of
firms, and consequently to a fragile performance of macroaggregates.
Although Minsky refers to dynamic motions of the economy and to business
cycle models in his writings on financial instability, he has not formalized his
ideas in the context of dynamical models (except in Minsky 1957). A major
effort, therefore, is undertaken in these chapters to formalize the role of debt
financing for macro theory in the context of a dynamic theory of firms, the
market, and macroaggregates. This task, however, cannot be done without some
technical tools borrowed from dynamical system analysis. Many chapters in this
volume apply advances in linear and nonlinear dynamical theory (particularly
nonlinear dynamics of high dimensions), limit cycle models for two dimensions,
and complex and chaotic dynamics for one-dimensional nonlinear difference
equations, in order to model the relation of the real and financial side of the
economy. The studies are divided into three sections.
In the first section, contributions appear which deal specifically with basic
linear and nonlinear models of the interaction between the real and financial sides
of market economies. Almost all pursue basic themes, in one way or another, of
macroeconomic fragility by modeling economic motions in a way inspired, for
example, by Kalecki (1971), Kaldor (1940), or Goodwin (1948, 1951, 1986) and
by examining the interaction of the real economy and financial markets initiated
by Minsky. Contributions for this part of the book are provided by Taylor and
O'Connell, Woodford, Franke and Semmler, Shaikh, and Dumenil and Levy. All
are methodologically similar. Though most of the chapters utilize basic economic
models as their background (for example, IS/LM versions of macrodynamic
models), they focus particularly on nonlinear dynamics when considering the
interaction of financial and asset markets with the product (or labor) market.
Based on a new innovative and relevant methodology, important economic results
are presented in this section of the book.
The second part of the book continues these themes and deals with stabiliza-
tion policy (fiscal and monetary policy) in the context of dynamic views of real
and financial motions in the economy. Most papers here are of macroeconomic
orientation, such as the chapter by Day which evaluates the effects of stabilization
policy in a model of complex and rich economic dynamics (in the context of an
economic chaos model). Asada also focuses on a macroeconomic approach by
including, in a seminal way, the financial and asset markets in a Goodwin-type
cycle model and discusses economic policy effects on this basis. Fazzari and
Caskey, starting with a critique of the standard IS/LM presentation of the mac-
roeconomy where usually debt and debt contracts are not accounted for, develop
INTRODUCTION xiii
a critique of neoclassical policy postulates. The chapter by Albin, commencing

with a microstudy of firms' behavior in a dynamic neighborhood, elaborates a
complex microdynamical model evaluating economic policy (particularly mone-
tary policy) on this basis. Both chapters by Day and Albin demonstrate how the
underlying motions of a complex nonlinear economy can change due to fiscal or
monetary policy actions.
The third part of the book is concerned with empirical evidence on debt and
financial instability, in particular with cyclical patterns of debt and macroactivi-
ties. Niggle, starting also with Minsky's theoretical contributions, develops
some empirical measures of corporate (cyclical) indebtedness from the Flow of
Funds Accounts and relates this to the fluctuations of macroaggregates. Wolfson
discusses empirically-oriented approaches to analyze the interrelation of debt and
macroeconomic fragility and elaborates on causes of financial crisis in certain
periods of the U.S. economy. Finally, the contribution by Jimenez and Nell
focuses on the main determinants of debt in a developing country. A case study of
a small country with an open economy is analyzed in order to include a view on
this topic from a Third World perspective.
As stated above, most contributions draw more on nonconventional traditions
than on standard textbook presentations of macrodynamics. In order to make the
reader more familiar with these perhaps lesser-known traditions, and for the
purpose of providing some background on the applications of dynamic analysis to
economics, the following sections present some of the relevant concepts. An
introduction to the theory oflinear and nonlinear economic dynamics and macro-
cycle models is given. This is followed by a discussion of some recent topics on
economic dynalJlics and financial instability. A selected bibliography on the
relevant technical and economic themes of the book are then provided for the
reader.
Linear and Nonlinear Dynamical Models

of Economic Fluctuations
This first part attempts to give a brief justification of why models of nonlinear
dynamics are utilized by the authors and why they seem to be empirically more
relevant. There is a long tradition in economic theory that utilizes only linear
dynamical models in macroeconomics (mostly in two dimensions, for example,
IS/LM versions of macromodels). Moreover, a common tradition in modern
macroeconomic writings, textbooks, and even econometric work in macrodyna-
mics is that economic macrosystems are basically considered stable at their
stationary (or steady-state) equilibrium values, and perturbations originating in
unexpected random shocks are thought to be transitory and lead to a return to
equilibrium after sufficient periods of time. In traditional theories, including the
rational expectation view of business cycles (Sargent 1979; Lucas 1981), it
generally has been accepted that stable dynamical models (i.e., stable around
xiv WILLI SEMMLER
their equilibrium solutions) are adequate representations of economic dynamics.

Numerous studies have attempted to demonstrate that cyclical co-movement of
data on production, income, and employment results solely from randomly dis-
tributed shocks on monetary aggregates, productivity of firms, or consumption
behavior of households. The first type of shocks is representative for monetary
cycle theories (Lucas 1981), and the two latter types are relevant in real business
cycle theories (cf. Kydland and Prescott, 1982; Long and Plosser, 1983). This
tradition of macroanalysis started with Ragnar Frisch in the 1930s and still
prevails in a large part of economic theory, for example in new classical econom-
ics (Sargent 1979; Lucas 1981; Barro 1984) and in econometric approaches to the
business cycle (more extensive references to these approaches are given by
Woodford in this volume).
Major new innovations in the theory of economic dynamics and cycles were
already initiated in the 1940s by the application of nonlinear dynamics to the
problems of fluctuating growth and erratic accumulation. The theory of nonlinear
economic dynamics originally was put forward in the work of Goodwin (1948,
1951), Kaldor (1940), and Kalecki (1971) and was also present in Hicks' (1950)
contribution to the theory of trade cycles. This theoretical tradition of endogen-
ously generated economic cycles has been revived recently, mainly beause of
recent mathematical advances in the theory of nonlinear oscillations (Hirsch and
Smale 1974; Marsden and McCracken 1976; Guckenheimer and Holmes 1983;
Gandolfo 1986) and chaotic dynamics (Li and Yorke 1975; Ruelle and Takens
1971; Smale 1967; Lasota and Mackey 1985).
The essential new view in the tradition of nonlinear modeling of economic
dynamics is that the equilibrium dynamics of the economy (around its stationary
or steady-state values) are inherently unstable arising from a Hopf-bifurcation for
certain parameters of the model. Economically, such instability properties in the
vicinity of the stationary or steady-state values of the model are well known from
Keynesian economics or the growth model of Harrod-Domar type, where an
unstable accelerator generates positive feedback effects in the neighborhood of
the equilibrium. This is essential for the dynamics of the models of Foley (1986) 2
and Day included in this volume. On the other hand, the departure of the economy
from its equilibrium value may-due to nonlinearities in the response of econom-
ic agents to economic data-generate globally stable economic systems. Such
globally attracting forces may, for example, arise from the financial or liquidity
side of the economy (cf. Foley [ 1986] and Day included here). Since the equilibri-
um position of the economy is normally conceived as unstable, the economy is
forced to generate stable cycles around the equilibrium (e.g., limit cycles).
Interesting variations of these types of analysis are represented by the chapter by
Dumenil and Levy, where a transition can occur between different regimes of
complex dynamics (cf. the contribution by Albin). Thus, recent applications of
nonlinear (and chaotic) dynamics to economic modeling seem to yield more
interesting and realistic results than the old linear dynamic models. Although in
INTRODUCTION xv
nonlinear models the dynamics are more complicated than in linear models, they
allow for stable, closed cycles or bounded (erratic) fluctuations, as in the type of
models presented by Day, Albin, and Woodford. The trajectories of the dynamics
of the latter type frequently are only describable by probabilistic concepts (cf.
Lasota and Mackey 1985).
The theoretical framework of nonlinear dynamics in economics as originally
initiated by the works of Goodwin, Kaldor, Kalecki, and Hicks has led to the
elaboration of different types of nonlinear cycle models: (1) nonlinear accelera-
tor-multiplier models (Goodwin, Hicks, Minsky), (2) nonlinear wage-share em-
ployment dynamics (Goodwin), (3) nonlinear income-investment dynamics (Kal-
dor), (4) nonlinear profit-investment models (Kalecki), and (5) nonlinear
overlapping generations models (Benhabib and Day 1982; Grandmont 1985). In
addition, as mentioned, nonlinear economic dynamics are not necessarily re-
stricted to cyclical behavior but can also generate bounded, erratic fluctuations as
in the deterministic chaotic models included in this volume.
The theoretical tradition referred to above has proven to be very fruitful for
macroeconomic theorizing, and recently, several further extensions of such ver-
sions of economic dynamics have been elaborated and published in economic
journals. As recent work has also shown, numerous seminal extensions and
applications of nonlinear theories of business cycles (or chaotic dynamics) have
been developed by a new generation of theoretically-oriented researchers. On the
basis of the aforementioned tradition of nonlinear dynamics, several further areas
of research became important. Among them are: (1) the microdynamics of non-
linear cycle theories, (2) growth, technical change, and employment problems in
the context of nonlinear short-run and long-run models, (3) economic dynamics
for open economies, (4) stabilization policy in nonlinear cycle models, (5) the
difference of time-continuous and time-discrete nonlinear models (chaotic dy-
namics), and (6) econometric tests of nonlinearities (or deterministic chaotic
properties) in time series data. Many of these extensions have been worked out in
recent literature. In this book, however, the authors focus mainly on the (possible
nonlinear) dynamical interaction of the financial and the real side of the economy.
Dynamical Models of Macrofluctuations

Dynamical models with money and finance of linear and nonlinear types have
already been developed in the context of the traditional IS/LM framework.
Authors representing this approach include Schinasi ( 1981, 1982), Torre ( 1977),
Stutzer (1980), Benassy (1984), and recently Taylor (1985) and Taylor and
O'Connell (1985). In recent discussions, the work of Taylor and O'Connell
(included in this volume) has served as a new starting point for study of the
interaction of the real and financial side of the economy as verbally formulated by
Minsky. Besides IS/LM versions of macromodels, such problems also have been
xvi WILLI SEMMLER
studied in a framework of accelerator-multiplier models with money and finance

(cf. Minsky 1957). For the majority of those studies, however, the investment and
financing behavior of firms and the lending and borrowing under conditions of
indebted units in an economy really had not been well integrated in macrodynami-
cal models.
Recently, the economic effects of debt financing of economic units, particular-
ly of firms, have been studied extensively in the financial and more empiri-
cally-oriented literature (cf. Brealy and Meyers 1984; Wojnilower 1982; Eckstein
and Sinai 1986; Wolfson 1986). In empirical work on recent trends in corpo-
rate finance, for example, it has been stated that debt financing has become
a dominant way of generating funds-slowly replacing self-financing and equity
financing of firms. This, at least, holds as a secular trend, but it has also be-
come an important way of raising funds over the business cycle (Brealy and
Myers 1984; Wolfson 1986). As the contributions by both Niggle and Wolfson in
this volume show, there is a strong secular as well as cyclical component in
the determination of the liquidity and debt variations of corporate financial
structure.
As previously mentioned in theoretical stupies such as those put forward by
Minsky (1975, 1982, 1986), it has been shown that high debt/asset ratios will
appear as a dangerous liability structure of firms, increasing the risk of bankrupt-
cy of economic units and possibly developing unstable macroeconomic trajector-
ies. Following the research of Minsky and taking into account the increasing
empirical relevance of debt financing of economic units, four different types of
approaches can be seen which relate the conditions of financing to macroeconom-
ic fluctuations.
The first type of model is solely concerned with exogenous credit constraints
and its impact on macrofluctuations. Within this framework, it is argued that
equity financing is not a viable alternative to debt financing. Thus, exogenous
credit constraints, rather than interest rates, are the effective determinants of
economic fluctuations (Stiglitz and Weiss 1981; Greenwald, Stiglitz, and Weiss
1984; Guttentag and Herring 1984). This idea of exogenous credit constraints
(already put forward by von Hayek and Hawtrey in the 1920s) is formalized in
dynamic models by Woodford (1986 and also Woodford in this volume), where
credit constraints appear as a fixed upper limit of borrowing which enters as a
parameter in a time-discrete macromodel generating nonlinear cycles among
other types of motions.
In a second type of model, the ease and drain of finance and liquidity is more
or less endogenously generated. Foley, in particular, has presented a series of
papers where this approach is taken to explain self-generated liquidity-growth
cycles (Foley 1986, 1987). Here, borrowing and lending are fully endogenized in
an interdependent economic system. In Foley's papers, borrowing, lending, and
capital outlay (the investment of firms) are essentially dependent on the profit rate
and on the endogenously generated liquidity of the enterprise system as a whole.
INTRODUCTION xvii
Nonlinear macroeconomic cycles therefore, are generated in the profit rate and
liquidity, where a liquidity drain appears at periods of high growth rates and an
ease of liquidity occurs at low growth rates of the economy. Thus, the drain and
ease ofliquidity become the attracting forces of the dynamic system, whereas an
unstable accelerator effect creates an unstable equilibrium. A similar relation
between an unstable accelerator and a stabilizing liquidity effect is also character-
istic of Day's model included in this volume. Several important features of
Foley's models are that the dynamic system is considered to be a closed one,
capitalist consumption is equal to zero, and there are no asset holders outside of
the firms.
A third type of model allows for multiple assets being held outside of the
investing firms. Specific features for economic dynamics arise here from chang-
ing asset composition, which responds to economic variables such as the rates of
return of firms, their expected rates of return, and the interest rate. Models of this
type are developed in Taylor (1985) and Taylor and O'Connell (1985), and also
in the basic model in Taylor (1982). There, however, capitalist firms are exclu-
sively externally financed through loans and equity issuance. The savings gener-
ated by rentier households, as suggested in Kalecki's macroeconomic writings,
are channeled through the banking system to investing firms. The main new
economic result in Taylor and O'Connell is that financial and asset markets can
render the macrodynamics totally unstable (not to be expected in standard IS/LM
models). This arises from the fact that in contractionary periods, for example, the
previous debt of firms and the additional shifting of asset holders into liquidity
(money) will generate a positive feedback effect during the downward movement
of the economy; the further away the macroaggregates are from their normal
position, the further they will depart from it. This dynamic system is also
globally unstable.
A fourth type of model can be developed as a variation of the Kalecki-Taylor
approach. However, this type explicitly takes into account the role of external
financing and indebtedness of firms in macrofluctuations. Though the models by
Taylor, and by Taylor and O'Connell develop rich macroeconomic dynamics,
including financial instability as proposed by Minsky in his financial crisis theory,
the role of firms' indebtedness and debt payment commitments is not studied
fully. As noted before, a high degree of indebtedness and debt payment commit-
ments of firms, however, seem to be essential causes for financial instability. The
role of debt for investment and borrowing behavior of firms, as well as for
lending by asset holders, is formalized and explored in the contribution by Franke
and Semmler in this book. A Kalecki-Taylor type of (nonlinear) IS/LM version of
a macromodel is expanded to a large dynamical system, which includes two
additional dynamical equations representing the evolution of the debt/asset ratio
and the state of confidence of investors. Among other scenarios, cyclical move-
ments of rates of return, growth rates, the interest rate, the state of confidence,
and the debt/asset ratio arise. In Woodford's contribution, particularly in the first
xviii WILLI SEMMLER
part of his chapter, the debt/asset ratio is made endogenous, feeding back to a
complex macrodynamics of the chaotic type. Both of the models by Franke and
Semmler and by Woodford refer to Kalecki's view that higher debt leads to an
increasing risk for firms-the "law of increasing risk."
In other contributions in this volume, such as the contribution by Shaikh, the
implications of both debt and equity finance for macrofluctuations is explored in
a framework inspired by the Marxian theory of money and credit. A framework
based on nonlinear differential equations for fast and slow dynamics is developed
which allows for an integrated treatment of Marxian, Kaleckian, and Keynesian
ideas on external financing of firms and macro fluctuations. In Dumenil and Levy,
a nonlinear dynamical macromodel with real and financial variables, derived
from the microbehavior of firms, is elaborated. It not only allows for financial
instability and cycles but also for stagnation as a possible scenario.
Based on the types of models presented in this book, the results for economic
policy are very intriguing. Particularly, as the chapters in Part II demonstrate,
stabilization policies of the monetary and fiscal variety are likely to generate
transitions between different types of dynamics when a nonlinear complex type of
motion of the underlying economy is allowed for. Well known and widely used
economic policies can generate quite unexpected outcomes, as shown in the
studies by both Day and Albin.
In sum, a number of central themes of the dynamics with money and finance
and their implications for macrofluctuations and economic policy are presented,
and the empirical measures and trends of debt financing are provided. It may be
appropriate, however, to indicate a further research strategy that could not be
pursued here fully. The fol lowing additional dimensions could be explored on
these topics: (1) microeconomic studies of borrowing and investment behavior of
firms and their effects in a system of interdependent dynamics; (2) empirical and
theoretical studies of debt financing of firms and macrofluctuations in open
economies; (3) an intercountry comparison of stylized facts concerning the evo-
lution of the financing structure of firms, debt burden, and debt payment commit-
ments and their impact on macrofragility; (4) major policy implications for
extended versions of such models (e.g., to open economies). Subsequent research
on dynamical methodology, extended models, stylized facts and trends, and
policy implications will help to introduce further theoretical, empirical, and
policy dimensions into the theme ''Financial Dynamics and Business Cycles.''
The contributions to this book should be considered an important advance in this
development.
Notes
1. The conference and the publication of its papers were made possible by a generous
grant from Manufacturers Hanover Trust, which bears no responsibility for any views
expressed in this volume. The organizers also appreciate very much the help and assistance
of Marsha Lasker and Karin Ray.
2. Foley's paper was originally presented at the conference but has been published
elsewhere in the meantime.
INTRODUCTION xix
Selected Bibliography
Barro, R. J. 1984. Macroeconomics. New York: John Wiley.
Benassy, J. P. 1984. "A Non-Walrasian Model of the Business Cycle." Journal of Eco-
nomic Behavior and Organization 5:77-89.
Benhabib, J., and R. H. Day. 1982. " A Characterization of Erratic Dynamics in the
Overlapping Generations Model." Journal of Economic Dynamics and Control
48:459-72.
Bosworth, B. 1971. "Patterns of Corporate External Financing." Brookings Papers on
Economic Activity 2:253-85.
Brealey, R., and S. Myers. 1984. Principles of Corporate Finance. New York: McGraw-
Hill.
Chang, W.W., and D. J. Smyth. 1971. "The Existence and Persistence of Cycles in a
Non-linear Model: Kaldor's 1940 Model Reexamined." Review of Economic Studies
38:37-46.
Clower, R. W. 1967. "A Reconstruction of Microfoundations of Monetary Theory."
Western Economic Journal 6: 1-9.
Dana, R. A., and P. Malgrange. 1984. "The Dynamics of a Discrete Version of a Growth
Cycle Model." Analyzing the Structure of Econometric Models. J. Ancot, ed., The
Hague: Nijhoff Publishing Company.
Day, R. H., and W. Shafer. 1985. "Keynesian Chaos." Journal ofMacroeconomics 7(3):
277-95.
Eckstein, 0., and A. Sinai. 1986. "The Mechanisms of the Business Cycle in the Postwar
Era.'' The American Business Cycle: Continuity and Change. Robert J. Gordon, ed.,
NBER, Chicago: University of Chicago Press, pp. 39-105.
Fisher, I. 1933. "The Debt Deflation Theory of Great Depressions." Econometrica
I:337-57.
Foley, D. 1986. "Stabilization Policy in a Nonlinear Business Cycle Model." Competi-
tion, Instability, and Nonlinear Cycles, Lecture Notes in Economics and Mathematical
Systems. W. Semmler, ed., New York and Heidelberg: Springer-Verlag, pp. 200-211.
Foley, D. 1987. Liquidity-Profit Rate Cycles in a Capitalist Economy. Journal ofEconom-
ic Behavior and Organization 8(3): 363-377.
Gandolfo, G. 1986. Mathematical Methods and Models in Economic Analysis. Amster-
dam: North-Holland.
Goodwin, R. M. 1948. "Secular and Cyclical Aspects of the Multiplier and Accelerator."
Employment, Income and Public Policy. Essays in honor of A. H. Hansen. New York:
W. W. Norton.
Goodwin, R. M. 1951. "The Nonlinear Acceleration and the Persistence of Business
Cycles." Econometrica 19( 1): 1-17.
Goodwin, R. M., and L. F. Punzo. 1986. The Dynamics of a Capitalist Economy. Boul-
der: Westview Press.
Goodwin, R. M. 1986. "Swinging Along the Autostrada." Competition, Instability, and
Nonlinear Cycles, W. Semmler, ed., Heidelberg and New York: Springer-Verlag,
pp. 125-31.
Greenwald, B., J. E. Stiglitz, and A. Weiss. 1984. "Informational Imperfections in the
Capital Good Market and Macroeconomic Fluctuations.'' American Economic Review
(May): 194-200.
Guckenheimer, J., and P. Holmes. 1983. Nonlinear Oscillations, Dynamical Systems and
Bifurcations of Vector Fields. Berlin and New York: Springer-Verlag.
Guttentag, J., and R. Herring. 1984. "Credit Rationing and Financial Disorder." The
Journal of Finance (December): 1359-82.
Grandmont J. M. 1985. "On Endogenous Competitive Business Cycles." Econometrica
5(September) :995-1045.
xx WILLI SEMMLER
Hicks, J. R. 1950. A Contribution to the Theory of Trade Cycle. Oxford: Oxford Universi-
ty Press.
Hirsch, M. W., and S. Smale. 1974. Differential Equations, Dynamical Systems and
Linear Algebra. New York: Academic Press.
Joseph, D. D. 1983. "Stability and Bifurcation Theory." Comportement Chaotique des
Systemes Deterministes, G. Iooss, ed., Amsterdam: North-Holland, pp. 351-79.
Kaldor, M. 1940. "A Model of the Trade Cycle." Economic Journal 50(March):78-92.
Kalecki, M. 1971. Selected Essays on the Dynamics of the Capitalist Economy. Cam-
bridge: Cambridge University Press.
Klein, L. R., and R. S. Preston. 1969. "Stochastic Nonlinear Models." Econometrica
37:95-106.
Kohn, M. 1981. "Loanable Funds Theory of Unemployment and Monetary Disequili-
brium." American Economic Review (December):859-875.
Kydland F., and E. F. Prescott. 1982. "Time to Build and Aggregate Fluctuations."
Econometrica 50(November): 1345-70.
Lasota, A., and M. Mackey. 1985. Probabilistic Properties of Deterministic Systems.
Cambridge: Cambridge University Press.
Lakin, W. D., and D. A. Sachez. 1970. Topics in Ordinary Differential Equations. New
York: Dover.
Li, T. Y., and J. Yorke. 1975. "Period Three Implies Chaos." American Mathematical
Monthly 82:985-92.
Long, J. B., and C. I. Plosser. 1983. "Real Business Cycles." Journal of Political
Economy 9l(February):39-69.
Lorenz, E. 1963. ''Deterministic, Nonperiodic Flow.'' Journal of the Atmospheric Sci-
ence 20: 130-4 l.
Lucas, R. E. 1981. "Understanding Business Cycles." Studies in Business Cycle Theory,
R. E. Lucas, ed., Cambridge: MIT Press.
Marglin, S. A. 1984. Growth, Distribution and Prices. Cambridge: Harvard University
Press.
Marsden, J. E., and M. McCracken. 1976. The HopfBifurcation and its Applications.
New York: Springer-Verlag.
Minsky, H. P. 1957. "Monetary Systems and Accelerator Models." The American Eco-
nomic Review 47(6):859-82.
Minsky, H. P. 1975. John Maynard Keynes. New York: Columbia University Press.
Minsky, H. P. 1982. Can "It" Happen Again? Essays on Instability and Finance. Ar-
monk, New York: M. E. Sharpe, Inc.
Minsky, H. 1986. Stabilizing an Unstable Economy. New Haven: Yale University Press.
Nell, E. 1986 .. On Monetary Circulation and the Rate of Exploitation. Thames Papers in
Political Economy. London: Thames Polytechnic.
Rose, H. 1969. "Real and Monetary Factors in the Business Cycle." Journal of Money,
Credit, and Banking (May): 138-52.
Ruelle, D., and F. Takens. 1971. "On the Nature of Turbulence, Communications."
Mathematical Physics XX: 176-92.
Sargent, T. 1979. Macroeconomic Theory. New York: Academic Press.
Schinasi, G. J. 1981. "A Nonlinear Dynamic Model of Short Run Fluctuations." Review
of Economic Studies 48:649-56.
Schinasi, G. J. 1982. "Fluctuations in a Dynamic, Intermediate-Run IS-LM Model:
Applications of the Poincare-Bendixson Theorem." Journal of Economic Theory
28:369-75.
Semmler, W. l 986a. ''On Nonlinear Theories of Economic Cycles and the Persistence of
Business Cycles." Mathematical Social Sciences 12(1):47-76.
Semmler, W. 1986b, ed. Competition, Instability, and Nonlinear Cycles, Lecture Notes in
INTRODUCTION xxi
Economics and Mathematical Systems. Heidelberg and New York: Springer-Verlag.

Semmler, W. 1987. "A Macroeconomic Limit Cycle with Financial Perturbations."
Journal of Economic Behavior and Organization 8(3):469-95.
Semmler, W., and M. Sieveking. 1988. Nonlinear Growth-Liquidity Dynamics with
Bankruptcy Risk. New School for Social Research, University of Frankfurt. Mimeo.
Smale, S. 1967. "Differentiable Dynamical Systems." Bulletin of the American Math-
ematical Society 73: 747-817.
Stiglitz, J., and A. Weiss. 1981. "Credit Rationing in Markets with Imperfect Informa-
tion." American Economic Review (June):393-410.
Stutzer, M. J. 1980. ''Chaotic Dynamics and Bifurcation in a Macro Model.'' Journal of
Economic Dynamics and Control 2:353-76.
Taylor, L. 1982. Structuralist Macroeconomics. New York: Basic Books.
Taylor, L. 1985. "A Stagnationist Model of Economic Growth." Cambridge Journal of
Economics 9:383-403.
Taylor, L., and S. A. O'Connell, 1985. "A Minsky Crisis." Quarterly Journal of Eco-
nomics 1OO(Supplement):871-886.
Tobin, J. 1965. "Money and Economic Growth." Econometrica 33(0ctober):671-84.
Tobin, J. 1969. "A General Equilibrium Approach to Monetary Theory." Journal of
Money, Credit and Banking 1:15-29.
Tobin, J. 1982. ''Money and Finance in the Macroeconomic Process.'' Journal of Money,
Credit and Banking XIV(2):171-203.
Torre, V. 1977. ''Existence of Limit Cycles and Control in Complete Keynesian System by
Theory of Bifurcations." Econometrica 45(September): 1456-66.
Wojnilower, A. M. 1982. "The Central Role of Credit Crunches in Recent Financial
History." Brookings Papers on Economic Activity 2:289-327.
Wolfson, M. 1986. Financial Crises. Armonk, New York: M. E. Sharpe, Inc.
Woodford, M. 1986. Self-Fulfilling Expectations, Finance Constraints and Aggregate
Wolfson, M. 1986. Financial Crisis. Armonk, New York: M. E. Sharpe, Inc.
Woodford, M. 1986. Self-Fulfilling Expectations, Finance Constraints and Aggregate
Fluctuations. Columbia University, New York. Mimeo.
I
Basic Models on Nonlinear
Dynamics and Financial
Instability
AMinsky Crisis
Lance Taylor and Stephen O'Connell
Hyman Minsky's ideas about financial crises are influential. For example, he
provides much of the theoretical foundation for Charles Kindleberger's (1978)
well-known book Manias, Panics, and Crashes. But for all his citations in the
specialist literature, Minsky's work has never been elaborated formally, and he is
scarcely noticed in the textbooks.
One reason for the neglect is that Minsky's theories are both microeconom-
ically detailed and institutional. In recent essays collected in Minsky (1982), he
works with at least four types of financial actors: households plus firms variously
engaged in "hedged," "speculative," and "Ponzi" finance. Shifts of firms
among classes as the economy evolves in historical time underlie much of its
cyclical behavior. This detail is rich and illuminating, but beyond the reach of
mere algebra.
What can perhaps be formalized are purely macroeconomic aspects of Min-
sky's theories. Two general assumptions characterize the crises he discusses.
The first is that total nominal wealth in the system is macroeconomically de-
termined, dependent on confidence and the state of the cycle. More of his flavor
is captured if we further postulate that asset choices by firms and households
are not coordinated. Firms build up physical capital, obtaining finance from
equity or loans from intermediaries. They can also build up their own net worth.
Households use intermediaries or equity to direct their savings toward firms.
However, there is no effective arbitrage between valuations of physical capi-
tal held by firms and financial capital held by households. The market valua-
tion of shares can deviate substantially from the book value of capital, with
the difference being absorbed by net worth. With total wealth fluctuating over
time, separate portfolio decisions by firms and households can interact to create
crises.
Comments by Hyman Minsky, an anonymous referee, and Dan Raff are gratefully
acknowledged.
3
4 LANCE TAYLOR & STEPHEN O'CONNELL
The second major assumption is that there is high substitutability among assets
in household portfolios under certain circumstances-there can be a flight to
money when conditions are ripe. How often this possibility arises is an empirical
matter. The crises on record show that it cannot be ruled out of court. When
panics occur, interest rates rise, investment is cut back, and profit rates fall. As a
consequence, the valuation of firms' capital assets declines and so does their net
worth. The stage is set for the debt-deflation process that Minsky and Irving
Fisher (1933) emphasize. Part of the process is extensive financial disintermedia-
tion and "disappearance" of assets. Endogenously varying levels of wealth in the
macro system permit debt deflation to occur.
The text followed in developing these two ideas is titled John Maynard Keynes
by Minsky (1975). This book has the advantage of stating Minsky's crisis story
against the backdrop of The General Theory and the distributional accounting of
Michal Kalecki (1971). We largely follow the latter's formulations, beginning
with a very simple model and then indicating extensions at the end of the paper.
On the production side of the economy, there is markup pricing at a constant
rater over the wage bill (representing prime cost). The nominal wage is w, and
the labor-output ratio is b. The price level P is given by
( 1) P = (1 + r)wb.
Minsky follows Keynes and parallels later model-builders such as Foley and
Sidrauski (1971) in assuming that there are separate capital- and consumer-
goods-producing sectors. That complication is dropped here for simplicity, so the
price of new investment goods is P.
Is it legitimate to impute this price to physical capital goods in place? If so, a
rate of profit r can be defined as
(2) PX - wbX rwbX T X

r=
PK (1 + r)wbK 1 + T K
where Xis the level of output, and K is the capital stock. Other pricing rules for
physical assets would of course produce different expressions for the rate of
profit-in particular, Minsky's analysis is based on prices for individual buildings
and machines. Indeed, he would go further and assert that the PK term in the
denominator of (2) is impossible to define after the Cambridge controversies. For
that reason, all his formulas are stated as levels, while the ones here are based on
division by PK. The trick simplifies differential equations for growth, at the cost
of begging serious questions about the valuation of capital stock.
Minsky's investment theory is built around expected returns generated by
physical capital in the process of production. In a stylized way, we can imagine
firms using a rule of thumb for investment that depends on anticipated profits and
A MINSKY CRISIS 5
a discount factor. The capitalized value of expected earnings per unit of invest-
ment is an appropriate shadow price (called Pk by Minsky) for the investment
decision. It can be written as
(3) Pk = (r + e)Pli,
where i is the current interest rate and e reflects the difference between the
anticipated return to holding capital and the current profit rate r. The variable e
carries a heavy burden in the story that follows. It represents expected high or low
profits, which in turn depend on the overall state of confidence. In Minsky's
view, financial and product market conditions, internal finance, and existing
liability structures all influence Pk and, in the present treatment, Q·
Minsky makes investment demand depend on the price differential Pk - P;,
where P; is the supply price of new investment goods (also subject to real and
financial perturbations). For present purposes P; is replaced by P, and the price
differential is
(4) Pk - P = (r + e - i)Pli
Algebra becomes simpler if we use the variant specification (in nominal terms):
(5) Investment demand = PI = [g 0 + h(r + e - i)]PK,
where 8o is a constant reflecting autonomous capital stock growth, and the

coefficient h measures firms' investment response to the expected difference
between profit and interest costs. The theory of equation (5) is quite orthodox. 1
Income streams generated by production are the wage bill wbX and markup
income rwbX (or rPK). Following Kalecki, we assume that all wages are con-
sumed. Profits are all distributed to rentiers, who have a saving rate s2 • The
aggregate saving flow is given by
(6) Saving supply = srPK = srwbX.
Excess demand for goods is just the difference between (5) and (6). After divid-
ing through by PK, the following condition for equilibrium in the commodity
market is obtained:
(7) 80 + h(r +e- i) - sr = 0.
If the profit rate r or the output level X increases when there is excess demand,
commodity market adjustment is stable ifthe conditions - h > 0 is satisfied-
investment must respond less to profit rate increases than saving. 3 Solving (7) for
r and plugging the result into the investment demand function gives a reduced
form for the capital stock growth rate g ( = II K) as
(8) s[go + h(e - i)]

g=
s - h
A fall in the interest rate or an increase in anticipated profits leads to a higher

growth rate. Since
(9) g = sr
from the saving function, the profit rate and capacity utilization go up as well.
The next step is to look at the asset side of the economy along the usual
portfolio balance lines. There is an outside primary asset F, or fiscal debt. It can
take the form of money (M) or short-term bonds (B), held by the rentiers
(workers' financial market participation is ignored, consistent with the assump-
tion that they do not save). The capitalized value of the plant and equipment held
by firms is PkK = (r + e)PK/i. Firms have emitted an outstanding stock of
equity E; its market price is p e determined below. The difference between the
I
value of capital stock and equity is firms' net worth N. 4 Their balance sheet (along
with that of the rentiers) appears in Table 1. In differential form, the firms'
balance sheet identity is
Table 1
Simplified Balance Sheets for Firms and Rentiers
Firms Rentiers
r + Q
i
M w
N
B
. .
(10) Pk! + PkK = Pe E + PeE + N,
where a dot above a variable denotes a time derivative. The liability counterparts
of new investment or capital gains on the existing stock are new equity issues,
higher equity prices, or increased net worth. We do not go into how firms decide
A MINSKY CRISIS 7
about issuing new stock; hence the adjusting variables are the price of equity and
net worth.
Total wealth of the rentiers is
(11)
A price for bonds does not enter in (11), since they are short term. The change in
rentiers' wealth over time is
. . . . .
(12) W = PeE + Pe E + M + B = PeE + srPK.
Their wealth increases from capital gains and financial saving.

At each point in time, rentiers allocate their wealth across assets according to
the following equations for market balance:
(13) µ(i,r + e)W - M = O,
(14) w,r + e) W - E = 0,
and
(15) - (3(i,r + e)W + B = 0,
where µ + ~ + (3 = 1. Only two of these three equations are independent. As

usual, we work with (13) and (14) for the money and equity markets with i and Pe
as the equilibrating variables, respectively. The excess supply function (15) for
bonds will be equal to zero when the other excess demand relationships satisfy the
same condition.
The arguments in the asset demand functions are the bond interest rate i and
the anticipated profit rate on physical capital r + e. Incorporating transactions
demand would require the use of XI K (or r, again) as an additional argument, but
this possibility is ignored for simplicity. The notion behind using r + e to
measure returns to equity is that wealthholders try to look through Wall Street to
"fundamentals" on the production side, instead of basing share purchase deci-
sions on the Dow Jones average Pe. A more elaborate theory of asset demand
would use the expression (r + e)PIPe + fie as the return to equity, where fie is
the expected growth rate of Pe. If, following the rational expectations school, the
actual and expected rates of inflation of equity prices were made equal (except for
a white noise error term), then (14) with (r + e)PIPe + fie as the return to
equity could generate a stock price bubble. Inverting (14) would make lie a
positive function of Pe, and the standard rational expectations saddlepoint solu-
tion could emerge. 5
We ignore this possibility because bubbles do not seem central to Minsky's
crisis theory, though he mentions them from time to time. 6 His argument would
be that under most (but not all) circumstances, shareholders simply do not agree
about expected inflation of the equity price. On average (though not for some) the
arbitrage opportunity is ignored; the possibility of capitalizing economy-wide
gains or losses on share prices is not exploited. 7 Folklore contends that Joseph
Kennedy got out of the stock market before the crash of 1929. Most other
participants did not, and their error generated a crisis of confidence of the type to
be discussed below.
With bubbles excluded, the key variable in (13) and (14) is the anticipated
corporate return r + e- Note from Table 1 that higher returns bid up firms'
valuation of their capital stock. The same is true of financial wealth, since from
(11) and (14),
(16) F
W=
1- w,r+ e)
An increase in rand e will drive up t and thus share prices and financial wealth
will rise. In effect, rentiers' net worth is determined macroeconomically from
their valuation of anticipated profits, feeding into market balances for asset
supplies and demands. The share price can be solved for as
(17) Pe = W(l - ~)](FIE);
in turn, Pe determines the change in firms' net worth given their investment and
issuance of new equity in (10).
From (16) it is easy to rewrite the money market excess demand function as
(18) µ,(i,r + e) = a[l - w,r + e)],
where a = Ml F is the share of fiscal debt issued as money.

Using subscripts i and r to stand for partial derivatives with respect to the
interest rate i and the expected profit rate r + e, we may write the differential
form of (18) as
(19) 71;di + 71,dr = -11,de + (1 - ~)da,
where
1/; = µ,; + a~;

A MINSKY CRISIS 9
and
A higher bond interest rate cuts back on demand for money, so that µ; is
negative. Since demand for equity also falls, ~;is negative, making TJ; < 0. The
partial derivative µr is negative, but an increase in r or e raises the demand for
nominal equity. From the standard assumption that assets are gross substitutes,
~r > Iµr I . However, if money and equity are close substitutes in asset demand,
the magnitudes of the two partial derivatives will be close to each other. Further,
if ex is a small enough fraction, then T/r < 0. For reasons to be made clear shortly,
we shall assume high substitutability between money and equity, so that the
portmanteau derivative T/r is indeed negative. 8
Note immediately from (19) that an open market operation to increase the
money supply would raise ex and reduce the interest rate for a given rate of profit.
From (3) and (17) there would be higher asset prices Pk and Pe; Minsky (1975)
devotes long passages to justify this result. An increase in the expected extra
profit rate e will reduce i when there is a high degree of asset substitutability. 9
Equations (7) for the commodity market and (18) for money form a system
analogous to the usual IS/LM construct. However, it should be recognized that
underlying (18) is the assumption that both the money and equity markets clear. In
equilibrium, the price of equity Pe and nominal wealth Ware determined along
with the profit and interest rates. As shown in Figure 1, we assume that the
financial market equilibrium schedule has a negative slope in (r,i) space, due to
strong substitution between money and liabilities of firms. The "story" is that if
realized or prospective profits increase, then rentiers wish to shift their portfolios
away from money and bonds and toward claims to real assets. With a sufficiently
strong shift away from money, the equilibration process requires a rise in the
equity price and hence, in wealth. Interest rates fall to make households content
to hold the existing stock of bonds at the increased level of wealth.
For short-run stability in our analog to the IS/LM system, the slope of the
financial market curve must be shallower, i.e., less negative, than the slope of the
commodity market schedule, as shown in Figure 1. An increase in e will pull
rentiers sharply enough toward equity to bid down the interest rate, as shown by
the dashed line. In the commodity market, a higher e stimulates investment
demand, thus increasing output and the rate of profit. Overall, the outcome is a
lower interest rate, a higher profit rate and a higher Pk-there is a positive linkage
between expected profits and the actual profit rate and rate of capital stock
growth. On the other hand, if prospects seem grim, a fall in anticipated profits
will lead rentiers to flee toward money, drive up interest rates, and strangle
growth. Tighter monetary policy (a lower ex) would have a similar effect, shifting
the financial market locus upward. The outcome would be a higher interest rate
and a lower rate of profit.
Figure 1. Responses of the Interest Rate and Profit Rate to an Increase

in the Expected Incremental Profit Rate e.
Interest rate 1
Financial
---
---.'."ark et
"'- Commodity
market
Prof it rote r
These mechanisms can generate a crisis. To see the details, the way in which
anticipated profits and monetary policy evolve over time needs to be specified.
The most plausible theory about the expected profit differential e is that it should
depend on the general state of the economy: e might increase, for example, when
the actual profit rate is high or the interest rate is low. Both hypotheses in fact give
the same dynamics, but the interest rate link is used here, since it involves easier
algebra. To do so, suppose that the "normal" dynamic story about e is given by
the equation,
(20) e= - f3(i - n.
When the rate of interest exceeds its "normal" long-run level r, expected profits
begin to fall.
To complete the dynamics, government policy behavior must be specified. In
standard Keynesian fashion, both monetary and fiscal policy have substantial
influence over the path of capital stock growth in our model. Minsky (1982)
offers lengthy discussion of the interaction of monetary and fiscal interventions in
a complex financial system. In the current model, the money-debt ratio a can be
written as
a=
M
F
M
PK
PK
F =; t~J
A MINSKY CRISIS 11
where f is the ratio of outstanding fiscal debt to the capital stock. Leaving fiscal
complications aside, we fix government expenditures as a proportion of the
capital stock and taxes as a proportion of expenditures. On these assumptions,
f is fixed, and government spending disappears as an autonomous component of
the capital stock growth rate g. The money-debt ratio then evolves according
to the rule,
(21) a= M - g,
so that for a fixed money growth rate M, ci falls as g increases.

The nonactivist monetary policy of pre-Keynesian days when financial panics
occurred with some frequency could be characterized as a choice of a fixed rate of
money supply growth. This sort of policy has a flavor of "leaning against the
wind,'' since money growth does not respond to changes in g. However, it is a far
cry from the activist policy pursued in many countries after World War II. The
shift of October 1979 in Federal Reserve operating policies toward more precise
targeting of money supply growth rates might perhaps be characterized as a move
from more complicated interventions toward a rule like (21). Minsky might
attribute the retreat from this policy in mid-1982 to growing realization on the
part of the monetary authorities that crises still can occur.
The system (20) and (21) has a steady-state equilibrium at i = rand g = M.
With partial derivatives from (20) in the first row, its Jacobian matrix takes the
form:
(22)
where the subscripts on i stand for derivatives through the IS/LM system, (7) and
(18), and the growth rate derivatives come from (8).
Equations (20) and (21) are potentially unstable. From Figure 1, an increase in
e
e lowers the interest rate and thus raises the derivative in (20). This positive
feedback does not necessarily dominate the system, since the Jacobian determi-
nant - {3iOlg e is easily seen to be positive (signaling possible stability).
The phase diagram appears in Figure 2, with arrows showing directions of
adjustment in the different quadrants. To explore the possibilities, assume that the
economy is initially in a complete steady-state equilibrium at point A. A momen-
tary lapse of confidence would cause e to jump down from A to a point like B.
Equally, a one-shot market operation to reduce the money supply would cause i to
rise. For a newly set (lower) value of OI, (20) shows that e would start to fall from
A, setting off a dynamic process like the one beginning to B.
If the authorities hold to a constant money supply growth M when the economy
is away from steady-state, then a below-equilibrium value of e is associated with
Figure 2. Adjustment Dynamics When a Fall in the Expected Incremental Profit Rate
e from an Initial Equilibrium at A Leads Finally to a Return to Steady-State.
Anticipated incremental
profit rate p
a=O
p=O
Ratio of money
to outside assets a
slow capital stock growth and a rising money-debt ratio a from (21). This
increase would reduce the interest rate and raise e.
If this effect were strong
enough, the economy would follow a path like the one leading through C and
return to equilibrium. A minor crisis occurs in the sense that the profit rate and
output fall, leading to a lower interest rate, higher investment demand, and
ultimate recovery.
But what happens ifthe (a,e) trajectory does not turn the corner at C? At the
micro level, the system enters a debt-deflation contraction such as described by
Irving Fisher (1933). Minsky (1982, 42) describes past examples as follows:
Whenever profits decreased hedge units became speculative and speculative

units became Ponzi. Such induced transformations of the financial structure
lead to falls in the price of capital assets and therefore to a decline in invest-
ment. A recursive process is readily triggered in which a financial market
failure leads to a fall in investment which leads to a fall in profits which leads to
financial failures, further declines in investment, profits, additional failure,
etc.
In terms of Figure 2, output and investment can fall forever, or at least until the
model changes. This is a true Minsky crisis, and it occurs when the derivative ie
is strongly negative and the slope of the e
= 0 locus in Figure 2 is shallow.
Going back through the algebra reveals that this condition applies when there is
A MINSKY CRISIS 13
high asset substitution. A reduction in e leads the interest rate to rise and the
profit rate to fall, driving rentiers into money and bidding up the interest rate
further. Expected profits fall still more, and the process never ends. An unstable
Minsky crisis looks like movement into a liquidity trap except that the interest
rate is steadily rising. From (3) and (17), the descent into the trap is accompanied
by plummeting capitalized quasi rents and equity prices-general disintermedia-
tion. Financial claims and counterclaims collapse as the microeconomic manifes-
tation of the crisis.
To follow the financial collapse in detail is beyond our scope here. However,
three observations are worth making. First, Minsky stresses the importance of
intermediaries in accelerating both boom and crisis by creation and destruction of
''layered'' financial structures. Table 2 gives an expanded balance sheet that may
illustrate what he has in mind. Firms now issue debts D1 along with equity as
liabilities. These are held by intermediaries as assets, along with quantities B;
and M; of outside bonds and (high-powered) money. They also have net worth Q
and liabilities (deposits) to the rentiers in amount D;. As far as the public is
concerned, these deposits are equivalent to money. Money supply is a variable
endogenous to the entire macro system, as argued by Keynesians such as Kaldor
(1982).
Table 2
Amplified Balance Sheets for Firms, Financial
Intermediaries, and Rentlers
Firms
r +e
PK
N
Intermediaries
D;
Q
Rentie rs
(M - M;) + D;
(B - B;) w
PeE
In the initial phases of an expansion, profit rates rise, and interest rates fall.
The partial derivatives of firms' net worth N with respect to these variables are
(23)
Nr = ~r [r~K (r~rl~)
- - - - PeE
J
1 1 - ~
and
(24) 1 (iU~)
- - - - PeE
]
1 - ~
Signs are ambiguous here, since Pk and Pe on opposite sides of the firms'
balance sheet both fall with i and rise with r. However, one would expect Nr to be
positive when the share of rentiers' financial wealth held in equity ~and the share
demand elasticity r~rl~ are relatively small. If r and P are related positively
through a rising markup or aggregate supply curve, Nr > 0 is still more likely.
Similar arguments suggest that N; < 0. If these conditions hold, then at the
beginning ofa boom firms' net worth will begin to rise. They will tend to borrow
against this increase, creating assets that intermediaries can then expand across
the economy. The process will reverse in the downswing, and the intermediaries'
overall importance will shrink. At the top of the expansion, the ratio of firms'
debt to their net worth rises, and they shift gradually (in Minsky's terminology)
from "hedge" to "speculative" and even "Ponzi" positions. The stage is set at
the micro level for financial collapse; ultimately some wave of failure sets it off.
Assets and liabilities of the intermediaries contract, as the value of capitalized
expected profits declines. The process carries with it bankruptcies and financial
hardship, especially for the "Ponzi" firms that had been happily emitting new
liabilities to cover ongoing interest costs.
Second, in his recent writings, Minsky (1982) stresses the importance of
government deficits and Federal Reserve interventions in cutting off the possibil-
ity of open-ended crises as discussed above. In Kalecki accounting incorporating
the government, we have
Profits = Investment + Government deficit - Current account deficit.
In crisis, investment falls, but the government deficit goes up. It can act as a
source of demand to prevent endless debt deflations. In a like manner, Federal
Reserve intervention to increase the growth rate of the money supply could
preclude crisis. Both fiscal and monetary stabilizers could be described formally
by extending our model to include government and central bank transactions
explicitly.
Third, bankruptcies of firms are an intrinsic aspect of the downswing. Reduc-
A MINSKY CRISIS 15
tions in investment demand as firms attempt to sell off capital assets to meet
inelastic cash requirements can make the ''commodity market'' curve in Figure 1
flat or upward sloping at low rates of profit. In this situation, monetary contrac-
tion can lead to unstable dynamics, even in the absence of high substitutability
between money and capital. For details see O'Connell (1983).
In closing observe that for empirical testing, the key mechanism in the crisis
theory here is the negative relationship of expected profits and the rate of interest
discussed in connection with Figure 1. This linkage in turn requires a substantial
degree of substitutability between equity and other assets in the aggregate portfo-
lio. Were there less substitutability, the financial market equilibrium locus in
Figure 1 would slope upward (as LM curves usually do) and an increase in e
would drive up i. Thee = 0 locus in Figure 2 would be answered by an immedi-
ate upward movement in that variable.
High substitutability plays a central role in other portfolio-based crisis mod-
els.10 It represents a certain absence of inertia in the financial system, as opposed
to a case where more sluggish responses to changes in returns underlie general
stability. Over time, asset substitutability may rise if the central bank regularly
has intervened as a lender oflast resort to avert potential crises. Taking the past as
a guide to the future, participants in financial markets may become accustomed to
exposed positions. Their portfolio switches may become more frequent and
substitution more acute when the economy is at the peak of the cycle, or dire
portents are in the air. If, under these circumstances, the central bank shifts to a
less interventionist policy line, the stage may be set for disaster. With sensitive
asset markets, financial crisis must always be considered as a live macroeconom-
ic possibility.
Notes
1. Using Pk as a shadow price for investment decisions, of course, resembles Tobin's

(1969) use of "q." However, we depart from Tobin by not carrying the q-calculation over
to the equity market. Separation of the investment decision from the price of equity is a
corollary of the independence of households' and firms' financial actions that was men-
tioned at the outset.
2. In principle, the saving rate could depend on wealth or some notion of permanent
income. As will become clear below, such a behavioral assumption would lead to a
positive dependence of s on e. The resulting aggregate demand effects would reinforce our
story and are omitted for simplicity.
3. See Taylor (1983) for discussion of stability and other properties of the present
model. O'Connell (1983) analyzes a model in which the commodity market stability
condition is violated at low levels of capacity utilization, giving rise to unstable dynamic
processes with a Fisher-Minsky flavor.
4. Large outstanding levels of corporate net worth appear to be characteristic of
modern capitalism. See Atkinson (1975, 129-31) for estimates for the United Kingdom
and a discussion of the difficulties this phenomenon creates for analysis of wealthholding
in general.
5. In formal terms, let </i(i,) be the inverse function of W,) with respect to its second
argument. Then from (14) with (r + e)PIPe + Ile as the return to equity we have
_ _ [
Ile - </J I,
F
PeE
+ PeE
J -
(R +
Pe
e)R
,
so that fie depends positively on Pe. For more on how such a relationship can generate
saddlepoint instability, see Burmeister ( 1980).
6. See also Kindleberger (1978). The textbook example of a rational expectations
bubble is the tulip mania in Holland more than 300 years ago. For an early exposition of the
theory, see Samuelson (1957).
7. Analogously, investment demand never responds with enough alacrity to potential
profit to driver + e and i into equality. Minsky (1975) cites borrowers' and leaders' risk
in the investment context and nowhere suggests the tulip mania triggers macroeconomic
capitalist crises.
8. If we included transactions demands in the model, they would make 1/r less
negative or positive. We assume substitution effects dominate.
9. Minsky (1975) prefers to treat the negative effect of e on i in terms of shifts in
liquidity preference. On page 123 we learn that "during a boom the speculative demand
for money decreases.'' Further, on page 76 if higher income from a boom ''is interpreted
as increasing the surety of income from capital-asset ownership, then the liquidity prefer-
ence function will shift, so that for a given quantity of money, the higher the income, the
higher the interest rate, and the higher the price of capital assets." In other words, for
given money and income, higher exected profits (which drive up the price of capital assets)
would have to be associated with a lower interest rate (because, again, speculative demand
declines). The implied sign change in the derivative 1/r from positive to negative as r rises
could be modeled in the present framework. Its main effect would be to increase stability
on the downswing and make an endless Minsky crisis of the type discussed herein
impossible.
10. For example, see Dornbusch and Frenkel (1982).
References
Atkinson, A. B. 1975. The Economics of Inequality. Oxford: Clarendon Press.

Burmeister, Edwin. 1980. Capital Theory and Dynamics. Cambridge: Cambridge Univer-
sity Press.
Dornbusch, Rudiger, and Jacob A. Frenkel. 1982. The Gold Standard and the Bank of
England in the Crisis of 1847. Paper presented at conference Retrospective on the
Classical Standard 1821-1931. Hilton Head, N.C.
Fisher, Irving. 1933. "The Debt-Deflation Theory of Great Depressions." Econometrica
1:337-57.
Foley, Duncan K., and Miguel Sidrauski. 1971. Monetary and Fiscal Policy in a Growing
Economy. New York: Macmillan.
Kaldor, Nicholas. 1982. The Scourge of Monetarism. Oxford: Oxford University Press.
Kalecki, Michal. 1971. Selected Essays on the Dynamics ofthe Capitalist Economy, 1930-
1979. Cambridge: Cambridge University Press.
Kindleberger, Charles P. 1978. Manias, Panics and Crashes: A History of Financial
Crises. New York: Basic Books.
A MINSKY CRISIS 17
Minsky, Hyman P. 1975. John Maynard Keynes. New York: Columbia University Press.
- - - . 1982. Can "It" Happen Again? Essays on Instability and Finance. Armonk,
New York: M.E. Sharpe, Inc.
O'Connell, Stephen A. 1983. Financial Crises in Underdeveloped Capital Markets: A
Model for Chile. Department of Economics, Massachusetts Institute of Technology.
Mimeo.
Samuelson, Paul A. 1957. "Intertemporal Price Equilibrium: A Prologue to a Theory of
Speculation. " Weltwirtschaftliches Archiv. LXXIX: 181-219.
Taylor, Lance. 1983. Structuralist Macroeconomics. New York: Basic Books.
Tobin, James. 1969. "A General Equilibrium Approach to Monetary Theory," Journal of
Money, Credit and Banking I: 15-29.
Finance, Instability, and Cycles
Michael Woodford
Neoclassical optimal growth models are found to have a stable stationary equilib-
rium for a wide range of parameter values, including those usually judged to be
empirically realistic. While examples are known in which the stationary competi-
tive equilibrium is unstable and equilibrium cycles exist (Benhabib and Nishi-
mura 1979, 1985), or even in which equilibrium dynamics are chaotic (Boldrin
and Montrucchio 1986; Deneckere and Pelikan 1986), these depend both upon
relatively extreme assumptions about the relative capital intensities of different
sectors and upon a sufficiently high rate of time discount. (For sufficient condi-
tions for the absence of cycles or chaos in optimal growth models, see Scheink-
man [1976] and Dechert [1984].) It follows from the stability of the stationary
equilibrium in the absence of exogenous shocks that business cycles (i.e., repet-
itive rather than merely transitory deviations from the stationary equilibrium)
must be explained as resulting from the response of the economy to continual
exogenous shocks. Repetitive aggregate fluctuations are identified with the sta-
tionary Markovian equilibria of a stochastic growth model, as in the work of
Kydland and Prescott (1980, 1982), where the exogenous shocks are stochastic
shifts in the production technology.
Critics of this sort of approach to business cycle theory have typically attacked
the fundamental methodological postulates of equilibrium modeling-it is denied
that markets should be modeled as clearing at all times, or that agents should be
assumed to possess perfect foresight (or ''rational expectations" in the case of a
stochastic equilibrium). Neoclassical economists who object to the new classical
macroeconomics typically hold that these postulates are appropriate for models of
long-run equilibrium, but that business cycles are short-run phenomena which
cannot be modeled in that way; more radical critics deny that these postulates are
appropriate in analyses of economic dynamics at all.
I wish to suggest instead that the logic of conventional new classical accounts
of the business cycle is less inexorable than such terms of debate seem to presume.
18
FINANCE, INSTABILITY, & CYCLES 19
Even if one grants the postulates of spot markets for all goods that always clear
and perfect foresight regarding all future spot prices, the standard results on the
stability of stationary competitive equilibrium need not result if one modifies
another crucial assumption of the new classical models that has received some-
what less attention-the existence of perfect financial markets.
Consider here an infinite horizon competitive equilibrium model that is ortho-
dox in all respects, except for the introduction of certain financial constraints.
These constraints seem to be reasonably descriptive of financial arrangements in
complex modern private enterprise economies, and they are closely related to
specifications that have often figured in non-neoclassical business cycle theories.
In this model, the stationary competitive equiliibrium may be unstable, and the
perfect foresight equilibrium dynamics may converge instead to equilibrium
cycles or to a chaotic attractor. This possibility depends neither upon multiple
production sectors nor upon a high rate of time discount, as in the examples of
equilibrium cycles discussed above.
Even when the stationary competitive equilibrium is not unstable, perfect
foresight equilibrium frequently is indeterminate for the model, i.e., for given
initial conditions, a continuum of nearby equilibria exists. This indeterminacy-
which never occurs in optimal growth models with perfect financial markets 1-
can also be regarded as a type of intrinsic instability of the competitive process.
Since different dynamic paths for the economy are equally consistent with fulfill-
ment of expectations, arbitrary events can cause agents' expectations to jump
from one such path to another. Change in expectations is then self-fulfilling. In
such circumstances, stationary stochastic fluctuations are possible in which
autonomous shifts in "animal spirits" constantly occur and cause fluctuations in
the rate of production and accumulation, but in which the economy is nonetheless
always in a "rational expectations equilibrium. " 2
An Economy with Concentrated Ownership of

Capital and No External Finance
The economy to be considered here is a stationary infinite horizon economy with

a one-sector production technology, infinite lived agents, endogenous labor sup-
ply, and competitive spot markets for all goods (i.e., for labor and for the single
produced good). It is thus like the sort of economy investigated by Kydland and
Prescott (1980), except that the existence of stochastic shocks to the production
technology are not assumed, but imperfections in the financial markets are as-
sumed.
The kind of financial market imperfections of primary interest are restrictions
upon the ability of economic units (households or firms) to finance expenditures
other than out of their own current or past income. That is, there is no assumption
of the existence of a single, economy-wide competitive market for claims to
future income (including claims to income contingent upon various possible
20 MICHAEL WOODFORD
future events, as in the Arrow-Debreu framework) to which all agents have equal
access. This will allow the marginal tradeoffs between present and future goods
to be different for different economic units at a given point in time. Cyclical
variation in the discrepancy between the marginal tradeoffs of different units in
the economy is crucial to the cycle models to be developed below. In the present
section, for simplicity, the extreme case of an absolute inability of any economic
unit to obtain external finance at all is considered. This means that no financial
assets are exchanged at all, so that each unit must spend an amount equal to its
income in every period.
In order for borrowing constraints to have any effect upon economic dynam-
ics, of course, it is necessary to assume that agents are not all identical. At certain
times some agents would in fact wish to borrow on terms at which other agents
would be willing to lend. Accordingly, instead of assuming a single representa-
tive infinite lived agent who both supplies labor and owns the capital stock, as in
standard neoclassical growth models, assume that two distinct types of infinite
lived agents exist, one for each of these two roles. Workers are assumed not to
accumulate capital goods, and furthermore, the two groups are assumed not to be
able to borrow from or lend to one another.
While there are other ways in which one might introduce a differentiation
among agents that would allow borrowing constraints to bind, this particular
differentiation seems particularly relevant to the organization of production in
modern capitalist economies. Such a differentiation of roles figures prominently
in many non-neoclassical business cycle theories. (Kalecki's is perhaps the most
obvious example, but many less formal discussions also treat separately the
decisions of capitalists and workers, or producers and consumers.) The constraint
upon financial intermediation assumed here is given less attention in traditional
accounts but is implicit in the treatment of the decisions of the two classes of
agents by authors such as Kalecki. If workers lend to capitalists through a
complete and fully competitive set of financial markets, they can effectively
accumulate capital themselves, so that the distinction between the two roles is
negated. (It is shown in the next section that some limited lending by workers to
capitalists does not much alter the dynamics predicted by the model. As the limit
upon such lending is progressively relaxed, however, one approaches an economy
which is equivalent to one with a single representative agent.) Furthermore, the
absence of borrowing or lending between the two groups implies that each
group's expenditures in a given period are constrained by the income of that
group. Some such financial constraint is thus implicit in all models which assume
that workers consume exactly their wages each period, and that capitalists accu-
mulate out of their profits. (Again, Kalecki is explicit about this, but many other
studies in economic dynamics have also assumed a link of this sort between
income distribution and the division of total expenditures into consumption and
investment.)
Various explanations are possible of the assumption that workers do not
accumulate capital and that no lending occurs between the two groups. One might
assume that financial institutions treat all agents alike, but they lend only to
borrowers with certain kinds of collateral. Capital goods are obviously much
more transferable than is labor power, for which reason the latter sort of
"wealth" makes poor collateral. If borrowing against future labor income is
impossible, it need only be shown that in equilibrium workers prefer to borrow
rather than save (at the rate of return earned by capitalists) in order to explain
both why workers own no capital (and hence cannot borrow), and why capitalists
do not borrow from workers. This, in turn, requires only the assumption that
workers discount the future more than capitalists.
Under yet another interpretation, concentrated ownership of means of produc-
tion results from the fact that only certain agents possess the knowledge required
to recognize profitable production opportunities and organize production appro-
priately. Wage earners would like to lend to these entrepreneurs (at the rate of
return that entrepreneurs receive upon their own capital), but they are unable to
effectively monitor the uses to which the funds are put or the amount of profits
actually earned. Hence they do not lend, and entrepreneurs' accumulation of
capital is constrained by their profits. One needs only to show then that in
equilibrium, wage earners do not wish to borrow from entrepreneurs; this occurs
if wage earners discount the future less than do entrepreneurs.
This interpretation is particularly reminiscent of Kalecki' s theory, in which the
macroeconomic consequence of income distribution that is most emphasized is
the effect of profits upon the quantity of entrepreneurial capital and hence upon
the amount of investment undertaken. The similarity is even greater when it is
shown, in the following section, that even if wage earners do lend to entrepre-
neurs, it can be assumed that there is a limit to permissible asset-liability ratios
(owing to the "principle of increasing risk") in order to obtain similar results.
However, the simpler case of an economy in which no lending occurs is treated
first.
To be specific, consider an economy in which there are two types of infinite
lived agents. Workers seek to maximize
00
E /- 1 [u(ct) - v(nt)]
t=l
where Ct is consumption in period t and nt is labor supply. Here 'Y is a discount

factor between zero and one, and u and v are twice differentiable functions
satisfying:
(i) U I > 0, U II < 0, VI > 0, VII > 0,

(ii) v '(0) = 0, v '(n) - oo for large enough n, and
(iii) u '(c) + cu (c) > 0.
11
Assumptions (i) are standard monotonicity and concavity assumptions. Assump-

22 MICHAEL WOODFORD
tions (ii) are stronger than necessary, but together with (i) insure that the single-
period problem
Max u(c)-v(n) s. t. c wn
c, n
has an optimal solution c,n > 0, for any positive real wage w. Let the labor
supply that solves this problem be denoted n = s(w). Assumption (iii) then is
necessary and sufficient for s(w) to be a monotonically increasing function, i.e., it
is the condition that leisure and consumption be "gross substitutes."
If workers neither lend to nor borrow from capitalists and cannot accumulate
capital goods themselves (presumably because of economies of scale at low levels
of production), then in fact Ct = Wtnt will be workers' budget constraint each
period, and so optimizing labor supply will be nt = s(wt), and consumption
demand will be Ct = Wts(wt). Accordingly, the supply curve s(w) summarizes all
aspects of workers' behavior that are relevant for the present model.
Suppose that there exists a constant returns to scale one-sector production
technology (at least, with constant returns over the range of variation that actually
occurs over a business cycle). The relevant aspects of this technology can be
summarized by specifying r(w), the level of quasi rents per unit of capital (assum-
ing employment of an optimal quantity of labor inputs) as a function of the real
wage. The demand for labor (per unit of capital goods held by capitalists) will
then be d(w) = - r '(w). It follows that for any quantity k > 0 of capital goods in
existence per worker, the market-clearing real wage will be the solution to:
(1) kd(w) = s(w).
If r is assumed to be a decreasing convex function of w (as follows from profit

maximization subject to a convex production set), then d(w) is a decreasing
function ofw, and a unique solution w(k) to (1) exists. Also note that w(k) will be
an increasing function of k.
Finally, suppose that capitalists seek to maximize:
00
E ,Bt-ilog qt
t=l
where qt is consumption in period t, and (3 is a discount factor between zero and

one. These preferences are chosen to simplify the calculations; they imply a
constant savings ratio (3 for capitalists, regardless of their exectations regarding
the future rate of profit. In the case that (3 is close to one, i.e., capitalists do not
discount the future very much, this assumption is probably not too restrictive. (In
any event, it is expected that the capitalists' savings ratio should never fall too far
below one.)
Without loss of generality, in the case of a one-sector technology, complete
depreciation of capital in each period's production is also assumed. Then each

period's capital stock can be identified with capitalist saving in the previous
period to obtain:
(2)
This gives each period's capital stock as a function of the previous period's capital
stock. Analysis of the possible equilibrium paths of capital accumulation then
reduces to the analysis of the properties of a one-dimensional map.
If r(O) > {3- 1 is assumed, as is necessary in order for the capital stock not to go
to zero asymptotically, we find that there exists a unique steady-state capital stock
k* > 0, such that if the initial capital stock per worker equals k*, (2) yields kt =
k* forever. However, this steady- state equilibrium need not be stable. If the right
hand side of (2) is written f(kt), then a necessary and sufficient condition for
instability is that f'(k*) < -1. (It can be shown that the assumptions above
preclude the case f'(k*) > 1.) In terms of the fundamental relationships of the
model, this condition can be expressed as follows.
Proposition 1. Let es.ea denote the elasticities of supply and demand for labor
respectively, evaluated at the steady-state equilibrium, and let sn, sk represent the
income shares oflabor and capital, respectively, also evaluated at the steady-state.
Then the steady-state is locally unstable if and only if
(3)
If condition (3) holds, then for almost all initial capital stocks k 1 , the equilibrium
capital stock fluctuates forever within a bounded interval [O,k+], without ever
converging.
The case described in Proposition 1 is depicted in Figure 1. The arrows
indicate the sequence of capital stocks (k 1 , k 2 , k 3 , k 4 , k 5 , k 6 ) for an initial stock k 1
only slightly less thank*. It is clear that kt can quickly diverge quite far from k*,
and that the path of kt in equilibrium may be quite erratic.
It is important to note the role of the financial constraint in allowing this kind
of instability to occur. In a standard one-sector optimal growth model, kt+l is
necessarily a monotonically increasing function of kt, and this (together with the
fact that f must cross the 45 ° line only once and from above) insures global
stability of the steady-state. This is because when the same "representative
agent" receives both wages and the quasi rents to capital goods, his income
equals total output, necessarily an increasing function of the capital stock. Then,
assuming that future consumption is a ''normal good,'' saving will be an increas-
ing function of the capital stock as well. In our model, on the other hand, capital
accumulation depends upon capitalists' saving, which depends upon capitalists'
income. Even though a larger capital stock must mean a larger total output,
capitalists' incomes may be smaller, if wages rise rapidly enough as the capital
24 MICHAEL WOODFORD
Figure 1
k1+1
45°
/
/
/
/I
/ I
/
I
I
/I I
I I
/ I I
/ I I
I I
I
I
/ I
k* k+ kt
ks ~ ka k, k2 ~
stock is increased. This explains why f may be decreasing over some values of k,
and why sufficiently low elasticities of both supply of and demand for labor are
necesssary for instability.
Proposition 1 does not describe the asymptotic behavior of the capital stock
(and hence of production), except to say that it does not converge to a constant
value; it may converge to regular periodic fluctuations, or it may be forever
aperiodic. It is particularly interesting to note the possibility of strongly chaotic
dynamics, by which we mean that:
(i) the map f has at most a countable number of periodic points and all of its
periodic points are unstable; and
(ii) for almost all initial capital stocks k 1 , there exists a probability measureµ
on [O,k +]which is absolutely continuous with respect to Lebesgue measure (i.e.,

the measure µ. can be described in terms of a density function g, so that
µ.(A) = \A g(k)dk for A any measurable set), and which describes the asymptotic
frequency distribution for kt (i.e., for A any measurable set with positive mea-
sure, the asymptotic frequency with which kt E A is equal to µ.[A]).
In such a case, the set of initial conditions in [O,k+] that result in bounded but
asymptotically aperiodic (i.e., chaotic) trajectories is of full measure. Hence the
chaotic equilibrium dynamics should be "observable.' ' 4 Furthermore, for almost
all initial conditions, the equilibrium dynamics will "appear stochastic" in the
sense that the long-run frequency distribution for kt is given by an absolutely
continuous measure.
Li and Yorke (1978) show that a sufficient condition for strongly chaotic
dynamics is a map f that (i) maps a bounded, closed interval into itself, (ii) is
piecewise C 2 , and (iii) has If' I > }.. > 1 at all points where the derivative is
defined. The following result gives sufficient conditions for this to occur in our
model.
Proposition 2. Suppose that s(O) > 0 and r '(O) > - oo (i.e., d(O) < oo ). In
addition, suppose that there exists a wage w+ > 0, such that
es(w) + ed(w) - r '(w)
<
w r(w) + {3- 1
for all 0 < w ::::; w+, and such that the limit of the left-hand side as w approaches
zero is defined and less than the limit of the right-hand side. Finally, suppose that
s(O) r'(O)
Then the equilibrium dynamics for the capital stock described by (2) are strongly
chaotic.
This proposition is proved in section 3 of Woodford (1988). Briefly, these
conditions insure that for all k less than a critical positive value, w(k) = 0, so that
fbecomes a straight line with slope {3r(O) > 1. In this region, there is unemploy-
ment because of the existence of insufficient capital goods. For values of k above
the critical value, the wage is positive, and it rises so fast with increases ink that f
is decreasing with a slope steeper than -1, at least for k no larger than k+.
The existence of strongly chaotic dynamics of this sort would not, of course,
require that wages ever fall to zero if it were supposed that the supply of labor
becomes perfectly elastic at some wage w > 0. Then the initial straight segment
off would have a slope {3r(w), but this would still have to be greater than one, if
the capital stock is not to approach zero asymptotically. Hence under conditions
similar to those described in Proposition 2, strongly chaotic dynamics are ob-
tained.
26 MICHAEL WOODFORD
Consequences of Limited External Finance
The above model is doubtless too extreme in assuming that all economic units
must spend exactly their income. In fact, while internal finance remains the most
important source of funds for firms even in an economy with such well-developed
financial markets as the United States, there is some reliance upon external
finance. Cyclical variations in the degree of external financing are one of the
kinds of regularities to be explained. In the present section, we show that similar
results can be obtained even when outsiders lend to firms, as long as there
continue to be limitations upon the degree to which firms can seek external
finance.
In the present section, the production technology remains as before, except
that firm-specific productivity shocks that cancel out in the aggregate are intro-
duced. To be precise, suppose that when an entrepreneur chooses to accumulate
capital goods in the quantity kt+ 1 , the number of effective capital goods with
which he is able to produce in period t + 1 will be akt+i. where a is a random
variable whose value is not known at the time of the accumulation decision. The
profit function r(w) now describes the (deterministic) production possibilities per
unit of effective capital. If a continuum of identical entrepreneurs with indepen-
dent drawings of a is assumed and (without loss of generality) one assumes that
E(a) = 1, then the aggregate production possibilities are deterministic and
identical to those assumed in the first section. (Agents who are able to organize
production will be referred to in this section as entrepreneurs to clarify the fact
that they are not the only agents who earn non-wage income.)
A competitive market for riskless debt to which all agents have equal access is
introduced. Because the firm-specific productivity shocks are private informa-
tion, debt contracts contingent upon these shocks are not enforceable. Hence
(since there are no aggregate shocks), the only kind of debt that will be issued is
straight debt, the promise to pay a fixed amount at a future date. It is assumed that
the penalties for default are such that debtors will repay whenever they have the
funds required for repayment.
Let entrepreneurial equity et refer to the net worth of an entrepreneur in period
t, after sale of the output produced in period t, payment of wages, and repayment
of any debt issued to finance that production. The entrepreneur chooses a level of
consumption qt, and a level of capital stock to accumulate kt+i; this will imply
issuing debt in the quantity kt+l + qt - et. When at+l is realized, l)is effective
capital holdings will be at+ikt+i. and he will earn quasi rents of at+ikt+ 1 r(wt+ 1).
If the real interest rate on one-period debt issued in period t is it+i, entrepreneur-
ial equity in period t+ 1 will then be
It can be shown that optimizing behavior for an entrepreneur with equity et,
facing an interest rate of it+ 1 , having perfect foresight regarding the quasi rents to
capital rt+l = r(wt+i), and knowing the distribution G from which at+l is drawn,
is to choose
(4)
where 8(z) denotes the solution to
(5) az - 1
E[ ] =0
(az - 1)8 +1
Equation (5) has a unique solution 8(z) ~ 0 for all 1 ~ z < a -l, where!!... is the
lower bound of the support of G (possibly zero). This solution function is easily
shown to be monotonically increasing in z; 8(1) = 0, while 8(z) - ex> as z - ~- 1 .
Finally, 8 is greater or less than one according to whether z is greater or less than
E(a- 1).
The function 8(z) indicates how the degree of leverage that entrepreneurs are
willing to undertake depends upon the ratio z = rt+ 1 /it+ 1 , i.e., upon the amount
by which the expected rate of profit exceeds the rate of interest at which they can
borrow. The ratio 8 is finite even when rt+l > it+i. because of the risk of a low
realization of CTt+ 1 , against which the entrepreneur is unable to insure because the
realization is private information. The argument is essentially Kalecki's (1939)
"principle of increasing risk. " 5 Dependence of the sort indicated in (4) of
investment upon the amount by which r exceeds i is familiar both from Kalecki's
models of aggregate dynamics and those of authors such as Foley (1986) and
Franke and Semmler (1986). For these purposes, the most important thing to
notice about (4) is that the level of available internal funds continues to be a major
determinant of capital accumulation, even when there exists a competitive market
for one-period debt, and even when 8 (a measure of the degree to which external
finance is used) is well above one. Because of this, the same sort of "profit
squeeze" mechanism described previously makes possible endogenous cycles
and chaos.
We will suppose that workers have the same preferences as before, but that
they may hold the debt of entrepreneurs (when 8 > 1) or borrow from entrepre-
neurs to finance greater current consumption (when 8 < 1). Because we now
allow Ct * wtnt, labor supply will not, in general, depend only upon the current
real wage. However, in the special case that u(c) = c, nt = s(wt) can be written,
even though consumption demand will no longer equal Wts(wt). In this case,
workers' demand for riskless bonds is perfectly elastic at the interest rate qt+l =
'Y- 1 . The investment equation (4) then becomes
28 MICHAEL WOODFORD
Because the functions() and ware monotonically increasing, and the function r is
monotonically decreasing, equation (6) has a unique solution kt+l = h(et), that is
monotonically increasing in et. This function indicates the way in which the
availability of internal funds continues to determine the level of capital outlays,
even when a competitive market for riskless debt exists.
It follows that the level of quasi rents per unit of effective capital in period t + I
is given by r(et) = r {w[h( et)]}, so that aggregate entrepreneurial equity per
worker, after repayment of debt, equals
Because each entrepreneur's decision (4) is linear in his equity, one need only
keep track of all the macroeconomic variables of interest. Hence, the analysis of
asymptotic dynamics reduces again to the analysis of a one-dimensional map.
The map f defined in (7) can exhibit behavior like that depicted in Figure I.
The main difference is that here it can be shown that f(e) > -y- 1{3e for all e > 0.
This is because, after consuming (I-met, an entrepreneur could obtain a certain
level of equity et+l = 'Y - 1 f3et by simply lending the remainder. If instead he
chooses to invest in risky production, it must be because the distribution of
possible values for et+l so obtained has a mean that is greater than this. Hence it
follows that f cannot be bounded above, for all e > 0. Nonetheless, attention can
be restricted to a bounded interval, as before.
Assume that {3 > 'Y, since it is obvious that otherwise et increases forever, until
eventually entrepreneurs no longer have any need of external finance. Then it can
be shown that
lim f(e)/e = -y- 1{3.

e - oo
It follows that there exists a bounded interval E = [O, e], such that fmaps E into
itself, and such that et eventually enters and remains forever within this interval.
Hence, we can again restrict our attention to the asymptotic dynamics within a
bounded interval.
If we assume again that the labor supply function s(w) is both very elastic at a
sufficiently low real wage (possibly zero) and very inelastic at a sufficiently high
wage, then the map f defined by (7) has the following general shape:
• for low levels of et, f is increasing with a slope > I (wages rise little as labor
demand expands, so quasi rents remain high)
• for higher levels of et, f is decreasing, possibly with a slope even < - I
(rapidly rising wages in response to increased labor demand mean that aggregate
quasi rents fall as the capital stock rises)

• for the highest levels of er. f is again increasing, with a slope approaching
{3-y- 1 , which is < 1 (expected quasi rents are so low that higher er no longer
means much of an increase in investment, the excess internal funds simply being
lent out at the interest rate -y- 1).
The invariant interval E may or may not include values of e in all three of the
above regions. It is evident that the steady-state must lie within the second region,
and that its stability depends upon whether or not f' < -1 at the steady-state. The
conditions under which this occurs are as follows:
Proposition 3. The steady-state equilibrium of an economy with limited exter-
nal finance of the kind described above is unstable if and only if
(8)
where eo is defined as
zO'(z)
eo -
O(z)
and z = r/i, and where all quantities are measured at the steady-state. When
(8) holds, er remains asymptotically within the bounded interval E, but never
converges to any constant value.
Proposition 3 gives sufficient conditions for self-sustaining endogenous fluc-
tuations to occur, although it does not indicate whether they are asymptotically
periodic or chaotic. Either is possible. In particular, it is possible in this case, as
in the previous section, to describe ranges of parameter values in which the
dynamics are strongly chaotic. This is possible, in this case as before, even when
the discount factors (3 and -y are chosen arbitrarily close to 1. Here, however, we
consider only the existence of self-sustaining fluctuations more generally.
It is useful to note that (8) reduces to (3) in the case that O(z) = 1 for all z. The
results here, then, are continuous with those of the previous section. It is also
clear from this that self-sustaining fluctuations will be possible at least in the case
of 0 sufficiently insensitive to z. In addition, note that (8) certainly does not hold
in the case that O(z) is very elastic (i.e., eo is very large). This indicates that, if
risk does not increase much with leverage so that a small change in z will induce a
large change in desired leverage, instability is impossible. Thus, the case of
perfect financial markets are approached in the appropriate limit as well. Finally,
it is interesting to note that for a given value of eo, a higher value of 0 at the
steady-state makes instability more likely. The mere existence of external finance
30 MICHAEL WOODFORD
does not mean that instability of the sort displayed in the previous section cannot
occur. Quite to the contrary-unless the desired degree of leverage is very sensi-
tive to the ratio z, a higher degree of leverage is associated with greater instabil-
ity. This latter conclusion is reminiscent of the conclusions of Franke and
Semmler ( 1986).
Consequences of Monetary Exchange
In this section, an even wider range of dynamical phenomena is demonstrated to

be consistent with perfect foresight equilibrium if an additional financial con-
straint representing the transactions role of money is introduced. In particular, it
will be shown here that it is possible to obtain quasi-periodic motion on an
invariant curve, and that it is possible for equilibrium to be radically indetermi-
nate.
Assume that money must be used in certain transactions. The introduction of a
cash-in-advance constraint, as advocated by Clower (1967), means that certain
incomes cannot be spent until the period after that in which they are earned. The
period is a time lag associated with the payments mechanism and is presumably
rather short. In particular, assume that wages are paid in money and that workers
must pay for the goods they consume. Therefore, the consumption expenditures
of workers each period are constrained by the money they hold at the beginning of
the period, representing the previous period's wages. Capitalists, on the other
hand, are assumed not to hold money as they can spend their profits in the period
in which they are earned; this might result either because producers allow each
other trade credit, settling accounts only at the end of each period, or because
banks extend within-period credit to agents with capital goods to pledge as
collateral. The cash-in-advance constraint for workers allows us to develop a
theory of the movement of money prices over the business cycle. The fact that
capitalists are not cash constrained means that price level movements are caused
by shifts in the desired level of investment as a fraction of total output, as in the
neo-Keynesian model of Marglin (1984). In this model, however, the rate of
investment desired by capitalists is not arbitrary, but represents optimizing be-
havior given perfect foresight. 6
Workers are assumed to have the same preferences as in the first section and to
face a sequence of period-by-period cash constraints
(9)
Here Mt represents the money holdings of workers at the beginning of period t,

which evolve according to the relation
where Pt is the money price of the good in period t and wt is the money wage. For
one or another of the reasons discussed above, workers do not save either by
purchasing capital goods themselves or lending to capitalists. 7 And as all of the
equilibria of interest here remain close to the stationary competitive equilibrium
(in which the price level is constant, so that the real return upon money holdings
is zero, whereas the rate of return required in order to induce workers to save is
'Y-l - 1 > 0), they are unwilling to save by holding money either. Hence workers
spend their entire beginning-of-period money balances each period, and con-
straint (9) always binds. It follows that PtCt = Mt and Mt+l = wtnt. Hence in
each period, workers choose a labor supply nt to maximize
subject to the budget constraint
The assumption of an infinite lived representative worker thus again turns out
not to matter. The consumption and labor supply decisions of workers are in fact
identical to those that would be made by overlapping generations of two-period-
lived workers, each of whom works only in the first period of life and consumes
only in the second, and who holds his savings in the form of money balances.
Thus this model provides a reinterpretation of the results regarding the possibility
of endogenous fluctuations in overlapping generations models obtained by
Grandmont, among others. 8 These models are sometimes criticized because the
mechanisms illustrated can only explain fluctuations that occur on time scales of
approximately the lifespan of an economic agent, so that these are not "business
cycle" models at all. The reinterpretation provided here answers this criticism. If
one interprets these models as models with finance constrained agents, then it is
plain that the time scale on which fluctuations occur has nothing to do with
demography.
Capitalists are also assumed to have the same preferences as in the first
section. Again it follows that they wish to save a constant fraction {3 of their net
worth, so that
(10)
(11)
With the assumption that the expected return on capital always exceeds that of
money, capitalists never wish to accumulate money rather than capital. Since in
the stationary equilibrium, the real return on capital is {3- 1 > 1 from (11), this
will be true for all equilibria close enough to the stationary equilibrium.
32 MICHAEL WOODFORD
Assume a fixed coefficients technology in which m units of period t labor are

combined with each unit of capital (period t -1 produced good) to produce a units
of period t produced good, which may be either consumed or accumulated for use
in period t + 1. Given this technology, labor demand is always
Also assume a{3 > 1. This is necessary in order for capitalists to be willing to
maintain a positive quantity of capital in a stationary equilibrium.
It is now possible to derive the complete set of conditions for a perfect
foresight equilibrium. Suppose, at the beginning of period one, capitalists own a
capital stock k 1 > 0, and workers hold the entire stock of outside money M > 0.
Given that capitalists never hold money, workers must end up holding the same
stock of outside money M at the beginning of every period. This implies that in
equilibrium
(13)
in all periods. Because of (10)-(11), total purchases by capitalists in period twill

equal 1/(3 kt+t so that the condition for goods market clearing in period t is
(14) M I
+ - kt+l = akt.
Pt f3
Finally, substitution of (12)-(14) into the first-order conditions for optimal labor
supply yields
(15)
f3
where V(n) nv '(n) and U(c) = cu '(c). This is an equilibrium condition
involving the evolution of the capital stock only. Any sequence of values for kt, t
= 2,3, ... , that satisfy (15) for given k 1 constitutes a perfect foresight equilibri-
um. For the given kt sequence, it is possible to construct unique sequences for the
price level, the wage, etc., that satisfy all of the above equilibrium conditions. (It
is necessary also that certain inequalities hold at all times, which are suppressed
here in the interest of brevity. Specifically, as noted above, the rate of return on
capital must exceed that on money, and workers must not wish to save money.
Again, these inequalities do in fact hold for all of the capital stock trajectories
near the stationary equilibrium that are studied below.) Analysis of the stability of
stationary equilibrium and of the existence of endogenous equilibrium cycles
simply requires an analysis of the solutions to the difference equation (15).
Given the above assumptions on preferences and technology, one easily estab-
lishes that there exists a unique steady-state capital stock k*. The perfect foresight
equilibrium dynamics in the neighborhood of the steady-state may be character-
ized through an examination of the linearization of (15) about the steady-state,
i.e.,
(16) kt+2 - k*] = [a/1 - /1E] rkt+l -k*l

[
kt+l -k* I 0 Lkt -k*J
where
n*v '(n*) + n* 2v "(n*)

E=--------
k*[u'(c*) + c*u"(c*)]
and n* and c* are the steady-state values of nt and Ct, respectively. Note one can
write
1 1
E = (a - - )(- + 1)
11 e
where e is the elasticity of labor supply with respect to wtf Pt+ 1 , evaluated at the
equilibrium steady-state. Assumption (iii) on the preferences of workers then
implies that e > 0.
Focusing on the case in which a/j > 1 and e > 0, it is easily shown that the
eigenvalues of the matrix in (16) are of the following sort:
(I) if (a/j/2) 2 > /jE and /jE > I, there are two real eigenvalues between zero
and one.
(2) if (a(3/2) 2 < (3E < 1, there are two complex eigenvalues of modulus less
than one.
(3) if (a/j/2) 2 < /jE and /1E > 1, there are two complex eigenvalues of
modulus greater than one.
(4) if (a/j/2) 2 > /1E > 1, there are two real eigenvalues both greater than one.
These various regions are displayed geometrically in Figure 2.
The following result is then a simple consequence of the stable manifold
theorem:
Proposition 4. In the case that the elasticity of the labor supply curve e satisfies
(17) 2 - a/j
e(--) > 1
a/j - 1
34 MICHAEL WOODFORD
then for each initial condition k 1 in a certain neighborhood of k*, there is a one-
dimensional continuum of perfect foresight equilibria all converging asymptoti-
cally to the stationary equilibrium. Thus, in this case, perfect foresight equilibri-
um is indeterminate. On the other hand, if the inequality in ( 17) is reversed, there
exists no perfect foresight equilibrium with kt always within a neighborhood of
k* that converges asymptotically to the stationary equilibrium, unless k 1 = k*
exactly. In this case, the stationary equilibrium is unstable. Note that either case is
consistent with the general assumptions made above; in fact, as shown in Figure
2, all four cases discussed in the previous paragraph occur in non-empty regions
of parameter space.
It is also interesting that for no parameter values does the equilibrium steady-
state have the "saddle path stability" property present in optimal growth models
in the case of a sufficiently low rate of time discount. 9 Hence, no matter what
parameter values are chosen, some type of macroeconomic instability is implied.
In the case that (17) does not hold, perfect foresight equilibrium cannot converge
to the equilibrium steady-state for generic initial conditions. The asymptotic
behavior of such perfect foresight equilibria cannot be characterized in general,
but in some cases, there will exist perfect foresight equilibria converging asymp-
totically to periodic or quasi-periodic motion on an invariant circle. The follow-
ing is an immediate consequence of the Hopf bifurcation theorem for maps:
Proposition 5. Consider a one-parameter family of economies indexed by a
parameterµ,, and suppose that a, {3, and e are all C 1 functions ofµ,, with 0 < a- 1
< (3 < 1 and e > 0 for allµ,. Suppose further that there exists aµ,* such that
{3(µ,*)E(µ*) = 1
{3'(µ*)E(µ,*) + {3(µ*)E'(µ,*) =;; 0.
Then there exists a continuous function µ(a) and a continuous family of closed
curves r(a), for a E (0, a), SUCh that as a - 0, µ,(a) - µ,*and r(a) shrinks to the
equilibrium steady-state, and such that for each a E (0, a), r(a) is an invariant
circle of the perfect foresight equilibrium dynamics for the economy indexed by
µ,(a).
When a certain regularity condition is achieved, 10 Proposition 5 implies that
there exist invariant circles for the perfect foresight equilibrium dynamics, either
for allµ in a left neighborhood ofµ*, or for allµ, in a right neighborhhood ofµ,*.
When the invariant circles near the steady-state exist for {3E < 1, only perfect
foresight paths that begin on the invariant circle remain on it, whereas when the
invariant circles exist for {3E > 1, all perfect foresight equilibrium paths begin-
ning close enough to the invariant circle converge to it asymptotically. Both cases
are possible. 11 In either case, there exist perfect foresight equilibria that asymp-
totically approach the invariant circle for all k 1 in a neighborhood of k*.
Clearly, the situation described in Proposition 5 is possible. The line labeled
Figure 2
8/3
both unstable
complex
I
I
I
I
I
I
I
I
I
/3E = 1 in Figure 2 indicates the points that satisfy the first condition, and almost
all of the points must also satisfy the second condition. Note also that it is possible
to satisfy these conditions with /3(µ *) arbitrarily close to 1. Thus in economies of
the kind studied here, as above, equilibrium cycles are possible even when the
agents who determine the rate of capital accumulation have an extremely low rate
of time discount.
Notes
l. See Woodford (1987, section 1).
2. On the general connection between indeterminacy of perfect foresight equilibrium
and the existence of stationary rational expectations equilibria in which stochastic fluctu-
ations occur that are unrelated to any shifts in tastes or technical possibilities, see Farmer
and Woodford (1984), Woodford (1984, section 4; 1986). For a general discussion of the
role of imperfect financial intermediation in allowing equilibria of this kind, see Woodford
(1987).
36 MICHAEL WOODFORD
3. For details of the argument, see Woodford (1987, section 2.B).

4. Thus the criticism made by Grandmont ( 1985, l 026-7) of some earlier examples
of chaotic dynamics in economic models will be invalid in such a case.
5. Keynes's (1936) discussion of "borrower's risk" refers to essentially the same
kind of considerations. For an exposition, see Minsky (1975, 106-16). Greenwald and
Stiglitz (1985) provide alternative theoretic foundations for such a relationship.
6. This does not, however, mean that there is no role for autonomous shifts in the
expectations of capitalists in the causation of aggregate fluctuations; the stochastic equilib-
ria referred to in note 2 can be interpreted as equilibria in which autonomous shifts in
"animal spirits" cause aggregate fluctuations.
7. Here absolute impossibility of external finance is assumed, as in the first section.
However, the results of this section can easily be generalized to the case of an endogenous
leverage ratio, as in the next section. Again, one finds that a high leverage ratio 0-as long
as the desired leverage ratio is not very elastic-makes it even easier to produce cycles.
8. For an example of a cash-in-advance economy that yields equilibrium conditions
formally identical to those of Grandmont (1985), see Woodford (1987, section 2.A). The
production economy treated here yields equilibrium conditions closely related to those of
the overlapping generations economy studied by Reichlin ( 1986).
9. See, for example, Magill (1977) and Scheinkman (1976).
10. The condition is that a certain expression involving a,m, /3 and the derivitaves ofu
and v be non-zero. See Guckenheimer and Holmes (1983, 162-3). The condition plainly is
valid for most parameter configurations on the locus of Hopf bifurcation shown in
Figure 2.
11. Reichlin ( 1986) shows that both cases are possible for an overlapping generations
model with production whose equilibrium conditions correspond to the /3 = 1 limit of
equation (15). By continuity, both cases are also possible in the case of /3 sufficiently
close to 1.
References
Benhabib, J., and K. Nishimura. 1979. "The Hopf Bifurcation and the Existence and
Stability of Closed Orbits in Multisector Models of Optimum Economic Growth."
Journal of Economic Theory 21:421-44.
- - - - . 1985. "Competitive Equilibrium Cycles." Journal of Economic Theory
35:284-306.
Boldrin, M., and L. Montrucchio. 1986. "On the Indeterminacy of Capital Accumulation
Paths.'' Journal of Economic Theory 40:26-39.
Clower, R. W. 1967. ''A Reconsideration of the Microfoundations of Monetary Theory.''
Dechert, D. W. 1984. "Does Optimal Growth Preclude Chaos? A Theorem On Monotoni-
city." Z. Nationalokonom. 44:57-61.
Deneckere, R., and S. Pelikan. 1986. "Competitive Chaos." Journal of Economic Theory
40:13-25.
Farmer, R. E. A., and M. Woodford. 1984. Self-fa/filling Prophecies and the Business
Cycle. CARESS working paper no. 84-12. University of Pennsylvania (April).
Foley, D. K. 1986. Endogenous Financial-Production Cycles in a Macroeconomic Model.
Barnard College. Mimeo.
Franke, R., and W. Semmler. 1986. Debt Financing of Firms, Stability, and Cycles in a
Dynamical Macroeconomic Growth Model, Disc. Paper No. 19, Fach. Wirtsch. Uni-
versity of Bremen, (Fall). Revised version appears in this volume.
Grandmont, J. M. 1985. "On Endogenous Competitive Business Cycles." Econometrica
53:995-1045.
Greenwald, B., and J. E. Stiglitz. 1985. Information, Finance Constraints, and Business
Fluctuations. Princeton University. Mimeo.
Guckenheimer, J. and P. Holmes. 1983. Nonlinear Oscillations, Dynamical Systems, and
Bifurcations of Vector Fields. New York: Springer-Verlag.
Kalecki, M. 1939. Essays in the Theory ofEconomic Fluctuations. New York: Russell and
Russell.
Keynes, J. M. 1936. The General Theory of Employment, Interest and Money. Reprint.
New York: Macmillan, 1986.
Kydland, F., and E. C. Prescott. 1980. "A Competitive Theory of Fluctuations and the
Feasibility and Desirability of Stabilization Policy,'' S. Fischer, ed., Rational Expecta-
tions and Economic Policy. Chicago: University of Chicago Press for NBER.
- - - . 1982. "Time to Build and Aggregate Fluctuations." Econometrica 50:
1345-70.
Li, T., and J. A. Yorke. 1978."Ergodic Transformations from an Interval into Itself."
Transactions of the American Mathematical Society. 235: 183-92.
Magill, M. 1977. "Some New Results on the Local Stability of the Process of Capital
Accumulation." Journal of Economic Theory 15:174-210.
Marglin, S. A. 1984. Growth, Distribution, and Prices. Cambridge, Mass.: Harvard
University Press.
Reichlin, P. 1986. "Equilibrium Cycles and Stabilization Policies in an Overlapping Gen-
erations Economy with Production." Journal of Economic Theory 40:89-102.
Scheinkman, J. A. 1976. "On Optimal Steady States of N-Sector growth models when
Utility is Discounted. "Journal of Economic Theory 12: 11-30.
Woodford, M. 1984. Indeterminacy of Equilibrium in the Overlapping Generations Mod-
el. Columbia University. (May). Mimeo.
- - - . 1986. "Stationary Sunspot Equilibria in a Finance Constrained Economy."
Journal of Economic Theory 40: 128-37.
- - - . 1987. "Expectations, Finance, and Aggregate Instability," M. Kohn and S.C.
Tsiang, eds. Finance Constraints, Expectations and Macroeconomics. Oxford Univer-
sity Press.
- - - . 1988. "Imperfect Financial Intermediation and Complex Dynamics," W. Bar-
nett, J. Geweke, and K. Shell, eds., Economic Complexity: Chaos, Sunspots, Bubbles,
and Nonlinearities. Cambridge University Press, forthcoming.
Debt-Financing of Firms, Stability, and
Cycles in aDynamical Macroeconomic
Growth Model
Reiner Franke and Willi Semmler
The economic effects of debt-financing of firms have recently been studied in

empirical as well as theoretical work. In empirical work on recent trends in
corporate finance, it has been stated that debt-financing has become a dominant
way of generating funds-slowly replacing self-financing and equity financing of
firms. This, at least, holds as a secular trend, but it has also become an important
way of raising funds in the business cycle (Brealey and Meyers 1984, 292;
Wolfson 1986).
In theoretical studies, such as those put forward by Minsky (1975, 1982), it is
demonstrated that high debt-asset ratios will appear as a dangerous liability
structure of firms increasing the risk of bankruptcy and developing a fragile
macroeconomic dynamics. At first glance, this might appear solely as a problem
of financial constraints. A number of earlier articles have already dealt with the
problem of liquidity and cash constraints (Clower 1967; Tsiang 1956; Kohn
1981a, 198lb). These constraints require that economic agents have the means of
payment available from previous sellings before any intended demand can be
made effective. As usually stated, however, firms can buy and sell without cash
provided that other firms (or banks) are willing to extend credit. Thus,
the problems that have subsequently been addressed are related more to credit
For helpful discussions and comments, we want to thank P. Flaschel, D. Foley,

H. Minsky, E. Nell, and L. Taylor. An early version of this paper was also presented
at a workshop on Dynamic Models at the University of Bielefeld, a conference
of the Eastern Economic Association, and at the Econometric/Macro Workshop of
the Universite de Montreal. Comments of the participants are gratefully acknowl-
edged.
38
DEBT-FINANCING OF FIRMS, STABILITY, & CYCLES 39
constraints than to liquidity constraints.

Recent models, therefore, have focused on the impact of credit constraints on
economic dynamics and macroeconomic fluctuations. Here we find two types of
models. The first one is concerned with exogenous credit constraints. It is argued
that equity financing is not a viable alternative to debt-financing and that exoge-
nous credit constraints (rather than interest rates) are the effective determinants
of economic fluctuations (Stiglitz and Weiss 1981; Greenwald, Stiglitz, and
Weiss 1984; Guttentag and Herring 1984).
A formalization of this idea can be found in Woodford (1986a, 1986b). He
designs a class of dynamic macro models in which credit constraints are built in
as a fixed upper limit of borrowing. In particular, depending upon the value of this
parameter, they can lead to nonlinear cycles.
By contrast, another strand of modeling seeks to fully endogenize debt-financ-
ing in an interdependent economic system. In Foley's seminal papers (1986,
1987), borrowing, lending, and capital outlays of firms are essentially governed
by the endogenously determined profit rate and liquidity. With no asset holders
outside the enterprise system and capitalist consumption being equal to zero, this
(nonlinear) closed model generates cyclical fluctuations in these macroeconomic
variables. On the other hand, there are capitalist firms which are externally
financed exclusively through loans and equity issuance underlying the models of
Taylor and O'Connell (1985) and Taylor (1985). The savings generated by rentier
households, as suggested in Kalecki.'s macroeconomic writings, are channeled
through the banking system to investing firms. Though these models develop a
rich macroeconomic dynamics-including financial instability as studied by
Minsky in his financial crisis theory-the role of firms' indebtedness and debt
payment commitments is not fully worked out. A high degree of indebtedness and
high debt payment commitments of firms, however, seem to be central in Min-
sky's theory of financial instability.
In this chapter, partly utilizing the asset-accounting framework put forward in
the articles just mentioned, debt-financing of firms will be explicitly taken into
account. In a fully interdependent model incorporating investing firms, savings
of rentier households, commercial banks, and the government, the endogenously
generated debt of firms (created through borrowing) is feeding back dynamically
to the investment behavior of firms, their borrowing of funds, the asset market,
the interest rate and the expected rate of return (representing the confidence of
investors with regard to future development). The impact of debt-financing of
firms on aggregate economic activity will be studied within this context.
First the model is introduced, which utilizes six key economic variables: the
interest rate, the gross and the net rate of profit (net of interest costs), the rate of
growth of the capital stock, the degree of the indebtedness of firms, and the
difference between current and expected profit rates (the "state of confidence").
Two sets of markets are considered, product markets and asset markets. Both are
supposed to be continuously in equilibrium, i.e., the economy evolves in a
40 REINER FRANKE & WILLI SEMMLER
sequence of temporary equilibria. This hypothesis, however, by no means pre-

cludes it from behaving in a cyclical manner. Though various simplifications are
introduced, the model will still exhibit considerable complexity.
The analysis itself is carried out in two steps. First, the temporary equilibria
are investigated. Under reasonable conditions, which are made explicit, it is
shown that these equilibria exist and are stable with respect to dynamical sub-
processes, so to speak, which refer to the very short-run. In this way, four
variables are obtained as implicit functions of the remaining two, which are the
degree of indebtedness and the state of confidence.
Next, the motions of these two variables are studied in the plane and standard
mathematical instruments (mainly, local stability analysis, phase diagrams, and
the Poincare-Bendixson Theorem) are utilized. On the basis of the thorough
analysis of the first step, considerable information can be inferred subsequently as
to the resulting movements of the other variables. Different assumptions regard-
ing the various reaction functions make it possible to work out essentially three
types of scenarios. Two of them lead to stability of the long-run equilibrium,
where all of the variables mentioned cease to change and the economy is uniform-
ly growing (in the first case stability is global, in the second at least local). The
third and most interesting scenario gives rise to persistent, but nevertheless
bounded, macroeconomic fluctuations of investment, economic growth, the debt-
asset ratio, the expected profit rate, the interest rate and the difference of net
profit rate and interest rate-all varying, partly with leads and lags, in a cyclical
manner.
The proofs of the various lemmata and theorems utilized in the text are
collected in an appendix. 1 Finally, it will go without saying that all functions
involved are assumed to be continuously differentiable.
The Model
Here, the full model will be set up. That is, the temporary equilibrium framework
is introduduced as well as the dynamical laws of change. The analysis is undertak-
en in the two ensuing sections.
Firms' Profits and Equities
Focusing on the flow and asset accounting of firms, the national account is
followed. Let PX be the value-added aggregated over the economy, with X the
flow of real output and P the corresponding price. Moreover, K denotes the
physical capital stock of firms, L the stock of loans that firms have received from
commercial banks, and i the current rate .of interest (interest payments are as-
sumed to be instantaneous). The gross and the net rate of profit, r8 and r,
respectively, are given by
PX - Wages PX - Wages - iL
r= r =
PK PK
Defining >-: = L/PK, they are linked together by the formula
(1) r=r+i>-
We seek to capture expectations of profitability by introducing the variable e.
It reflects the difference between the anticipated returns to holding capital and the
current profit rate. To be precise, r + e is the expected net rate of profits, and
r + e is the expected gross profit rate (according to this interpretation, expecta-
tions with respect to changes in the interest rate or the liability structure are
neglected). e may be also identified with the overall "state of confidence"
concerning the future development. As will be seen, it is to play an important role
in the dynamic formulation of the model. The demand price PK of capital, i.e.,
the capitalized value of expected earnings per unit of investment can be defined:
(2) r+e
PK= - - P
Occasionally, there will be reference to
(3) r + e
+ >-
In addition to loans, firms issue a stock of equity, E. Given K, PK, and L. the
formation of share prices p e is determined by the equation
(4)
This implies that the net worth of firms is identically zero. Moreover, as in Taylor
and O'Connell (1985, 876), bubbles in the stock market are excluded since they
do not seem central to Minsky's theory of financial instability.
With equation (3). we can readily derive
(5) r + e
PeEIPK =
which is another equation that will prove useful further below.

Table 1
Balance Sheets in the Economy
Assets Liabilities
Central Banks
High-powered F DC Deposits of commercial banks
money (interest-free bank reserves)
Commercial Banks
Bank reserves DC Do Interest-free deposits from the
public
Loans to firms L Di Interest bearing deposits from
the public
Firms
Capital stock PKK L Loans from commercial banks
(valued at the
demand price) PeE Equity
Public
Deposits with Do w Wealth
commercial
banks Di
Equity PeE
Asset Markets
The asset side of the economy and in particular the portfolio balance sheets are
outlined in Table l. In addition to firms, we have identified households of rentiers
(the public), commercial banks, and a central bank. By hypothesis, workers do
not save. Interpreting equation (4) also as a balance sheet of firms (so that their
net worth is zero), the table shows the choice of economic state variables and their
definitory relationships. 2 Cancelling assets against offsetting liabilities over the
economy as a whole immediately leads to
(6)
Desired asset holdings of rentiers are represented by functions d0 = d 0 (r + e,i),

e = e(r + e,i) comparing the different (expected) rates of return. At each point
of time rentiers wish to have their total wealth W split up into
D0 = d 0 W, Di = (1 - d0 - e)W, PeE = eW.

It will be assumed that these plans can always be realized. Of course,
e, = oelo(r + e) > 0, e; = oe/oi < 0.
As policy variables of the monetary authorities we consider <P = F/PK, the

reciprocal of the velocity of base money with respect to the capital stock, and the
credit multiplier µ > 1 which relates bank reserves to private deposits held with
the commercial banks, i.e.,
Now, taking equation (5) and substituting eW for PeE gives W/PK = (r +
e)/ei. Taking account of the equality µF = µDc = D 0 + Di = (1 - e)W, we
infer
(7) 1 - e r + e
µ¢ =
e
(if e > 0) or, equivalently,
(8) r +e
e=
On the other hand, we have (F + PKK)/PK = <P + (r + e)/i + Afrom equation

(3), which combined with (6) givesµ¢ = (1 - e)[r/J + (r + e)/i + A]. Equat-
ing this to (7) and solving for e leads to
(9) r +e
e =
r + e + i(A + <P)
Comparing (8) with (9) we conclude: provided that all asset markets are in
(stock) equilibrium (and r + e and i are distinct from zero), the firms' ratio of
indebtedness A and the policy variables µ and <P are connected by the equality
(10) (µ - 1) <P - A = 0
It is assumed that the government does not attempt to influence A and it allows
its (positive) debt F = Dc to grow at the same rate as the capital stock, so that¢
remains constant. (Incidentally, the same is assumed in Taylor and O'Connell
[1985, 880), whereas in Taylor [1985, 393] <Pis endogenously determined.) With
respect to the supply of deposits, it imposes no restrictions on commercial banks.
Consequently, because of the clearing of asset markets, µ will always adjust to
fulfill ( 10) identically.
With these additional suppositions regarding government policy, it follows
that asset markets are completely characterized by equation (9). It is analogous to
the usual LM schedule. Note, however, that the function representing house-
holds' liquidity preference, d has been eliminated. There shall be no further
0 ,
reference to it in the rest of the chapter, its role will be implicit only.
Investment and Goods Market
Planned investment of firms, I, is represented by a real function

h = h(r + e - i) which measures the firms' responses to the expected differ-
ence between net profit and interest costs (see also Taylor and O'Connell 1985,
873). The basic argument for this device is desire to take account of risk. On the
one hand, Kalecki (1937, 84-85) writes, " ... the rate of investment decision is
an increasing function of the gap between the prospective rate of profit and the
rate of interest'' and the difference between the ''prospective rate of profit, and
the rate of interest, is equal to the risk incurred."
On the other hand, it is not clear from this passage whether the gross or the net
rate of profit is meant. For this purpose, the latter is chosen. Recalling that the net
profit rate is r - i:A, it can be argued in accordance with Minsky's writings
(e.g., 1975, 87) that high indebtedness and high debt payment commitments (i:A)
make new investments riskier; firms may even have to turn to selling assets ifthe
risk of default rises due to dwindling net cash flows. It is in this limited sense that
the structure of financing enters into the investment decisions. For simplicity, a
significance of}.. beyond this point will be neglected. Nevertheless,}.. will play an
important part when the financing decision of firms is considered, which will be
captured by another function.
Employing the function h, investment demand in nominal terms is
(11) PI= h PK
Certainly, h = dhld(r
I +e i) > 0.
If PG is the excess of government expenditure over taxes, and the budget

deficit is sole~y financed by the creation of base money (cf. Tobin 1982, 178),
then PG = F obtains, which, because of the hypothesis P = constant, <P = con-
stant, implies
d(PK)ldt
PG/F = F/F = K/K = l/K
PK
Hence,
(12) PG=<t>PI
Besides the assumption of nonsaving workers, suppose that rentiers only save
and consume from their current income, i.e., from the flow of gross profits, rPK
(they are equal to immediately distributed net profits of firms plus interest
payments, which are mediated by the commercial banks). For simplicity, let the
corresponding savings propensity s be fixed throughout all adjustments
(o < s::s 1). 3
Finally, we postulate that all demands are satisfied. Then, from PC = total
consumption = wages + (1 - s)rPK, the identity PC + PI + PG = PX
= Wages + rPK, and from equation (12) we deduce (1 + <t>)PI - srPK = 0.
Thus, referring to (11) and dividing through by PK,
(13) (1 + <t>) h (r + e - i) - sr = 0
In effect, (13) allows for an IS-curve for the clearing of the commodity market
in the system.
Dynamics
Whereas in the previous subsection, the investment plans of firms are consid-
ered, it now has to be clarified how this is financed. That is, firms have to decide
whether to issue new shares and/or raise additional loans (or pare them down). In
what follows, we particularly want to concentrate on the effects of debt-financing
on economic dynamics. We make debt-financing of firms dependent on their
present liability structure, :>.., on the one hand, and the difference between the
expected gross profit rate and the rate of interest on the other.
There will be no particular problem with the dependency on }.. . Since a high
ratio of indebtedness increases the risk of bankruptcy, a firm finding itself in such
a situation will cut down its demand for additional loans, or otherwise the banks
will impose some credit constraints on it. Specifically, lenders might even reject
any further application for credit (cf. Greenwald, Stiglitz, and Weiss 1984;
Woodford 1986a). The way in which financing is influenced by a firm's rate of
return may require more discussion. The following simple argument is offered.
Facing a risky investment project, a firm compares the expected returns accruing
to it with the riskless alternative of receiving interests from deposits with a bank.
Since the former are profits net of interest costs, it will compare the expected net
profit rate r + e with the interest rate i. When raising credits to finance part of
this new investment, of prime interest to the bank is not so much its profitability,
but a sufficient margin of safety that the firm will survive. In other words, the
bank is concerned that the proceeds of the firm will be sufficient to repay the
principal and to cover the interests incurred. So, when granting a loan, the bank
will take into account the excess of the expected gross profit rate, r+ e, over
the interest rate. 4 To sum up, consider the borrowing of firms as essentially
determined, directly or indirectly, by credit constraints of banks. Nevertheless, as
will be seen, they might be very flexible. Formally, they are represented by a
function b-where b stands for (net) borrowing-which is to govern the growth
rate of debt-financing,
LIL = b(r + e - i,A.)

According to the above considerations, the signs of partial derivatives will be
br = ab!a(r + e - i) ;;:::: 0, b}._ = ab!aA. :5 0. The two arguments of the func-
tion b may well differ in their importance. There are several possibilities how
they are related to each other, i.e., to what extent b will increase or decrease if one
of them is changing. This gives rise to different scenarios, which will be studied
in the Long-Run Dynamics section. . .
Taki!lg account of P = constant, the change of A. itself is given by A.IA. = LI
L - KIK. 5 So letting g denote the rate of growth of the capital stock (cf.
equation [11)), we obtain
(14) g = h(r + e - i)
(15) ~ = A[b(r + e - i, A.) - gl

It remains to be considered how expectations regarding future profits evolve
over time. They have been represented by the parameter e which indicates the
state of confidence. It should depend on the general state of the economy, in
particular on the actual profitability of real investment and the burden of interest
payments. We suppose that e increases if the difference between the net rate of
profit and the interest rate is great, and e decreases if it is only small. Further-
more, the intensity of these changes may depend on the degree of indebtedness of
firms. Whereas the difference of net profit rate and interest rate (r-i) indicates the
liquidity position of firms (increased liquidity giving rise to an increase in e)' the
degree of indebtedness is considered as a measure of risk of default, and the rise
of A will cause e to fall. This generalizes the formalization in Taylor and
O'Connell (1985, 880). In formal terms,
(16) e = v(r - i, A.)

where Yr = av/a(r - i) > 0, V}._ = av/a}.. :5 0.
These are all the elements of the model. The constituent variables are r, r, i, g,
A., and Q· Their motion is determined by the equations (1), (9), (13)-(16) (recall
that</> was assumed to be fixed). (1) and (14) can be readily substituted, so that
there will be four equations (El)-(E4) for the four variables r, i, A, and e.
(El) r+e
e(r + e, i) - =0
r + e + i (A + </>)
(E2) (1 + </>) h(r + e - i) - s (r + i>.) = 0
(E3) >. = >. [b(r + i>. + e - i, >.) - h(r + e - i)]
(E4) e= v(r - i, A)
The evolution ofA and e is directly governed by the differential equations (E3)
and (E4). The determination of rand i is only implicit: given Aand e, they adjust
instantaneously (and simultaneously) such as to clear the commodity market and
the asset markets, i.e., such as to satisfy equations (El) and (E2). Market
equilibria being established, a subsequent-so to speak-change of A and e will
also bring about a change in the two rates of profit and interest. This, in turn,
influences the changes of A and e, etc. The study will reveal further that system
(El)-(E4) is indeed a complete and consistent description of the dynamics in the
economic model.
When both (El) and (E2) are satisfied, we shall speak of a temporary equilibri-
um (with respect to A, e given). Accordingly, (El)-(E4) generate a sequence of
e
temporary equilibria (in continuous time), and r and i will cease to change if
').. = 0 and = 0. If this should happen, the economy finds itself in a position
of long-run equilibrium or steady growth.
Analysis of Temporary Equilibria
The aim in this section is to obtain the rate of interest i and the net profit rate r as
functions ofA and e, denote them by r = R(A, e) and i = J(A, e), such that the
two market clearing equations (El) and (E2) are identically fulfilled. We proceed
in two steps. First, besides Aand e, r is also considered as given. We look for an
interest rate that brings about equilibrium on the asset markets, i.e., that estab-
lishes equation (El), and designate it i = j(>.,e, r). In the second step, with
respect to a given pair (A, e), a net profit rater is sought such that putting i =
j(>.,e ,r), equation (E2) is satisfied. If this r is designated R(A, e) and r = R(>.,e), i
= J(A, e): = j[A, e, R(A, e)] is inserted in (El) and (E2), then, by construction,
both equations will be simultaneously fulfilled.
In the course of this study, several assumptions on the behavioral functions e =
e(r + e,i) and h = h(r + e - i) will be put forward. Some of them are almost
inevitable to ensure existence of the two functions R(A, e) and J(A, e). Others will
help simplify the analysis and the exposition. In particular, they will permit some
definite, global results concerning the signs of the partial derivatives of R and J.
Step 1. Suppose that>., e, as well as rare given, 0 :5 A :5 1, r + e > 0. Define
r+e
f = €(A, e, r, i) :
r + e + i (>. + <P)
By allowing the degree of indebtedness to fluctuate between zero and one,

assume that there is a maximum risk of bankruptcy at A = 1 beyond which firms
would not extend (or lenders would not concede to) their indebtedness. Most
likely there is, as Woodford (1986a) argues, a maximum A < 1 up to which firms
(or banks) would allow the debt ratio to rise. The results would not be affected by
this in any essential way.
Given the (independent) variables>., e, and r, (El) is satisfied for those values
of i at which the two functions i - e(r + e, i) and i - E(A, e, r, i) intersect.
Existence of such a point, along with the property that the interest rate falls short
of the expected net rate of profit, is ensured by the first assumption. It is
illustrated in Figure 1.
Assumption 1. For all values of r, e with r + e > 0 the following holds:

(i) 0 < e(r + e,0) :5 1 and there is a point i such that e(r + e, i) > €(0, e, r, i)
(ii) i ~ r + e implies that e(r + e, i) < f(l, e, r, i)
Assumption 1(i) expresses the idea that rentiers are sufficiently inclined to
hold equities. According to (ii), however, this willingness appreciably declines if
the expected rate of return from shares gets close to the interest rate (this can be
explained by risk-aversion of the majority of households).
Thus we can present
Lemma 1. Let>., e. and r be given, 0 :5 >. :5 1, r + e > 0, and suppose that
Assumption 1 holds true. Then there exists a rate of interest i > 0 such that (E 1) is
satisfied and i < r + e for all these i.
As a matter of fact, if e(r + e, 0) is less than unity then there are at least two
positive interest rates-i 1 and i 2-bringing about (El) (in Figure 1 they are
marked with respect to >. = 1). So the question arises which one to choose.
Moreover, at future stages in the analysis it will become important whether at
these points e; > oE:!oi (which is the case with i 1), or whether e; < odoi (as with
i 2). This question is decided by means of the following consideration.
Let L, PK, µ, r, and e be given and fixed, and take i as an independent variable.
Referring to the behavioral functions d0 and e, the supply of deposits D 0 + Di by
rentiers can be expressed as a function of i, D = D(i) = D 0 (i) + D;(i) (in the
Figure 1
e(i, e. r, ·)
e(r + e. ·)
/ e(O, e. r, ·)
/
i,
proof of Lemma 2 it will be shown that D(i) = (r + e)[l - e(r + e, i)]PK/

[e(r + e, i)i]. By the same token, De = Dc(i) = D(i)/J.t. Now, suppose that at
some i0 , D(i0 ) < Dc(i0 ) + L = D(i0 ) /J.t + L results. In this situation, we make use
of the idea that competition among commercial banks forces some of them to
raise the interest rate they offer to households in order to attract more deposits.
Others will follow, so that there will be a general rise in the rate of interest until
D(i) = D(i)/i.t + Lis reached (for a similar argument, cf. Kohn 1981b, 864).
Note that stock market transactions of shares will have simultaneously taken
place. Thus, a very short-run process in which the interest rate is the only
adjusting variable has been sketched. It finally drives excess demands on the asset
markets to zero. Of course, the process can only work if D(i) is an increasing
function. It is just this stability condition (in a local sense) that furnishes the next
assumption. Beforehand, by relating D(i) and its derivative D '(i) to the points of
intersection i 1 and i2 in Figure 1, Lemma 2 ensures that it is indeed meaningful.
Lemma 2. Let L, PK, J.t, r, and e be given (0 ::5 L/PK = >. ::5 1, r + e > 0,
J.t = (A + </>)!</>,i.e., J.t satisfies equation [10]). Then i* > 0 is a rate of interest
establishing (E 1) if and only if D(i *) = D(i *)/J.t + L, where D(i) is the function
defined in the text above. Moreover, D '(i*) > 0 is equivalent to
e; = oe(r + e, i*)loi < oi:(>., e, r, i*)loi.

On the basis of this lemma, the rate of interest which is most to the right out of
the two (or even more) points of intersection of the curves e(r + Q,.) and E(A,
e,r,.) is chosen. Figure 1 shows that necessarily E; ::5 iJE/iJi. In Assumption 2,
simply suppose that this inequality is strict.
Assumption 2. For all values A., e, r with 0 ::5 A. ::5 1, r + e > 0, and with
respect to i = j(A., e,r), defined as the greatest number i that brings about
e(r + e, i) = E(A., e, r, i), the following strict inequality holds,
e; = iJe[r + e, j(A., e,r)]/iJi < iJe[A., e, r, j(A., e, r)]/iJi
It is evident that the just defined function j(A., e, r) is continuous (even continu-
ously differentiable). 6 If the additional Assumption 3 is employed, definite state-
ments on the sign of its partial derivatives can be made. This will simplify
calculations considerably in the further analysis. In essence, Assumption 3 says
that locally around i = j(A., e,r), the function e is of the form e = e(r + e - i).
The ensuing Lemma 3 completes the first step of this section.
Assumption 3. For all 0 ::5 A ::5 1, r + e > 0, and with respect to the function
j(A., e, r) defined in Assumption 2, the following equality holds,
e; = iJe[r + e, j(A.,e,r)]/iJi = - iJe[r + e, j(A., e, r)]/iJ(r + e) = -er

Lemma 3. Let Assumptions 1-3 be satisfied. Then the function j(A., e, r) defined
in Assumption 2 is continuously differentiable and
oj/or = ojloe = A;!Ar > 0,
ojliJA. = i(r + e)I Ar > 0,
where Ar: = Ao er - (r + e)(A. + ¢) > 0,
A;: = Ao er - i (A. + ¢) > 0,
and A0 : = [r + e + i(A. + ¢)]2.
Step 2. With respect to a given pair (A., e), 0 ::5 A. ::5 1, we are now prepared to
look for a point r0 making the function r - F(A., e, r) vanish, where
(17) F(A., e, r): = (1 + ¢) h[r + e - j(A., e,r)] - s [r + j(A., e, r)A.]

Assume that even in the presence of very high differences between the expected
net rate of profit r + e and the corresponding interest rate i = j(A., e, r) that
clears the asset market, the planned rate of growth of the capital stock is bounded
from above (since, e.g., the volume of employment cannot increase at an infinite
speed). It follows that F(A., e, r) will be negative for all r sufficiently large.
Hence, the IS-equation (E2) can be satisfied if a profit rate r exists that causes
F(A., e, r) > 0. This positivity means that, at least for some medium values of r
(or r + e, respectively), firms' willingness to invest is sufficiently strong-in

relation to s[r + j(>-., e, r)A] (observe that by virtue of Lemma 3 this term is
falling if r is decreasing). Along with the previous assumptions, remarkably no
more than these two ideas are needed to also infer uniqueness of the root r as 0 ,
well as the sign of the derivative oF!or at this point.
Assumption 4. For all values of A and e (0 :5 A :5 1) the following holds:

(i) There is a real number ii such that h(r + e - i) :5 ii for all i > 0, r + e
> 0.
(ii) There exists a valuer such that r + e > 0 and (1 + <!>) h[r + e - j(>-., e,
r)] > s[r+j(>-., e, r)>-.].
Lemma 4. Let Assumptions 1-4 be satisfied. Then for every pair (>-., e),
0 :5 A :5 1, there exists a net rate of profit r 0 > - e bringing about F (A, e, r 0 ) =
0. It is uniquely determined .and oF(>-., e, r 0 )/or < 0.
oF/or < 0 could be interpreted as a condition ensuring stability in a dynamic

process in which A and e remain fixed, and the profit rate r responds to excess
demands, while the interest rate is continuously clearing the asset markets. A
similar idea can be found in Taylor (1983, 19-20; cf. also Taylor and O'Connell
1985, 874). There, however, the rate of interest also is given and does not change
during the (very short-run) process. Moreover, the negative slope of the function
that corresponds to this F(>-., e,.) (with i fixed) is introduced as an assumption. By
contrast, in our treatment oF/or < 0 is not assumed but results from a stability
assumption that refers to dynamic adjustments in a different set of markets. These
are the asset markets with the rate of interest as the (only) adjusting variable.
Since the latter are commonly regarded as having the fastest dynamics, it may be
argued that our choice of a stability assumption (i.e., Assumption 2 along with
Lemma 2) is preferable.
As stated earlier, we now only need to set R(>-., e) = r 0 , with respect to the r 0 of
Lemma4, and then define J(>-., e) = j[>-., e,R(>-., e)]. It is obvious that by putting r
= R(>-., e), i = J(>-., e), both (El) and (E2) will be satisfied for all>-. and e. On the
basis of Lemmata 1-4 and their proofs, the partial derivatives of J and Ras well as
of R + e - J can be determined. Since g = h(r + e - i) and h' > 0, the latter
provides information about ceteris paribus changes of the growth rate of the
capital stock. The results are given in the following theorem.
Theorem 1. Suppose that Assumptions 1-4 apply. Then there exist two continu-
ously differentiable functions R: [0,1] x R - R, J:[O,l] x R - R+ + (R+ +is
the set of strictly positive real numbers) such that (El) and (E2) are satisfied if
we put r = R(>-., e), i = J(A, e). R(>-., e) + e is always positive. Over the whole
domain, the signs of the partial derivatives are as follows:
oR!oA. < 0, oRloe < 0, oJloe > 0, and oJ/oA. has the same sign as the
expression, -(1 + </>)(r + e)h' - s[A; - (r + e)] (A; as defined in Lemma
3).
If g(A., e) denotes the rate of growth of the capital stock in a temporary
equilibrium (El), (E2) induced by A. and Q, then
sgn ogloA. = sgn (oR/oA. - oJ/oA.) < 0
sgn ogloe = sgn (oRloe + 1 - oJloe) < 0
Remark I. An apparent puzzle is the possible derivative oJ/oA. < 0. The first
reaction to a rising A. occurs directly on financial markets: in order to satisfy the
increased demand for loans, the rentier households have to increase D 0 + Di,
i.e., (1-e) must rise ore must fall, subsequently giving rise to a higher i (cf.
Lemma 3, ojloA. > 0). This, however, leads to an imbalance in the IS relation.
Saving, S = s(r + iA.)PK, will rise with A. and i increasing, and investment, PI =
h(r + e - i)PK, will fall.
Thus, PI < S is to be corrected. This could occur by a fall in r causing a
simultaneous fall in PI and S. If the reaction of h tor is not too strong (i.e., h' is
small) the balance of PI and Swill be achieved. If, however, h' is large, then the
fall in interest rate i has to be sufficiently strong, possibly below its previous
value, in order to adjust Pl to S. Note that since financial markets have been
ignored, this line of reasoning may be incomplete. A mathematical analysis is
inevitable to predict how the rate of interest in a temporary equilibrium will
change if, ceteris paribus, the ratio of indebtedness increases.
Remark 2. Though g only stands for the growth of the capital stock, under the
additional assumption of a constant mark-up rate T, its changes are closely related
to changes in the level of economic activity, i.e., to capacity utilization X/K: X/K
rises (falls) if and only if g rises (falls). This can be demonstrated in two steps.
First, equation (13) points out that g = h(r + e - i) increases if and only if the
gross rate of profit, rs, does. Secondly, movements of rs are in step with those of
XIK. To see this, use the mark-up princing rule, PX = (1 + T)Wages, in the
r
determination of with T the mark-up over wages. It leads to
PX - Wages T X
rB =
PK
Thus, given T, changes in the profit rate r (respectively in the growth rate g)
reflect changes in the utilization of capacity.
Before turning to a dynamic analysis, a proposition is presented concerning
the impact of a change in the monetary policy parameter <P on economic activity.
Of course, it is of a purely comparative-static nature. The result may be constrast-
ed with Tobin (1965).
Theorem 2. Suppose that the economy is in a temporary equilibrium (El), (E2),

that the monetary authorities increase <P, and that without any change in Aand e, a
new temporary equilibrium comes about. Then under Assumptions 1-4, the new
rate of growth of the capital stock will have decreased, og(A., e, <P)fo</> < 0, in
obvious notation.
Without further assumptions on the relative size of h' and er, no further
statements are possible with respect to the resulting change of the equilibrium
values of r and i.
Long-Run Dynamics
With the aid of Theorem 1, it is now possible to write equations (E3), (E4) as a
closed dynamical system of its own. In compact form,
>.. = U (>.., e)
(*)
e = v (>.., e)
where
U(A, e): = >..{b[R(A, e) + e - (1 - A) J(A, Q), A] - h[R(A, e) + e
J(A., e)]}
V(A, e): = v[R(A, e) - J(A, Q), A]
First consider borrowing by firms, i.e., the function b = LIL and impose the
following assumption on it. With expectations of positive net profits presup-
posed, the planned rate of growth of loans (or the one conceded by banks) exceeds
the growth rate of the capital stock if the degree of indebtedness }.. is low, and it
falls short of it ifA is already very high. Recall that the degree of indebtedness}.. is
also an indicator of the risk of bankruptcy. If }.. is high, investors may restrict
themselves concerning new borrowing, or banks may constrain credit for firms
with a high>.. Conversely, credit constraints are loosened ifA is low. This explains
the subsequent Assumption 5(i). Its second item looks more technical and is
mainly made for convenience. It serves to rule out any bifurcation of the curve U
= 0.
Assumption 5(iii), on the other hand, is a require'Ilent on the function v which
governs adjustments of expectations. It can be explained as follows. Theorem 1
infers that, with respect to each fixed }.. = ~. the expression r - i = R(~.
e) - J(~. e) is strictly decreasing in e. The same is true for V(~, e). Now, if e is
sufficiently high, the difference r - i will have become so low, perhaps negative,
that this optimism can no longer be justified (note that oRliJe - iJJliJe < -1).
Hence e = V(~. e) < 0. In the presence of a considerably negative Q, the
situation is reversed: r - i will have increased so much that previous pessimism is
no longer maintained and e begins to rise, V(~, e) > 0. It then follows that there
exists a e ebringing about V(>.., e) = 0. In Assumption 5(iii) this is postulat-

ed for the extreme values >.. = 0 and >.. = 1.
Assumption 5.
(i) For all values r, e. i with r + e > i, there is a>.. > 0 such that
b(r + i>.. + e - i, >..) > g = h(r + e - i), whereas
b(r + i>.. + e - i, >..) < g = h(r + e - i) if>.. = 1.
(ii) If U(A, e) = O, then either cJU(A, e)lcJA *
O or cJU(A, e)loe *
O (or
both).
(iii) There are numbers eo and ei, such that V(O, eo) = 0 and V(l, e 1) = 0.
Lemma 5. Let Assumptions 1-5 be satisfied. Then to each eER, there exists at
least one A, 0 < A < 1, such that U(A, e) = 0. The set
MA: = {(A, e)f[0,1] x R: U(A, e) = O}
consists of one or more (isolated) smooth curves in the (A, e) plane. (Precisely, at
each point (>.., Q)EMA, Acan be locally represented by a continuously differentia-
ble function of e. A= A(e), with A(Q) = >..and U[A(e), e] = 0 identically, or, if
this is not possible, it is e that can be locally represented by a continuously
differentiable function of A).
To each AE[O, 1], there exists one and only one e with V(A, e) = 0. The set
Me: = {(A, e) E [0,1] x R: V(A, e) = O}
is a single connected smooth curve which is downward sloping.
From Lemma 5, we conclude that the two sets MA and Me have at. least
one point (A0 , e0 ) in common. This will be a stationary point of system(*): A = 0
and e = 0. With Assumption 6 below, further study will be based on the
hypothesis that e0 = 0. Before this, briefly consider how to proceed if e0 * 0.
Since the analysis will reveal that (A0 , e0 ) cannot be a saddle point, two cases are
distinguished. First, let (A0 , e0 ) be repelling. If, in addition, this rest point is
unique, then the global analysis can run along essentially the same lines as
Scenario 3 below, though the dynamics, and in particular the limit cycles, may be
less regular. On the other hand, suppose that (A0 , e0 ) is an attractor and that the
process is already very close to it. This implies that r and e have practically
ceased to change, and that the expected net rate of profit r + e persistently
deviates from the realized one. Hence, sooner or later the expectations have to be
revised, which in the formulation of the model means that the function v(.,.) has
shifted. In this case, virtually a new dynamic process is set up and new motions
are initiated. If the stationary point (A0 , e0 ) of this modified system is again an
attractor, where e0 is different from zero, this procedure might be repeated, etc.
Of course, a shifting of the function v(.,.), or even ofanother function, might also
occur before the process gets near a rest point. In fact, this seems to be the correct
way to view what is actually happening in a dynamic economy.

Nevertheless, to reveal and discuss unambiguously the tendencies inherent in
the economy as modeled above, the assumption that the formation of expectations
is (already) consistent is employed.
Assumption 6. If(>.., e) is a point where the two sets M>,. and Me defined in
Lemma 5 intersect, then e = 0 at that point.
Lemma 6. There is one and only one point (A*, e*) of process (*) such that U(A.*,
e*) = o and V(A.*, e*) = o.
All the information for an analysis of the dynamics is available but one. It
concerns the reaction coefficients of the involved behavioral functions and their
relative size, i.e., the derivatives or partial derivatives of e(. ,.), h(.), and b(., .).
Remarkably, they shall be needed at one point only, the long-run equilibrium (>.. *,
e*). In conjunction with the previous assumptions, it will be possible, neverthe-
less, to arrive at global results. With respect tor*= R(A.*, e*),i* = J(>..*, e*),
these derivatives are
e, = e,(r* + e*' i*)-reactions of households in their share holdings to

changes of expected net profits;
h' = h '(r* + e* - i*)-investment reactions of firms to changes of the differ-
ence between expected net profits and interests, r + e - i;
b, = b,(r* + i*>..*+ e* - i*, >..*)-reactions of firms in debt-financing to
changes of the difference between expected gross profits and interests' r + e -
i;
b>,. = b>,.(r* + i*>.. + e* - i*,>..*)-reactions of firms in debt-financing in
response to changes in their present degree of debt-financing.
It can now be shown that different assumptions on the relationships between
these adjustment parameters lead to different motions of process (*) and possibly
even to a different stability behavior. Three scenarios of economic dynamics will
be distinguished. In technical terms, they correspond to the cases (1) U>,. < 0,
Ue<O, (2) U>,.<0, Ue>O, (3) U>,.>0, Ul?>O (where U>,. =au(>..*, e*)!a>..,
Ve =au(>..*, e*)lae). The fourth possibility, U>,. > 0, Ve < 0, shall be left
aside since, as can be verified shortly, it requires a special conjuncture of the
parameters. To grasp the economic meaning of these inequalitites, the explicit
calculations ofU>,. and Ue are given. Also the partial derivatives ofV(.,.) will
turn out to be of some interest.
Lemma 7. With respect to>..= A.*, e = e* = 0, r = R(>..*, e*,), i = J(>..*, e*),

and to e,, h ', b,, b>,. just specified, define (as in Lemma 3),
A;(e,): = [r + e + i(>.. + ¢))2e, - i(>.. + ¢) > 0
A,(e,): = [r + e + i(>.. + ¢))2e, - (r + e)(>.. + ¢) > 0 and furthermore,
A1: = (r + e - i)(}. + </>) > o

A2: = [(r + e - i)</> - i}.](1 + </>)
A3 : = s[(l - </>)(r + e) + i(}. + </>)] > O
Ap: = (1 + </>)A 1h '+s[Ar(er) + AA;(er)] > 0
Then the partial derivatives of U and V, evaluated at (A, e) = (A*, e* ,), are,
-1
There are several possibilities for the adjustment parameters er, h ', bn and bA
to fall within one of the three scenarios. Only the most striking ones will be
illustrated. To begin, er will be called "small" (or "large") if in both UA and U e
the coefficients associated with br. [A;( er) - (r + e)] as well as
[AA;(er) - A 1], are negative (or positive, respectively-recall that A;(er) > 0).
Then the following cases are considered, where again, all statements are confined
to the steady-state position.
Scenario I: UA < 0, U e < 0. Households are rather inflexible when deciding the
proportion of shares to hold. Firms, on the other hand, are reserved in their
investment reactions-compared to their adjustments of debt-financing in re-
sponse to changes in gross profitability over interests. Their response to the debt
structure, however, is not relevant. In short, er is small, h' is small in relation to
bn and the size ofbA does not matter. More precisely, the coefficients ofbr in UA
and Ue are both negative. h' is so small that, in Ue, A 1h' falls short of (-1)
[AA;(er) - Ailbn and similarly in UA.
Scenario 2: UA < 0, U e > 0. The outst~nding feature of this scenario is the high
responsiveness of firms to their debt structure. They (or their banks) are so
cautious that the present flow of borrowing is sharply reduced if the degree of
indebtedness, A, only slightly increases. That is to say, bA is strongly negative.
How negative it is depends on the other parameters. For example, er may be
small and investment reactions may be strong (h' large), such as to render U e
positive. Or, alternatively, households may respond very quickly in their distribu-
tion of assets, i.e. , er is large, so that U e > 0 in any case. The second term in UA
will probably be positive here, but this is supposed to be outweighed by the size of
b>..
Scenario 3: U>. > 0, U e > 0. As indicated in the signs, U>. > 0, U e > 0, the
increase in debt as well as the rise of expected rates of return exert a positive
feedback effect on the degree of indebtedness of firms. This will result under the
following conditions. A change in the debt structure has only a minor influence on
firms' borrowing (b>. is small), whereas share holding of rentiers is quite respon-
sive to relative changes in the rates of return, i.e., er is large. This makes U e
positive. If, on the other hand, A 2 is positive, the same is true for the term in the
curved brackets in the formula for U>.. If A 2 is negative, it is hypothesized that
either br is so small that A 2br + A 3 is positive. Should the latter expression be
negative, reactions in investment are so reserved that s[A;(er) - (r + e)]br
dominates [A 2br + A 3]h '. Finally, b>. is so close to zero that the positivity in the
formula is preserved. This third scenario is economically the most interesting
one, because it generates persistent cycles. It will be discussed further below.
In the following three theorems, the stability (or instability) results that can be
derived for the three scenarios are given.
Theorem 3. Suppose that Assumptions 1-6 hold and that U>. (>. *, e*) < 0, U e(>. *,
e*) < 0. The long-run equilibrium position (A*, e*) is globally asymptotically
stable. Considered locally, (A*, e*) is a stable node, i.e., all trajectories approach
the equilibrium directly (not in spirals). The essential characteristics of the
motions of process (*) are given in Figure 2.
Remark. From Lemma 5, it is known that the curve Me is strictly falling

(everywhere). M>. is strictly falling in the vicinity of(>.*, e*). At a greater
distance, however, a positive slope may result. In this case some trajectories will
display a "bouncing" behavior, as is illustrated in Figure 3. Similar things may
happen in the economies treated below. It is appropriate to point out the possibil-
ity of the existence of these short-term, or intermediate, cycles on this occasion.
For an unambiguous formulation of Theorem 4 and 5, the concept of a limit set
is referenced. The basic idea is already expressed by the very name of it. The
precise definition is given in the mathematical appendix.
Theorem 4. Suppose that Assumptions 1-6 hold and that U>. (>. *, e*) < 0, U e<>- *,
e*) > 0. Then (>. *, e*) is a locally asymptotically stable equilibrium of process
(*). Moreover, if a trajectory starts from a point with>. > 0, then its limit set is
either the steady-state(>.*, e*) itself or a closed orbit in the(>., e) plane. The
Figure 2
>..
0 e* =o
Figure 3
Figure 4
>..
essential characteristics of the motions are given in Figure 4.

Figure 4 and the data given are not yet sufficient to indicate whether (A*, e*) is
a stable center, i.e., whether it is (at least locally) approached in a cyclical
manner. This will happen if
(18)
(see the proof of the theorem 4). If the inequality is reversed, then from four
directions the trajectories tend directly towards (A*, e*) (see Coddington and
Levinson 1955, 384). On the other hand, whether (A*, e*) also will be a global
attractor depends on the position and the slopes of the curves MA and Me relative
to each other.
Actually, the stability tendencies that are revealed by Theorems 3 and 4 should
not be overestimated. By holding the price level constant, possible destabilizing
deflationary (and inflationary) effects are disregarded (cf. note 5). On the other
hand, the assumed formation of expectations-via the adjustment function v-is
already consistent, i.e., e = 0 if the process comes to a halt (see Assumption 6
and the Remark preceding it).
As indicated above, perhaps the most interesting case within the framework is
obtained if scenario 3 prevails. The following result can be derived.
Theorem5. SupposethatAssumptions 1-6holdandthatUA(A*, e*) > 0, Ue(A*,

e*) > 0. If expectations of future profit rates are very responsive to changes in
r - i, that is, ifthe coefficient Vr is so large that I Ve I > UA, then the steady-
Figure 5
+ - +
+
0 e
state (>, *, e*) is locally asymptotically stable. If these adjustments are compara-
tively sluggish, such that I Ve I < U>,,., it is locally repelling.
The limit set of each trajectory that starts at a point with >.. > 0 is either (>.. *,
e*) itself or a closed orbit. A phase diagram typically looks like that of Figure 5.
The theorem says, in particular, that in the unstable case each (nontrivial)
trajectory will tend towards a closed orbit. Sufficient conditions for the genera-
tion of persistent cycles exist. The central ones, we repeat, are as follows. By way
of Assumption 5(i) the firms' degree of indebtedness,>.., is globally bounded by
zero and one, and the equilibrium point itself, (>..*, e*), is a repeller (locally
totally unstable). The latter property is implied by the inequalities U>,,. > 0, U e
> 0 and I Ve I > U>,,.. The conditions for U>,,., U e > 0 are listed in the
statement concerning the third scenario. The most important ones are that near
the rest point (>.. *, e*), the firms' borrowing responds only weakly to the level of
firms' debt, whereas households respond strongly with their equity holdings to
the net profit rate. 7
In addition to the the motions of the constituent variables >.. and e of the
reduced process (*), the resulting behavior of some other key variables are of
interest. The way in which the respective peaks and troughs follow each other can
be noted so that the different stages of a cycle can be singled out. This can be done
by making reference to the functions and properties that previously have been
established. In order to avoid unnecessary complication, the choice will be
restricted to three additional variables: the rate of interest, i = J(>.., e,), the
difference between the actual net rate of profit and the interest rate, r - i =
R(>-.e) - J(>-.e), and the growth rate of the capital stock, g = h(r + e - i).
Recall that, according to Remark 2 subsequent to the formulation of Theorem 1,
the movements of g can be identified with the changes in the level of economic
activity or capacity utilization, as well as with the realized gross rate of profit r.
Figure 6 shows the succession and the relative position of the upper and lower
turning points of all these variables (a proof can be found in the mathematical
appendix. To be precise, it has been supposed that aJta>-. < 0 not only at(}..*, e*),
which readily follows from Theorem 1 and the scenario's characterization, but on
the entire(}.., e) plane. This assumption is by no means necessary. It has merely
been employed to simplify the argument). However, it has to be stressed that this
is the only information Figure 6 is able to provide. In particular, it is not meant to
indicate that the time intervals [tk, tk + i] have equal length. In fact, nothing is
known about their relative lengths. And, the upswings and downswings need not
be as regular as they have been sketched.
Recall, finally, that the dynam~c development of}.. is not only regulated by the
growth of loans of firms, b = LIL, but also by the growth rate of the capital
stock (see equation [15]). In case the elasticity ofb with respect to}.. and r is small
( I b}._ I and br small), in essence, the development ofA is passively regulated by
the growth of the capital stock; more precisely, it is regulated inversely by the
growth rate g. This is fully confirmed by Figure 6.
The model presented here examines the impact of debt-financing of firms on
aggregate macroeconomic activity. In particular, it explores the complexity of
macroeconomic dynamics resulting from a high degree of debt-financing of firms
and internal mechanisms by which the latter is reduced again. The analysis
proceeds in two steps. First, slightly modifying a framework already employed
several times in the literature, the product and asset markets are modeled and
their short-run stability properties are investigated. In this context, the ratio of
indebtedness of firms and the economy's "state of confidence" (the difference
between present and expected profit rates) are considered to be exogenously
given. The corresponding temporary equilibria of these markets are not assumed.
They are demonstrated to exist and to be (locally) stable-under some reasonable
conditions that are made explicit.
In a second step, the debt-asset ratio and the state of confidence are endogen-
ized and, based on the hypothesis of continuous market clearing, it is formulated
how they change over time. Merging the two steps, the interrelated motions of six
economic key variables are obtained. In addition to the debt-asset ratio and the
state of confidence, they are the current gross and net rates of profit, the interest
rate, and the growth rate of the capital stock.
Though the resulting dynamics are quite complex, different assumptions on
the diverse reaction functions make it possible to work out essentially three types
of scenarios. 1Wo of them exhibit local (if not global) stability with respect to a
unique long-run equilibrium of steady growth. The other one, the most interest-
ing case, generates persistent fluctuations in the aforementioned variables,
Figure 6
ts lg
g g
~----i-r- i
Is lg
tending to a closed orbit (so that a growth cycle results). Conclusively, it should
be stressed in particular that the method utilized here does not need to assume
exogenously fixed upper limits oflending and borrowing. Rather, the model gives
a fully endogenous treatment of all the variables involved.
Notes
1. Because of limited space the appendix with the mathematical derivations are not
included in this version. The appendix is available from the authors at request.
2. There are several similarities to the economy-wide balance sheet in Taylor (1983,
93), whereas, however, in Taylor (1985, 392) the capital stock of firms is valued at P,
instead of Px.
3. Savings by private wealth holders increase the stocks of D0 , Di, and PeE. In what
proportions is already grasped by the functions d0 and e introduced in the preceding
subsection.
4. The analysis in the following sections is by no means dependent on this device.
There would be only minor changes if the net profit rate entered as an argument into the
borrowing function, i.e., if below b = b(r + e - i, >-),instead ofb = b(r + e - i, >-)
5. Setting P = constant leaves aside the (possibly additional) destabilizing effects of
debt deflation in our dynamics, which should be considered subsequently (cf. Minsky
1975, 54, Tobin 1975).
6. To be mathematically precise, this is ensured by the Implicit Function Theorem. For
further details see the proof of Lemma 3 in the appendix.
7. In more general terms, as formulated in Lemma 7, we can state that in the determi-
nation of u}.. > 0, the coefficients for the firms' investment response to (r* + e* - i*)
and the firms' borrowing response to r outweigh the response of firms' borrowing to the
level of debt. On the other hand, for U e > 0, it is required that the linear combination of the
coefficients for firms' borrowing response to r and investment response to
(r* + e* - i*) generate a positive feedback effect on e These two positive feedback
effects together with I v 12 I > u>-. will turn the equilibrium into a repeller and generate
globally persistent cycles. The conditions for those cycles to occur bear a close resem-
blance to the conditions of the equilibrium dynamics sketched by Kalecki in his business
cycle theory ( 1937), but in a much simpler framework and only for a stationary economy.
References
Brealey, R., and S. Myers. 1984. Principles of Corporate Finance. New York: McGraw
Hill.
Clower, R. W. 1967. "A Reconsideration of Microfoundations of Monetary Theory."
Coddington, E. A., and N. Levinson. 1955. Theory of Ordinary Differential Equations.
New York: McGraw Hill.
Foley, D. 1986. "Stabilization Policy in a Nonlinear Business Cycle Model." Competi-
tion, Instability, and Nonlinear Cycles, Lecture Notes in Economics and Mathematical
Systems. W. Semmler, ed. New York and Heidelberg: Springer-Verlag.
Foley, D. 1987. Liquidity-Profit Rate Cycles in a Capitalist Economy. Journal of Econom-
ic Behavior and Organization 8(3):363-377.
Greenwwald, B., J. E. Stiglitz, and A. Weiss. 1984. "Informational Imperfections in the
Capital Market and Macroeconomic Fluctuations." American Economic Review
(May):194-200.
Guttentag, J., and R. Herring. 1984. ''Credit Rationing and Financial Disorder.'' Journal
of Finance (December): 1359-82.
Kalecki, M. 1937. "A Theory of the Business Cycle." The Review of Economic Studies
(February):77-97.
Kohn, M. 198la. "In Defense of Finance Constraint." Economic Inquiry (June): 177-
195.
Kohn, M. 198lb. "A Loanable Funds Theory of Unemployment and Monetary Disequili-
brium." American Economic Review (December):859-879.
- - - - . 1982. Can "It" Happen Again? Essays on Instability and Finance. New York:
M. E. Sharpe, Inc.
Stiglitz, J.E., and A. Weiss. 1981. "Credit Rationing in Markets with Imperfect Informa-
tion." American Economic Review (June):393-410.
Taylor, L. 1983. Structuralist Macroeconomics. New York: Basic Books.
- - - - . 1985. "A Stagnationist Model of Economic Growth." Cambridge Journal of
Economics 9:383-403.
Taylor, L., and S. A. O'Connell. 1985. "A Minsky Crisis." Quarterly Journal of Eco-
nomics 100:871-86.
Tobin, J. 1965. "Money and Economic Growth." Econometrica 33(0ctober):671-84.
Tobin, J. 1975. "A Keynesian Model of Recession and Depression." American Economic
Review 65: 195-202.
Tobin, J. 1982. "Money and Finance in the Macroeconomic Process." Journal of Money,
Credit and Banking XIV(2):171-203.
Tsiang, S .C. 1956. "Liquidity Preference and Loanable Funds Theories, Multiplier and
Velocity Analysis: A Synthesis." American Economic Review (September):540-64.
Woodford, M. 1986a. Finance, Instability and Cycles. Paper prepared for conference
Economic Dynamics and Financial Instability, New School for Social Research,
March. Revised version included in this volume.
- - - - . 1986b. Self-Fulfilling Expectations, Finance Constraints and Aggregate Fluc-
tuations. Columbia University, New York. Mimeo.
Wolfson, M. 1986. Financial Crisis in the U.S. Economy. New York: M. E. Sharpe, Inc.
Accumulation, Finance, and Effective
Demand in Marx, Keynes, and Kalecki
Anwar Shaikh
This chapter develops a new approach to the theory of effective demand. The
familiar relationships between aggregate demand, supply, and capacity are linked
to a corresponding relationship between finance and debt. These cross-links
provide a natural foundation for a macroeconomic model of internally-generated
cyclical growth. The scenario which results from the model will be very similar
to the classical and Marxian descriptions of normal accumulation, with supply
and demand fluctuating erratically around a cyclical growth path with an endo-
geneous trend. Moreover, whereas current theories of effective demand generally
need to resort to exogeneous factors such as technical change, population growth,
or bursts of innovation in order to explain economic growth (Mullineaux 1984,
87-89), this classical/Marxian approach will be able to explain growth endogen-
eously through the normal rate of profit.
The framework developed in this chapter is grounded in Marx's schemes of
reproduction, in Chipman's (1951) illuminating treatment of Keynesian flows,
and in the pioneering elaborations of the Marxian schema by Dumenil (1977) and
Foley (1983). The results are distinct from either of the two major traditions in
modern macroecomics, since neither Say's Law (aggregate production generates
a matching demand) nor Keynes's/Kalecki's Law (aggregate demand induces a
matching supply) is assumed. On the contrary, as in Marx, both aggregate supply
and demand are found to be themselves regulated by more basic factors (Kenway
1980; Foley 1983). Because capitalist production is fundamentally anarchic, this
regulation process is always characterized by constant shocks and discrepancies.
Nonetheless, the inner mechanisms of the system continue to operate. The end
The framework used in this paper was first presented at the International Conference
on Competition, Instability, and Nonlinear Cycles at the New School for Social Research,
New York, March, 1985.
65
66 ANWAR SHAIKH
result is a turbulent and erratic pattern in which supply and demand cycle
endlessly around an endogeneously-generated growth trend (Bleaney 1976, ch. 6;
Shaikh 1978, 231-2; Garegnani 1979, 183-5).
It is important to note that the present analysis is concerned solely with the
relationship between effective demand and accumulation in the absence of any
changes in technology or potential profitability. These themes are central to
Marx's schemes ofreproduction, to Keynes's theory of output, employment, and
effective demand, and to Kalecki's theory of effective demand and cycles. More
importantly, such considerations are a necessary prelude to the analysis of factors
which may modify the path of accumulation and even transform it into a general
crisis.
A Framework Linking Aggregate Demand, Supply,

and Finance
This section will develop a general framework linking aggregate demand, supply,
and capacity to their duals in finance and debt. The aim is to make this framework
broad enough to encompass the basic approaches in Marx, Keynes, and Kalecki,
while still keeping it manageable. Therefore, the price level, money wages, and
the rate of interest will be held constant, since their variations are not central to
the above approaches. Similarly, we assume that the aggregate consumption of
workers' equals their wages, so that aggregate personal savings derive only from
capitalist personal income. However, there is no assumption of any a priori
balance between aggregate demand and supply in the short run (as in Keynes and
Kalecki), nor between aggregate capital disbursements and internal finance (as in
Marx's schemes of reproduction). Indeed, it is one of the central themes of this
paper that the linked imbalances in the above two domains play a crucial role in
regulating the overall reproduction process.
Aggregate Demand, Supply, and Capacity
Following Marx, the period of production is taken as the basic unit of time, and it
is assumed that the difference between inputs purchased and used in each period
is small enough to be treated as a relatively small random variable (which will be
reintroduced in the simulations). Inputs entering production at time t - I lead to
output at time t. By definition, potential profit on production (the money form of
aggregate surplus value) in period tis the difference between the money value of
aggregate supply Qt in period t and the sum of materials costs Mt- 1 , labor costs
Wt-l and depreciation DEPt-l on inputs used to produce current output. The
money value of aggregate supply in period t can be written as
(1)
ACCUMULATION, FINANCE, & EFFECTIVE DEMAND 67
Current aggregate demand D, is composed of the current demand for materials

Mt and for new plant and equipment (gross fixed investment) JG,, for desired
additions to final goods inventories CINV,, and for workers' and capitalists'
consumption CONW, and CONR,, respectively. All of the above items require
actual expenditures, except for CINV, which represents the portion of output
which capitalists would like to retain in final goods inventories in order to attain
some desired inventory level. When supply and demand do not balance, the actual
change in final goods inventories-which equals the difference between gross
output (additions) Q and gross sales (deductions) M + I + CONW + CONR-
will differ from the desired change CINV.
(2)
Excess demand E, in any period t can be defined as the difference between

aggregate demand and aggregate supply. Note that when excess demand is posi-
tive, realized profits will be greater than potential profits.
(3)
Combining equations (1) - (3), recalling that workers'consumption CONWt

equals their wages W,, and grouping like terms, yields
(4)
where
A, = (Mt - M,_ 1 ) + (Wt - W,_ 1) = accumulation of circulating capital
= net accumulation of fixed capital
= net fixed investment 1
CINV, = desired accumulation in final goods

inventories
Equation (4) could be expressed in terms of the more familiar balance between
total accumulation expenses ("ex ante investment"), A + I + CINV, .and total
nonconsumed surplus product ("ex ante savings"), P - CONR; but, this would
be misleading for several reasons. First, the so-called total investment would then
be a hybrid of actual accumulation in circulating capital A (investment in inven-
tories of raw materials and goods-in-process) and desired accumulation in final
goods inventories CINV, both of which tend to be ignored in conventional
accounts. Second, the so-called total savings would then merely represent the
excess of the surplus product over and above the personal consumption demand of
68 ANWAR SHAIKH
the capitalist class, which in no way corresponds to any quantity of money

revenue withdrawn from immediate expenditures ("saved"). Indeed, the ac-
counting device of representing the money value of this nonconsumed surplus
product as the sum of the personal savings of capitalists (which do represent
money withdrawn from immediate expenditure [Keynes 1964, ch. 16]) and the
"retained earnings" of the business sector (which do not not necessarily corre-
spond to any money revenue withdrawn from expenditure [see equation 8 below])
simply conflates the relation between the supply/demand for commodities and the
sources/uses of funds. This conflation obscures important connections between
these two domains and are, therefore, treated separately.
Since we are abstracting from changes in technology, wages, and working
conditions, aggregate capacity (normal capacity output) N, will be proportional
to the aggregate fixed capital stock Kf,: N, = vKf,, where v = the constant
capital-capacity ratio (as in Harrod). Defining capacity utilization u, as the ratio
of output to capacity (so that u > 1 implies above normal capacity utilization), we
can write
(5)
The last step is to consider the effects of circulating and fixed investments on
output and capacity, respectively. Given the constant fixed capital-capacity ratio v
assumed above, the change in capacity is proportional to the level of fixed
investment (since this is the change in the fixed capital stock).
(6)
On the other hand, given some real period of production which will be taken as
the unit of time, current output Q, and current potential profit P, are the results of
inputs M,_ 1 and w,_ 1 purchased and used in the previous period. Given a
constant profit margin on costs, m = P,/(M,_ 1 + W,_ 1), the change in current
potential profits is proportional to the change in past inputs. Since the latter is
simply the circulating capital investment in the last period (see equation 4), it can
be written
(7)
Equation (7) expresses the connection between circulating investment and the
expansion of production. This relation is often neglected these days, even though
it has always been an integral part of classical and Marxian schema. Modern
national income accounts tend to lose sight of circulating investment because they
adopt the convention of treating current expenditures for materials and labor
(M, + W,) as the production costs (intermediate inputs) of current output Q,
(BEA 1980, 6-9). This implicitly assumes a zero time of production, which is
tantamount to assuming away the production process altogether.

Equations (4)-(7) define the fundamental equations of aggregate production
and effective demand. It is important to note that any a priori balance between
demand, supply, or capacity is never assumed.
Aggregate Finance and Debt
In treating finance, assume that firms pay dividends R to capitalists, who in turn
consume a portion, CONR, and save the remainder, SAVR. Also assume, as does
Kalecki (1965, 97) that firms borrow these personal savings of capitalists, SAVR,
by issuing stocks or bonds to that amount. This is more or less equivalent to
Marx's assumption that capitalists draw their personal consumption directly out
of profits, leaving the rest available for use by the firm (Marx 1967, vol. 2, ch.
21). Any borrowing or lending above this amount is then assumed to be mediated
by the banking sector. Additionally, assume that this banking sector is willing and
able to fulfill the needs of its borrowers or depositers without having to change the
interest rate. This assumption is made merely in order to duplicate the Keynesian
and Kaleckian assumption that bank finance can be freely acquired (or lending be
freely accepted) at some given rate of interest below the potential rate of profit. In
an important and insightful paper, Asimakopulos points out that Keynes and
Kalecki justify their treatment of planned investment as unconstrained by (i.e.,
independent of) the current flow of savings precisely through the assumption of
"freely gotten finance" (Asimakopulos 1983, 222-27). By adopting the very
same assumption, hopefully it will be clear that the basic differences between
Marxian and conventional theories of effective demand have nothing to do with
the presence or absence of credit. 2
The need for external finance arises because the projected expenditures of
firms may exceed the projected internally available sources of funds. Borrowing
must, therefore, precede the actual expenditures it aims to finance (Robinson and
Eatwell 1973, 218-19). In general, this borrowing will be assumed to consist of
two parts: direct borrowing of current capitalist savings SAVRt through the issue
of new stocks or bonds and bank borrowing Bt for any needs beyond this level.
Total Borrowing = Total Planned Uses - Total Internal Sources
Bank Borrowing + Stock/Bond Issues = Planned Uses - Sources

(8) Bt + SAVRt = Planned Uses - Sources
The output Qt forthcoming over any period t is determined by the materials

and labor set into motion in the previous period. Of this current projected output,
the amount CINV t represents the desired additions to inventories of final goods,
so that it is the remainder which is projected for sale. Since any financial receipts
70 ANWAR SHAIKH
of principal and interest on past lending by firms are treated as negative finance
charges on the side of uses of funds, the total projected internal sources of funds
of the business sector in period t simply equals projected sales, Qt - CINV t·
Over the same period, the total planned uses of funds must encompass five basic
categories: circulating capital expenditures for materials Mt and wages Wt to be
purchased in this period (in order to produce output for the next period), fixed
capital expenditures for gross investment in plant and equipment IGt. the pay-
ment of finance charges Ft which represent currently due principal and interest
charges on past borrowing (or when negative, the current receipt of principal and
interest revenues on past lending), disbursements of dividends Rt to capitalists
for whom they will serve as current income, and any planned changes in money
reserve levels CMRt. It should be noted that since the money reserves of firms
may be fed by past borrowing, government increases in the money supply, or
even by increases in the supply of a money commodity such as gold (as in Marx),
the term CMRt represents any desired adjustments over and above these other
sources of changes in money reserves. Thus equation (8) becomes
Combining equations (1) and (9) results in
Bt + SAVRt= Mt + Wt + IGt + Ft+ Rt + CMRt

- [(Mt-1 +Wt-1)+DEPt-1 +Pt) - CINVtJ
= [(Mt - Mt-1) + (Wt -Wt-1)) + (IGt

+ CINVt + Ft + CMRt - (Pt - Rt)
(10') Bt+ SAVRt= [At+ It +CINVt] + [Ft+CMRt - (Pt - Rt)]
The second term in brackets on the right side of equation ( l 0 ') is the difference
between financial uses Ft + CMRt and retained earnings Pt - Rt. Therefore,
retained earnings correspond to financial leakages from expenditures only when
all "investment" (At + It + CINV t) is deficit financed (i.e., financed entirely
through borrowing, Bt + SAVRt). Since this will not generally be the case, it is
incorrect to treat retained earnings as a form of business savings.
Finally, noting that capitalist revenue R = consumption CONR + saving
SAVR, equation (IO') can be rewritten
All of the above quantities represent planned expenditures and projected

revenues, as anticipated at the beginning of period t. But if it can be assumed that
short-run expenditure plans are revised between periods, not within them, and
that short-run revenue estimates are relatively accurate (in a stochastic sense),
then from equation (4) the first term in parenthesis on the left hand side of
equation (10) is simply excess demand Et plus a small random variable (which is
reintroduced during the simulation process). Thus,
(11)
where,
Bt bank borrowing by firms
Er excess demand
Fr finances charges (principal due + interest due)
CMRr desired changes in money reserves
Note that all these terms may be either positive or negative, with correspond-
ing interpretations.
Equation (11) is the fundamental equation of finance. It says that the bank
borrowing of the business sector must cover its own planned deficit finance of
current expenditures (which will then show up as excess demand Er), plus finance
charges due on past borrowing, plus any funds needed to adjust money reserves to
desired levels. The terms Fr and CMRt play a particularly important role here,
because they reflect the feedback of past events on current borrowing.
Equation (11) above can also be read in another way.
(11 ')
The term in parentheses on the right side is the net bank borrowing of the
business sector, since it is the difference between current new borrowing Br and
current repayments of principal and interest Fr. Equation (11) then reveals that
when excess demand is zero, any desired adjustments in money reserves (in light
of any direct injections of new money) must be covered by net bank borrowing.
In a growing system, this implies a growing level of net borrowing, though this
may well be a constant proportion of total profits or total output. More important-
ly, any excess demand E must therefore be fueled by an injection of bank credit
over and above the amount required for money reserve adjustments. But any such
additional borrowing implies future finance charges. Thus episodes of excess
demand carry the seeds of their own negation, because the net injections of credit
which fuel them also carry over into the future as accelerated leakages. This
feedback will play a vital role in bounding the growth cycles of the system.
It is interesting to note that the above feedback effect was essentially ignored
by both Keynes and Kalecki when they formulated their respective theories of
effective demand. What is more, even after criticisms of their work led them to
admit that they had implicitly relied on ''credit inflation'' (Kalecki) or increased
"bank finance" (Keynes) as the crucial foundation of their explanation of in-
72 ANWAR SHAIKH
creases in business activity (Asimakopulos 1983, 223-26), neither author ever

really analyzed the impact of this ''credit inflation'' on the level of business debt.
Instead, both ended up focusing on its impact on the level of the interest rate, thus
deflecting attention away from the magnitude of business debt associated with it.
This left the interest rate as the principal regulator of investment decisions, as is
evident in Keynes. This same emphasis on the interest rate has been revived
recently by several authors (Taylor 1985; Foley 1987) as a means of breaking out
of the impasse generated by the apparent instability of growth within convention-
al theories of effective demand. But while the influence of interest rate move-
ments is clearly important, it is not necessarily the central factor regulating
accumulation. It will be seen that even when the interest rate is assumed to be held
constant, say through some "appropriate" set of state policies, the feedback
between finance, debt, and accumulation will turn out to be sufficient to stabilize
accumulation. The resulting theory of effective demand is very much in the
classical/Marxian tradition, with the internal profitability of the system driving
accumulation and the consequent debt burden constraining it. Any such construc-
tion vitiates all claims that there is an inherent contradiction between theories of
effective demand and classical and Marxian theories of growth. The next section
will therefore develop a simple model embodying the above principles.
A Macro Model of Internally Generated

Cyclical Growth
The model developed below focuses on profits, investment, savings and finance,
because these are the critical variables in the debate surrounding the relationship
between effective demand and accumulation. The adjustment of inventory and
money reserve stocks will not be treated here, because they play a relatively
secondary role in the basic analysis in Marx, Keynes, and Kalecki, and because
space limitations preclude the necessary development.
An important aspect of the approach is the distinction between fast and slow
variables. Slow variables are assumed to have decision periods longer than those
of corresponding faster variables (e.g., years instead of months), so slow deci-
sions are effectively cast in terms of moving averages of the faster variables.
Although one can conceive of many different sets of variables with each set
operating at its own intrinsic speed, the present analysis is confined to just two
speeds. The basic fast variable will be the proportion of potential profit (surplus
value) which is invested in circulating capital. In Marxian terminology, this is the
rate of accumulation in circulating capital, and it regulates the relation between
supply and demand. The corresponding slow variable will be the rate of accumu-
lation in fixed capital, which regulates the relation between supply and capacity.
In what follows, the (relatively) fast adjustment process is modeled first and
then the (relatively) slow one. They may be thought ofas "short-run" and "long-
run" adjustments, provided that two things are understood. First, the corre-
sponding time horizons are defined within this framework and may not corre-
spond to those implicit in other frameworks. Second, the short- and long-run
balance points are not equilibrium points in the conventional sense, but rather
centers of gravity around which the system cycles.
The (Relatively) Fast Adjustment Process
The relationships between aggregate excess demand, bank borrowing, and in-
vestment in circulating capital in equations (4), (7), and (11), respectively, form
the core of the fast adjustment process. Noting that we are abstracting from stock
adjustments, we can write
(12)
(13)
(14)
The next step is to define the interrelationships among the terms of the above
equations. Assume that the ratio of capitalist consumption to potential profits is a
constant c (on the grounds that dividends are proportional to profits and capitalist
consumption is in turn proportional to dividends), and that the rate of accumula-
tion in fixed investment k is a constant in the short-run, since it is a slow variable.
Finally, the essential link between past borrowing and present debt service is
captured by assuming that all borrowing or lending by firms must be paid back at
a constant interest rate i within one period. Accordingly,
(15)
(16)
(17) Ft = (1 + i)Bt-1
The remaining step is to model the behavior of the rate of accumulation in

circulating capital a = A/P. This ratio expresses the strength of the tendency to
expand production and is generally determined by various factors ranging from
the level and trends of past profits to various expected gains and costs. There need
be no specific assumption about the determinants of the level of the rate of
accumulation. Instead, simply assume that firms attempt some arbitrary rate of
accumulation, which they then modify based on the results of their attempt.
Specifically, assume that if any arbitrary initial attempted accumulation rate
results in a level of internally available finance above potential profit (surplus
value), then the accumulation rate in the next period will be higher. The opposite
74 ANWAR SHAIKH
holds when internal finance falls below potential profit. In this way, the rate of
change in the rate of accumulation becomes linked to the financial strength of the
firm.
At the beginning of any period t, firms must assess their internally available
finance and formulate their borrowing and expenditure plans for the period. As
the internally available and borrowed funds are actually expended, the resulting
demand serves to realize a particular level of aggregate profit. Thus, realized
profits in period tare themselves the result of expenditures undertaken in period t
(Kalecki 1965, 45-46). It follows that only the profits realized in period t-1 can
enter into finance which is internally available at the beginning of period t.
Actual internally available finance at the beginning of period t is defined as
profits realized in the previous period t-1 minus any debt service payments which
firms are obligated to pay over the coming period t. Aggregate profits in t-1 are
realized by aggregate purchases (A+ I +CONR)t-t, and from equation (4) these
equal the sum of potential profits and excess demand (P+E)t-l· Debt service
payments over period tare given by equation (7). Thus at the beginning of period
t, the internally available internal finance is
Xt = (Realized Profits in t-1) - (Debt Service in t)

= (P+ E)t-1 - Ft = (P+ E)t-1 - (1 +i)Bt-1
The accumulation reaction function then states that the change in the rate of
accumulation in circulating capital is proportional to the percentage of the excess
of internally available finance over potential profit (surplus value).
(18) (A/P)t - (A/P)t-1 = h[(E - (1 + i)B)/P]t- 1
Equations (12)-(18) describe the complete short-run model. At this point, it is

useful to consolidate the above equations, express all terms as proportions of
potential profit, and write them in their continuous time equivalents, so as to
facilitate subsequent proofs. Combining (12), (15), and (16), letting e = excess
demand as a proportion of potential profit, a = the rate of accumulation in
circulating capital, recalling that c and i are constant and k is constant in the
short-run because it represents t~e slowly changing rate of accumulation in fixed
capital, and using the notation P to signify the instantaneous rate of change of P,
etc., it follows that
(19) e=a+k+c-
(20) PIP =ma
(21) a= h[e - (1 +i)b]
Combining equations (14) and (17) gives Bt = (1 +i)Bt-l +Et. Translating

. .
into continuous time, B+ B = (1 +i)B + E+ E, where i now stands for the
instantaneous interest rate. Dividing through by potential profit P, letting b =
B/P, and noting that b = B/P - ( P/P)b and e = E/P - ( P/P)e, the equation
of finance can be expressed as
. . B/P + b= (l+i)b + E/P + e

b+ ( P/P)b + b = b + ib + e + ( P/P)e + e
. .
(22) b = e + ib + e + ( P/P)(e - b)
Equations (19) - (21) can be reduced still further. Since c is constant, and k is
constant in the short-run, (19) implies e= a, which can then be substituted into
(21). On the other hand, P/P = ma from (20) and a = e + d from (19), where d
= 1 - (c + k), all of which can be used to rewrite (22). On this basis, the result is
two nonlinear differential equations which describe the essential mathematical
structure of the fast adjustment process.
(23) e = h[e - (1 +i)b]
(24) b = e+ (l+i)e + (md - i)(e - b) + me(e - b)
where e, b represent excess demand and borrowing as fractions of potential

profit, respectively, and
c the constant propensity to consume out of profits
k the rate of accumulation in fixed capital (constant in the short run)
d 1 - (c + k)
m the constant profit margin on costs
the constant rate of interest
The short run adjustment process summarized above has several remarkable
properties (proofs are in the Appendix). First of all, it has only one stable critical
point ate = 0 and b = 0, which implies that the system automatically converges
around a generally growing path on which supply and demand balance (e = 0) and
accumulation is self-financing (b=O). This path is none other than the aggregate
equivalent of the Marxian expanded reproduction path implied by the parameters
c ,k (of which simple reproduction is a special case). 3 Second, the stability of this
model is assured by the simple and plausible economic requirement that in the
vicinity of expanded reproduction, the funds reinvested by firms be capable of
earning an incremental rate of return which is greater than or equal to the rate of
interest. In other words, active capital should be capable of earning at least as
much passive capital. Third, subject to the above condition, the model is ex-
tremely robust, because it is stable for all positive values of the reaction coeffi-
cient h and is cyclically convergent for all plausible values of h.
The above properties imply that from any arbitrary initial situation, the model
will converge in a cyclical fashion towards aggregate expanded reproduction. But
this does not imply that the system will ever be in expanded reproduction, because
76 ANWAR SHAIKH
Figure 1. Excess Demand and Debt.
0.6
0.5
Excess demand
0.3
0.1
0.0
-0.1
-0.3
-0.5
-0.6
-0.8
0 25 50
Time
once the effects of the anarchy of capitalist production are simulated by subjecting
the model to recurrent random shocks, the system cycles endlessly around ex-
panded reproduction without ever coming to rest upon it. The simulation results
are shown in Figure 1 in which excess demand, e, and the debt burden, b, cycle
erratically around the balance point of zero. Figure 2 shows how this translates
into the fluctuation of actual profit (realized surplus value) around potential profit
(produced surplus value). Taken as a whole, these figures exemplify Marx's
conception of expanded reproduction as the inner tendency-the regulating aver-
age-of the erratic path of the actual system.
The short-run model has several other interesting properties. To begin with,
since excess demand e is approximately zero in the short-run, equation (19)
implies that
(25) a = 1 - (c+k) = d
When averaged over short-run fluctuations, the rate of accumulation in circu-

lating capital, a, is inversely proportional to the propensity to consume, c, and to
the fixed investment accumulation rate, k. This means that even though an
exogeneous rise in either c or k may initially stimulate the system, the financial
Figure 2. Produced and Realized Profit.
1000.0
900.0
800.0
700.0
600.0
500.0
400.0
300.0
.______
200.0 Realized
profit
100.0
0.0
0 12.5 25
Time
drag created by the additional debt will end up lowering a by the same amount, at
least over the average short-run cycle.
The average short-run rate of return on fixed investment is also inversely
proportional to c and k. Defining this as
r = Pl Kf = (PIP)l(IIP)
where l/P = k, PIP = ma from equation (18) and a is given by (25), so that
m(l - c - k) m(l - c)
r = mdlk = - m.
k k
Once again, an exogeneous rise inc or k may initially raise the short-run rate of
return on fixed investment by initially stimulating a, but will end up actually
lowering it as the new short-run average level is established.
Lastly, it can be shown that the rate of capacity utilization will be roughly
constant over the average short-run cycle at some level which will, in general, be
78 ANWAR SHAIKH
different from normal capacity utilization. While this is reminiscent of the stan-
dard Keynesian and Kaleckian conclusion that there is no short-run mechanism
which will make actual output equal "full employment" (i.e. normal capacity)
output, it is worth noting that our results hold for a growing system, whereas
those of Keynes and Kalecki hold solely for a static level of output. To derive our
result, note that from equation (5) the fixed capital/capacity ratio v = Kf/N is
constant, while from equation (7), the constant profit margin on costs m implies a
constant profit margin on output n = P/Q. The levels of capacity, output, and
capacity utilization, respectively, can be written as
(26) N = Kf/v
Q =Pin
u = QIN = (llrn) P/Kf
where rn = n/v = the normal-capacity rate of profit on fixed capital.

To analyze the short run behavior of u = QIN, note that
N =:= Kf/v = I/v = (k/v)P and

. p (P/P)P maP m(l - c - k)P
Q=- =------
n n n n
Thus Q = pN, where

mv[(l - c)/k - 1]
p=
n
is constant in the short run. Integrating both sides, Q = pN + (Q 0 - pN 0 ),
where the term in parentheses is the constant of integration evaluated at some time
t0 • Rewriting,
u = QIN = p + (u 0 - p) N 0 /N
from which it is clear that as the system grows and N rises over time, u ap-
proaches p. A rise inc or k will, therefore, tend to lower the average short-run
level of capacity utilization by lowering its asymptote p.
The (Relatively) Slow Adjustment Process
In the analysis of the short-run, the rate of accumulation in fixed capital k is taken
as given on the grounds that it represents a slow variable. But since the short-run
level of capacity utilization will generally differ from the normal level of capacity
utilization, it is to be expected that k will slowly react to any such discrepancy.
Defining a longer unit of time T (e.g., years instead of months) to accommodate
this slow adjustment process, the reaction function for the rate of accumulation in
fixed capital k is written as
(27) k/k = g(u - 1)

where u = the level of capacity utilization (normal level = 1)

g = the reaction coefficient (a positive constant)
The effects of such a reaction function depend on the counter-response of

capacity utilization to k. Now, from equation (25) it is known that the fast
adjustment process will lead to the rough equality a + k = 1 - c. Suppose the
short-run level of capacity utilization is above normal, so that k begins to slowly
rise. From the point of view of the short-run adjustment process, k has risen to a
new higher level. This rise may initially stimulate effective demand and raise a.
But as the new short-run center of.gravity is established, a will fall to accomodate
the new higher short term level of k. Thus, an acceleration in the the growth of
capacity will end up decelerating the growth of actual production, so that the
capacity utilization level will tend to fall back toward normal (or even past it).
This tendency is in striking contrast to the knife-edge instability usually found in
conventional effective demand models. It is, on the other hand, implicit in most
classical and Marxian analyses of accumulation. To formalize it, diff~rentiate the
expression u = (llrn) PIKf given in equation (26), recalling that PIP = ma :=
m(l - c - k) from equations (6) and (25), respectively, while by definition Kf
= I and I/P = k,
ulu = PIP - KflKf = ma - (llP)(PIKt) = m(l - c - k) - (rn)ku
(28) ulu = m(l - c) - mk - (rn)ku
Equations (27) and (28) define a system of two nonlinear differential equations
representing the slow adjustment process through which the level of capacity
utilization reacts back on the rate of accumulation in fixed capital, and vice versa.
The above long-run adjustment process has the striking property that it is
stable around the normal capacity utilization level u = 1 (see proofs in the
Appendix). This critical point is the only stable one. Its stability holds for all
positive values of the reaction coefficient g and is oscillatory for all plausible
values of g, as long the system is at all profitable. This means that for any single
displacement, the system tends to oscillate back toward the normal level of
capacity utilization. More importantly, in the face of random shocks representing
a multitude of concrete factors and disturbances, the system tends to cycle
endlessly, alternately overshooting and undershooting the normal level of capac-
ity utilization. Note that since the adjustment of the fixed investment share is
denominated in time units T, while that of the circulating investment share is
denominated in some shorter time unit t, it follows that the period of the fixed
investment cycle is likely to be greater than that of the circulating investment
cycle. Figure 3 below shows the simulation results for capacity utilization u in
relation to the normal level u = 1. Figure 4 shows the corresponding behavior of
profit on actual production and of profit on normal capacity production (profit on
"warranted" output).
* * *
80 ANWAR SHAIKH
Figure 3. Capacity Utlllzatlon.
1.5
1.4
1.3
u
1.2
1. 1
1.0
0.9
0.8
0.7
0.6
0.5
0 100 200
Time
Figure 4. Normal Profit and Actual Profit.
500.0
450.0
400.0
350.0
300.0
250.0
200.0
150.0 Profit on normal

capacity output - - - - - -
100.0
- P r o f i t on
actual output
50.0
0.0
0 25 50
Time
The aim of this chapter has been to present a new approach to the question of the
role of effective demand in accumulation. The first step in this direction was an
attempt to create a simple framework which was general enough to encompass the
essential differences between Marxian, Keynesian and Kaleckian approaches to
this issue. Issues which are not central to the above approaches (such as the
effects of workers' savings, of the adjustments in inventory and money stocks, or
of the difference between short-term and long-term debt) were ignored, while
others which did play a central role in one or the other of the main approaches
(such as the constancy of prices, wages, and interest rates in Keynesian and
Kaleckian [KK] theory, or the link between investment in circulating capital and
output growth in Marxian theory) were retained. Since Marx's schemes of repro-
duction abstract from aggregate borrowing or hoarding (Bleaney 1976, 106-7),
while KK theories are crucially dependent on the assumption that finance is
"freely available" at a constant rate of interest (Asimakopulos 1983), it was
particularly important to retain this latter assumption in order to establish that it
was not a decisive factor in distinguishing the two sets of approaches. What did
turn out to be decisive were the crucial links between accumulation expenditures,
finance, bank credit, and the burden of debt.
But the question of credit is only half of the story. An equally important
difference arises in the analysis of accumulation. Dynamic analysis, as in Marx
and Harrod, tends to see growth as an inherent aspect of production and invest-
ment plans, so that it is their trend which is seen as reacting to market feedback.
In contrast to this, both Keynes and Kalecki adopt a notion of essentially passive
firms aiming at attaining a given level of output. Production plans are implicitly
static, and it is the level of production (rather than its trend) which is taken to
respond to feedback. 4 Mathematically speaking, Marx's (and Harrod's) reaction
functions tend to be formulated in terms of proportions or rates of growth,
whereas those of Keynes and Kalecki tend to be cast in terms of absolute levels of
variables. This is a difference which becomes quite crucial in the analysis of
macroeconomic growth.
In latter part of this chapter, the above considerations were used to develop a
simple but powerful macroeconomic model of cyclical growth. The proportion of
potential profit devoted to expansion of output was assumed to respond posi-
tively to the level of excess demand and negatively to the burden of debt ser-
vice payments. This was shown to give rise to persistent short-run cycles cen-
tered around expanded reproduction in the Marxian sense. Over the longer
run, the proportion devoted to expansion of capacity was assumed to rise when
capacity utilization was above normal (and fall in the opposite case); this sim-
ple assumption was found to lead to persistent long-run cycles centered around
normal capacity utilization (the Harrodian warranted path). The overall model
thus generates two distinct cycles which oscillate around a growth trend ultimate-
ly regulated by the intrinsic profitability of the system. Unlike most modern
approaches, no recourse is made to external factors such as technological change
82 ANWAR SHAIKH
or population growth in order to explain the basic growth trend, 5 and there is no
presumption that the system tends to achieve the full employment of ll!hor (as
opposed to the normal utilization of fixed capital). In this sense, the model
presented here is a concretization of the theory of effective demand implicit in the
classical/Marxian tradition (see Shaikh 1978).
Many aspects of this approach remain to be developed. For instance, the
introduction of sustained government deficit spending introduces a new factor, in
that it seems to give rise to a corresponding sustained excess demand. This seems
to suggest a formal basis for a link between deficit spending and inflation, at least
under conditions of normal growth. Similarly, a falling potential rate of profit
seems to produce qualitatively new behavior, in that the stable growth cycles
analyzed here are eventually undermined and turn unstable at the point where the
mass of profit-of-enterprise becomes stagnant. Both of these results are very
suggestive of classical and Marxian arguments. Lastly, it is possible to generate
deterministic limit cycles instead of the stochastic ones explored here by specify-
ing slightly different functional forms for the two accumulation reaction func-
tions. The important thing is that the general approach adopted in this paper
seems to provide a very fruitful and dynamic alternative to traditional theories of
effective demand.
Appendix: Analysis of Stability
Stability of the Fast Adjustment Process
The fast adjustment process is characterized by equations (23)-(24) above. Defining z

= e - b, they can be rewritten as
(29) e= - hie + h(l +i)z
(30) z= - (l+i)e - (md - i)z - mez
where m = the constant profit margin

i = the interest rate
d = l - (c+k), in which c = the constant propensity to consume out of
profits
k = the constant-in-the-short-run rate of accumulation in fixed capital
h = the reaction coefficient for the circulating capital accumulation function
and m, i, c, k, and hare positive by definition, and dis positive as long as the average
short-run rate of return on fixed investment, r = md/k, is positive (see the discussion
following equation [25)).
The above system has two critical points:

e=z=Oand
me = (1 + i)/i (1 +md), mz = - 1 - (1 + md)i/(l + i).
Its Jacobian J is
-hi h(l + i)
J -[
-(1 + i) - mz -(md - i) -
Linearizing around the second critical point shows that its determinant reduces to
DetJ2 = -h(l +2i +mdi) < 0, since i,m,d are all > 0. This means that the second
critical point is unstable. On the other hand, linearizing around the first critical point e
= z = 0, gives
TR JI = - [hi + (md - i)]
DET Jl = hi(md - i) + h(l + i)(l + i) = h[l + 2i + mdi]
Since h, m, d and i > 0, DET JI > 0. Then a sufficient condition for (local)
stability is md ~ i, because this ensures that TR JO < 0 (Hirsch and Smale 1974, 96).
What is more, it can be shown that the discriminant of this system is negative for all
plausible values of the reaction coefficient h (e.g., for i between .02 and .20 and md
between i and 3i, any value of h between .027 and 144 will yield a negative discrimi-
nant), so that convergence will generally be oscillatory. Lastly, the phase diagram of
uus system of equations (omitted for brevity) indicates that the basin of attraction of
the stable point is very large, since it encompasses both the positive e-space and the
positive z-space. Only for initial points in which both e and z are sufficiently negative
will the model exhibit instability.
Now consider the economic content of the stability condition md ~ i. From
equation (25), th~ short-run regulating rate of accumulati?n a = d and from equa-
tion (20) ma = P/P, thus the stability condition becomes P/P > = i. Now consider
the funds that businesses reinvest in their own operations. It has been assumed that
dividends Rare proportional to potential profits (surplus value) P, so that retained
earnings RE = P - R = P(l - x), where x = the dividend payout rate. The
correspo!lding net incremental return to these reinvest~d funds is the increase ~n
profits P minus the increase in payouts of dividends R and finance charges F.
But since b = ~at the critical point in question, accumulation is self-financing on
average, and F=O at the critical point. Thus the incremental rate of return on
reinvested funds is
r* = (p R)/(P - R) P(l - x)/P(l - x) = P/P
It follows that the stability condition md ~ i is equivalent to the basic economic

84 ANWAR SHAIKH
requirement that the funds reinvested in business earn a rate of return greater than or
equal to the interest rate.
Stability of the Slow Adjustment Process
The structure of the slow adjustment process is given by equations (27)-(28). These
are reproduced below in slightly different form.
(31) k=-gk+gku
(32) ii= m(l - c)u - mku - (rn)ku 2
Here, k and u are the variables, and all the others are positive constants:
g = the reaction coefficient for the rate of accumulation in fixed capital

m = the profit margins on costs
rn = the rate of profit on fixed capital at normal capacity (the normal fixed
capital rate of profit)
c = the capitalists' propensity to consume out of potential profits < 1
Once again, there are two critical points:
u = 1, k = k* = [m(l - c)]/[m+rn] > 0 and

u = 0, k = 0.
Forming the Jacobian of this system,
J =[ g(u -
-[m+rn]U
1) gk
m(l - c - k) - 2(rn)ku
J
Linearizing aro!!nd the second critical point u = k = 0, the system is found to be
locally unstable, since Det J2 = - gm(l - c) < 0. On the other hand, in the vicinity
of the first point u = I, k = k*, since k*[m+rn] = m(l - c), the Jacobian, its trace
and and its determinant reduce to
gk* ]
-rnk*
TR JI = - rnk* < 0, DET JI = gk*[m + rn] > 0
which implies that the first critical point is locally stable. Moreover, the convergence
implied by this stability is generally oscillatory, because plausible values of m, c, and

rn yield a negative discriminant for all but the very smallest values of the reaction
coefficient g (e.g., for all m, c, rn between .1 and .5, a reaction coefficient g > .05 is
more than sufficient to guarantee oscillatory behavior).
Since capacity utilization cannot be negative, u ~ 0. The corresponding phase
diagram of this system (omitted for brevity) shows that any trajectory in the positive
quadrant will converge on the first critical point u = 1, k = k*. The slow adjustment
process is then stable around the warranted path.
Notes
l. Strictly speaking, net investment is the difference between gross fixed investment
and current retirements IRt. rather than current depreciation allowances DEPt. But the
difference between the latter two is not important here.
2. Bleaney points out that Marx abstracts from all credit and hoarding, which means
that an increase in investment must be financed by a corresponding decrease in some other
form of demand, such as capitalist consumption. This explains why there is no multiplier
in Marx's analysis of the schemes of reproduction, even when investment changes to
accommodate the transition from simple to expanded reproduction, etc. From this,
Bleaney leaps to the conclusion that the introduction of credit into the Marxian schema
would "lead logically to the Keynesian solution" (Bleaney 1976, 107). Our analysis
makes it clear that his conclusion is quite unwarranted.
3. In Marxian expanded reproduction, supply equals demand and borrowing equals
zero for each of the departments of production, and hence also for the aggregate. Here, we
focus solely on the aggregate.
4. Only in an aside on Harrod does Kalecki modify his basically static focus to try to
account for' 'an expanding economy.'' But the analysis is very awkward and seems largely
designed to support Kalecki's earlier conclusions concerning the inherent static tendency
of accumulation in the absence of external factors such as technical change (Kalecki 1962).
5. Kalecki (1965) relies on technical change and external markets to explain growth;
Goodwin (1986) relies on exogeneous population growth and technical change; while
Foley (1985) relies on the growth in the exogeneous money supply.
References
Asimakopulos, A. 1983. "Kalecki and Keynes on Finance, Investment, and Saving."
Cambridge Journal of Economics 37(Sept./Dec.).
Bleaney, M. 1976. Underconsumption Theories. New York: International Publishers.
Bureau of Economic Analysis (BEA). 1980. Definitions and Conventions of the 1972
Input-Output Study. Washington, D.C. (July).
Chipman, J. S. 1951. The Theory of Inter-Sectoral Money Flows and Income Formation.
Baltimore: John Hopkins Press.
Dumenil, G. 1977. Marx et Keynes Face a la crise. Paris: Economica.
Foley, D. 1983. "Say's Law in Marx and Keynes." Mimeo.
- - - - . 1984. "Money, Accumulation, and Crises." Mimeo.
- - - - . 1987. "Liquidity-Profit Rate Cycles in a Capitalist Economy." Journal of
Economic Behavior and Organization 8(3):363-77.
Garegnani, P. 1979. ''Notes on Consumption, Investment and Effective Demand: a Reply
to Joan Robinson." Cambridge Journal of Economics 3: 181-87.
86 ANWAR SHAIKH
Goodwin, R. M. 1986. "Swinging Along the Turnpike with von Neumann and Sraffa."
Cambridge Journal of Economics 3:203-10.
Hirsch, M. W., and S. Smale. 1974. Differential Equations, Dynamical Systems, and
Linear Algebra. Orlando: Academic Press.
Kalecki, M. 1939. Essays in the Theory of Economic Fluctuations. London: Allen and
Unwin.
- - - - . 1962. "Observations on the Theory of Growth." Economic Journal
(March): 135-53.
- - - - . 1965. The Theory of Economic Dynamics. New York: Monthly Review.
Kenway, P. 1980. "Marx, Keynes, and the Possibility of Crisis." Cambridge Journal of
Economics 4(1): 23-36.
Keynes, J. M. 1939. The General Theory of Employment, Interest, and Money. New
York: Harcourt, Brace, and World. 1964.
Marx, K. M. 1967. Capital. New York: International Publishers.
Minsky, H. 1982. Can "It" Happen Again?: Essays on Instability and Finance. Armonk,
New York: M.E. Sharpe, Inc.
Mullineaux, A. W. 1984. The Business Cycle after Keynes: A Contemporary Analysis.
New Jersey: Barnes and Noble.
Robinson, J., and J. Eatwell. 1973. An Introduction to Modern Economics. London:
McGraw-Hill.
Shaikh, A. 1978. ''An Introduction to the History of Crisis Theories.'' U.S. Capitalism in
Crisis. New York: Union for Radical Political Economics: 219-41.
- - - - . 1984. "The Transformation from Marx to Sraffa." Ricardo, Marx, Sraffa.
Ernest Mandel, ed. London: Verso.
Taylor, L. 1985. "A Stagnationist Model of Economic Growth." Cambridge Journal of
Economics 9(4): 383-403.
The Real and Financial Determinants
of Stability:
The Law of the Tendency
Toward Increasing Instability
Gerard Dumenil and Dominique Levy
The primary aim of this chapter is to present a theory of business cycles and their
secular evolution, which is elaborated within a dynamic model of capitalist
production. We will show that this model, described below, can produce a
number of distinct "regimes" such as balanced growth, overheating, stagnating
growth, or crises. The business cycle will be interpreted as a short-run sequence
of such regimes. This theory allows for differentiation between various stages in
the actual evolution of business cycles in the history of capitalism, from the
traditional nineteenth century pattern to the contemporary phenomena observed
today. This ability to make historical differentiations is a major advance over
previous attempts to model the business cycle which have treated every historical
stage in the evolution of the business cycle as equivalent. Finally, this evolution of
the business cycle is linked to Marx's law of the tendency for the rate of profit to
fall and a corollary to this law is formulated which will be called ''the law of
increasing instability.''
Although the analysis of crises presented below is not directly derived from the
work of Smith, Ricardo, or even Marx, we nevertheless contend that it is classi-
cally inspired in the sense that it is based on a dynamic model whose assumptions
are classical, and because it clarifies many aspects of Marx's view of business
cycles in relation to the law of the tendency for the rate of profit to fall. In
A preliminary version of this study has been published in the Economic Forum, Univer-
sity of Utah, Winter 1986-87. We thank M. Glick for his aid in the translation of this text
into English.
87
88 GERARD DUMENIL & DOMINIQUE LEVY
particular, it provides an answer to the difficult question, "Why does profitabil-

ity matter?"
Our framework of analysis was developed in a previous study of the classical
analysis of competition and the formation of prices of production (Dumenil and
Levy 1983, 1987a). The framework of these original studies was multi-commod-
ity (two or three commodities). However, in the present study, a single-commod-
ity approach is utilized for the sake of simplicity. Although this can be interpreted
to mean a switch from a micro to a macro approach, the basic principles and the
model remain unchanged.
An important aspect of the present study is that the financial determinants of
the cycle are also analyzed: the issuance of money, repayments of loans and
interest, business failures, and monetary policy. In addition, the relative impact
of real and financial determinants is discussed.
This chapter will consist of four parts:
• Real Stability and Instability. In this part, the model which underlies the
business cycle analysis will be presented. Minimum reference will be made to
monetary and financial factors. Nevertheless, the fundamental aspects of the
analysis can be derived independently of monetary considerations.
• Financial Stability and Instability. This part introduces a macroeconomic

model with money and the specific results obtained in relation to the treatment of
money are established.
• Business Failures and Stagflation. In this part, some elaboration is added to

the model with money, in particular a treatment of firm and bank responses to
situations of tight liquidity. Business failures are also explicitly incorporated into
the model.
• Historical Tendencies. This part attempts a formulation of a theory of the

forms of business cycles, from its early forms in the nineteenth century to the
contemporary patterns of events in contemporary capitalism. Following this
formulation, the relationship of the theory of business cycles with the law of the
tendency for the rate of profit to fall is introduced.
Real Stability and Instability
The object of this first section is to define a minimum framework of analysis in

which the issue of the stability of capitalist production can be analyzed. The
fundamental principles which govern what we call disequilibrium microeconom-
ics are presented, along with the results obtained in a macroeconomic model
derived directly from this microeconomics.
THE TENDENCY TOWARD INCREASING INSTABILITY 89
Disequilibrium Microeconomics
Our research stems from the attempt to reinterpret the analysis of the competitive
process by the classics (Smith 1776; Ricardo 1817; Marx 1863). 1 In the classical
view, economic agents react to the observation of disequilibrium. Capitalists
seeking maximum profitability for their capital, for example, move a fraction of
it if they notice that the rate of profit is superior in other industries. This
profitability differential is what we call ''the evidence of disequilibrium.''
Observation of disequilibrium - Modification of behavior
Therefore, the formalization of the classical analysis relies on the modeling of the
behavior of the agents in a situation of disequilibrium.
This approach to the analysis of economic activity (although quite natural and
evident in many literary commentaries on economic activity) is ignored by
mainstream economic theory, with very little exception. In the traditional
perspective, the behavior of economic agents is modeled with the assumption that
equilibrium always prevails. We will denote this perspective "equilibrium
microeconomics," which can be counterposed to "disequilibrium microeco-
nomics.''
This latter paradigm was applied by the classical economists to three types of
situations:
1. The observation of differences between rates of profit induces the mobility

of capital among industries.
2. The observation of the inequality of supply and demand leads to the modifi-
cation of prices. Excess supply is an inducement to diminish prices, and the
reverse applies for excess demand.
3. The observation of the inequality between supply and demand also leads to
the direct adjustment of outputs, without waiting for the classical indirect trans-
mission:
Modification of prices - Change in profitability - Migration of

capital - Modification of the output
This direct mechanism, which we called "direct control of quantities," was

emphasized by Ricardo (1817, ch. 4).
These adjustments are not made by an auctioneer prior to real economic
activity, as in the Walrasian paradigm of the tatonnement. Instead, classical
adjustment occurs in the course of real economic life (production, transactions,
consumption, etc.). 1\vo types of agents participate in these adjustments. First,
enterprises decide the level of outputs and the modifications of prices. Second,
the mobility of capital is insured by the "centers for the allocation of capital." In
real capitalism, such centers exist under a number of forms. Enterprises engaged
in the production of several products are one example of such agents, while
holding companies, banks, and individual investors are other examples.
The evidence of disequilibrium is always assessed individually, not globally,
on the basis of limited information. For example, enterprises become aware of
excess supply or demand through the levels of their inventories.
When fixed capital is explicitly treated in the modeling of the above behaviors,
two variables play a crucial role: the ratio of capacity utilization of fixed capital
(denoted u) and the ratio of inventories to outputs (s). The behavior of enterprises
tends to stabilize the value of these two variables. However, enterprises do not
strive to obtain full capacity utilization or zero inventories. In a world fraught
with uncertainty, enterprises attempt to reach certain values of the two ratios
which they consider optimal. We call these values "normal values." The normal
value of the ratio of capacity utilization (for example, 80%) is denoted, u, and the
normal value of the ratio of the inventory to output (for example, one month of
output) is denoted s.
Two further points are added to the three points made by the classics and
described above:
4. Enterprises do not necessarily use funds in the exact amounts which are
allocated to them by the centers. The level of investment which corresponds to the
full use of the allocated funds is what we call "normal investment." However,
depending on the level of their ratio of capacity utilization of fixed capital, they
can only use part of these resources or, to a certain extent, may overstep the
limitations of the allocation.
5. In an extended model which will be presented below, a treatment of the

issuance of money will be incorporated. The behavior of the agent which assumes
this function is analyzed in a similar manner reacting to the observation of
disequilibrium. The issuance of money is adjusted on the basis of two variables:
• Disequilibrium in the ratio of capacity utilization of enterprises (excess or

underutilization of capacity) which expresses the desire on the part of enterprises
to demand financial resources.
• Inflation.
On the basis of such foundations, we have built several models. The first
model of this type demonstrated the relevance of the classical analysis of competi-
tion (Dumenil and Levy 1983). An analytical treatment of this model with two
commodities and no fixed capital was provided in that study. A number of further
developments were proposed in Dumenil and Levy (1987a). In particular, a
treatment of two centers for the allocation of capital and the existence of two
processes of production for the same commodity in one industry were added. In
Dumenil and Levy (1986a), we provided an analytical treatment of the stability of

classical long-term equilibrium with any number of centers and enterprises. The
existence of fixed capital was taken into account in a number of other contribu-
tions which we will refer to throughout the present study:
1. The so-called ''micro model.'' Three industries are considered which

produce the elements of fixed capital, the elements of circulating capital, and
consumption goods, respectively. There are 13 variables in this model (originally
15, which can be reduced to 13 as a result of the homogeneity of the equations).
This analysis was done by computer simulation (Dumenil and Levy 1985b). An
analytical treatment of a similar model with 10 variables was given in Dumenil
and Levy (1986b).
2. The "macro model." This model is a replication of the micro model for
one commodity. The mechanisms related to the mobility of capital among indus-
tries are abstracted from. Therefore, only two variables exist: u and s. An
analytical treatment of this model is presented in Dumenil and Levy (1987b).
3. The "macro model with money." This development of the previous model
embodies the explicit treatment of money. The number of variables is now equal
to four, u ands, and two new financial variables (the ratio of the stock of money of
the enterprise to the sum of its productive capital and inventories, denoted m, and
the equities of the bank, also divided by the sum of the productive capital and
inventories of the enterprise, denoted e). The main results will be introduced
below.
It is apparent that these basic models can be the basis for a large number of
further developments some of which will be reported below. It is possible to
endogenously determine an exogenous parameter or to add a constraint, depend-
ing on the demonstration made. For example, business failures can be explicitely
treated in the macro model with money, the rate of interest can be given or
determined endogenously, the supply of labor can be limited, etc.
The study of a dynamic model of this type implies the formulation of a relation
of recursion which defines how the values of the variables in one period can be
derived from their values in the previous period. For example, the "macro
model'' above leads to the determination of:
llt+t = U(ut,St)
St+t = S(ut,St)
The functions U and S depend on the parameters which define the technology,
the functions of reactions of the agents, and the ratio of accumulation in each
period. Two issues must be investigated, the existence of equilibria and their
stability. In the mathematical sense of the term, an equilibrium is a set of values of
the variables (u* ,s*) such that:
u* = U(u* ,s*)
s* = S(u* ,s*)
Such equilibria can also be called "stationary states." This concept in a

growth model corresponds to that of "homothetical growth." The issue of the
stability of an equilibrium is more complex. Do we find convergence toward this
equilibrium when the computation defined by the recursion is initiated with given
values of the variables (u 0 ,s0 ) and repeated until infinity? Stability usually de-
pends on the value of the parameters and the initial values of the variables. The
convergence region (also called the "region of attraction") is the set of all initial
values of the variables for which convergence can be obtained.
If several such equilibria exist and are stable, then this means that the economy
described by this model can display different types of stationary states, which can
be called "regimes."
The Equilibria, The Pitchfork
The primary property of these models is that they can, under certain circum-
stances, lead to convergence toward a situation which we call "normal equilib-
rium" (N). This is important because it corresponds to the competitive outcome
described by the classics. When several industries are considered, the classical
notion that the rates of profit are equalized in the long-run is achieved. The ratios
of capacity utilization of fixed capital, as well as the ratios of inventory to output,
attain their normal values. 2
A second property of these models is that other equilibria, in addition to the
normal equilibrium, can exist and can be stable:
1. Overheating (0). In this situation, the ratio of capacity utilization is above
normal, and the ratio of inventory to production is inferior to normal. The rate of
growth is also above normal. Inflation also characterizes a situation of overheat-
ing.
2. Stagnating growth, or a Keynesian situation (K). In this regime, the ratio of
capacity utilization and the rate of growth are below normal, while the ratio of
inventory to production is above normal. Prices are decreasing (except in situa-
tions of stagflation, which will be considered below).
3. Depression (D). In a depression, the economy collapses toward zero if
nothing prevents the dramatic decline.
The stability of these equilibria and the configuration of their region of attrac-
tion is primarily conditioned by the value of a parameter which models the
behavior of enterprises. This parameter is denoted E and measures the intensity of
the reaction to scale up or down the level of capacity utilization, in response to the
change in inventories (E measures this intensity only in a vicinity of normal
equilibrium).
In the macro model without money, the existence and stability of the equi-
Figure 1. Existence and Stability of the Equilibria: The Macro Model without Money.
E
N
The existence and stability of the equilibria are primarily determined by, E, a parameter which
measures the intensity of the reaction of enterprises to disequilibria in the ratio of their inventories to
outputs. The equilibria which exist and are stable in the corresponding intervals of the values of E are
indicated in the.figure: Normal Equilibrium (NJ-Overheating (OJ-Stagnating growth, or Keynes-
ian situation (KJ-Depression (DJ.
libria depend on the values of e in the following manner (see Figure 1):
1. Depression exists and is an equilibrium for all values of e.

2. When e inferior to a given value e1 , normal equilibrium exists and is stable.
In this situation, the convergence region is large. Thus, only a major disturbance
can change the economic course toward another equilibrium (depression).
3. For e between e1 and e2 , both overheating and Keynesian equilibria exist and
are stable. Their convergence regions are limited in one direction. This means
that in spite of the size of the region, in general, it can be easily destabilized when
a shock pushes the economy in one particular direction.
4. For e higher than e2 , Keynesian equilibrium is no longer stable. For these
values of e, only overheating remains stable (abstracting from depression).
In the micro model without money, e can not be too small. Below a certain
threshold, the adjustment of relative prices and of the proportions of outputs will
not be achieved.
A further aspect of the relationship between equilibria and e is that, in Keynes-
ian equilibria and situations of overheating, the equilibrium values of the ratio of
capacity utilization and ratio of inventory to outputs, depend on the value of e. For
example, comparing two situations of overheating reached for different values of
e, the ratios of capacity utilization will be different. A higher e will produce a
higher ratio of capacity utilization. Therefore, it is possible to plot this relation-
ship between e and the corresponding equilibrium value of u or s. The results are
presented in Figure 2. This figure clearly illustrates the existence of a bifurcation
for e = e1 • The configuration obtained is that of a pitchfork (see Guckenheimer
and Holmes 1983).
Figure 2. The Pitchfork: The Macro Model without Money.
····-················-·····-···········-·····-···········-······································-·····-·····
0
N
u= .8
: .... . . . .
.5 ! ··-·-. ... ... ......
........
I
E1 E2 I D
.1 .2 .3 .4
This is an extension of Figure I. The equilibria are represented by their value in u. A similar fig-
ure could be presented for s. The cotifiguration obtained is that of a pitchfork.
The classics, especially Marx, have never contended that capitalism is globally
stable, in spite of the role played in their paradigms by the formation of prices of
production. This former view would be an extreme form of an apology for
capitalism and corresponds to the old but still influential free-market conten-
tion-laissezfaire, laissez passer. In other words, an equilibrium exists and when
disturbed, it will be autonomously restored, provided that no perverse behavior
(for example, rigidities) interferes.
Capitalism is only partially stable, and the degree of stability depends on given
circumstances. N globally stable would be equivalent to the impossibility of
crises. N always unstable would be equivalent to the irrelevance of the classical
analysis of the formation of prices of production. The existence of various
regimes and the limited character of stability (nonglobal stability) is, in our
opinion, a fundamental characteristic of capitalism. The pattern of equilibria
which has been revealed must be considered as a powerful device for the analysis
of capitalism. The remainder of this study is an attempt to support this proposi-
tion.
The fact that several regimes exist and are surrounded by given regions of
attraction shows that the economy can switch from one regime to another, as a
result of shocks to the variables or changes in the values of the parameters
(reactions functions, technique, etc.). The study of these movements introduces a
theory of business cycles. Business cycles correspond to specific chains of
regimes which switch from one to the other.

Before elaborating on the theory of business cycles described above, we will
articulate the monetary and financial aspects of the analysis and then treat the
issues which are related to liquidity squeezes.
Financial Stability and Instability
In the above investigation, the analysis focused on the real determinants of

stability as opposed to the monetary and financial dimensions of the analysis. This
second part now deals with the latter issues. The monetary and financial frame-
work of analysis embodied in the model (macro model with money) are present-
ed, followed by a discussion of the forms of interaction between the real and
monetary determinants of stability.
Money and Finance in the Model
In the workings of a real economy, as well as in the model, many transactions

require the use of money. For example, purchases and sales on the commodities
market are settled through transfers of money. But, money is also necessary in
order to realize financial transactions: the migration of capital among industries,
the payment of wages and interest, the distribution of dividends, and the repay-
ment of debts. Moreover, since the economy is growing, the quantities of money
which are required for such purposes must also increase. For this reason, the
issuance of money must be incorporated as a necessary component of the model
with money.
We define two agents in the ''macro model with money,'' one enterprise which
produces the only commodity, and a bank. Since this bank is unique, it must
simultaneously be conceived as an ordinary bank and a central monetary institu-
tion, such as the Federal Reserve. The money which is created by this bank is
comprised of paper money and deposits accessed by transfers or checks. Howev-
er, we will not distinguish between the two types of monetary assets. We assume
that no interest is paid on deposits in the banks. We further abstract from recipro-
cal loans among enterprises (since only one enterprise is considered), as well as
loans to and from households and the state. Money is exclusively issued by the
bank and never originates from other sources, such as private treasuries or
foreign exchange.
When considering the issue of money, the only sound method of analysis is
through the balance sheets of enterprises and banks. Two sides of the balance
sheet are implied in the expression "money and finance." Money refers to the
asset side and finance refers to the liability side. The balance sheets of the
enterprise and of the bank in our model are presented in Figure 3. 3 The notation is
as follows:
Figure 3. Balance Sheets of the Enterprise and of the Bank:

Stock of Money Held by Final Consumers.
/P/ /Ee/
-
/Eb/
/C/
/Le/ /Lb/ _-_-_-_-] / M0 /
JMbJ
/Me/
/Pl Stock of productive capital I Cl Stock of commodity capital

IM" I Stock of money capital IE" I Equities of the enterprise
/L 8 / Loans received by the enterprise /Lbl=IL 8 / Loans by the bank
IF!' I Equities of the bank
IM'I Stock of money outside !M'! =IM"I + IM'I Stock of money
of the circuit of capital issued by the bank
I Pl + I Cl + IM" I is "Total Capital" IP/ + IC/ is "Engaged Capital"
Stock of money of the enterprise Equities of the bank

m = e =
Engaged capital Engaged capital
The enterprise must decide on price, the capacity utilization of its stock of
fixed capital, and its level of investment. Prices and capacity utilization are
decided in the manner described in the first part of this chapter. Prices are
modified on the basis of the level of involuntary inventories. The ratio of capacity
utilization is fixed on the basis of its previous value and the level of involuntary
inventories.
The decision concerning the ratio of utilization (decision to produce) only
depends on the "real" variables. This decision is not a function of the availability
of money, except in situations of extreme liquidity squeezes, when the enterprise
is rationed and cannot execute its plan for production (even with zero investment)
because of a lack of money.
On the contrary, the decision to invest is made on the basis of the funds
available for investment. After production has been decided and the necessary
purchasing power secured, the remaining amount of money determines the ability
to invest (we abstract here from precautionary behavior which would induce
enterprises to conserve stocks of money in order to honor future commitments).
THE TENDENCY TOWARD INCREASING LIABILITY 97
This chain of events can be made more explicit by the following sequence of
actions which begins at the opening of the market:
1. The inputs are purchased, as well as the elements of fixed capital corre-
sponding to the new investment, and wages are paid. Money flows back to the
enterprise.
2. The enterprise repays its debts which have matured, both principal and
interest.
3. The enterprise pays dividends. Specifically, by this transaction we mean
the net balance of the payment of dividends and the purchase of newly issued
shares by dividend re~eivers.
4. New loans are issued by the bank to the enterprise. The new money is
added to the already existing stock held by the enterprise.
5. The enterprise decides on its level of utilization of fixed capital and pur-
chases the inputs necessary to set in motion the existing fixed capital (prior to the
new investment) at the decided level of capacity utilization.
6. The enterprise decides on the level of new investment on the basis of the
size of the remaining stock of money.
The three first transactions are mandatory actions taken by the enterprise.
Only the fourth event is voluntary and depends on the initiative of the enterprise.
The demand for loans affects their total volume. The supply ofloans on the part of
the bank is not limited by the availability of resources, since only one form of
money exists which can be created by the bank. However, the issuance of money
must be limited in some way. It is very difficult to conceive of an ex ante control
procedure. One method might be that the bank would estimate an amount of
"normal" transactions in some sense, and thus compute the value of the stock of
money exactly corresponding to these transactions. Such a procedure would lack
realism, however. For this reason, the issuance of money is controlled ex post in
our model, as explained earlier. In the decision to issue money, the two indicators
of disequilibrium are the eagerness of the enterprise to borrow, based on its ratio
of capacity utilization, and the evidence of inflation (or deflation).
Parameter r is the debt-equity ratio for the enterprise. The rate of change of the
price level is j. The decision to issue money through credit is governed by:
in which gi and g,. are two strictly increasing functions of their arguments and are
equal to one when their arguments are equal to zero.
The main results which can be obtained in this model are the following:
1. It is possible to reach equilibrium by the simultaneous convergence of real

and financial variables.
Figure 4. The Macro Model with Money: Convergence toward Normal Equilibrium.
20
15
10
5
i = 2.5 10- 4
0 '---~~~~~...._~~~~~ ........~~~~~-"~~~~~--''--_.
0 200 400 600 800
This figure presents the convergence of the two rates of return on equity: enterprise r, and bank ,. ,
toward equality at a level greater than the rate of interest. This rate ofprofit is the rate ofprofit over
the real components of capital IP/+ !Cl, if the stock of money out of the circuit of capital is equal to
zero.
2. A variety of regimes can be obtained as in the previous model which

excluded money: normal equilibrium, overheating, stagnating growth or
Keynesian situations, depression (if we abstract from business failures).
In normal equilibrium, the financial variables defined in the first part stabilize
to specific values. In overheating situations, mis superior to its value in normal
equilibrium, and e is lower. In the Keynesian regime, the situation is inversed.
Moreover, it is interesting to note that the rate of profit over equity of the bank is
equalized to that of the enterprise, as shown in Figure 4.
This last result represents an important finding concerning the relationship
between the rate of profit and the rate of interest. The rate ofprofit is not equal to
the rate ofinterest paid on loans. This result is not due to a specific assumption of
the model but is a necessary feature of a financial system in which money can be
issued by private agents and in which the interest charged on loans is greater than
the interest paid on deposits. First, the rate of interest on deposits is smaller than
the rate charged on borrowings (it is assumed to be equal to zero in the model).
Second, IM1'1 is the sum of notes and all deposits (checking, saving, etc.); interest
is paid only on some of these elements.
Consider the balance sheet of the bank. The income of the bank is the interest
paid. If i is the rate of interest, this income is equal to i /Lb I. The rate of return on
equity is therefore:
The rate of return on equity for the enterprise is re, and its rate of profit over
engaged capital is r. When equilibrium prevails and the stock of money out of the
circuit of capital is zero, rb = re = r, thus:
1E'1
i = r-b- and i < r if IM" I > 0
IL I
The mere existence of money (issued by a capitalist institution) implies that the
rate of interest is inferior to the rate of profit by an amount determined by the
relative quantity of money required within the circuit of capital.
Real and Monetary Determinants of Stability
A crucial result, in what has been presented so far, is that the inclusion of money
and finance in the model does not alter its basic properties. This finding stresses
the primary importance of production and the real aspects of the workings of
capitalism. However, this feature does not imply that money is "neutral." In this
section, we will briefly investigate some dimensions of the relationship between
real and monetary determinants, as revealed by the model.
Money and Finance as Triggers
In a model with regimes, the sequence of economic events can be altered by the
influence of' 'shocks.'' When money is integrated in such a model, one important
form of shock is a monetary shock. Therefore, money can propel the economy
from one regime to another.
An example of such monetary influence concerns the rate of interest. Consider
an economy in an overheating situation-a regime in which the convergence
region is very limited in one direction. A sudden rise in the rate of interest can
destabilize the system and propel it into a recession, bounded by a Keynesian
situation, or a collapse. Both real and monetary determinants play a role in this
phenomenon. The "real" aspect of the mechanism defines the underlying config-
uration on which the impact of the movement of the rate of interest can be
analyzed. The propensity of capitalist production to generate endogenous reduc-
tions in productive activity constitutes the fundamental factor. Money only acts as
a trigger. This type of determination certainly played a role in the nineteenth
century and still represents an important aspect of contemporary capitalism.
The shock which triggers the switch may also concern the monetary aggre-
gates (sharp curtailment of the issuance of money, or sudden reflation) rather than
the rate of interest. The change in monetary policy in the U.S. economy since
1979 clearly illustrates the crucial role played by such events.
Money and Finance and the Value of The Reaction Parameters
A second and more complex aspect of the interaction between real and monetary
determinants is the effect of a number of interrelations which are excluded from
the model. We will consider only one example where the parameter E is dependent
on the value of the rate of interest.
Parameter E measures the intensity of the enterprise reaction to scaling up or

down its activity when its inventories differ from their normal value. It is easy to
understand that this type of reaction is influenced by the rate of interest. A high
rate of interest is an inducement for enterprises to manage their inventories tightly
(high E). Short-term borrowing is costly, and conversely, excess liquidity can be
deposited in order to yield a considerable return (in real life, not in the model).
Similarly, a low rate of interest, in relation to the cost of scaling down activity, is
an inducement to ignore rising inventories and to wait for the restoration of
demand.
Financial Instability
Another important aspect of the interaction between real and financial determi-
nants is their possible divergent paths. While the model can be stable from the
point of view of the real mechanisms, it can be disrupted by the influence of the
financial mechanisms. This general idea will only be briefly sketched on the basis
of Figure 5. In the first part of the experiment described in this figure, the rate of
interest is low. The normal equilibrium is stable. As we increase the rate of
interest (but keeping it below the rate of profit), normal equilibrium becomes
unreachable and the economy begins to decline and is driven downward. In our
opinion, the result of this experiment is not a common pattern of events in
advanced capitalist countries; however, it may correspond to the situation of
many third world countries where the burden of debt is becoming unbearable.
Business Failures and Stagflation
This part of the study introduces a group of possible developments of the model
related to the reactions of the enterprise and the bank to a situation of liquidity
squeeze within the productive system. It also develops a number of ideas concern-
ing business failures which were raised in the first part. The various reactions
which can be associated with the squeeze are presented, followed by an explana-
Figure 5. The Macro Model with Money: Financial Instability.

u
.5
i = 0.0025
O'--~~~~~...._~~~~~-"-~~~~~---'~~~~~~...__.
0 200 400 600 800

In the first part of the experiment described in this figure, the rate of interest is low. The normal
equilibrium is stable. In the second part of the experiment the rate of interest is higher. Normal
equilibrium can now no longer be reached, and the economy is declining.
tion of the way in which business failures are accounted for in the model. Next, an
important aspect of contemporary economic policy is analyzed: the trade-off
between inflation and business failures under stagflation.
LiquUJ.ity Squeeze
As a result of borrowing, the enterprise is committed to paying back its debt with
interest. This obligation is proportional to the amount of debt incurred. In the
model, the enterprise must pay out at each step, a given fraction, o,
of its total
debt with the interest, i.e., a total /Lel(i + o). This quantity must be compared at
each period with the stock of money which is held. If the stock is deficient, then
the enterprise goes bankrupt.
As long as the stock of money is clearly larger than the commitment to pay, the
behavior of the enterprise can be as described earlier. However, if the situation
becomes tighter, specific responses can be initiated by the enterprise and the
bank. Three such reactions have been modeled:
1. The enterprise modifies its behavior concerning the determination of

prices. Its sensitivity to stockpiling is maintained, but an element of rigidity is
introduced and inflation is created. The enterprise tends to alleviate the financial
pressure by increasing prices. It has a choice between diminished activity (which
would be avoided by diminished prices) and the inability of honoring past com-
mitments. Stagflation originates from this choice of response. This is usually
described as cost-push inflation, although in our view its origin is primarily
financial.
2. The enterprise can retain funds for the purpose of future debt payment and
payment of interest, forgoing additional finance of its activity.
3. Lastly, the monetary authorities can initiate policies to maintain economic
activity. Two forms of such policies have been considered:
• We call the first form of policy ''bolstering. ' ' The bank increases the money
supply in reaction to the demand emanating from the enterprise.
• We name the second form of monetary policy "rescuing." In this case,
money is specifically created in order to salvage the enterprise facing bankruptcy.
Money is specifically created in amounts which allow the enterprise to meet its
immediate obligations. Since the bank is the only lender in our model, this money
is only created to be immediately destroyed when the enterprise pays back the
bank. This paradoxical behavior allows the enterprise to survive.
The inflationary tendency is modeled in a macroeconomic perspective. We
define a variable, 11, which accounts for the degree of liquidity shortage:
Jiff
"=
A limit value exists, ii of 11, below which at least one share of the economy is
confronted with liquidity problems. The equation which defined the determina-
tion of the price by the enterprise was:
It is now modified by the addition of a second function:
in which h is an increasing function of its argument for the positive values of this
argument, and h = 1 for the negative values of its argument.
In a similar manner, the issuance of money, modified to include the "bolster-
ing" policy, is determined by:
in which hv possesses the same properties as hv above.

To model the "rescuing" policy requires the development of a further frame-
work of analysis which will be developed below.
Business Failures
In a model in which only one enterprise is considered, it is difficult to study a

phenomenon such as bankruptcies. Normally, an analysis of bankruptcy would
require the existence of a large number of enterprises, since once a unique
enterprise in a one enterprise model is bankrupt, the analysis ends. In the present
study, we will conserve the macroeconomic framework of analysis and consider
the enterprise in the model, as the average of a large set of enterprises which
differ in regard to l, their ratio of indebtedness:
l =
IP/ + /Cl + IM"!
First, the computation of the flows of money necessary for the payment of debt
and interest is made. Then the share of the enterprises capable of paying is
determined, depending on the distribution of enterprises in l. The model is built
in such a way that it is impossible for all enterprises to default simultaneously.
The assets and liabilities of the proportion of enterprises which fail are deleted
from the economy. 4 The loans which are not repaid are computed as losses for the
bank and diminish its equities.
The fraction of the system with high indebtedness violently disappears in this
process. As a result, the financial structure of the economy tends to be restored.
The consideration of business failures introduces an important complement to
the general pattern of the equilibria presented in the first part of this study (the
pitchfork). A new regime appears as an inferior situation in which the system is
sustained by the constant devaluation of capital. We denote this regime F (see
Figure 6).
Policies, The Inflation/Failures Trade-Off
Previously, two reactions of the bank to a situation of liquidity squeeze were

introduced: "bolstering" and "rescuing." The effects of these policies on the
pattern of the equilibria can be described on the basis of the representation of the
pitchfork in Figure 6. Both policies are successful to a limited extent since the
lower branch of the fork is shifted upward.
If enterprises react to a liquidity squeeze by a tendency to increase prices, as
described before, stagnating growth and inflation can coexist on the lower branch
of the fork. This situation corresponds to "stagflation." Under such circum-
stances, policies of bolstering or rescuing operate in an inflationary context.
Monetary policy must be neither too restrictive, nor too loose. A strict control
increases the bankruptcies or might create a collapse, while a loose policy
accelerates inflation.
The issuance of money is controlled in the model by the parameters of reaction
Figure 6. The Macro Model with Money and Failures: Bolstering and Rescuing.
·u
N
u= .8
.5 .
D
0 '---~~~~~~~~~~~~~~~~~~~~~~~--
0 .1 .2 .3 .4
Same as the legend ofFigure J, plus: Inferior situations with business failures. The effect of bolstering
and rescuing is to shift this branch upward from (a) to (b).
for the bank. If we vary the parameter which measures the sensitivity of the
bank's reaction to monetary difficulties for a given degree of aversion to infla-
tion, we obtain a trade-off between inflation and failures, as shown in Figure 7.
Historical Tendencies
In this last part, a theory of the historical evolution of the business cycle in
relation to Marx's law of the tendency for the rate of profit to fall is formulated.
First, the relationship between Marx's law of the tendency for the rate of profit to
fall and the framework presented above is sketched. This analysis provides an
interpretation of the various forms of the business cycle in the history of capital-
ism. In this context, the relationship between the evolution of capitalism and the
development of economic theory is presented.
The Law of the Tendency toward

Increasing Instability
In this study, we will not discuss either the factual (Dumenil, Glick, and
Rangel 1987a) or theoretical controvers:y (Okishio 1961) around Marx's law of
the tendency for the rate of profit to fall. The purpose of this section is to establish
the following thesis:
A tendency toward a stricter management of capital (an increasing parameter
Figure 7. The Inflation/Failures Trade-off.
o~~~~~~~~~~~~~~~~~~~~~~~~___,~
0 25 50 75 100
In the macro model with money and failures, the impact of policies of bolstering and rescuing in a
stagflationist environment are shown. Varying the intensity of the policy, a trade-off between inflation
and failures is created. j is the rate of inflation, and IJ the proportions ofenterprises which go bankrupt
at each period.
e) can be associated with the historical tendency for the rate ofprofit to fall. This
stricter management results in a tendency toward increased instability. More-
over, most countertendencies to the falling rate of profit can not offset this
propensity toward instability (this movement of e), but instead are the expression
of this modification in the functioning of capitalism.
Actually, two parameters exist in the present form of the model. They both
account for the intensity of the reaction of enterprises to the evidence of disequili-
bria. The first concerns enterprise reactions to stockpiling, the second measures
the reaction to the degree of capacity utilization. High levels of these es corre-
spond to swift and strong reactions to the evidence of bloated or deficient inven-
tories, or excessive or inferior levels of capacity utilization. These strong reac-
tions indicate a form of "tight" management. This strict concern becomes a
necessary feature of the management of enterprises as the rate of profit tends to
decline. The combination of a diminished rate of profit with a relatively high rate
of interest implies the transformation of management in the direction of control-
ling costs and minimizing the quantities of money IM" I, inventories I Cl, and
productive capital IP/, necessary for their operations. This transformation of
management is a crucial aspect of the evolution of capitalism (the consequences
concerning the class structure are not addressed here).
Intensive utilization of fixed capital plus strict management of inventories and
liquidities are important countermeasures to the falling rate of profit. They offset
considerably the fall in profitability, as can be shown empirically. But the crucial
point to understand in this regard is that these countermeasures do not restore the
stability ofcapitalism. On the contrary, they are responsible for what we called in
previous work the "instability in dimension" (Dumenil and Levy 1985b). When
enterprises manage their inventories more carefully, promptly cutting production
in response to excess levels of inventories, they contribute to the restoration of
their profitability. However, such actions jeopardize the general stability of the
system, since any downward signal is transmitted to the whole system (dimin-
ished purchase of inputs, diminished wages, etc.). This type of epidemic can
either develop into crises or stabilize the economy at inferior levels of capacity
utilization.
It is important to recall that Marx associated the tendency for the rate of profit
to fall with accelerated accumulation-not stagnation-and with increasing insta-
bility (Dumenil 1977). The model provides a theoretical justification for this
view.
It must also be stressed here that these forms of management have very
positive effects concerning the control of the proportions of prices and outputs in
capitalist economies. In this sense, individual interests coincide with the general
interest. But, this same capacity to efficiently react to disequilibrium is responsi-
ble for the instability of capitalism in dimension. For this reason, capitalism is
very stable in ''proportions'' and unstable in ''dimension. ''
This property explains why the free-market perspective concentrates its atten-
tion on proportions which are the positive aspect of capitalism. Equilibrium
microeconomics in a general equilibrium model rules out any possibility of
deficient demand, and this explains why the Keynesian revolution was necessary.
Keynes disregarded the issue of proportions and only considered the problem of
dimension in a macroeconomic framework of analysis in which Say's law is
rejected. This property also explains why two paradigms exist in Marx's analysis,
that of stability, at the beginning of Volume III of Capital where the convergence
of prices toward prices of production is analyzed (as well as the adjustment of
outputs), and that of instability, in recurrent references to crises throughout his
work.
This formulation of a law of the increasing instability of capitalism, in relation
to the reaction parameters of enterprises and the tendency for the rate of profit to
fall, introduces a theory of the historical evolution of the forms of the business
cycle, an issue which will be addressed now.
The Forms of the Business Cycle
The pitchfork analysis which has been presented in this study stresses a number of
basic features concerning the origin of business cycles. Moreover, it suggests the
existence of different forms of the business cycle corresponding to successive
stages of capitalism. We will consider four such forms: the nineteenth century,
the between-two-wars, the nineteen-fifties, and contemporary configurations.
The Business Cycle as a Succession of Regimes
Figure 8 presents four possible configurations of the business cycle. In the first
case, figure (a), the value of Einsures the stability ofnormal equilibrium. After an
important shock, the economy progressively moves back to normal equilibrium.
In the case of(b), the economy switches from normal equilibrium to overheating,
then is destabilized toward a Keynesian state, and moves back to normal equilib-
rium. This latter chain requires a change in Eor a lateral shift of the pitchfork. In
the case of (c), the value Eis again given, as in the first instance, but at a higher
level (beyond the bifurcation). The level of activity flips from overheating to a
Keynesian state and vice versa. In the last case, (d), it is a Keynesian state which
is destabilized. Beginning from a situation of stagnating growth (instead of the
traditionnal overheating as in (b), a recession occurs.
The model which is presented in this study is not a model of the business cycle.
However, it is possible to develop this model to obtain endogeneous switches
from one regime to another (for example, E can be transformed into an endoge-
neous variable).
The Nineteenth Century Configuration
If one considers the profile of economic activity in nineteenth century England

(Gayer, Rostow, and Jacobson-Schwartz 1953), it is easy to distinguish two
subperiods. The first period, which can be called "Ricardian," is characterized
by a steady growth which is only interrupted by limited and short lived stages of
recession (cf. Figure 8 [a]). Ricardo referred to these as states of "distress." A
second period might be called "Marxian." In this period a fullfledged business
cycle appears and recurs reliably (cf. Figure 8 [b]).
The traditional pattern of the business cycle, as described by Marx, corre-
sponds to a repetitive chain of events which requires about ten years to thoroughly
unfold. The economy is drawn from one stage to the other in a specific scenario.
Beginning in the trough, the recovey ushers in a phase of balanced growth which
lasts several years. Then a phase of overheating is initiated and lasts approximate-
ly a few months or more. Suddenly the economy is destabilized and declines at
high speed, only to stagnate at low levels of activity. Capital is devalued in the
recession and the whole circuit is reinitiated. There is very little inflation, on
average, during the entire process. It is easy to reinterpret this pattern of events
on the pitchfork. The story goes from a Keynesian state, to normal equilibrium,
to overheating, and into a collapse which stabilizes in the Keynesian situation.
The cycle can be presented as a succession of switches from one regime to the
other. In other approaches to the study of business cycles, the circuit is directly
Figure 8. Four Configurations of the Business Cycle.
~
u u
_____ ................................................................. .r ............................................. .. ... .......... -.. -.... --------... ----.. -.... -.......... ------.. --.. -----------.. -.. ---
...
••
••
•
......_
~ f
u a b
u
-.... -.. -......... --....................................................... -- .... ---... -.......... ---.. ---............ -- -- -............ --- ............................................................................................ -..... --.... ---.. --
c d
This figure illustrates the fact that the business cycle can be analyzed (using the pitchforks presented in Figures 2 or 6) as a succes-
sion of switches from one regime to another. Various configurations can, thus, be obtained.
Figure 9. Business Cycles: The Nineteenth Century Configuration.
u = .8
.5
0 '--~~~~--''--~~~~--'~~~~~--'~~~~~___.~..,
0 500 1,000 1,500 2,000
In this traditional form of the cycle, the value off is usually smaller than f 1 • Only during limited
periods, as a result of the effect of a shortage or a bottleneck, such as deficient labor power, f will
increase and the economy will switch from normal equilibrium to overheating. During this regime, the
economy is destabilized and collapses. It stagnates for a limited period of time, but eventually the
constraint is relaxed and f restored. The economy moves back to normal equilibrium, and so on.
described in relation to a single mechanism; this constitutes an important method-

ological difference between our approach and others. 5 In this theory of the
succession of regimes, inflation is also zero in the average (zero in normal
equilibrium, inflation in overheating, deflation in Keynesian situations).
Since normal equilibrium lasts several years, the value of€ is usually inferior
to €1 • High values are only cyclically reached when the economy encounters a
constraint which affects the behavior of enterprises. When the constraint becomes
effective, the consequence is an increase of€ (stricter management). This is the
process which Marx describes in Vol. III of Capital, where he discusses the
overaccumulation of capital, as well as in the study of the reserve army in Vol. I.
In the analysis of overaccumulation, the rate of profit falls as a result of an
increase in wages. Another constraint is the inability of the credit system to adjust
to the growth of production. In such a case, the rate of interest increases and€
rises temporarily.
The cyclical movement of € can be modeled without specifying the nature of
the constraint. In the example in Figure 9, we assume that the economy is
confronted with such a constraint. The rate of growth of the constrained variable
is inferior to that of the economy in normal equilibrium. The existence of this
restriction has two consequences. First, € is sensitive to the pressure created by
the constraint and varies according to the distance between the actual situation
and the limit produced by the constraint. When the economy moves closer to the
constraint and E increases, it switches to a regime of overheating. Second, when
the economy reaches the constraint, it is autonomously destabilized and tumbles
down. At the bottom of the collapse, the pressure of the constraint is no longer
felt and E is restored, the recovery initiated, and the same process is ready to be
repeated.
The Between-Two-Wars Configuration
The cycle which occured in the 1920s is very similar to that described
above, although the general magnitude of the process is different. Over
that cycle, E increased considerably, and the overheating was sharp and lasted
for a relatively long period. The depression was exceptional when the col-
lapse occurred. The monetary and economic authorities did not respond with
the appropriate expansionary policies (see Dumenil, Glick, and Rangel 1987b).
As a result, a very high level of bankruptcies occured. Laissez faire, laissez
passer was an attitude that carried disastrous consequences under such circum-
stances.
The Nineteen-Fifties Configuration
Any student of the post-World War II business cycle can not help but notice that
the character of the cycle was highly modified. During the fifties, normal
equilibrium does not appear attainable. Instead, the economy is constantly
moving from excessive spurts of growth to recessions, producing cycles of
very short duration. This new situation can be easily interpreted as a new range
in the variation off. As a result of the transformation undertaken during the
war, Eis now superior to E1 and fluctuates in this new range but does not recover
to its previous values in normal equilibrium. The cycle corresponds to con-
stant switches from overheating to Keynesian state and vice versa (cf. Fig-
ure 8 [c]).
These circumstances produce a stop and go situation, with zero inflation
on the average. Demand policies appear to be very effective and well adapted
to the new economic conjuncture. However, the economic authorities seem in-
capable of defining the right measure of the therapy (a right measure does not
exist).
The result of a series of such shocks in the model are presented in Figure 10,
and thus, a new form of cycle is generated. In this simulation, the shocks up-
ward or downward are automatic and their intensity depends on the size of the
discrepency between the actual and the normal value of the ratio of capacity
utilization.
In the first half of the sixties, the ratio of capacity utilization soars. Strong
and steady growth is restored temporarily, and the business cycle seems to
Figure 10. Business Cycles: The Fifties Configuration.
u = .8
.5
0 20 40 60 80
The value off is fixed, and demand shocks (budget deficits) are autonomously generated depending on
the dijfetence between the actual ratio of capacity utilization and its normal value. The characteristic
pattern of the stop and go is reproduced.
disappear permanently. Autonomous stability is combined with stabilizing policy.

Figure 11 compares two reversions back to normal equilibrium following a
strong disruption of equilibrium. Stabilizing policies in a context of autonomous
stability are very effective.
We will now consider the course of events which accompanies the new-and
dramatic-fall in profitability in the second half of the sixties.
The Contemporary Configuration
With the decrease in profitability of the late sixties, a new shift forward in the
value of e occurs. In this situation, the economy is constantly attracted by an
inferior regime (with approximately zero industrial growth). Movements take
place around this inferior equilibrium. There are recurrent upsurges which can
never be consolidated, as well as recessions which are not preceded by phases of
overheating. Three important features of this period must be outlined:
1. A high ratio of business failures is a constant feature of the economy (above

the minimum level necessary for the adjustment of the proportions of prices and
outputs in capitalism).
2. The survival behavior described above leads to inflation (see the Liquidity
Squeeze discussion).
Figure 11. Demand Polley.

u
1 ·····································································································································
.5
0 '--~~~~~-"-~~~~~-"-~~~~~-"-~~~~~--"~~
0 100 200 300 400

The underlying conditions are those of normal equilibrium. The economy has been the victim of a
strong perturbation of equilibrium. The two curves in the figure allow the comparison of the autono-
mous return to equilibrium (a) and the return accompanied by a systematic demand adjustment (b).
The efficiency of the policy is clearly exemplified.
3. The inferior level of economic activity in which the economy is confined is

constantly sustained by economic policy. Therefore, the inflation/failure trade-
off described before is a basic element of the description of the economy.
Under such circumstances, the configuration of the business cycle is again
different. The economy stagnates in a sustained inferior state of activity (with
inflation). It is destabilized downward by a demand shock, without the phase of
overheating, and then recovers to the same inferior regime (cf. Figure 8 [d]).
Economic Theory and the Forms of the
Business Cycle
In the above section, a theory of the evolution of capitalism in relation to the
tendency for the rate of profit to fall and the analysis of the cycle has been
established. A distinction between different stages of the evolution of the business
cycle was made, and in each stage, the specific configuration of the particular
cycle was examined. In the present section, we will use this previous analysis to
briefly suggest a relationship between the evolution of the business cycle and the
history of economic theory. Schematically, each stage discussed above can be
associated with a corresponding stage in the development of economic thinking:
1. A characteristic feature of the cycle in the nineteenth century is that the
economy remains in normal equilibrium for a number of years. Therefore,
normal equilibrium is actually perceived by the economic agents as a natural
course of events. The other episodes are dramatic, usually of short duration, and
are considered as "exceptions" or "accidents." In fact, most of the ideological
representations associated with the Walrasian paradigm of equilibrium corre-
sponds to normal equilibrium. In this simple view, an equilibrium exists and is
stable. Only perverse behavior on the part of some economic agent can prevent
the rapid return to equilibrium. With the exception of money (mentioned by
Walras), state intervention is unjustified. In spite of their violent character, crises
appear as episodes of restoration. Normal equilibrium in the nineteenth century
thus displays all the components of the free-market creed. However, the episode
of the 1930s shows very clearly how the defense of this conception can be
disasterous outside of its sphere of applicability (i.e., when the conditions for
normal equilibrium are not met).
2. Stabilizing policies (budget deficits), corresponding to the Keynesian para-
digm of the boosting of the economy, are actually shocks. They have a positive
influence under many circumstances when the switch from one regime to the
other is desired. These policies are effective in a situation of normal equilibrium
which has been disturbed, although they are not absolutely necessary. These same
policies can also obstruct a movement toward depression. However, their effect is
circumscribed by the particular equilibria which exist and are stable. This is
clearly evidenced by the stop and go configuration, which we introduced to
describe the fifties, in which the fine tuning can never be achieved. In spite of the
difficulties of the stop and go policy, demand policy appears very effective to the
policy makers, and the fact that they constantly overshoot the target confirms this
effectiveness. For this reason, the postwar period was fertile ground for the rise
of the Keynesian view in the U.S. when normal equilibrium was restored in the
first half of the sixties. The miracle seemed at hand and fine tuning was put on the
agenda. This period coincided with the final triumph ofKeynesianism in the U.S.
3. In the seventies, a new revolution in economic thinking was accomplished.
Attention became focused on monetary policy. This view coincided with the new
situation in the economy. The only available economic alternative offered was
between a very distant overheating, which was historically accepted only during
war periods, 6 and the present sustained inferior state of affairs. Demand policy is
no longer adequate. The aim of policy was no longer to provoke a switch from
one regime to the other, but to sustain stagnating growth within "tolerable"
levels. The nature of the situation is well represented by the failures/inflation
trade-off which has been presented. It defines the program of monetarism.
Keynesian demand policy does not have any impact on the trade-off between
failures and inflation. In the analysis of monetarism, the criticism of Keynesian-
ism and the emphasis on monetary policy must not be confused with the decision
to make the fight against inflation the number one priority (as decided in 1979).
Similarly, the interpretation of this theoretical revolution as something related to
the new factual situation is not a validation of the theory itself, just as its success
in the sixties does not prove the truth of the Keynesian theory.
* * *
In conclusion, it is possible indeed to build a theory of crises and business
cycles on the basis of the classical analysis of competition. The same mechanisms
which account for the ability of capitalism to manage its ''proportions,'' i.e., to
determine adequate relative prices and outputs, are responsible for the fragile
structure of capitalism concerning its dimension (the general level of economic
activity). Prompt reaction to the signals of disequilibrium, when they indicate the
necessity of moderating activity, can initiate a backward spiral of decreasing
demand from within the productive system (an endogenous diminishing of de-
mand). This propensity to recession is conditioned by the level of profitability, in
relation to the rate of interest and more generally by the constraint that the
financial system applies to enterprises. High pressure (low profitability and
financial constraint) implies strict management on the part of enterprises, and
thus, instability.
This view has a lot in common with Marx's analysis of crises in Capital. First,
in both cases, the problem of crisis is approached in terms of instability. The
situation of the reserve army or financial factors are presented as possible mecha-
nisms which trigger the destabilization of the system. Second, the movement of
the rate of profit is the crucial factor which determines the fundamental condi-
tions for stability in both analyses.
Concerning the interaction between monetary and financial factors, it appears
that the real determinants of stability are fundamental. Monetary factors can play
a role, which under certain circumstances can determine the future of the system.
However, such mechanisms must be understood on the basis of an analysis of the
primary determinants of stability.
Following our interpretation, Marx's law of the tendency for the rate profit to
fall can be extended. The law of the tendency toward increasing instability is a
natural completion of the former law. A crucial point in this analysis is that
increasing instability cannot be avoided by the operation of many of the counter-
tendencies which, to a large extent, offset the fall in profitability. On the contrary,
increased instability is the product of these countertendencies. Therefore, it
appears clear that the present difficulties of contemporary capitalism can still be
viewed as new forms of the expression of its inner contradictions.
It is possible to be more specific in the presentation of this thesis, since
successive stages in the forms of the "instability problem" can be identified.
Four such stages have been described in this chapter, the nineteenth century, the
between-two-wars period, the fifties and sixties, and contemporary capitalism.
Each configuration justifies a different economic perspective: free-market nonin-
tervention, Keynesian demand policies, and monetary polices.
Notes
1. See Dumenil and Levy (1983, 1985a, 1985b, 1987a).
2. Depending on the model studied, this result has been demonstrated either analyti-
cally or using computer simulations.
3. See Dumenil (1977).

4. Marx called this the "devaluation of capital."
5. See Kalecki (1935), Samuelson (1939), Goodwin (1951), Semmler (1984), and
Foley (1985).
6. Specifically, overheating was accepted during the two world wars. It justified
measures which were opposed to the free-market ideology during World War II.
References
Dumenil, G., M. Glick, and J. Rangel. 1987a. "The Rate of Profit in the United States;
From the Turn of the Century to the Nineteen Eighties." Cambridge Journal of
Economics 11(4):331-60.
- - - - . 1987b. "Theories of the Great Depression: Why did Profitability Matter?"
The Review of Radical Political Economics 19(2).
Dumenil, G. and D. Levy. 1983. "La Concurrence capitaliste: Un processus dyna-
mique," J.P. Fitoussi and P. A. Muet, eds. 1987.
- - - - . 1985a. "The Classicals and the Neoclassicals, A Rejoinder to Frank Hahn."
- - - - . 1985b. "Stability and Instability in a Dynamic Model of Capitalist Produc-
tion" (abridged version). W. Semmler, ed., pp. 132-69. 1986.
- - - - . 1986a. The Stability of Long-Term Equilibrium in a General Disequilibrium
Model. Working paper 8717. Paris: CEPREMAP.
- - - - . 1986b. The Analytics of the Competitive Process in a Fixed Capital Environ-
ment. Paris: CEPREMAP. Forthcoming in Manchester School 57 (1):34-57.
- - - - . 1987a. "The Dynamics of Competition: A Restoration of the Classical Analy-
sis." Cambridge Journal of Economics 11(2): 133-64.
- - - - . 1987b. "The Macroeconomics of Disequilibrium." Journal of Economic
Behavior and Organization 8: 337-95.
Dumenil, G. 1977. Marx et Keynes face a la crise. Paris: Economica.
Fitoussi, J. P., and P.A. Muet, eds. 1987. Macrodynamique et desequilibres. Paris:
Economica.
Foley, D. 1985. Liquidity Profit Rate Cycles in a Capitalist Economy. New York: Barnard
College.
Gayer, A. D., W. W. Rostow, and A. Jacobson-Schwartz. 1953. The Growth and Fluctu-
ations of the British Economy (1790-1850). Oxford: Oxford Clarendon Press.
Goodwin, R. M. 1951. "The Nonlinear Accelerator and the Persistence of Business Cy-
cles." Econometrica 19(1): 1-17.
Guckenheimer J., and P. Holmes. 1983. Nonlinear Oscillations, Dynamical Systems, and
Bifurcations of Vector Fields. New York, Berlin: Springer-Verlag.
Kalecki, M. 1935. "A MacroDynamic Theory of Business Cycles." Econometrica
3(3):327-44.
Marx, K. 1863. Capital. Vol. 3. New York: First Vintage Book Edition. 1981.
Okishio, N. 1961. "Technical Change and the Rate of Profit." Kobe University Economic
Review 7: 86-99.
Ricardo, D. 1817. The Principles of Political Economy and Taxation. London: Dent and
Son Ltd. 1960.
Samuelson, P. A. 1939. "Interactions between the Multiplier Analysis and the Principle of
Acceleration." Review of Economic Statistics 21(2):75-78.
Semmler W. 1984. On Stability and Instability in Classical Economics. New York: New
School for Social Research.
Semmler, W., ed. 1986. Competition, Instability, and Nonlinear Cycles Lecture, Notes in
Economics and Mathematical Systems. No. 275. Berlin: Springer-Verlag.
Smith, A. 1776.The Wealth of Nations. London: Dent and Son Ltd. 1964.
II
Stabilization Policy
in Nonlinear Dynamical Models
Comparative Monetary and Fiscal
Policy Dynamics
Richard H. Day
According to the dynamic Keynesian story, rising employment and capacity

utilization during the upswing of the business cycle lead eventually to rising
interest rates due to the crowding out on money markets as the transactions
demand for money rises. This eventually causes a fall in investment and declines
in aggregate demand, employment and interest rates. Although the latter will
tend to stimulate investment and consumption, recovery can take some time
because of excess capacity.
Using the standard Hicksian model of this process but with supply adjustments
on commodity markets, Day and Shafer (1985) have shown that such fluctuations
will persist with either a periodic or nonperiodic character when induced invest-
ment is strong enough. In a subsequent study (1987), they prove that erratic
cycles can occur with positive measure, i.e., they can be "observable" in princi-
ple. In that case, GNP behaves like a stationary stochastic process even though
the model is deterministic. Day and Lin (in preparation) have shown that all of
these possiblitities are consistent with the normal qualitative properties of de-
mand and are robust for parameter values that fall within the ranges that have
emerged in various econometric discussions.
One of the salient findings of this work is the complex response that macro
behavior exhibits in response to a change in parameter values. As the intensity of
induced investment changes, or as the multiplier and monetary effects vary, a
stable stationary state can shift into or out of a cycle that can be strictly periodic,
This paper was begun at the Industrial Institute of Economic and Social Research,
Stockholm, in the summer of 1983, continued at The Netherlands Institute for Advanced
Study in Wassenaar the following year, and completed at the Modelling Research Group,
Department of Economics, USC, Los Angeles, 1985. The simulations and computer
graphics were prepared by T. Y. Lin. The final draft was prepared at Deer Harbor,
Washington, in July 1987.
119
120 RICHARD H. DAY
have periodic turning points with random amplitude, or have both randomly
fluctuating turning points and amplitudes.
What implications do these findings have for considerations of monetary and
fiscal policy? This chapter shows how changes in the money supply (M), govern-
ment expenditure (G), or the average tax rate (r) can trigger the economy into or
out of stable stationary, periodic or nonperiodic behavior. Moreover, the change
of a policy parameter in a given direction can have switching directions of
influence. For example, a steady increase in government expenditure could shift
the economy from a stable stationary state through cycles of varying periods,
nonperiodic fluctuations, and back into a stable stationary state.
This fact introduces a source of uncertainty that arises not from the erratic
nature of political and other "exogenous" influences, but entirely from intrinsic,
nonlinear interactions of the financial and real sectors. It provides a new hypoth-
esis concerning the notorious difficulty of anticipating the real world effects of
given monetary or fiscal changes.
Obviously, it will be necessary to explore this hypothesis in more general
models where the special assumptions used here are relaxed. But when the
general price level is changing rather modestly, when output, employment, inter-
est rates and investment are changing relatively rapidly, and when the phenom-
enon of crowding out on money markets during a regime of tight money is
evidently at play, the present analysis should be relevant, even if only as a crude,
first approximation under these very special conditions. Certainly, it exhibits
some features of the overall picture, and it seems worthwile to investigate the
issues in a tractable theoretical setting. The analysis is familiar to every econo-
mist and requires a minimum of mathematical and computational machinery.
It has become common in recent years to examine questions of macro policy
within the context of intertemporal equilibrium in overlapping generations (OG)
models. It is now well known that such models can also produce nonperiodic
cycles as shown in Benhabib and Day (1981, 1982) and elaborated in Grandmont
(1985). The former show that fluctuations can persist with constant monetary
growth, and the latter shows that government policy can stabilize chaotic cycles.
In constrast, the present analysis is conducted in a disequilibrium context, which,
in spite oflimitations, has some features that are possibly closer to reality than the
OG approach, at least under some conditions. In particular, its ingredients of
consumption, investment, and monetary demand have been looked at empirically,
and something is known about the signs and the range of values specific param-
eters should have.
The Model
The standard linear demand for money is
(1)
COMPARATIVE MONETARY & FISCAL DYNAMICS 121
where LO, k, and A are parameters. Given a fixed supply of money, M, and
immediate money market clearing, r = r' + (k/A)(Y - Y**), where Y** = (M
+ >.ti L0 )/k is the income level at which the interest rate is equal to r0 • If we
assume that this value bounds interest from below then,
(2) lr0 , 0 :5 Y :5 Y**

r = L m(Y; M): =
r0 + (k/>.)(Y - Y**) Y** :s Y :s M/k.
In what follows, assume that rO = 0. In Keynesian terms, the interval (0, Y**)
is the liquidity trap where changes in aggregate income have little impact on
interest rates. Above Y**, the interest rate is sensitive and the crowding out effect
of the transactions demand for money drives interest rates upward. 1
The linear investment function is
(3) I = 1° + (3Y - 'Yr.
It should also be augmented by a nonnegativity constraint if disinvestment is not

important in the short run. It then takes the form
(4) l(r,Y): =max {O, (3(Y - Y') - 'Yr}.
If Y' is positive, it is a threshold below which excess capacity is so great that

orders for investment goods are zero even though interest rates may be very low.
Above the threshold Y ', income has a stimulating effect. 2
Although all of the above is quite conventional, it is at this point that we
introduce a crucial novelty. Substituting (2) into the investment function so as to
eliminate the interest rate yields an ''IY'' relation that incorporates both accelera-
tor and monetary crowding out effects. The resulting investment-income func-
tion has 3 or 4 branches depending on the strength of the monetary interaction.
The first branch occurs in the range 0 :5 Y :5 Y' when induced investment is
zero because of excess capacity. (Autonomous investment may still be positive.)
The second branch occurs for incomes between the investment threshold Y ' and
Y** where interest is at its lower bound, but the stimulating effect of income on
induced investment is operating. The slope here is (3. The third branch occurs
after the transactions demand pushes the interest rate up out of the liquidity trap,
and the interest depressing effect of interest comes into play. The slope here is a:
= (3 - k')'/A, so it must be less than the slope in the earlier regime. If a is positive,
the monetary effect is weak. The investment function will look like Figure la. If
a is negative, then the monetary effect is strong. The picture is shown in Figure
lb. In this case, the rise in interest rates required to clear the money market
depresses investment, overpowering the stimulating effect of income increases.
This crowding out is enhanced by increases in k or 'Y or by decreases in Aor (3, or
if the ratio of interest rate effects ')'IA is greater than the ratio of income effects
(3/k. It implies that eventually investment is driven to zero when income reaches
122 RICHARD H. DAY
the point Y*: = [(k-y/A)Y** - {3Y']/a, where Y** = (M-L0 )/k. For incomes
above this point, induced investment is eliminated and only autonomous invest-
ment remains.
This analysis deals solely with the case in which the monetary effect is strong
as shown in Figure lb. In this situation, the investment function can be written
(5) 0, 0 ::5 y ::5 y I
) {3(Y - Y'), Y' ::5 Y ::5 Y**

J = H(Y): =
a(Y* - Y), Y** ::5 Y ::5 Y*
0, Y* ::5 Y ::5 yu
The regime switching character of the JY function is a consequence of the

linear forms for money and investment demand (1) and (3). It could be thought of
as approximating a smooth function where the marginal income effect is fairly
stable over wide ranges but changes more rapidly in certain critical regions. The
interval [0, Y '] approximates a situation where investment is very low due to
extreme excess capacity. The ''liquidity trap'' range of income [0, Y**] approxi-
mates a region where the interest rate changes very little with changes in Y; the
range [Y**, Y*] corresponds to a region where the interest rate is sensitive to
growing money demand; the range above Y* is the area where endogenous
investment is reduced severely because of high interest rates. Day and Lin (1985)
show that the approximation is good and that the qualitative behavior of the
general and the piece-wise linear models are similar.
The model is completed by the consumption function and the components of
autonomous demand. Let the former be
(6) C = c0 + aY, where a = (1-r)a
in which a is the marginal propensity to consume (MPC) and r is the average tax
rate. Define the constant expenditure term A = J0 + C0 + G, where J0 and C0 are
autonomous consumption and investment, and G is government expenditure.
Also, let µ, be an intensity parameter that enables us to vary exogenously the
importance of induced investment in aggregate demand or to explore proportion-
al changes in the parameters of investment and monetary demand. This parameter
can also be thought of as a speed of adjustment in investment.
Assuming a one-period adjustment lag, the difference equation for GNP is
obtained,
(7) A + aYt, 0 ::5 Yt ::5 Y'

A - {3Y' + [a + µ,{3]Yt, Y' ::5 Yt ::5 Y**
A + aY* - [a + µ,a]Yt, Y** ::5 Yt ::5 Y*
A + aYt• Y* ::5 Y t ::5 yu
Figure 1. The IV Function with Intrinsic Monetary Effect.
c:
Q)
c:
Q)
E E
Ui Ui
~
-= ~
Y' y•• Income Y' y•• Y* Income
(a) Weak monetary effect (b) Strong monetary effect
;::;
w
124 RICHARD H. DAY
The bound yu is the smaller of full employment income, full capacity income or
M/k.
From (7) and the definition of a, it is obvious that an increase inµ, enhances the
stimulating effect of income on investment in both the second and third "bran-
ches" or regimes of the aggregate income function. (A decrease, of course, has
the opposite influence.) In the third regime, where the crowding out effect plays a
role, the slope of aggregate demand is (1 - r)a + µa. Given a = {3 - (')'IA)k, an
increase ofµ, can be interpreted as increasing {3, 'Y or k, or decreasing A.
The implication for aggregate demand 9(Y) of weak and strong monetary
effects are shown in Figure 2. Of course, ifµ, = 0 so that induced investment is
unimportant, the model boils down to the simple stable multiplier process. If 'Y is
zero, so that investment is completely insensitive to interest rates, the same can be
said except that the multiplier involves the marginal effect of income on invest-
ment. If this is sufficiently high, it could lead to unbounded divergences from the
simple stationary state (in which the system supposedly would break down). But
if 'Y > 0, then the monetary interaction is present; these possibilities are shown in
the two diagrams. If the interaction is weak, then the usual multiplier stability
analysis can be undertaken. (One must be careful to determine in which regime or
regimes the stationary state occurs.) It is only when the monetary effect is strong
that all the complex dynamics briefly described in the beginning of the chapter
can come into play. It is in this situation that the comparative static analyses can
no longer be assumed to suffice. Instead, a comparative dynamic analysis is
necessary.
Bifurcation and Comparative Policy Dynamics
The methodology for a comparative dynamic analysis, nontheless, is similar.

In comparative statics, one studies how stationary states change in response
to shifts in a parameter. Such stationary states, if stable, give the asymp-
totic behavior of the dynamic model. If the stationary states are unstable,
they will not represent asymptotic behavior. Instead, model trajectories will
converge to a set of points called an attractor. This set will be finite if
the asymptotic behavior is a strictly periodic cycle; it will be uncountable
ifthe behavior is chaotic, and it will be an uncountable set with positive measure
ifthe asymptotic behavior is ergodic. In the latter case, the model will behave like
a stationary stochastic process. Sometimes these trajectories have periodic turn-
ing points but random amplitudes; sometimes both their turning point periods
and amplitudes are random. Two remarkable features of nonlinear dynamic
models that have been understood with rapidly growing precision in the past two
decades are (a) the robustness of these behaviors, that is, their occurrence for
"large" sets of parameter values, and (b) the intricately complex dependence of
attractors on parameter variations: models shift from stationary states through
cycles of various orders into and out of chaotic and ergodic regimes. The
Figure 2. Aggregate Demand with Monetary Interaction.
I I
I
I ,,,"" I
,,,,."" J/'
.,,.t ,,,.,,,..,,,. I
I,,,,.,,,,."" I
,,,,.( I .J""... I
"lJ "lJ ,.. I I
c: I I c:
ca ca
E I I E I I
Q)
~ I I 0 I I
I I
I I : I
I I I I
I I I I
I I I I
I I I I
I I I I
Income Income
(a) Weak monetary effect (b) Strong monetary effect
;:::;
VI
126 RICHARD H. DAY
method used to obtain the critical values of parameters where the quali-
tative behavior of a model changes, described in terms of its attractors, is called
bifurcation analysis. A detailed exegesis of the method and complete analytical
results are given in Day and Shafer (1987). Detailed numerical examples are
given in Day and Lin (1989). In those studies, bifurcation analyses were per-
formed for shifts in the importance of induced investment demand. Similar
analyses for shifts in the traditional macro policy instruments, government ex-
penditure, G, the tax rate, T, and the money supply, M, are presented
here.
First, consider shifts in government spending and refer to Figure 3a. The
parameter G influences the constant term representing autonomous expenditure
in aggregate demand. There are three stationary states. Suppose GNP is "stuck"
on the smallest of the three, and it is stable. Therefore, a small increase in G will
have the usual stable multiplier effect shifting the stationary state gradually
upward. When the total exceeds the level G 2 , the aggregate demand curve 2
results. Here a qualitative jump occurs. There is now only one stationary state.
Because the slope of aggregate demand is negative with absolute value greater
than one, fluctuations must be perpetuated. Indeed, using the theory developed in
the references cited, it can be seen that after a brief period of growth, ergodic
fluctuations occur and GNP will follow a stationary stochastic process. Further
increases in G will be associated with changing long-run distributions of GNP,
and irregular fluctuations will continue to occur. If G is increased enough, the
unstable stationary state will disappear. Then a stable stationary state is reached
and further measures in expenditure will again be associated with the usual stable
multiplier processes.
Assume GNP is stuck at the stationary state, and G is at a very high level.
Decreases in G will be associated with stable decreases in stationary GNP until
the point G 3 associated with the demand curve 3 is reached. After this, stochastic
fluctuations will appear. If G is decreased continuously, these will persist until G0
is reached and the stable multiplier process will re-emerge. Note that changes in
G have entirely different impacts on the qualitative behavior of GNP in the range
(G0 , G 3) than for values outside this range. These values, therefore, are critical
bifurcation points.
Next, consider changes in the tax rate. Such changes tilt aggregate demand by
changing the slope of the aggregate consumption function. Beginning with the
same base situation as before, critical bifurcation points for Tare readily identi-
fied, as shown in Figure 3b. Again, the attractors shift from stable stationary
states reached by the usual multiplier process to ergodic sets associated with
nonperiodic fluctuations that are stationary stochastic processes.
Finally, consider changes in the money supply. The switching points Y** and
Y0 depend on the money supply parameter but not on the slopes of the aggregate
demand segments. Changes in M, therefore, induce parallel shifts in aggregate
demand in the third regime, as shown in Figure 3c. 1\vo bifurcation points are
Figure 3. Comparative Polley Dynamics.
Increasing G Decreasing r Decreasing M

0 "C 0 "C
-0
c c lij /
al al
E E t E
CD CD CD
c c c
Income Income Income
(a) Government investment (b) Tax rate (c) Money supply
N
-.J
128 RICHARD H. DAY
readily identified. When M 2 is exceeded, the maximum value of aggregate

demand exceeds yu and the model must shift to a non-Keynesian regime. When
M < M0 , only a single, stable, stationary state exists. In between these values,
chaos reigns. All the complex dynamics that have been mentioned can occur. Of
course, all this rests on the particular values of the parameters of aggregate
demand chosen for the base situations. As we emphasized above, however,
ergodic behavior is robust: there are continuous ranges of values for the policy
instruments that yield nonperiodic fluctuations with positive measure.
Comparative Dynamics I: Counterfactual

Simulations for Three Different Periods
These findings can be illustrated by constructing bifurcation diagrams. Such

diagrams provide an estimate of the attractor as a function of a given parameter.
Given a base situation, a trajectory can be computed and the values of GNP
plotted. By eliminating the first part of the computed history, say, the first fifty or
one hundred points, an estimate of the asymptotic behavior of the trajectory
remains. Then the parameter can be increased or decreased slightly and a new
trajectory computed and plotted, and so on, for the range of parameter values of
interest.
In the paper cited above, Day and Lin provide three examples for the param-
eters of this model corresponding to the great depression (Period I), the early
sixties (Period II) and the late seventies (Period III). In each case, the parameters
are the same except for 'Y, the interest effect on investment, which is adjusted to
reflect the accumulation of capital stock. Of course, A (which contains G), Y ',
Y*, Y**, LO and Mare adjusted to reflect shifts in autonomous expenditure and in
the money supply. The parameters are shown, their derivation is discussed, and
the estimated aggregate demand functions are illustrated in their paper. The
general appearance of all three aggregate demand functions is like that of Figure
2b, except that succeeding functions are much bigger, reflecting the growth of the
economy through the half century involved.
Using these as the base situations, three sets of bifurcation diagrams were
computed, one for changes in G, one for changes in r, and one for changes in M.
These are reproduced in Figures 4, 5, and 6. First, consider changes in govern-
ment expenditures as shown in Figure 4. For all three periods the computations
show the emergence of complex behavior in the range (G 0 , G 2). Within this
range, the multiplier process is unstable. In Period I, shown in Figure 4a,
persistent fluctuations shift from ergodic behavior, where GNP is distributed
throughout an interval, to stable cycles, to periodic cycles that have random
amplitudes and back to ergodic behavior. The patterns shown in the shading give
an indication of how the density of GNP values vary with changing G. Where the
shading is dark, the density is larger and where it is light, the density is smaller.
Period II shows a strikingly different pattern of response. Throughout most of
Figure 4. Comparative Dynamics on Government Expenditure (or Autonomous Investment and Consumption).
I ~ ~
GO G2 GO G1G2 Ga GO
(a) I: 1930-34 (b) II: 1961-65 (c) Ill: 1975-78
~
Figure 5. Comparative Dynamics on the Tax Rate (or the Marginal Propensity to Consume). ....,
0
.... .. ·~... ~t;

·~. ~. ..', ,~. ~."'~~ ·Y.i
··\:@r~i~ 74r
~,..<... _
~.; ll::--J!\:
I ~-'
tl,~ \~iR~
\ ~.~
(a) I: 1930-34 (b) II: 1961-65 (c) Ill: 1975-78

RICHARD H. DAY 131
the expenditure range, a fluctuation with two-period turning points occurs. Be-
tween G0 and G 1 , this cycle has a random amplitude, whereas, from G 2 to G 3 ,
behavior is strictly periodic. Notice that the amplitude of the cycle increases from
G0 through G 2 and decreases to G 3 , where a stationary state emerges. Notice also
that in the short interval (G 1 , G 2), four and probably higher order cycles exist,
although these are not easily seen because of the scale.
Period III displays GNP values which beyond G0 distribute themselves
throughout an interval which changes very little with further changes in expendi-
ture. Except for extremely low expenditure levels where a very high unemploy-
ment stationary state prevails, fluctuations are influenced only in the details of
their frequency distribution.
Next, consider tax policy. The bifurcation diagrams for the three examples are
shown in Figure 5. As before, each diagram is strikingly different. In Period I,
the complex change in the qualitative behavior of GNP in response to increases or
decreases in the tax rate are evident. In Period II, a small increase in the tax rate
could have caused a drastic, discontinuous fall in GNP. Decreases in the tax rate
would have had only a gradual impact on average GNP, with slowly growing then
declining fluctuation amplitudes. In Period III, the qualitative effect of tax rate
changes is more like that in Period I.
In Figure 6, the three bifurcation diagrams are given for changes in the money
supply. Again, the qualitative influence of policy change is strikingly different in
the three examples.
The vertical line gives the money supply used in the base period siumlation.
Qualitative Dynamics
In order to get a better understanding of the complex macroeconomic response to

policy changes, a closer look at the relationship between the qualitative behavior
of GNP and the parameters of demand is needed. For this purpose, six parameter
sets will be used that represent widely varying estimates reported in the literature.
These parameters are described in the Appendix at the end of the chapter. All lie
within the ranges of values discussed in the literature.
The aggregate demand functions are shown in Figure 7. For Cases 1 through
5, there is a single stationary state, while for Case 6, there are three. In
Cases 1 and 2, when µ = .25, the stationary state is stable. In Cases 3
through 5, however, it is unstable, and in Case 6, the smallest one is stable,
but the other two are unstable. This is seen in column 4 of Table 1 of the Ap-
pendix which gives c = 8 '(Ym), where ym is the largest stationary
state.
If the parameters are those of examples 1 and 2, then irregular fluctuations
could be propagated only in the presence of exogenous, random shocks. This is
the situation usually assumed by macroeconomists to hold in the real world. In
examples 3 through 6, the situation is different: fluctuations persist because
132
II)
ID
I
co
(J>
>.
"a
Q.
:I
rn
~
c
0
-
:::E
G
s:.
c
0
ri
e
;.
Ill
~
i
t
E
8
cO
Q)
:;
CJ)
u:::
Figure 7. Aggregate Demand for Alternative Yalues of a andµ.
Y' Y** Y Y* Y' Y** Y Y*

(a) µ = 0.25, a = 0.36 (b) µ = 0.25, ex = 0.64
13.6°/o decreose-.......L-'-....,
in M
Y' Y** Y Y* Y' Y** Y Y*

(c) µ = 0.50, ex = 0.36 (d) µ = 0.50, ex = 0.64
Y' Y** Y* Y' Y** y•

(e) µ = 0.75, ex = 0.36 (f) µ = 0.75, ex = 0.64
134 RICHARD H. DAY
ym is unstable. All trajectories must lead into a trapping set. This set is the
interval [Ymin, ymax]. ymin is aggregate demand when endogenous investment
is crowded out by the interest rate effects. ymax is the aggregate demand when
endogenous investment is at its maximum attainable level. These values are ymin
= 8(Y*) =A+ aY* and ymax = 8(Y**) =A - {3Y' +(a + µ(3)Y**. This set
is indicated by the boxes inset in Figures le through 1f.
By using the transformation y = g(Y): = (Y - ymin)l(Ymax - ymin) for
Y in the interval [Ymin, ymax], an equivalent map on the unit interval [O, 1] is
l
given by
(8) f(y): 1 - by** + by, y t: [O, y**]

= g(y): 1 + xy** - cy, y f [y**, y*]
h(y): = - ay* + ay, y " [y*, l]
where a = (1 - 7)a
b=a+µ(3and
c = a + µa above, and
y* = g(Y*) = y** + lie
g(y**) = y**.
Of course, T(O) = 1 -by** and T(l) = a(l -y*). The parameters are summa-
rized in Appendix Table 3.
Figure 8 presents the graphs of equation (8) for the unstable examples under
consideration. These are indicated by solid lines in each figure. All have three
linear segments with the same tilted-z shape, and as all are unstable, all must
exhibit trajectories that fluctuate. Where do these trajectories go?
To answer this question, consider the iterated maps where T"(y) is defined by
T0(y) = =
y, T 1 (y) T(y) and Tn+ 1 (y) = T(r(y), n = 1, 2, 3, ... In Figure
8, the iterates Tn, n = 1, 2, 4 are shown. The original map T1 = T has three
segments. Successive maps have more segments as if the original map had been
stretched, then folded to give a sequence of jagged teeth, somewhat like the blade
of a saw.
Figure 8 corresponds with example 3 when a = .36. In addition to ym, the
stationary state, there exists a unique two-period cycle which is determined by the
fix points ofT2. It is unstable because dT2(y)ldy evaluated at the periodic points
ym 21 and ym 22 is - be = -1.86. T3 (which is not shown) has no fix points except
ym, but T 4 has four fix points in addition to ym, ym 21 and ym 22 . These are denoted
by ym 4 i, i = 1, ... 4, which are stable because dT4(y)ldy = -ab 2 c - - .70
evaluated at the periodic points ym 4 i, i = 1, ... 4. In a world described by this
example, a regular predictable four-period business cycle would emerge unless
perturbed by random shocks.
Figure 8b shows the iterates one, two and four for example 4 whenµ = .5 and
Figure 8. Successive Iterated Maps on the Trapping Set.
I ',
I -
I
• I
·.I
... ...
I·.
I ·.
----- T4 .i
o2 per cycle :1
.4 per cycle ; I
: I
: I
,-- .... _,...,"'\

;
:I
I I • f
• I I : I
•• I ~; I
I •• ;\ :
I ·.... ·••·••... _.· '/\I
(a) a = 0.36, µ = 0.50 (b) a = 0.64, µ = 0.50
..
.......
: :
.. ·
i \ .. ··
(c) a = 0.36, µ = 0.75 (d) a = 0.64, µ = 0.75

a = .64. The second iterate, like that for example 3, has an unstable two-period
cycle. But now T 4 has become steeper, not flatter, which also possesses four
additional fix points. Indeed, T 4 is expansive over the entire interval [O, 1], that is,
its derivative, wherever it is defined, is greater than unity in absolute value. Such
maps generate nonperiodic, or chaotic, trajectories that can be characterized by
probability functions or cumulative distributions (Day and Shafer 1985, 1987).
Although cycles of every order 4n, n = 1, 2, 3 ... exist, they are unstable and
136 RICHARD H. DAY
irregular fluctuations emerge for almost all initial conditions, i.e., with prob-
ability one for initial conditions drawn at random from [O, 1].
A closer look at the map T 4 shows that it possesses a very interesting property
in addition to expansivity. It possesses four invariant sets indicated by the inter-
vals Ji, i = 1, .. .4 in Figure 8b. This means that T4(1i) =Ii, so that every fourth
period, a trajectory visits the same invariant set. Moreover, Ii+t = T(li). The sets
Ii can, therefore, appropriately be called "periodic sets" and have the following
interpretation. If
(9) Yt E I i then Yt+i E

I(i+j)mod4 , J· -_ 1, 2 , ...
Examples 5 and 6 are shown in Figures 8c and 8d. The first of these possesses
a stable two-period cycle, while the second possess stable periodic sets within
which fluctuat Ions are erratic, as in example 4. If Ji, i = 1, 2 are the periodic sets,
thel'I if
(10) Yt E I i then Yt+i E

I(i+j)mod2 , J· -_ 1, 2 , ..
has the same stochastic-like character noted for example 4.

To put it another way, examples 4 and 6 reveal a kind of periodic fluctuation
with more-or-less random amplitudes, but where the latter are confined to well-
defined, predictable sets. The amplitudes are not random in the usual stochastic
sense that they are determined by random shocks, but rather in the sense that for
predictions farther and farther into the future, calculations based on the known,
immediate past value become more and more hazardous; tiny errors due to round-
off are amplified explosively so that exact prediction is possible only in the set-
valued sense.
Although it is extremely difficult to derive analytical expressions for the
density functions characterizing the ergodic property exhibited by examples 4
and 6, they can be approximated by computation. In Figure 9 the numerical
histograms for the examples 3 through 6 are shown based on 10,000 iterates of
equation (8). In the stable periodic Cases 3 and 5, all of the density is piled up on
the intervals containing the periodic points; whereas for examples 4 and 6,
contrastingly, the density is distributed throughout the invariant periodic sets.
Within these sets, GNP has a nonperiodic character, wandering close to periodic
cycles of various orders but always wandering away from them in an unpredict-
able way. Of course, random shocks which allow impulses of sufficiently large
magnitude would upset even this degree of predictability.
Comparative Policy Dynamics II:

Policy Effect Reversals
Using the information from the preceeding section, it will be shown how the
profile of aggregate demand is modified within the trapping set by changes in the
Figure 9. Density Distributions for Trajectories In Cases 3-6.
(1, 112.005347) (1, 1/12.22826)

# iterates: 3000 # iterates: 3000
# intervals: 50 # intervals: 50
00,00 State-variable
(.;) A stable four period cycle, (b) Four periodic sets,

µ =.5, c; = .36 µ = .5, c; = .64
(1, 1/1.002674) (1, 1/5.681818)

# iterates: 3000 # iterates: 3000 ·
# intervals: 50 # intervals: 50
g
Q)
f
:::i
:;
Qj
a:
00,00 State-variable 00,00
(c) A stable two period cycle, (d) Two periodic sets,

µ =.75, c; = .36 µ =.75, c; = .64
policy instruments. The four unstable cases of the preceeding section will be used
as examples. First, consider monetary policy as represented by an exogenous
change in the money supply M. In order for aggregate demand to have the tilted-z
shape shown in Figure 8, it can be shown that the following expression must hold:
(11) ymin < Y** < Y* < ymax.
Let Am indicate the change in the variable it precedes caused by a change in the
money supply, which we denote AM. Then it is easy to see
(12) Amymin = a17AM
AmY** = (l/k)AM
AmY* = 17AM
Amymax = (b/k)AM
138 RICHARD H. DAY
where 7/ = kµ/(>.a) and b = a + µ{3 as before. The parameters Y ',a, b, and care
unaffected. As noted in section 2, a change in the money supply involves a
parallel shift in the third segment of the aggregate demand function in (7), now
defined in the interval [Y** + Am Y**, Y* + AamY*]. This corresponds to a
shift in the segment g(y) = 1 + c** - cy when the trapping set is transformed to
the unit interval. For (11) to hold for the new money supply M + AM, we must
have
(13) ymin + a71AM < Y** + k- 1 AM

< Y* + 71AM < ymax + bk- 1AM.
For any AM satisfying this set of inequalities, the qualitative dynamics

described in the previous section are unchanged. Of course, the exact amplitudes
of the periodic cycle in example 3 or the range of amplitudes in the periodic sets
of example 4 will be modified so that monetary policy can have the effect of
exacerbating or attenuating the business cycle. More interesting, however, are
changes that violate (13). If the last inequality on the right is reversed or replaced
by an equality, the tilted-z shape gives way to single-peaked shape determined by
two regimes:
(14) 1 - by** - byt, Yt € [O, y**]

= 1 + cy** - cyt, Yt € [y**, l]
where y** = 1 - 1/c. Note that ifb and care greater than unity, T 1 is expansive.
On the other hand, if the left-most inequality in ( 13) is reversed or changed to an
equality, then the tilted-z givt;<> way to a "check mark" shape defined by
(15) lg(yt): = 1 - cyt, Yt € [O, y*]

Yt+2 = T2(Yt): =
h(y): = -ay* + ayt,Yt f [y*, 1)
where y* = lie. Because 0 < a < 1, T 2 is not expansive. It is easy to show that if
akc > 1 for some k, then there may exist an expansive map of order k. If ac < 1,
then no iterate is expansive. (See Day and Shafer, 1987). For the switch from
(8) to (14) to occur, set
(16)
If 7J > 0, a sufficient increase in the money supply will cause the switch, while if
7/ < bk- 1 , a sufficient decrease will cause the switch. (Of course, if 7/ =
bk- 1 ,monetary policy cannot bring about the switch in question.) In either of
these cases, a large enough change in M shifts aggregate demand from a nonex-
pansive tilted-z form to an expansive, single-peak form, given that both band c
are greater than unity as they are for examples 3 and 4. The effect of such policy
changes is, therefore, to trigger the economy from a stable cycle (examples 3 and
5) or a set periodic chaos (examples 4 and 6) to a chaotic behavior in which
nonperiodic, and more or less random fluctuations in amplitudes are spread
throughtout the entire trapping set. In the former, there is perfect predictability
and in the latter the set, predictability is lost. 3
In example 3, an increase in the money supply of 6.6% is sufficient to cause
the destabilizing switch from a stable four-period cycle to turbulent fluctuations.
In example 4 where we have set periodicity, the switch occurs for a decrease in the
money supply of at least 13.6%. 4 Figure 10 shows the new empirical density
functions (histograms) for these two examples. Panels a and b correspond to
panels 9a and 9b, respectively. Changes in a policy instrument can eliminate, or at
least reduce, chaos in the sense of reduced variance of GNP, as can be seen by
beginning with these new situations and then changing the money supply. But
again, changes in control which are opposite in sign are required to have this
stabilizing effect given the difference in parameter values.
With regard to fiscal policy, changes in government expenditure cause
changes in A.
(17) .::lg ymin = ymin + .::lG

.::lgY** = 0
.::lgY = 0
dg ymax = ymax + .::lG.
Using these in the inequality expression (11) in a way analogous to the discussion
of changes in M, increases in government expenditures such that
(18) .::lG 2:::: Y** _ ymin
switch the economy to the T 2 type (Equation 15), while decreases such that
(19) .::lG ::5 Y* _ ymax
switch it to T 1 (Equation 14). When the economy is periodic as in Case 3,

increases in expenditure satisfying (18) simply reduce the amplitudes of the
periodic cycle, while decreases satisfying (19) bring about ergodic, strictly tur-
bulent fluctuations.
When the economy is originally set periodic, as in example 4, periodic chaos
is changed to turbulent, nonperiodic fluctuations by decreases satisfying ( 19),
while increases satisfying (18) can cause either periodic cycles or chaotic
140
Figure 1O. Density Distributions Induced by Changes In the Money Supply.

(panel a compares with Figure 3a, panel b with Figure 3b)
(1, 1/16.79104)
# iterates: 3000
# intervals: 50
...
(0, 0) State-variable
(a) Distribution of GNP values induced from a stable four period cycle by an increase in
the money supply (Case 3).
(1, 1117 85714)

#iterates: 3000
.. # intervals: 50
i)'
c:
CD •
--
f
:::J
-• -. .
~
-~
Qi
a:
~
. . . .
.
~
• ~
~
(0, 0) State-variable
(b) Distribution of GNP values induced from a stable set-periodic cycle by a decrease in
the money supply (Case 4).
behavior depending on the values of a and c.

Recall that a is the marginal propensity to consume unadjusted for the average
income tax, i.e., a= (1 - r)a. A change in the tax rate in amount tu induces a
change in the adjusted marginal propensity to consume of -aA.r. Therefore,
(20) ymin = (1 - aAr)Y*

Y* =0
Y** = 0
ymax = (1 - aAr)Y**.
Now changes in the tax rate can cause switches in the qualitative dynamics,
perhaps pushing the economy into or out of a chaotic, set periodic, or strictly
turbulent regime. The possibilities in the several examples can readily be derived
in a manner exactly analogous to the discussion of monetary or expenditure
policy.
* * *
Although the calculations have been tedious, some striking new insights have
emerged:
(1) Policy changes can trigger drastically different patterns of reponse de-
pending on their pre-existing levels and their direction.
(2) A policy may have no effect for small changes and then, after a higher or
lower threshold is reached, trigger great changes in behavior.
(3) Changes in autonomous expenditure, prices, and capital accumulation
must bring with them great changes in the qualitative response of GNP to policy
changes, even when all the other parameters are fixed.
(4) Complex dynamics occur for a wide range of behavioral parameter and
instrumental variable values.
Evidently, policymakers may have even more to contend with in attempting to
guide an economy than has already been recognized by macroeconomists. Com-
parative policy analysis may need to be concerned not just with the effects of
monetary and fiscal parameters on the long-run level of output and employment.
It may also need to recognize the potential ability of policy changes to shift output
adjustments into cyclical or nonperiodic fluctuations, or contrastingly, to stabi-
lize such instabilities when they emerge. It may also need to recognize the
disconcerting possibility that the direction of influence on the economy of a given
policy instrument under one set of conditions may be in the opposite direction
under another set of conditions.
A better understanding of these possibilities in more general dynamic models
where some of the restrictive assumptions of the present one have been relaxed
would seem to be an important priority for further research.
There are still many questions with incomplete answers. Does the economy
142 RICHARD H. DAY
possess a structure that can be unstable when prices adjust to excess supplies and
demands, when capital accumulates, and when consumption and investment
decisions respond over a considerable period of time to unfolding and anticipated
conditions? Can this instability be moderated by appropriate fiscal and monetary
policies? Answering these questions is what macroeconomics is all about. It is an
unfinished task. Hopefully, the potential importance of nonlinear structural rela-
tionships as a source of macroeconomic trouble has been demonstrated here, a
source relatively neglected in the years that have intervened since Hick's, Good-
win's, and Kaldor's seminal contributions to the subject.
Appendix
Hall (1977) estimated the adjusted marginal propensity to consume to be a = .36.

Some discussants argued for a still lower value, but several others, notably Gramlich
and Klein, suggested a much higher value. (Hall, 121.) Branson (1979) gives a value
of .72, while Morley (1983, 67) gives a value of .65. Both of these imply a negative
value for autonomous consumption in 1977, the year for which Hall's data were
calculated. For a = .64, this is not a problem, so we use that value.
For the investment function, Hall estimated the base values a = l.36 and 'Y =
83.8. He considered alternative values representing possible fractions of complete
adjustment possible within a given year as derived from Jorgensonian investment
considerations. This is similar in the present context to a choice of values ofµ between
zero and one. Hall presented evidence for a value ofµ = .25. Gramlich and Klein,
however, (op cit., p. 121) suggested a much higher value. We shall use .50 and .75 to
represent this alternative view. Klein asserted (see Hall, op. cit., p. 121) that it was
an econometrically established fact that the marginal effect of income on aggregate
demand, which in our notation would be a + µ{3, was roughly 1.5. Given a = .36 and
(3 = l.36, then µ- .84. Or, using a = .64, µ- .63. Thus, the figure of .5 is
moderate and the value .75 is not extreme. Sims' argument (Hall, op. cit., p. 105) for
a positively sloped IS curve is, in effect, also an argument for a large value forµ. The
parameters describing marginal effects are completed by the demand for money
which Hall estimated to be k = .135 and A = 2.0. These possibly proved to be the
most controversial of all. Modigliani, for example, argued for parameter values many
times lower (op. cit., p. 111) and Goldfeld suggested lower values too, though not by
so exaggerated an amount (op. cit., p. 119). In the present piece-wise linear version
of the standard model, these parameters enter aggregate demand only in the interest
sensitive regime, but this regime will be badly represented if data for only a narrow
range of interest rates were available. (For a discussion of this point see Day and Lin).
Although Hall addressed what are essentially dynamic issues, he used a tradition-
al, comparative static analysis based on local linearity in the neighborhood of the
Keynesian disequilibrium which he assumed was stable. For these reasons he only had
to consider marginal effects of policy variables. But (7) need not be stable. In order to
obtain its solutions, we have to estimate the threshold or ''kink'' parameter Y ', Y**,
Table 1
Marginal Income Effects for Alternative Values of a and µ

(1) (2) (3) (4)
Case a µ b = a + µ{3 c = a + µu
1 .36 .25 .70 .71

2 .64 .25 .98 .43
3 .36 .50 1.04 1.79
4 .64 .50 1.32 1.51
5 .36 .75 1.38 2.86
6 .64 .75 1.66 2.58
Table 2
Level Parameters for Alternative Values of a and µ
Case co A Y' Y** Y* ymax ymin
1 379 661 424 1276 1545 1410 1218

2 0 282 424 1276 1545 1388 1273
3 379 661 733 1276 1447 1489 1183
4 0 282 733 1276 1447 1467 1209
5 379 661 836 1276 1415 1569 1170
6 0 282 836 1276 1415 1547 1187
Table 3
Parameters of the Transformed Map*
b c a Y* * Y* T(O) T(1)
Case 3 1.04 1.79 .36 .304 .863 .684 .047

Case 4 1.32 1.51 .64 .260 .922 .657 .045
Case 5 1.38 2.86 .36 .266 .615 .633 .139
Case 6 1.66 2.58 .64 .247 .634 .590 .235
*The parameter of aggregate demand for the four unstable cases are transformed to the unit interval.
and Y* because of its non-linearity. To do this we must estimate the "level"

parameters L0 and C 0 that enter the demand for money and demand for goods
functions.
The relevant data in real terms for 1977 were Y = 1350, C = 865, I = 210, G =
282, M = 232 and r = 5%. To estimate C 0 , we use equation (6). Then A= C 0 + G.
To estimate L0 , assume we are not in the liquidity trap, so L0 = M + >..r - kY. From
(4), 1° = - {3Y' = I - {3Y + rr. Dividing by - f3 gives Y '. From Y** =
(M-L0)/k and using (4), we get Y* = [(7 k/A.)Y**-{3Y']/a.
The resulting parameters are shown in Tables 1 and 2.
144 RICHARD H. DAY
Notes
l. When income exceeds M/k, the model breaks down. A switch in regime would be
necessary then to explain how changes in the price level, the transaction velocity, or the
money supply take place so as to permit continued functioning of the system. In this study,
attention is confined to parameter values that avoid this problem.
2. IfY' is negative, then I0 = - /lY' is positive and can be thought ofas autonomous
investment. If Y** > Y ', then investment has an interest sensitive range. For this to
happen, we must have Y** = (M + >..rm - L0 )/k > Y' or given that rm = 0, M - L0 >
kY '. In what follows we assume that this is true.
3. Actually, changes in policy instruments can lead to a breakdown in the trapping set.
Then trajectories can "escape" the bounds given by ymin and ymax with GNP declining
to the smallest stationary state. This actually happens in example 6 when we try to
duplicate the comparative dynamic analysis carried out for example 4. The result is that
when the money supply is reduced sufficiently, fluctuations eventually die out, and GNP
declines to the smallest, stable stationary state.
4. In the latter case, increases in the money supply would simply expand the domains of
the periodic sets and, hence, the range of amplitudes of the four-period, erratic fluctu-
ations. In the former case decreases in the money supply will eventually cause a switch in
regime to the inverted-peak profile. Because ac < 1, expansivity is not induced, and the
monetary effect is simply to reduce the amplitude of the strictly periodic cycle. In this
sense, the economy is stabilized but at the cost of higher average unemployment and excess
capacity.
References
Benhabib, J., and R. Day. 1981. "Rational Choice and Erratic Behavior." Review of
Economic Studies 48:476-95.
- - - - . 1982. "A Characterization of Erratic Dynamics in the Overlapping Genera-
tions Model." Journal of Economic Dynamics and Control 48:459-72.
Branson, W. 1979. Macroeconomic Theory and Policy. 2nd ed. New York: Harper and
Row.
Day, R., and T.Y. Lin. 1987. "A Keynesian Business Cycle," in: E. Nell and W. Semmler
(eds), N. Kaldor and Mainstream Economics, Festschrift for N. Kaldor forthcoming.
London: Macmillan Press.
Day, R., and W. Shafer. 1985. "Keynesian Chaos." Journal of Macroeconomics 7:277-
95.
- - - - . 1987. "Ergodic Fluctuations in Deterministic Economic Models." Ad-
vances in Dynamic Economics. A. Medio, ed. (Special Issue). Journal of Economic
Behavior and Organization 9.
Grandmont, I. 1985. "On Endogenous Competitive Business Cycles." Econometrica
53:995-1045.
Hall, R. 1977. "Investments, Interest Rates, and the Effects of Stabilization Policies."
Brookings Papers in Economic Activity 1:61-12 l.
Morley, S.A. 1983. Macroeconomics. Chicago: The Dryden Press.
Monetary Stabilization Policy
in aKeynes-Goodwin Model
of the Growth Cycle
Toichiro Asada
Why do periodical alternations of booms and slumps occur in a capitalist econo-

my? A classical answer to this question is a Marxian theory of the business cycle,
which is based on the notions of industrial reserve army and class struggle. 1
According to this theory, the rate of profit is determined by the ''rate of exploita-
tion,'' and the latter is determined by the relative bargaining power of capitalists
and workers. It may be said from the macro-economic point of view that the rate
of exploitation approximates to the profit-wage ratio which reflects the relative
share of profit in national income. This theory supposes that workers' bargaining
power is inversely proportional to the rate of unemployment. That is to say, when
the rate of unemployment is low, workers' bargaining power is relatively strong
so that the relative wage share increases, which implies that the rate of profit
decreases. The decrease of the rate of profit induces the decrease of the rate of
accumulation, so that the rate of unemployment increases. An increasing rate
of unemployment weakens workers' bargaining power, which implies a recov-
ery of profitability so that accumulation is activated. Hence, according to this
theory, a mechanism of an endogenous business cycle is embedded in the capital-
ist economy through the changes in the relative bargaining power of capital and
labor.
It is well known that Goodwin (1967) formalized this Marxian business
cycle idea. According to Go9dwin, the above mentioned mechanism can be
described by the following system of dynamical equations.
This is a revised version of the paper which was published in The Hitotsubashi Review
91 (3) (March 1984) in Japanese.
145
146 TOICHIRO ASADA
(i) Z/Z = fi(E); ft I> 0, ft (E*) = 0, 0 < E* <
(ii) EIE = fz(Z); h I< 0, fz(Z*) = 0, 0 < Z* <
where Eis the rate of employment (1 minus the rate of unemployment), and Z is
the relative wage share in national income. Formally, this is equivalent to the
famous Volterra-Lotka system of "predator and prey" in mathematical biology,
and the solution curves of this system become closed orbits around the equilibri-
um point as in Figure 1. 2
Although Goodwin's formulation is quite exciting and useful, his original
model is insufficient to incorporate some characteristics of modern capitalist
economies, because it has some classical or anti-Keynesian flavors.
First, this model neglects the presence of the government sector. Second, it
presupposes the full utilization of the capital stock, so that it excludes by assump-
tion the Keynesian unemployment due to insufficient effective demand. Third,
the existence of the investment function-which is independent of the saving
function-is not allowed for. Fourth, money and financial assets are not incorpo-
rated into the model explicitly.
In this chapter, we explicitly incorporate the monetary sector and a Keynesian
investment function into Goodwin's growth cycle model, and investigate the
implications of the government's monetary stabilization policy. This modifica-
tion adds some Keynesian flavors to Goodwin's classical model, so that the model
may be called the Keynes-Goodwin model of the growth cycle. 3
The Model
The simplifying assumptions of the model are:

1. It is a one sector model.
2. Capital depreciation is neglected.
3. Economic agents are divided into two classes, i.e., capitalists and workers.
Workers spend all of their disposable income, and capitalists save a part of
their disposable income.
4. The goods market and money market are always cleared.
5. The population of workers and labor productivity grow exponentially at the
exogenously given rates. The output-capital ratio in the case of full capacity
utilization of capital is constant. 4
The variables in the model are defined as follows:
X real output.
=
K real capital stock.
=
I = real investment demand.
G = rt?al government expenditure.
g =KIK = rate of capital accumulation.
KEYNES-GOODWIN MODEL OF GROWTH CYCLE 147
Figure 1
L -----l---
------------,
r : I
I
I
I
I
_J
{j = capacity utilization ratio of capital stock (0 ~ {j ~ 1).

L = labor employment.
Ls = la~or supply.
n 1 = £ 3 /C = constant = growth rate of workers' population.
E = L/Ls = rate of employment = 1 - rate of unemployment
(0 ~ E ~ 1).
w = money wage rate.
p = price level.
w = wlp = real wage rate.
11" = pip = rate of price inflation.
"ll"e = expected rate of price inflation.
bi =
X/K = output-capital ratio (i =
constant = output-capital ratio in
case of full capacity utilization of captial).
bv = LIK = labor-capital ratio (v = labor-capital ratio in case of full capac-
ity utilization of capital).
£ = LIK = vii = labor-output ratio (reciprocal of labor productivity).
ni = - ft£ = constant = rate of technical progress.
tw = average tax rate on wage income.
tr = average tax rate on profit income (0 ~ tw ~ tr < 1).
Sr = capitalists' average propensity to save (0 < Sr ~ 1).
Z =
w L Ip X =
wf = relative share of pre-tax wage in national income
(0 ~ z ~ 1).
148 TOICHIRO ASADA
Zn = (1- fw) w L Ip X = (1 -tw) wf = relative share of after-tax wage in

national income (0 =5Zn =5 l -tw).
r = (p X-w L) Ip K = o(x-wv) = ox (l-wf) = ox (1-Z) =pre-tax
rate of profit.
rn = (1-t,) (pX-wL) I pK = (1-t,) ox (l-Z) = (1-t,) ox {l-Znl
( 1 - fw)} = after-tax rate of profit. 5
e = nominal rate of interest.
M = nominal money supply.
The equilibrium condition for the goods market is expressed as follows:
(*) X = (1-tw) wL + (1-s,) (1-t,) (X-wL) +I+ G
Dividing both sides of this equation by K and rearranging terms, we have
= g + h; g = IIK, h = GIK. 6
Now, assume the following investment function, which can be derived from
Tobin's version of Keynesian investment function (the so called "q theory" of
investment). 1
Substituting this equation into eq. (**), we have the following IS equation:
(1)
Next, the equilibrium condition for the money market (LM equation) may be
formulated as
where <P(e, 1re) is the "Marshallian k" (the reciprocal of the velocity of circula-
tion of money). Dividing both sides of equation (****) by K, we have
(2) m = ox <f>(e, 1re); m = Ml(pK).
The other basic equations in our model are as follows:
(3) wlw = ft.E) + 1re; f'(E) > 0
(4) E = L!L5 = &fxK!L5

(5) Zn = (1 - fw)wL/pX = (1-tw)W f Ip
(6) 1r = pip
(7)
(8)
(9) flf = -n 2 ; n 2 = constant.

Equation (3) is an "expectation-augumented" Phillips curve. This equation
says that workers' bargaining power reflects the rate of employment.
Equations (4) through (6) are the definitions of the rate of employment, after-
tax wage share, and the rate of price inflation, respectively. Equation (7) implies
that the investment contributes to the increase of capital stock. Equations (8) and
(9) say that the rates of increase of the labor force and labor productivity are given
exogenously.
Thus, there are nine independent equations and twelve endogenous variables
(E, Zn,P• w, 1r, 1re, e, K, Ls, f, oand M). Therefore, "three degrees of freedom"
remain in this system.
Goodwin-type Growth Cycle
If we close the system by adding the following set of equations, the system
bears a Goodwin-type growth cycle.
(10) 0 = l
150 TOICHIRO ASADA
(11) 7f'e = 11'
(12) MIM = µ, + a(e - e); a :=::: 0
Equation (10) implies that the full capacity utilization of capital stock is
always attained. Equation (11) is the so called perfect myopic foresight
hypothesis of inflation expectation. It says that "every current price rate of
change is known with certainty" (Burmeister 1980, 85). Equation (12)
formalized the government's monetary policy rule. The rule with a = 0 is called
a monetarist rule which fixes the rate of growth of the money supply. The rule
with a > 0 is called an activist or Keynesian rule which subcribes to change the
rate of growth of the money supply proportionally to the change of the nominal
rate of interest.
Equations (1) through (12) can be reduced to the following more compact
system,
(13) (i) H(Zn) = g(Zn, e - 7r) + h

(ii) m = i</>(e, 11')
(iii) mlm = µ, + a(e - e) - 7r - g(Zn, e - 7r)
(iv) Zn!Zn = wlw - 11' - nz = f(E) - nz
(v) EIE = g(Zn, e -11') - (n1 +n2)
where H(Zn) = i [1 - (1-sr) (1 - tr) - {sr(l-tr) + Ur-fw)} Zn!

(1 - tw)l. g(Zn, e-11') = g(i(l-tr) {l-Zn/(1-tw)}, e-11') and the
endogenous variables of this system are Zn, E, m, e and 7r.
The classical nature of this system is evident. In the short-run, Zn, E and mare
given, and the short-run equilibrium values of e and 7r are determined by
equations (13) (i) and (13) (ii). It follows from equation (13) (i) that if Zn is
given, the level of the rate of investment (g) which assures the full capacity
utilization of capital stock is determined uniquely, and this investment level is
attained by the adjustment of the real rate of interest (e - 7r). In other words, the
real rate of interest is determined in the goods market independently of the money
market. This is a fundamental feature of the classical macro model. Furthermore,
the postulate of classical dichotomy applies to this system. Namely, in this
system, the movements of main real variables are determined independently of
the money market. Now, let us prove this assertion.
First, solving equation (13) (i) with respect to e - 7r, we have
(14) e- 7r = 4>(Zn); 4> '(Zn) = {H'(Zn) - gi}I 82,

(-) (-) (-)
where g1 = og/o Zn and g2 =

oglo (e - 11"). If the sensitivity of investment with
respect to the change of the rate of profit is so small that I g 1 I is small enough,
then we have cl>' (Zn) > 0. From this point, we shall assume that cl>' (Zn) > 0.
Substituting equation (14) into equation (13) (ii), we obtain
(15)
Solving this equation with respect to 71", we have
(16) 71" = 11"(Zn, m); 11"1 = 011"/oZn = - </>1cl>'/(</>1 + </>2) < 0,

11"2 =011"/om = l/i(</>1 + ¢2) < o.
From equations (14) and (16) we also obtain
(17) e = cl> (Zn) + 71" (Zn, m) = e (Zn, m);

el = oeloZn = <P' + 11"1 = c/>2</> '/(¢1 + </>2) : : : : o,
e2 = oelom = 11"2 <O.
Substituting equations (13) (i) into equation (13) (v), and substituting
equations (16) and (17) into equation (13) (iii), we have the following
fundamental dynamical system:
(i)
(ii) E= {H(Zn) - (h + n1 + n2)} E = F2(Zn, E)

(iii) m= [µ + a {e(Zn, m) - e} - 71" (Zn, m)
- H(Zn) + h] m = F3 (Zn, m)
It can be seen by inspection that equations (SA) (i) and (SA) (ii) in this system
form a subsystem which is independent of equation (SA) (iii). This feature of the
system can be visualized in Figure 2. In this figure, arrows show the directions of
causalities. This figure clearly shows that paths of the rate of employment and
wage share are independent of the money market. Moreover, a subsystem which
consists of equations (SA) (i) and (SA) (ii) is the same as Goodwin's (1967)
growth cycle model. Hence, the working of the system can be analyzed by
utilizing Goodwin's method.
First, the stationary solution (long-run equilibrium) of the system (SA) (i)-(ii)
is given as
(18)
(19)
152 TOICHIRO ASADA
Figure 2. Directed graph of (SA).
.
1 rQ-
--------- 7 1
I I
I I
I Zn I
I I
/ /
I I
I I
I I
I I
I
I
I
I
I
I
I
I
I
I
I
I
I
/ C). /
I
I
0
I
/ E /
I I
I I
LI _________ .JI
Assume that 0 < Zn* < 1 - lw and 0 < E* < 1. In this case, the phase
diagram of the system becomes like Figure 3.
Goodwin (1967) showed that the solution curve of this system corresponding
to the given initial condition [Zn(O), E(O)] becomes a closed orbit around the
equilibrium point as follows.
From equations (SA) (i) and (SA) (ii), we have
(20)
or
Integrating this equation we obtain
(22) r,o(Zn) = j {H(Zn)IZn}dZn - (h+n1 +n2) log Zn

= j lf(E)I E} dE - n 2 log E + C 't(E) =
where C is the integral constant.
From this equation, it can be seen that
Figure 3
E=O
1~------------------ --------,
Zn>O
Zn>O
r
E<O
E>O
L
1 Zn<O
a
_j -i
Zn<O Zn<O
E>O E<O
o'--~~~~---11-~~~~~~~-----~~~~--1-~Zn
Zn* 1-tw
(23)
and
(24) 'Y'(E) = {f(E)-nn}IE ~ 0 - E ~ E*.
As Figure 4 shows the amplitude of the cycle depends on the initial condition
[Zn(O), E(O)]. The more remote the initial condition is from the equilibrium point,
the greater is the amplitude. The time patterns of the rate of employment(£), the
after-tax wage share (Zn), and the real rate of interest (e - 71") in this model are
illustrated in Figure 5. 8
Now, it is evident in this model that the monetary policy of the government is
ineffective in stabilizing the real variables, but, it is effective in stabilizing the
monetary variables such as m, 71" or e. Next, this assertion is proven.
The stationary solution of the system (SA) (i)-(iii) can be written as
154 TOICHIRO ASADA
Figure 4
E
'K-~=~·~I______.
'A--- -------~--' I
I
N I III
I
I
I
'l'(E)---~~-'------'.__~~.__~~~___.~___..__~_._~......_~ Zn
Zn* 1-tw
'l'(E)=qi(Zn)
-=====~
qi(Zn)
Figure 5
E, Zn, p-7C
I p-7C
I .,."'z
"-+-.
........ ~-""'--:/" n
I
I
Q'--~~~-'-~~~-'-~~~--'~~~~-'-~~~~~
IJ III IV
(25)
where <P* =
<P(Zn *). 9 Linearizing the system (SA) around the equilibrium point
(Zn*, E*, m*), gives
(26) F12*
Zn - ZnJ
0 [ E - E*
0 m -m*
where, F12*=f'(E*)Zn* > 0, F21*=H'(Zn*)E* < 0, FJ1*=[ae1*-11"1*

(+ or (-)
0 )
- H'(Zn*)] m*, and FJ3* = (ae2*-11"2*)m* = (a-1)11"2* m*.
(-) (-) (-) (-)
The characteristic equation of this system is
(27) Li(>.) = -F12*

>.
0
Hence, the characteristic roots are
= (a-1) 11"2* m*, ± i .J - H' (Zn*) f' (E*) Zn* E*

(-) (+)
where i = .J -1. Therefore, the solution of the system (26) can be expressed as
follows.
(29) (i) Zn(t) =Zn* +Bi cos (8t+t:1)

(ii) E(t) = E* + B2 cos (Ot+1:2)
= m* + A e"- + B3 cos
'\ lt
(iii) m(t) (Ot+t:3)
where A1 = (a - 1) 11"2* m* and 8 = .J -H'(Zn *) /'(£*) Zn* E* > 0.

(-) (-) (+)
In addition, the values of the constants A and Bi (j = 1, 2, 3) depend on the initial

condition [Zn(O), E(O), m(O)].
Equation (29) can be considered to be an approximate solution of the system
(SA) near the equilibrium point. It is clear from equation (29) that ifthe monetar-
ist rule of money supply (a = 0) is adopted, >. 1 > 0, so that the movements ofm
get to be explosive, and in this case the rate of inflation ( 11") also moves explosive-
156 TOICHIRO ASADA
ly. 10 The government can minimize this sort of price instability by adopting an
activist money supply rule. In fact, if a > 1, the explosive factor in equation
(29) (iii) vanishes. But, even in this case, the cyclical movements of monetary
variables caused by the "real" factors are inevitable.
The period of the cycle (7) in equation (29) is given as
(30) T = 21ff() = T( I H'(Zn *)I .J'(£*), Zn*, £*);

aTJalH'I <O, aTJaf'<O, aTJaZn*<O, aTJaE*<O
where,
(31)
Hence, the greater the senitivity of wage change with respect to the change of
employment if') and capitalists' propensity to save (sr), the shorter is the period
of the cycle.
Variable Propensity to Save
The workings of the real variables in the previous section's model are the same as
those of the original Goodwin ( 1967) growth cycle model in spite of the fact that
the monetary sector and the independent investment function are introduced. As
is well known, Goodwin's growth cycle model is "structurally unstable in the
sense that small variations in the parameters will alter the properties of the
system" (Desai 1984, 256). 11
In this section, this proposition is exemplified by performing a particular
perturbation in the previous section's model.
Now, assume that
(32)
and keep all of the other assumptions of the model intact. Equation (32) implies
that the increase of the expected rate of price inflation, (7r"), induces the increase
of capitalists' propensity to consume from the present income (1-sr), since the
increase of 1fe implies the acceleration of the decrease of the expected purchasing
power of money income. 12
In this case, the system (13) in the previous section is modified as
(33) (i) H(Zn, 7r) = g(Zn, e - 7r) + h

(ii) m = x<P(e, 7r)
(iii) mlm = p. + a(e-e) - 7r - g(Zn, e-7r)
(iv) Zn!Zn = j(E) - n2
(v) El E = g (Zn, e - 7r) - (n1 +n2),
where
(34)
1-tw
Hi = iJH/iJZn = - [x(l-tr) sr(?r) + (tr-tw)]/(1-tr)<O,

Hi = iJHliJ'K = - i(l -tr) [1-Zn/(1 -tw)] sr'( 7r) < 0 (if Zn < 1 -tw).
From equation (33) (ii) we have
(35) e = e(m, 'K); ei =oeliJm = ll(x </>1)<0,

e1 =oeliJ'K = - </>2/</>1 :s;O.
Substituting this equation into equation (33) (i), we have
(36) H(Zn, 7r) = g[Zn, e(m, 7r) - 7r] + h.
Solving this equation with respect to 7r, we obtain
(37) 7r = 7r(Zn, m); 'Kt =iJ'KliJZn = (H1-g1)/[g2(e1-l)-H1J.

(-) (-) (-)(-) (-)
7r2 =iJ'KliJm = gze1/[H2+g2(l-e2)] < 0.

(-)(-)(-) (-) (-)
Again in this case, 'Kt < 0 if the sensitivity of the investment with respect to
the rate of profit Ig1 I is relatively small. From now on, we shall assume that
'Kt < 0.
Substituting equation (37) into equation (35), we obtain
(38) e = e(m, 7r(Zn, m)) = e(Zn, m);

eZn = iJeliJZn = e1 'Kt ~ 0,
(- or ) (-)
0
1
em = iJeliJm = ei +e2 'Kz = ( - - </>2
i
7r2) /¢1.
( - ) ( - or ) ( - ) ( - or ) ( - ) ( - )
0 0
158 TOICHIRO ASADA
Now, assume that em < 0. This assumption will be rationalized if the sensi-
tivity of money demand with respect to the change of expected rate of inflation
I<P2 I is relatively small.
Substituting equations (37) and (38) into equation (33) (i), and further substi-
tuting equation (33) (i) into equations (33) (iii) and (33) (v), the following "fun-
damental dynamical system" results.
(i)
(ii) E= {H[Zn. 7r(Zn, m)] - (h + ni + n2)} E = F2 (Zn, E, m)
(iii) m= {µ + a[Q(Zn, m) - e] - 7r(Zn, m)
- H[Zn, 7r(Zn. m)] + h}m
= F3(Zn, m)
In this system, the classical dichotomy no longer holds, because there is a
feedback effect from the monetary sector to the real sectors so that the system
ceases to be "decomposable" (see Figure 6).
Now, the (local) stability of the system is investigated by assuming that an
economically meaningful stationary solution (long-run equilibrium) exists. 13
The Jacobian matrix (J*) evaluated at the long-run equilibrium point of the
system (SB) can be written as
(39)
]* = [ F21*
0 F12*
0
F31* 0
where F12* =f'(E*) Zn* > 0, F21* = (H1*+H2*7r1*) E*, F23*

(-) (-)(-)
= H2* 7r2* E* > 0, F31* = [aezn*-(l+H2*)7r1*-H1*]m*,
(- ) (- ) (+ or ) ( - ) ( - ) (- )
0
and F33* = [aem*-(I+H2*) 7r2*]m*.
(-) (-) (-)
The characteristic equation of this system is
(40) .::l(A) = IAI - ]*I

-F12*
;>..
0
Figure 6. Directed graph of (Se).
where
0
(41) ai = - -F12*F21*, a3 = F12* (F21* FJ3*-F23* FJ1*).
( +) (?) ( +) (?) (?) ( +) (?)
The Routh-Hurwicz conditions for stable roots are given by
(42) (i) a1 = -F33* > 0

(ii) a3 = F12*(F21*F33* -F23*F31*) > 0
(+) (+)
(iii) a1a2 - a3 = F12* F23* F31 * > 0.
(+) (+)
These conditions are equivalent to the following set of conditions:
(43) (i) F31 * > 0

(ii) F33* < 0
(iii) F21 * F33* - F23* F31 * > 0
Now, the implications of these local stability conditions can be investigated.

Condition (43) (i) is satisfied if IH2 * I is relatively small. This condition is
160 TOICHIRO ASADA
satisfied also in the case when ezn > 0 and ex is sufficiently large. The condition
that IH 2 * I is relatively small is equivalent to the condition that the sensitivity of
capitalists' propensity to save with respect to the expected rate of inflation
Is/(7r*) I is relatively small (see equation [34]). The condition that ex is suffi-
ciently large implies that the government's monetary policy must be sufficiently
activistic.
Now, let us suppose that Is r' (7r*) I is so small that IH2* I < 1. In this case,
condition (43) (ii) is violated so that the system becomes locally unstable if the
monetarist money suppy rule (ex=O) is adopted. But, even if IH2*1 < 1, the
government can assure the condition (43) (ii) by setting ex > 0 sufficiently large.
Finally, consider the implication of condition (43) (iii). Suppose that condi-
tions (43) (i) and (43) (ii) are satisfied. Then, a necessary, condition for the
inequality (43) (iii) to be satsifed is F 21 * < 0. If IH 2 * I is sufficiently small
(i.e., IH2* I < H 1*/71" 1*), this condition is satisfied. Furthermore, F21 *
(-) (-)
F33 * - F23 * F 31 * becomes a linear increasing function of ex if F 21 * < 0 and
ezn *is relatively small, reflecting the fact that I</:> 2* I is relatively small. In this
case, the government can assure the condition (43) (iii) by setting ex > 0 suffi-
ciently large, even if this condition is not satisfied when ex = 0.
The result of the above analyses can be summarized as the following:
Proposition 1.
(i) Suppose that Isr'( 7r*) I is relatively small. Then, the equilibrium point of
the system (SB) becomes locally unstable when the monetarist money supply rule
(ex = 0) is adopted.
(ii) Suppose that Is/(7r*) I and I</:>2* I are relatively small. Then, the equilibri-
um point of the system (SB) becomes locally stable when the monetary policy rule
is sufficiently activist (i.e., ex is sufficiently large).
Next, the condition for cyclical solution around the equilibrium point is
investigated. The characteristic equation (40) can be rewritten as
(40)' >-3 + b>. - c = a(ex)(>- 2 + b)
where a(ex) = F33*; a'(ex) = em* m* < 0, b =- F12* F11*, and

( +)
Suppose that ISr '7r*) I is so small that a(O) > 0, b > 0 and c > 0. Moreover,
for simplicity, suppose that ezn * = 0, so that c is independent of ex. In this case,
the relation of equation (40) ' is illustrated in Figure 7. U(>-) in this figure denotes
the left side of equation (40) ', while V(>.; ex) denotes the right side when policy
Figure 7
u.v
U(,t)
parameter a is given, where 0 < a 1 < a 2 < a 3 < .... We can see from this
figure that (i) the characteristic equation (40)' has a pair of complex roots so that
the cyclical fluctuation around the equilibrium point occurs when m (0,
a 4 ) U (a 5 , oo ), and (ii) equation (40)' has no complex root so that the cyclical
fluctuation around the equilibrium point does not occur when m[a 4 , a 5 ]. These
conclusions and the results of the stability analysis are summarized in Figure 8.
Adaptive Expectation Hypothesis
In this section, the perfect myopic foresight assumption of inflation

expectation (equation [11]) is replaced by the following adaptive expectation
hypothesis.
(44)
In addition, for simplicity, assume that Sr is constant through time. Then,

the system (13) is modified as
162 TOICHIRO ASADA
Figure 8
cyclical noncyclical cyclical
0 1\_~~~~-v-~a-4~~~_,/\'--~~a-5~--~~~ • a
unstable stable
(45) (i) H(Zn) = g(Zn, e - 7re) +h

(ii) m = i<P(e, 7re)
(iii) mlm = µ, + a(e - e) - 7r - g(Zn, e - 7re)
(iv) Zn!Zn = wlw - 71" - n2 = f{E) + 11"e - 71" - n2
(v) EIE = g(Zn. e _ 11"e) - (n1 +n2)
(vi) ~e = -y(7r-7re); -y>O.
From equation (45) (i) we have
(46) e - 1re = cf>(Zn); cf> '(Zn) = {H'(Zn) - gi}lg2.

(-) (-)(-)
Also in this section, it is assumed that cf> '(Zn) > 0. Substituting equation (46)
into equation (45) (ii), we have
(47) 11"e = 7re(Zn, m); 7ri = 07ref()Zn = - 4'1 cf> '/(4'1 +¢2) < 0,
7rz = a11"elam = l/i(¢1 +cf>2) < o.
From equations (46) and (47), we obtain
(48) e = cf>(Zn) + 7re(Zn, m) = e(Zn, m);

Qt= oeloZn =cf>'+ 1re = ¢2cf>'/(¢1+cf>2)~0.
e2 = aeiam = 7rz < o.
Then, the system (45) can be summarized as follows:
(Sc) (i)
(ii) E= {H(Zn) - (h+n1 +n2)} E = F2(Zn, E)

(iii) (7r1e/-y) Zn + (llm+11"2el-y) m = µ, + O! {e(Zn. m)-e}
- 7re(Zn. m) - H(Zn) +h = F3(Zn, m)
Now, the following proposition can easily be proven (the proof is omitted).
Proposition 2.
A set of local stability conditions of the system (Sc) is given by (i) the
small expectation coefficient 'Y and (ii) the large monetary policy parameter a.
Note that the model of this section is reduced to the model of the Goodwin-
type growth cycle if the expectation coefficient is infinite, because
lim 71" 6 = lim (11" - 1reI 'Y) = 11".
-y- + 00 -y- + 00
* * *
The main source of instability in the models presented in this chapter stems
from the money market. Suppose, for example, that the expected rate of inflation
is increased. Then, the nominal rate of interest tends to increase in pursuit of the
increase of the expected rate of inflation. This accelerates the increase of the
velocity of circulation of money (the reciprocal of the Marshallian k) through the
substitution from money to bonds. The increased velocity induces the growth of
the rate of inflation, which entails higher expectations of the rate of inflation. In
the Goodwin-type model, this instability is locked in the monetary sector, while in
the models that follow, instability spreads to the the real sectors through the
variable propensity to save or the discrepancy between expected and actual rates
of inflation. To offset this instability, the monetarist money supply rule is ineffec-
tive but activist rule is required.
In the models presented in this chapter, the budget constraint of the govern-
ment plays no dominant role.
The budget constraint of the government may be formulated as
. .
(49) pG+R-T=q B+ M
where R = nominal payment of the interest on bonds

T = nominal income tax
q = market price of the bond
B = existing bond stock
For simplicity, assume that the bond is consol type. Then, R = Band q = 11 e.
In this case, we can rewrite equation (49) as follows.
. .
(50) G + Blp - Tip = Bl eP + Mlp
Although this relation also must be met in the models in this chapter, this
equation does not affect the dynamics of variables such as E, Zn or m. However, if
164 TOICHIRO ASADA
consumption from the interest on bonds and the wealth effect on consumption are
introduced, the dynamics of the system are no longer independent of equation
(50), because this equation becomes another root of the transmission of instability
from the monetary sector to real sectors. Obviously in this case, monetarist policy
rule is ineffective to stabilize the economy, so that some sort of activistic stabili-
zation policy rule must be required. 14
Notes
I. See Marx (1967) Part VII and Sylos-Labini (1970) Chap. II.
2. See Hirsch and Smale (1974) Chap. 12.
3. Although the object of this chapter is somewhat similar to that of Di Matteo's
(1984) paper, it was independently written and the analytical details of both are consider-
ably different. As for the various developments of Goodwin's growth cycle model, see, for
example, Balducci, Candela and Ricci (1984), Desai (1973, 1984), Desai and Shah (1981),
Flaschel (1984), Glombowski and Kriiger (1984, 1986), Goodwin (1983, 1984), Medio
(1980), Pohjola (1981), Sato (1985), Van der Ploeg (1983, 1984), Velupillai (1979),
Wolfstetter (1982) et. al.
4. This implies that the technical progress is the "Harrodian-neutral" type.
5. From the expression ofr and rn. "distribution frontiers" are drawn (see Figure Fl
and Figure F2). If we assume full capacity utilization of capital, we haver = x (1-Z),
which is nothing but the Sraffian equation in our model (in Sraffa's 1960 notation, r = R
(1-w)).
6. Note that [s,(1-t,) + (t, -tw)J I (1 -tw) > 0 since 0 ::5 fw ::5 t, < 1. By the way, equa-
tion(*) can be rewritten as rn =(g+h-t)/s" where t= TIK= [twwL+t,(X-wL)]IK. This
equation relates the after-tax rate of profit (rn) to the rate of accumulation (g), the rate of
government deficit (h-t) and capitalists' propensity to save (s,), which is a basis of
Kaleckian theory of income distribution (see Kalecki 1971).
7. As for the proofs, see, for example, Tobin (1969) and Yoshikawa (1980).
8. Note that under the assumption <I> '(Zn) > 0, e - 7r moves in the same direction
as Zn (see equation (14)).
9. We assume that a * 1.
10. This statement follows from equation (16).
11. As for the formal definition of the structural (in)stability, see, for example, Hirsch
and Smale (1974 Chap. 16), Medio (1980) and Valupillai (1979).
12. This type of saving function was introduced by Okishio (1979).
13. Economically meaningful stationary solution implies that the stationary solution
(Zn*, E*, m*) of the system (SB) with the properties 0 < ~ < 1, 0 < E* < 1 - lw
and 0 < m*.
14. As for the formal analysis of stabilization policy in which the government's budget
constraint plays an essential role, see Asada (1987).
References
Akashi, S., and T. Asada. 1986. "Money in Kaldorian Cycle Theory." The Economic
Review. The Institute of Economic Research, Hitotsubashi University 37:169-77.
Asada, T. 1987. "Government Finance and Wealth Effect in a Kaldorian Cycle Model."
Journal of Economics (Zeitschrift far NationalOkonomie) 47: 143-66.
Balducci, R., G. Candela, and G. Ricci. 1984. "A Generalization ofR. Goodwin's Model
with Rational Behaviour of Economic Agents." R. M. Goodwin, M. Kriiger and A.
Figure F1 . Pre-tax distribution frontier.

r
Figure F2. After-tax distribution frontier.
fn
( 1-t,) ox
1-t~
166 TOICHIRO ASADA
Vercelli, eds. Nonlinear Models ofFluct1<ating Growth. Berlin, Heiderberg, New York
and Tokyo; Springer-Verlag.
Burmeister, E. 1980. Capital Theory and Dynamics. Cambridge: Cambridge University
Press.
Desai, M. 1973. "Growth Cycles and Inflation in a Model of the Class Struggle." Journal
of Economic Theory 6:527-45.
Desai, M. 1984. ''An Econometric Model of the Shares of Wages in National Income: UK
1855-1965." R. M. Goodwin et. al., eds., op. cit.
Desai M., and A. Shah. 1981. "Growth Cycles with Induced Technical Change." Eco-
nomic Journal 91:1006-10.
Di Matteo, M. 1984. "Alternative Monetary Policies in a Classical Growth Cycle."
R. M. Goodwin et. al., eds., op. cit.
Dornbusch, R., and S. Fischer. 1978. Macroeconomics. New York: McGraw-Hill.
Flaschel, P. 1984. "Some Stability Properties of Goodwin's Growth Cycle: A Critical
Elaboration." Zeitschrift far NationalOkonomie 44:63-9.
Flaschel, P. 1984. "The Inflation-Biased 'Natural' Rate of Unemployment and the Con-
flict over Income Distribution." R. M. Goodwin et. al., eds. op. cit.
Foley, D. K. 1986. "Stabilization Policy in a Nonlinear Business Cycle Model." W.
Semmler, ed. Competition, Instability, and Nonlinear Cycles. Berlin, Heiderberg,
New York and Tokyo: Springer-Verlag.
Friedman, M. 1977. Inflation and Unemployment. Occasional Paper No. 51. London: The
Institute of Economic Affairs.
Glombowski, J., and M. Kruger. 1984. "Unemployment Insurance and the Cyclical
Growth" R. M. Goodwin et. al., eds. op. cit.
- - - - . 1986. "Some Extensions ofa Classical Growth Cycle Model." W. Semmler,
ed. op. cit.
Goodwin, R. M. 1967. "A Growth Cycle" C. H. Feinstein ed. Socialism, Capitalism and
Economic Growth. Cambridge: Cambridge University Press.
Goodwin, R. M. 1983. "A Note on Wage, Profits and Fluctuating Growth Rate."
- - - - . 1984. "Disaggregating Models of Fluctuating Growth." R. M. Goodwin et.
al., eds. op. cit.
Hirsch, M. W., and S. Smale. 1974. Differential Equations, Dynamical Systems and
Linear Algebra. New York: Academic Press.
Kalecki, M. 1971. Selected Essays on the Dynamics of the Capitalist Economy. Cam-
bridge: Cambridge University Press.
Keynes, J. M. 1936. The General Theory of Employment, Interest and Money. London:
Macmillan.
Marx, K. 1967. Capital Vol. I New York: International Publishers.
Medio. A. 1980. "A Classical Model of Business Cycle." E. J. Nell, ed. Growth, Profits
and Property. Cambridge: Cambridge University Press.
Okishio, N. 1979. "Theoretical Frame of Monetarism." The Economic Review. The
Institute of Economic Research, Hitotsubashi University 30:289-99. (In Japanese).
Pohjola, M. T. 1981. "Stable, Cyclic and Chaotic Growth: The Dynamics of a Discrete-
Time Version of Goodwin's Growth Cycle Model." Zeitschrift far NationalOkonomie
41:27-38.
Robinson, J. 1956. The Accumulation of Capital. London: Macmillan.
Rose, H. 1967. "On the Nonlinear Theory of the Employment Cycle." Review of Eco-
nomic Studies. 34 (April):153-73.
Rowthorn, R. E. 1980. Capitalism, Conflict and Inflation. London: Lawrence and Wi-
shart.
Sargent, T. 1979. Macroeconomic Theory. New York: Academic Press.

Sato. Y. 1985. "Marx-Goodwin Growth Cycles in a Two-Sector Economy." Zeitschrift
far Nationa/Okonomie 45:21-34.
Sraffa, P. 1960. Production of Commodities by Means of Commodities. Cambridge:
Cambridge University Press.
Sylos-Labini, P. 1970. Problemi de/lo Sviluppo Economico. Editori Laterza & Figli, Bari.
Japanese edition, translated by H. Onoe, Heibonsha, Tokyo, 1973.
Tobin, J. 1969. "A General Equilibrium Approach to Monetary Theory." Journal of
Money, Credit and Banking 1:15-29.
Van der Ploeg, P. 1983. "Predator-Prey and Neoclassical Models of Cyclical Growth."
Zeitschrift far Nationa/Okonomie 43:235-56.
Vander Ploeg, F. 1984. "Implications of Workers' Savings for Economic Growth and the
Class Struggle." R. M. Goodwin et. al., eds. op. cit.
Velupillai, K. 1979. "Some Stability Properties of Goodwin's Growth Cycle," Zeitschrift
far Nationa/Okonomie 39:245-57.
Wolfstetter, E. 1982. "Fiscal Policy and the Classical Growth Cycle." Zeitschrift far
Nationa/Okonomie 42:375-93.
Yoshikawa, H. 1980. "On the 'q' Theory of Investment." American Economic Review
70:739-43.
Qualitative Effects of Monetary Policy
in "Rich" Dynamic Systems
Peter S. Albin
It is widely presumed that monetary policy has informational effects on business

expectations as well as allocative effects. The former influence may affect system
dynamics particularly, in already volatile settings. This chapter reports on on-
going computer experiments aimed at isolating the qualitative informational
effects of procyclical (accommodative) and countercyclical monetary interven-
tions. The modeling approach is described first and is followed by an explanation
of the associated schema for complexity classification of dynamic systems. The
next section contains the specification of monetary controls. The results and
interpretations are then presented. The analysis shows that intervention accord-
ing to conventional rules can: a) alter the intrinsic dynamics of the system, e.g.,
from periodic to aperiodic (chaotic) or the reverse; b) delay natural equilibrium-
seeking tendencies; c) worsen the tradeoff between approximating an aggregate
target and reducing fluctuations around the target; d) worsen the tradeoffbetween
approximating an aggregate target and achieving "selective" objectives. A
"countercyclical" rule is not necessarily superior to an "accommodative" rule
under these conditions nor does "finer tuning" necessarily lead to finer results.
Although the models specify only myopic expectations and intervention rules, the
analysis generates a strong presumptive case with respect to higher-order specifi-
cations as well.
The Experimental Setting
The experimental setting is a computer model simulating variations in the invest-

ment of individual firms who draw information on the likely path of the system
from observation of the actions of their economic neighbors. The system's
dynamic behavior in the absence of policy intervention is examined first. Detailed
168
MONETARY POLICY IN "RICH" DYNAMIC SYSTEMS 169
properties of the system are given in Albin ( 1987); its salient characteristics are
summarized below. 1
1. The economy consists of N firms organized in M industries. It is conve-

nient to operate with 100 firms in 5-firm industries.
2. Each firm sets a three-option investment plan contingent on expectational
information as to whether the business climate will be unfavorable, normal, or
favorable. Normal is the firm investment level that aggregates to the level of
capital formation which supports a balanced growth path in the sense of Solow
(1956) and as translated to a nonlinear system by Day (1982). The unfavorable
and favorable climates can be thought of as worst-case and best-case scenarios.
3. The data needed to calculate the three separate plan options are assumed
to be price-dimensioned and obtainable by the firm without additional cost in the
ordinary course of doing business. These data and the mechanics of calculating
optimal plans are of standard type and are not considered here.
4. The strategic information needed to choose the appropriate plan option is
costly and available only with some delay. The firm is assumed to base its strategic
decision on "leading indicators" drawn from observation of the immediate past
actions of its "economic neighbors." This is the most timely and least costly
source of strategic information.
5. The "economic neighborhood'' can overlap the industry but is not identi-
cal to it. Neighborhood relationships represent variously intra-industry relation-
ships among competitors, supplier-customer relationships, and/or inter-industry
input-output relationships.
6. The neighborhood is specified as an index set. Thus the index set (-2, -1,
1, 2) defines for firm J the neighbors (J-2, J-1, J + 1, J + 2). If all firms have the
same index set, the system is homogeneous and the industry distinction is purely
nominal.
7. If the neighborhood set varies according to the firm's position within its
industry, the system is heterogeneous. JYpically, one can designate one or two
firms as "leaders" who draw information from outside the industry, while the
remaining firms react to the leaders and within-industry competitors. Heteroge-
neous models can capture a variety of oligopolistic behaviors and "consonant
rivalry" (Kuenne, 1979).
8. The firm selects a negative, zero, or positive increment-designated - 1,
0, + 1-to its normal investment level. The selection is made according to an
algorithmic rule for reacting to the investment actions of its neighbors as ob-
served in the immediately preceedi~g period. The rule is a model parameter
which is varied experimentally. Since there is only a relatively small number of
observable actions within the neighborhood, it is feasible to simulate all possible
integer rules, thus generating the full strategy set of the firm (which set includes
both meaningful and nonsensical behaviors).
9. In the simulations described here the rules are restricted to functions of
170 PETER S. ALBIN
the algebraic sum of neighbor actions. In anglicized BASIC, a rule takes a form
like: IF THE NET TOTAL OF NEIGHBOR ACTIONS IN THE PREVIOUS
PERIOD IS 3 OR 4 THEN SELECT + 1 (a positive investment increment) ELSE
IF THE NET TOTAL OF ACTIONS IS -3 OR -4 THEN SELECT -1 (a
negative increment) ELSE SELECT 0 (normal investment).
10. Firms are arrayed on a line segment and the left-most and right-most
industries are treated as experimental boundaries. In the cases reported here, the
boundary industries are constrained to normal investment at all times.
11. For convenience, firms are assumed to be approximately equal in size. In
addition, the investment increment can be assumed to be of constant dollar
amount for all firms. Thus SUM, the algebraic sum of firms' actions can be
construed as either an "index of business sentiment" or with a scale transforma-
tion as a measure of "excess aggregate investment."
12. The system is started by distributing an initial pattern of non-zero shocks
to firms selected probabilistically. Therafter, the system evolves deterministical-
ly. The distribution of shocks is varied experimentally.
Figure lb gives representative outputs for a homogeneous, 100-firm model in

which the neighborhood and pattern of initial shocks are fixed, and the rule is
varied experimentally. Firms are arrayed horizontally. Above normal, normal,
and below normal investment levels are displayed as black, white, and grey,
respectively; and each line of patterns represents a time step of the system. To
help in identifying characteristic patterns, Figure la gives outputs for a simpler
two-state model (above normal= black; normal or below normal= white). The
displays can be thought of as low-detail snapshots of the contents of a data bank
giving values of a firm-level variable. The organizing principle is that found in
many standard data sets: firm records organized according to industry, and
industry records according to product or process adjacency (as in SIC codes).
Looking ahead to Figure 2, the conventional plots to the right of the snapshots
give the time series of SUM around the base line of zero excess aggregate
investment.
Complexity Classification of Dynamic Behaviors
Four distinct types of dynamic behavior appear in the figures (indicated by

numerals and notes). It has been determined that the same rule will consistently
produce the same qualitative behavior for all nontrivial initial configurations.
Each distinctive visual presentation, or characteristic signature, has been shown
(Wolfram, 1984; Albin, 1987) to correspond to a qualitative level of computa-
tional complexity that embraces dynamic properties, statistical characteristics of
generated data, and linguistic properties. The latter are critical since they deter-
mine the scale of data and computational resources needed to comprehend a
system and project its future path.
Figure 1a. The Four Qualitative Types in Two-state Models.
CRITERIA 2 2 2
·'4- • -.:a:: -.-. -:i..- ·,lt-._-r·
..~.. - .. ...a::..• •
Type 1
CRITERIA 3 4 5 CRITERIA 2 3
Type 2
'.::j
-
CRITERIA CRITERIA 2 3 3
.... -.J
N
.,,,,.,,__,,,_~-· .__.._-'- _,,_- - -_, ~ ...L-
~. . -~_-'- -'-....._ _._-'-...J,,,..~~ ~_._ _._.......,
...... --'-'--'- "'--'- • -'- _,_ • _._ ~.a.. - ~ .._ - L -_, _.__ -'-t
-'- - -'- ....._- ....._ _,_-• .a.. • ..&....1.-.L......_ -'- ...a... • _._ •
~~ _....~ ~~_.___._- -_,___, L.~ ~
...
-"- 4- ~-::-_._ -s....~-~...._~~ ....._.... L._,_""'-'--'-
_.__ _._ &.-...:.- -'- - _,_ ...._ _, _._ __._ ....._ -L.
. . . - .....__.._
~ '-
_, L- .J-
_,_
~ ....&- -~
-=- ....._-z..........-'- - ...._....._-'--A--'-
.............._ _..__._ . . . . .
..__,_ .._-'- .......:... ......._.__,_
-
-:;_........~_,_":_-'-J-_, ._-'-_._ '-........__,_-A--'- Type 3
-
J- .....:...._ ~....... -L-_,_ _._ -'- ~""""-
:::::_.___.__._~ ___,_ - ' -~...&.. ~....._ _.__,_ _,_-'-
....._ ~ ....1.... -J.- - - - ' - _,_ - ' - _._ _,,_
_,_ -=-.._~_.._~ -L-..............i....
......_-'- -1--'-:::::
_,_-4- -L-~- -a..-'- _,_ -"'- -'- ....._
...L-. ~-lo- _.::-_. _._:::::: :::::_,_-&....
.~-:- ~- :- . ....-'--'-~~.........._._ .
- - ·· - .. .... _,_ -'-~ - --_.__
-1-~ - L - _.__._
-'--~ •• · · - -L- ...:..--~~~~-L.....L-~_._
J.:""- .. ..:....-1- • _,_~~ . . : - ._..&.-'- •
. _...,,;""'-".
-• .......-,.....I. L'~...!r
1lr: .....·....,..-....,. ._.·<- ..
-- · ·~~ -.-'-
~~y..'"'W".I L~._J_.'V°J t. •• .:~.~
......_ ~~'- ...I . . . . ·-_, ' ....... I . . ; ... ·-_.___
T"W',;~_,---,
..._..... L _. . '-"'W"'
L_. c'.~ _, . -~ ~ _';l:'lwir_-W ..,....,,. ·~-.r' "".""'....;
e:J ·-............ .J. -v-...: _J. -· . . . -
•• - - L. ."W""·· _, ~ ....-.........,.. ~
-=-·....
..... ~-=·......J. ..,&,.. - * ...., .......... --- *-:_ .........::.:,
~ ... :--...,.;..~ -~- . ... .1. .......- ..... ·- ......_...._ ....... .
Type 3 ......... ~::... ......'¥".... ... .........._ ......
L .....- • • • . :"llP': '- _,~ '- ..,&,.. .._- _. J ........._
-' l.....,...J L '- . ..1 ~ .... ..-.·
L ... '- ....Lt.-_ '- I • • ._- ~.. L. o.,.,; .... ----=-
... ~ • 'w .,.. I ~ 'W" ' - ...._. • - ......
_.;j
'-u" I . . . . ...... • J ·- .. - ....- ..... ·-
_....... ,_ ~ ~·-,, L ._,..:;i.J .._.._.... - .... ~
L.....
~
........
. ..,;.,_.. ...
L~...- ~
.....- .. ... ,-_. ':-'-
.. . . ~*- ·---
... -..+..:
..:..L- - ....
..... _.......;..__.•
~........ ·:!':-':fv •-: - _.. ~
_, ~:..._,"1P_~J L...... + ..:.
.. ..... ,._ °* '-_-·¥-J L ..... .
.. ..... -' ~ -sir I .... • - · ·-_i"'F ...I 'v'. ~ •- ~
..... ... -.tr: .... - "r' ._ ...!
-..sw+" "IP." ::.&;"•• r:ifr ...
~ ......~.. * "'"* ·-.-· ·- __,.... . . ' ..... . . .
·- ~· ·:.{· r:-. ·-:::'
:.Jr .....
.17'"• -=;;; :-:...,.,·.~ ·
.......~....,
·. ..~J ,1,:...;
,~· ' bV. :'~±:~:l~.1..:: ·-·-·
!i""-. r.i -~ ~·,c. ..,,·. .. ~ , .............
-· ....·.++·!ii:.:. ··~ a.:., ,.,,,_.~.:,,.
~.i.~.u ,+_~ l'" .;.:a. . -. r.::,i·-~:i.~'a.·
- ,,_ •., 1 r.- _, • ~.J: ·, ~:o.A..;- .-,T~i
~ _,,... ::~ e~ ·.~...
·~-
'
... ~.. .::..... ~
_, .......,..j.W
·:fi·T'..... F
• "' - • • ·- - ;.... •• ~ • Ir •• ;·..- •• ••
-~-~ ~-'='-... ~:s:t&...c.;
;;...~M· =. . Type 3
~~:.i.:rl'I
~~}:-.~:=-• ... 0
. .•
Ir &. .I 11 ~ .=.:_:•41
: ~· t" :';"'"i-:1:
µ;,. -1:itll
:~ ~'.). ~ ::.~~~-:~:;: :i; .. .
-·.,~~lr.~~ 7¥-~¥g.;-'M.g,.
. L).;~:z:~· ··:.!.;~;
M. ;!¥:r;>:;..~
)!:~ ~;,:.;..~.~:.i;,~' ·!~- ·=-ti:-;-c ~~~.:::
.-~"--: ;:· ~ 1,; ':;1·~ - -~ e...;..... ~.o- ,:t..~
...=
~ ~... im-~ r+'!t~ ;~ . •. -·.r.....
CRITERIA 2 4 4 4 5
.
...,. •• ,~... ~- ~:.~ .J."·~ -~~·;::;~:,:r ·
u:r ·~~ ....·.. · ~
... . .
,.l:-;-t. j~.i.·
.~ .-...., ···..e -... ·-:·:·· ' .
• • • • I I • •
-'=" i •
t :r.'): I
.:.i ·:·iL.
_,r.:;;;.....
·.;. ·.::.:·,..
Type 4
Notes: In each panel, values for 100 sites (firms) are printed horizontally for 80 vertical time steps. Plotted sums are not given in these panels-but
see Figure 2, below. The "criteria" are rule parameters: thus "CRITERIA l 4 5" represents the rule "print as black if l or 4 or 5 of your neighbors
printed black in the previous time step." All cases are generated for an identical pattern of starting values which was initially produced randomly.
Note that the case on the right for type 2 is "transitional." It yields complicated cycles with periodicities that vary with the initial pattern-thus, it
does not emulate a limit cycle in the strictest sense. In othu topologies this rule generally produces aperiodic output. Incidentally, this rule: ''print black
if l or 2 or 3 neighbors previously printed black," is a quite plausible formulation in expectational models. -"'.....
Figure 1 b. Qualitative Behavior of Three-state Models.
_,
~
CRITERIA 2 3
Type 2
CRITERIA 3 3 CRITERIA 2 3 4
:: =~
l
Type 3
~
~~li
I lW
CRITERIA 2 2 CRITERIA 2 2
Type 3
CRITERIA 3 4 CRITERIA 2 4 5
· 4,• ~ .!'·· --~:·
'.r:-: • I -T'"
-:- J.~ ·-
·~·
..rV"'"-
·-,·m:·.::.·,.·
1
Type 4
·~
l!Ililli
rn .
~
niml:
Notes: 'JYpe 1 cases (all black, white, or grey) are not shown. The case in row 1 is type 2 . The case on the right in row 2 is transitional as noted in Figure la. The
other type 3 cases display ''pseudo sectors'' of like activity. The case to the left in row 4 is transitional in the line-segment topology; it displays long transients with -.I
Ul
type 4 properties.
176 PETER S. ALBIN
Qualitative 1jpes of Dynamic Behavior
Properties of four behavior types are given in summary form.
1. Uniform Stable Behavior. All sites (firms) quickly assume a common value
indicated by a uniform black, white or grey field.
Dynamics: The system represents a stable equilibrium with all trajectories of
summed site values attracted to a single limit point.
Computational Complexity: After transients die out the system can be replicat-
ed or projected using only a simple calculator or a computer without memory.
Linguistic Category: Patterns of white, grey, and black dots representing site
values can be thought of as strings of symbols which can be processed according
to linguistic rules. The manner in which various initial strings are processed into
uniform data is characteristic of a (Chomsky) regular language.
2. Simple Stability or Periodicity. Particular sites or local groups of sites take

on constant or simply-periodic values. This behavior is observable as horizontal
or vertical banding of greater or lesser complication.
Dynamics: The system is described as stable periodic. Summed site values
form limit cycles. A system evolving from a finite configuration ofnonquiescent
sites remains finite.
Computational Complexity: The system can be simulated on a finite memory
device.
Linguistic Category: The processing of symbols representing site values con-
stitutes a context-free language-e.g., a given string oflocal site values regardless
of where it is situated in the snapshot diagram always produces the same future
strings.
3. Aperiodic Behavior. In one manifestation, the characteristic signature ap-

pears "noisy" with no apparent regularities. Closer inspection confirms the
appearance of a wide variety of dot patterns in strings of arbitrary length. In the
technical literature, this presentation has prompted rigorous description via ergo-
dic theory (Are all possible strings encountered?) and entropic measurement (Do
all possible dot patterns occur with equal frequency-is the system maximally
disordered?). In another manifestation, the characteristic signature displays re-
current patterns, such as the self-similar "blocking triangles," and a reduced set
of dot patterns. This presentation is confirmed by entropic measurements and
prompts calculation of fractal dimensions.
Dynamics: The system is aperiodic and time irreversible. Summed values
suggest chaotic trajectories around strange attractors. An unbounded system
expands into vacant sites and ''pseudosectors'' with one value predominant in
multistate systems may appear and persist. As noted, strings of site values may
exhibit statistical regularities; these properties are unique to each generating rule
(firm strategy) and are insensitive to the initial pattern of shocks.
Computational Complexity: The memory requirements for a simulating device

grow without restriction. These requirements are "irreducible" also meaning
that there is no "analytical" shortcut to full system emulation.
Linguistic Category: The evolution of the system is "context sensitive,"
meaning that a given string of site values may produce different successor strings
at different locations in the snapshot diagram.
4. Irregular and Persistant Behavior. The characteristic signature consists of

several irregular forms which interact in unexpected ways. Different initial
patterns may produce dramatically different snapshot diagrams.
Dynamics: Depending on the initial pattern of shocks, this type of irregular
system may grow, cycle, contract or die out. Intricate local structures may
persist, die out, or propagate leading to intricate interactions with other local
structures, and "time irreversible" systems suggest historical development. Few
statistical regularities can be detected in time series of system aggregates; these
may exhibit shifts which mimic structural changes.
Computational Complexity: Local structures act, in effect, as components of a
universal computer. The system is computation irreducible; furthermore, limit-
ing system behavior and statistical properties for arbitrary imput are effectively
unpredictable.
Linguistic Category: The "human unpredictability" of the system is captured
by the Chomsky category of "unrestricted language"-formally corresponding
to the arbitrarily large simulation device required to emulate the system.
Projective Properties
For the economist, an intuitive understanding of the classification system can be

gained by asking how a forecaster for a firm might decode a time series of local
observations to derive a closed-form representation or to project the system and
thus improve upon the current myopic rule for forming expectations. For exam-
ple, there would be little problem in handling the type 1 uniform equilibrium. It is
clear that one need only observe a few time slices to confirm uniformity and that
this information would suffice to project the system arbitrarily far into the future.
In terms of computational resources, one would need only a pocket calculator
(without memory) to multiply the equilibrium value by t and so effect a growth
extrapolation t years into the future.
For the type 2 periodic system, a time slice of observed data thick enough to
contain the full record of a cycle would suffice for projections. These data would
fit within a finite memory device. Thus, for a cycle ofperiodp, in order to project
the system to period t arbitrarily far in the future, one needs only the capacity to
calculate a modular division of (tip) and look up the corresponding value in the
memory device. Perhaps, one might wish to observe more cycles to confirm the
dynamics, but the principle is clear: there exists an observation of finite length on
178 PETER S. ALBIN
the past behavior of a determinate system that tells all that is necessary to project
the future behavior of the system with simple means.
This commonly-held presumption on resource bounds for observation and
extrapolation no longer holds for the third and fourth types of systems. Put one
way, in order to predict the t'h state of the system, the entire record to t - 1 is
needed; put another way, there is no bounded store of data which tells all that is
necessary to know about the system; put still another way, the marginal value of
an additional observation does not decline. The computer needed to represent
such systems must have memory that grows in size with the length of the projec-
tion; furthermore, the computer needed for the type 4 system must be without any
effective restriction on its logic. The distinction between type 2 and type 3
systems is particularly important for the theory of economic expectations. In a
type 2 system, an economic agent needs only to observe a full cycle of data within
a restricted band of adjacent sites. Once a particular pattern of site values recurs,
the whole future history of that local band can be extrapolated. In a type 3 system,
the recurrence of a particular local pattern is of limited use in forming projec-
tions. The further future in the local band varies according to "context," the
values at distant sites. The data and computatioital resources needed to project the
"context sensitive" case are far greater than those needed for the "context free"
setting and greatly exceed those generally assumed for practical expectations
formation. The "context free/sensitive" designations allude to an isomorphic
system for classifying linguistic complexities-the Chomsky ordering.
Modeling Considerations
The four qualitative categories exhaust the dynamic potentialities for recip-
rocally interactive systems. The characteristic signatures appear in systems
with more states (color printing is required), narrower or broader neighbor-
hoods of interaction, higher-order lags, mixtures of rules and neighborhood
boundaries, and additional dimensions (Smith, 1971; Albin, 1975; Wolfram,
1986). This particular modeling scheme was chosen for a number of reasons. The
three-state plan specification corresponds quite closely to the familiar (worst-
case, normal, best-case) scenario format for expectational data and is convenient
for black and white printing. The five-firm neighborhood is suggestive of an
industry size in which strategic interaction is likely. Finally, the model is small
enough so that all strategic rules (of the integer type specified) can be investigat-
ed. This is probably the simplest model that is "recognizably economic" in its
information cost assumptions, consistent with a general equilibrium growth
framework, and capable of generating the full range of dynamics. The model was
stripped of obscuring detail, hence the austere assumptions. Note, finally, that
this type of system is characterized by ''complexity tradeoff,'' wherein a particu-
lar level of dynamic richness can be obtained in narrow-neighborhood models
with many states or in broad-neighborhood models with fewer states. Thus, the
dynamic behaviors of a fully-elaborated model can be no richer than those

exhibited here.
Dynamics and Expectations
Limit-point, and limit-cycle trajectories are, of course, quite familiar in the

literature pertaining to monetary dynamics. The strange-attractor (chaos) case is
of comparatively recent interest-mostly following work by Day (1982)-but is
produced with some awkwardness and considerable obscurity by curvature tricks
in nonlinear systems. The extremely rich irregular dynamics of the fourth type
are not considered in the literature at all, although a natural consequence of a
market economy acting as a mesh of interconnected computers. How important
are the rich cases? In models of appropriate scale-2 to 5 expectational states, 3-
to 7-firm neighborhoods-the extremely-rich type 4 cases are almost as likely to
occur as limit-point equilibria (for approximately 6% - 10% of the rules in a given
model). The limit cycle appears somewhat more frequently (generally, between
10 % and 20 % of the time), while the varieties of chaos are the most likely to occur
and increase in frequency as model size grows.
Are there a priori grounds founded in economic theory for anticipating one or
another type of dynamic behavior to prevail? I think not. A case can be made for
just about any one of the rules as a strategy for reacting to specific expectational
data. So long as there are advantages (timeliness, reliability, availability) to be
gained from drawing on low-cost local data channels and direct observation,
these sources will be used. The assumption made in each simulation that all firms
in the economy adhere to the same rule is, however, patently implausible. It is
employed in the spirit of experimentation. I conjecture that in an actual market
economy only a fraction of firms are situated in industries where there is play for
reciprocal interactions. Furthermore, such industries may differ as to the prevail-
ing rule, and rules may alter over time either spontaneously or in an evolutionary
way. Thus, the actual system might be viewed as a photo mosaic made up of
several snapshots with accordingly complex composite dynamics.
Industry Structure
According to the above conjecture, there may be zones of differing intrinsic

dynamic behavior within the economy. Furthermore, tendencies for self-organi-
zation into expanding or contracting ''pseudosectors'' have already been seen
within several snapshots. What about industry structure per se? Included in the
policy simulations to come are heterogeneous specifications which assume the
strategic leadership of key firms, along with restrictions on channels whereby
firms within an industry draw information on conditions outside the industry.
Briefly, the four complexity types reappear. However, sectors of like behavior
persist longer, and frequently the system is easier to stabilize in the manner to be
next described.
180 PETER S. ALBIN
Policy Interventions
Active stabilization policy, in general and in the context of our model, involves
attempts to achieve one or more of the following objectives:
1) centering the trend of the economy on a target path (taken here to mean the
line of zero excess aggregate investment);
2) reducing the amplitude of fluctuations in key aggregates (thus reducing the
real burdens of excess-capacity and tight-capacity production;
3) reducing sectoral imbalance within the aggregate (zero excess aggregate
investment achieved with a sectoral boom and offset by depression in other
sectors is likely to reduce the potential growth rate of the system;
4) reducing within-sector variability (thus reducing excess startup and shut-
down burdens at the level of the firm or industry);
One might think of these objectives as priority ordered. In practice, though,
loss functions might be such that the system is better off sacrificing a primary
objective, e.g., straying somewhat from the target path may yield better results
than centering on the path at the cost of wider fluctuations. In brief, the objectives
may also be viewed as ends whose attainment can involve tradeoffs because of
system interdependencies. In addition, two ancillary outcomes of intervention
which might be viewed as policy significant under certain circumstances will be
considered. The circumstances are where authorities co-operate and the actions
of one authority may facilitate the operations of another. These outcomes are:
5) transforming the qualitative dynamics of the system; and
6) selective simplification of dynamics.
The notion here is that if monetary intervention can turn a chaotic economy
into a periodic one or reduce the complication of pre-existing periodicity, the
tasks of another policy agency may thereby be made easier-or, perhaps, the
economy becomes less sensitive to shocks which otherwise might have to be
sterilized.
In modeling policy interventions, an "external entity," e.g., the monetary
authority, is specified as the shared neighbor of all firms in the system. This entity
transmits a common signal to each neighborhood, in effect, through the agency of
the banking system. Firms interpret this signal as an expectational datum. Recall
that firms prepare three-contingency investment plans that incorporate assump-
tions regarding long-term real interest rates and monetary conditions. Thus,
capital-allocative effects are already built into the plans. The choice of plan
option in the short-run rests on a reading of the local and aggregate business
climate. The current monetary signal is viewed as an early warning indicator to
be included with observations of the immediate past actions of the firm's econom-
ic neighbors.
There is, to be sure, a degree of artificiality in this attempt to isolate the
cyclical and expectational from the capital-theoretic-no provision is made for
plan revision based on realized experience. The scheme does, however, have the
virtue of simplicity and captures salient aspects of the signaling function.
The authority bases its signal (policy) on observation of the immediate past value
of excess aggregate investment. It follows either a myopic countercylical rule
(tightening credit in response to excess investment) or a procyclical rule. In
different experiments, the policies are applied with greater or lesser sensitivity to
the margin of excess (coarse or fine tuning).
Simulating Monetary Interventions
The computations are performed as follows:
1. A 100-firm economy is simulated; first, for a homogeneous model and a

selection of rules, then for several different heterogenous models and the same
rules.
2. The economy is allowed to run without intervention for 20 periods in order
to display its basic dynamics and characteristic signature. A control regime is
instituted in period 21 and then "inverted" in period 41.
3. The control is based on SUM, the algebraic sum of ( -1, 0, + 1) actions of
all firms. Thus, SUM can range (for a 100-firm economy) from -100, connoting
total slump, to + 100, connoting exuberant boom. SUM = 0 can result either
from all firms uniformly investing at the normal level or from above-normal
outcomes offsetting below-normal outcomes. In the present implementation of
the system, the authority does not differentiate between these cases (which surely
matter in a real economy) nor are there real output effects from persistent over- or
under-investment.
4. SUM is plotted to the right of the characteristic diagram. The dashed line
indicates zero net aggregate investment.
5. The monetary control MON can be set at ( -1, 0, + 1) according to
experimental rules. The current value of MON is plotted as a dot, blank, or black
square, respectively, just to the left of the rule line at the right of the diagram.
Properties of the System and Experimental Protocols
MON is treated by each firm as the equivalent of a unit of expectational informa-

tion received from firms in its neighborhood. Thus, if a firm would expand
investment if three of its neighbors signaled positively in the preceeding period, it
will now expand investment if MON = 1 and two firms signaled positively.
In the first group of experiments, an "accommodative" rule is applied in
period 21: if SUM > 0, SUM = 0, SUM < 0, then MON = -1, 0, + 1,
respectively. The rule is inverted in period 41 to become "countercyclical." In a
second grouping of experiments, MON = 0 only if SUM deviates from 0 by more
than 5 in absolute value. In the third and fourth groups of experiments, the
countercyclical rule is applied first with different degrees of fineness. 2 In a final
group of experiments, interventions are initiated in period 1 for models whose
inherent dynamics are of type 1 or 4.
Figure 2. Qualitative Effects of "Monetary" Interventions Working Through Cellular Automata Expectational Models.
~
-
CRITERIA 1 3 4
-:~-
~~: :!\: ,~: ]~ ~
}.... ~.... ~:;.E
...---.&4
--~ 1: {
a
CRITERIA 2 3 4 CRITERIA
~...._ _.r..:~_.. I "ra · . .•.. ·:· tL:-'-~~ : ;~:~~ -:
~
~~-&E Jl :}~~~~<~+-~1;.:I~
jf ri...·::.. ._ :: ~ ... .
--.
--.
--.
l ~i~~~~:~r~r.. f : --.
--.
- .
b c
183
CRITERIA 2 2 2 CRITERIA 2 4 5 ~
- ~. " "-.. -
· ~~TfliY~1Xi~i~1~:H~~~ L ?i~&,~:~!g~~jf~ c
h
Notes: The plot to the right of each panel gives the time path of ' 'aggregate investment' ' with the ''target' ' or ''O' ' level representing the volume of investment
consonant with a Solow growth path. In panels 2a-2g, the system develops without intervention for 20 periods. In 2a-2d, a ' 'countercyclical'' policy is instituted
in period 21 and an "accommodative" policy in period 41. In 2e-2g, the sequence of regimes is reversed (but for models in which the firm's own state is not an
argument of its expectation function). In both regimes, an intervention is triggered if actual investment differs from target investment by more than 4 percent.
In 2a, an initially chaotic regime is "centered" on target by the intervention but at a cost of wider fluctuations. In 2b, a periodic economy becomes aperiodic
with wider fluctuations, but again, improved centering. In 2c, fluctuations widen, centering deteriorates, and the economy becomes pronouncedly cyclical. 2d
comes close to being a success for "stabilization": a chaotic system increases measured entropy, corrects a tendency towards depressed output, centers on target
but with a relatively high fluctuation range. In each of2a-2d, the later shift to "accommodative" policy results in greater homogeneity and often a shift to or back
to aperiodicity. Cases 2e- 2g also show dramatic shifts in qualitative type, e.g., type 2 to type 3 in case 2e and shifts between different periodic subtypes in 2f.
Panels 2h and 2i show what happens when a "countercyclical" regime is instituted immediately after the initial shock-leaving no time for the standard
behavior to develop. In 2h, a rule which ordinarily results in type 1 equilibrium leads to complex persistent forms which mix features of other models. In 2i, a type
4 system displays a number ofunusual forms not otherwise observed. It appears that the imposition of a global control can lead to any of the 12 possible transitions
between Chomsky-Wolfram types and to many subtypes as well.
Results and Preliminary Interpretations
At this pilot stage of the research, visual classification of model dynamics and
verbal descriptions should suffice. We see from inspection of the pattern changes
in Figure 2 instances in which the monetary intervention leads to increases in
dynamic complexity (e.g., from limit cycles to chaos [2e]) and decreases in
complexity (e.g., from chaos to simple limit cycles [2t]). In fact, all possible
transitions between the four complexity types have been generated by the imposi-
tion of a stabilization rule or its inversion. Careful inspection also suggests that
interventions can also lead to higher or lower "complication" within a dynamic
complexity classification (e.g., shortening or lengthening of cycle periods, more
or less chaotic disorder). This casual observation is supported by calculations of
system entropies but further experimentation is needed on this point. One tenta-
tive conclusion is offered, however. On the record of several hundred experi-
ments covering a wide variety of firm decision rules and authority stabilization
protocols, interventions are highly likely to affect the qualitative dynamic proper-
ties of the system. Such effects change the statistical properties of the system,
often in ways that would encumber firm or industry forecasters or other policy
makers.
Incomplete Stabilization
But what of conventional stabilization? The moment of intervention is easily

detected in the time-series graph of SUM to the right of each characteristic
diagram. A typical sort of result is that illustrated in 2a and 2b where the
economy is better centered but at the tradeoff cost of wider fluctuations in the
aggregate. In many other instances, the tradeoff cost of wide fluctuations within
the aggregate is encountered. The outcome in 2c, worse centering and dramati-
cally wider oscillation, occurs infrequently but is a thought-provoking illustration
of the dangers in fine tuning. A few of these dangers are exemplified in 2h and 2i,
which illustrate systems that are stable in isolation but not if "stabilized" from
outside. Actually, the result in 2d, escape from a low fluctuation "recession" to a
fluctuating centered economy, is as near to a stabilization success as any ob-
served. It is a rarity in economies whose basic dynamics are of type 3. Accommo-
dative policy (Figures 2e, 2f, 2g) tends to drive the system off center, as one
might expect; but it also tends to eliminate fluctuations within the aggregate, thus
making a later countercylical intervention more effective.
Economic Implications
Results at this point must be judged as preliminary and conjectural; nonetheless, a

few themes emerge from the analysis.
1. A monetary instrument based on a macroeconomic criterion, when mixed
186 PETER S. ALBIN
with micro or local ''leading indicators,'' can alter the qualitative dynamics of an
expectations-driven economy.
2. The alterations may increase or decrease the complexity and complication
of system dynamics. System behavior appears to depend more on the way firms
use expectational information than on the procyclical or countercylical intent of
the intervention.
3. Preliminary procyclical interventions may be needed to stabilize a system;
countercylical interventions may increase aggregate or within-system disorder.
Interventions may also work in the manner usually ascribed to them-but do so
rarely. What happens depends primarily on firm-level expectational procedures.
4. The coarseness or fineness of filter rules for intervention matters-usually,
but not necessarily in the manner expected.
In brief: monetary intervention does alter the dynamics of systems whose
dynamics are already rich, but it does not do so in a way that suggests definitive
rules for policy. Although the models here abstract from realism to a considerable
degree, the results support the view that the expectational effects of monetary
interventions are uncertain and potentially perverse.
Notes
1. The underlying mathematical structure of the model is that of a one-dimensional
cellular automation. Wolfram (1986) is the primary source of foundation papers and
detailed bibliography concerning this form. Also see Albin (1975) for a general discussion
of cellular automata models in economics.
2. These rules are quite naive. In future work, I will experiment with higher-order
autoregressive schemes and built-in artificial intelligence geared to devising the best
adaptive rule.
References
Albin, P. S. 1975. Analysis of Complex Socio-Economic Systems. Lexington, Mass.:
Lexington Books.
Albin, P. S. 1987. "Microeconomic Foundations of Cyclical Irregularities or Chaos."
Mathematical Social Sciences 13: 185-214.
Chomsky, N. 1959. "On Certain Formal Properties of Grammars." Information and
Control 2.
Chomsky, N. 1963. "Formal Properties of Grammars." Handbook of Mathematical
Psychology 2. Pp. 323-418. New York: John Wiley and Sons.
Day, R. H. June 1982. "Irregular Growth Cycles." American Economic Review 72
(June): 406-14.
Farmer, D., T. Toffoli, and S. Wolfram, eds. 1984. Cellular Automata. Amsterdam:
North Holland.
Kuenne, R. E. 1979. "Rivalrous Consonance and the Power Structure of OPEC," Kyklos
32:695-717.
Li, T., and J. A. Yorke. 1975. "Period Three Implies Chaos," American Mathematical
Monthly 82 (December): 985-92.
Mandelbrot, B. 1982. The Fractal Geometry of Nature. New York: W. H. Freeman and
Co.
Martin, 0., A. M. Odlyzko, and S. Wolfram. 1984. "Algebraic Properties of Cellular
Automata." Communications in Mathematical Physics 93:219-58. Reprinted in Wol-

fram, 1986.
Smith, A. R. 1971. "Cellular-Automata Complexity Tradeoffs." Information and Con-
trol 18:466.
Solow, R. M. 1956. "A Contribution to the Theory of Economic Growth." Quarterly
Journal of Economics 70:65-94. Reprinted in Sen, A. ed. 1970. Growth Economics,
pp. 161-92. Harmondsworth, U.K.: Penguin Books.
Wolfram, S. 1983a. "Statistical Mechanics of Cellular Automata." Review of Modern
Physics 55. Reprinted in Wolfram 1986.
- - - . 1983b. "Universality and Complexity in Cellular Automata." Physica D.
Reprinted in Wolfram 1986.
- - - , ed. 1986. Theory and Applications of Cellular Automata. Singapore: World
Scientific.
Debt Commitments and Aggregate Demand:
ACritique of the
Neoclassical Synthesis and Policy
Steven Fazzari and
John Caskey
In the General Theory of Employment, Interest, and Money (1936), Keynes

analyzed the persistent, massive waste of productive potential so evident during
the Great Depression. Keynes's theory questioned the viability of laizzez-faire
capitalism. The "invisible hand" of self-interested market activity, that seemed
to coordinate economic behavior effectively at the micro level, did not guarantee
the full utilization of productive resources in the system as a whole.
In the postwar period, economic theorists responded to the Keynesian chal-
lenge by presenting a view of aggregate economic activity that integrated the
seemingly revolutionary results of The General Theory into neoclassical general
equilibrium analysis. This "neoclassical synthesis," in which Keynesian results
are interpreted as a special case of neoclassical orthodoxy, came to dominate
mainstream academic thinking, especially in the U.S. In this view, the concept
that market economies can be fundamentally flawed was transformed into the
notion that "real world" institutions create rigidities in the adjustment mecha-
nisms of market economies. These rigidities may cause the natural, endogenous
stabilizers of the system to act slowly when the general equilibrium is disturbed.
The economy could have Keynesian characteristics in the short-run, but the
system ultimately returns to its full employment general equilibrium state.
The neoclassical synthesis is based on a simple model of the institutional
structure of modern capitalism. Financing arrangements are not specifically
considered. The feedback from the financial side of the economy to the real side
works only through the interest rate that equates the supply and demand for a
188
CRITIQUE OF NEOCLASSICAL SYNTHESIS & POLICY 189
narrowly-defined money stock. These assumptions keep the model clear and
simple. But do they allow enough detail to make the neoclassical synthesis
relevant to the analysis of modern capitalism?
This is the question addressed in this chapter. The analysis focuses on whether
the stability results of the neoclassical synthesis hold up in a model of the market
economy that specifically recognizes the effects of debt contracts on aggregate
demand. The conclusions show that in spite of its widespread acceptance, the
neoclassical synthesis provides a fragile theoretical basis for the analysis of
modern market economies. Its central result, the endogenous stability of the
system, is called into question when the institutional details of financial structure
are considered in more depth. The ideas presented here also lead to a critique of
many current monetary policy views.
The Stabilizers of the Neoclassical Synthesis
The theory behind the neoclassical synthesis predicts that the economy will
eventually reach full utilization of productive resources through two disequilibri-
um adjustment processes. First, a supply-side process assures that when unem-
ployed resources exist, prices will fall. Second, a demand-side process translates
deflation into higher aggregate demand that pushes the system back to full
capacity. Both of these adjustment processes are essential for the theoretical
predictions of the neoclassical synthesis to hold. Each of them are briefly consid-
ered.
Suppose a resource, labor for example, is in excess supply. Its nominal factor
price, the money wage, falls in a competitive market. When wages fall, marginal
cost falls below marginal revenue, output supply expands, and the price level of
goods falls. The essential result from the supply side is that under-utilization of
resources generates deflationary pressure.
In Keynesian theory, however, a contraction in aggregate demand initially
causes unemployment. According to the neoclassical synthesis, the deflation
caused by excess supply stimulates aggregate demand through two channels.
First, there is the "Keynes effect." Lower prices increase the real value of
outside money. The increase in the real supply of liquidity reduces the premium
agents are willing to pay to hold money, and the interest rate falls. Lower interest
rates stimulate additional expenditure, and aggregate demand rises. The ''liquid-
ity trap" causes potential problems here because increases in the real outside
money stock may lead to negligible declines in the interest rate. But this can be
overcome by the "Pigou effect." With higher real wealth caused by deflation,
households directly increase their expenditure. The stability of the system need
not depend on interest rate reductions alone.
The adjustment process is summarized graphically in Figure 1. After aggre-
gate demand shifts from D0 to Di. excess supply in factor markets pushes factor
prices down. This increases aggregate supply for a given price level of output,
190 STEVEN FAZZARI & JOHN CASKEY
Figure 1
Price
Level S; = S,
S*
o,
01
v,-v· Output
causing the aggregate supply curve to shift outward, and it creates downward
pressure on the price level. Due to the Keynes and Pigou effects, falling prices
increase aggregate demand, so expenditure rises to reduce the excess supply. The
process continues as long as under-utilized resources exist, and the system con-
verges to full capacity, denoted by Y* in Figure 1. Thus, to the extent that prices
do not adjust instantaneously, the neoclassical synthesis model allows for short-
run output losses due to insufficient effective demand, but endogenous stabilizers
operate to restore full employment general equilibrium in the long-run. 1
The Macroeconomic Effects of Debt Financing: A

Critique of the Synthesis
While the neoclassical synthesis represents a widely held view in macroeconom-

ics, there have been criticisms of both the supply and demand sides. 2 In this
chapter, we join the critics by questioning the stabilizing effect of deflation on
aggregate demand. Our approach explicitly recognizes that debtors have payment
commitments with fixed nominal terms. An unanticipated deflation or reduction
in the inflation rate will reduce nominal cash flows relative to agents' previous
expectations. Thus, the "margin of safety" for debt payments declines. We argue
that this can reduce agents' expenditures, so when prices fall, aggregate demand
may also fall. If this is the case, the dynamic adjustment of an otherwise standard
neoclassical synthesis model can be very different from the stable process de-
scribed in the last section.
Debt contracts set up a stream of payment commitments through time. The
parties to contracts make these commitments with certain expectations in mind.
Both borrowers and lenders have a plan as to how cash flows will be generated to
meet the obligations. For example, if high inflation is expected, borrowers will be
more willing to commit to higher future cash payments because they expect the
high rate of inflation to increase their future cash flows. These expectations
become embodied in the debt contracts. They are implicit in the contract's terms,
and they cannot be changed until the contract expires or is abrogated.
What happens if events differ from the expectations held at the time payment
commitments were determined? Most of the literature has focused on distribu-
tional effects. Because actual prices differ from the price expectations embodied
in debt contracts, the real value of the payment commitments changes, and the
real distribution of wealth between debtors and creditors will differ from what
was anticipated in the contract. With a reduction of prices, the debtor loses just
what the creditor gains. For this reason, these distributional effects have general-
ly been assumed to cancel in the aggregate. 3
But more is happening than just a redistribution of weal th between debtors and
creditors. When the system receives a deflationary shock, the margin of safety
between debtors' cash flows and their cash commitments declines, and the prob-
ability of insolvency rises. Because insolvency is costly, debtors have an incentive
to avoid it by reducing discretionary expenditure. Even if debtors would choose
bankruptcy, their creditors will use all available means to force "austerity mea-
sures" on them. 4
We should not expect that debtors' expenditure reductions in this situation will
be offset by increased creditor expenditure. While lenders' real wealth increases
by an amount equal to debtors' losses from the deflation, the riskiness of that
wealth also increases due to the increased likelihood of bankruptcy. Lenders'
liquidity preference will increase in the face of increased risk. This will change
the relative price structure of assets. The greater demand for safe liquid assets,
government bonds for example, may drive down their yields, ceteris paribus, but
lower interest rates on these assets will not lead to increased expenditure. On the
other hand, as lenders shift away from the risky loans used to finance consumer
and business expenditure, interest rates on these loans will rise. This can signifi-
cantly curtail expenditure. 5
Unanticipated price declines or reductions in inflation can also depress expen-
diture due to a breakdown in the financial intermediation process. The literature
on credit rationing has shown that adverse selection problems can cause financial
intermediation to be curtailed when the riskiness of lending increases. If this
happens, as it did during the Great Depression in the U.S., it will also depress
demand. 6
These effects all point in the same direction. A decline in prices or the rate of
inflation relative to previously anticipated levels need not stimulate aggregate
demand. The net effect on aggregate demand depends on whether our cash flow
effect dominates the Keynes and Pigou effects emphasized by the neoclassical
synthesis. There is no reason to assume a priori that the aggregate demand curve
is downward sloping. 7
If the demand depressing effects of falling prices dominate the expansionary
tendencies caused by the Keynes and Pigou effects, macroeconomic analysis must
be carried on with an upward sloping aggregate demand curve. This wreaks
Figure 2
Price
Level
havoc with the neoclassical synthesis stabilizers discussed in the previous section.
Suppose we accept the standard microfoundations of the supply-side and assume
that unemployment causes factor prices to fall after demand declines. As shown
in Figure 2, this outward shift of the aggregate supply curve with an upward
sloping aggregate demand curve makes unemployment worse, as price reductions
depress demand. Also, the greater the deflationary pressure, that is, the faster
money wages fall in response to unemployment, the worse will be the result.
From an initial situation of depressed demand, falling wages and prices aggravate
the situation due to the existence of nominal debt contracts. This directly contra-
dicts the neoclassical synthesis view that the real effects of changes in aggregate
demand can be attributed to sticky or inflexible money wages.
When aggregate demand includes the cash flow effect based on nominal debt
commitments, the analysis of macroeconomic adjustments with a fixed aggregate
demand curve, as in Figure 1, is no longer adequate. For changes in the price
level will also change nominal cash payment commitments as agents recontract
debts with new expectations. After a negative shock that lowers the price level,
new debt contracts will be based on lower price expectations. Thus, debtors and
creditors will be able to tolerate a lower price level without reducing expenditure.
This shifts the aggregate demand curve outward. Holding the supply curve
constant, outward shifts in the demand curve push the system in the direction of
full employment.
On net, the direction output moves when there is unemployment depends on
the combination of the supply-side and demand-side adjustment processes. When
the demand curve slopes upward and prices fall, downward wage adjustments
reduce output while downward revisions in the price expectations embodied in
debt contracts stimulate output. The net result depends on which of these move-
ments dominates. Figure 3a and Figure 3b present some of the possibilities after a
negative demand shock pushes output below its full employment equilibrium
level to Y 1 • Figure 3a shows a case in which the demand shift dominates the
Figure 3a
Price
Level
Y* Output
Figure 3b
Price
Level
Y* Output
supply shift to push output back toward its full utilization level (Y*) in period 2.
In figure 3b, the supply shift dominates and output falls further from Y* in
period 2.
In Caskey and Fazzari (1987), the dynamic paths that can result from this
kind of model are analyzed. The adjustment paths can be cyclical and they
may be asymptotically unstable. If the demand curve is upward sloping, great-
er downward flexibility of wages leads to less stable macroeconomic dy-
namics. Again, these results directly contradict the neoclassical conven-
tional wisdom that attributes the real effects of demand movements to ''sticky''
wages.
Implications of the Alternative View
The analysis in the previous section shows that deflationary pressure resulting
from slack in the macroeconomy is not necessarily stabilizing. When compared to
the neoclassical synthesis, this conclusion leads to significant differences in the
view of how a market economy functions at the aggregate level. It also provides
new insights into the linkages between macro performance, economic institu-
tions, and economic policy. Some of these issues are considered in this section.
At the theoretical level, the most striking implication of our results is the
possibility that the natural adjustment processes in a market economy may not be
sufficiently strong to overcome destabilizing forces, even in the long-run, typical-
ly the domain of neoclassical general equilibrium results. The fact that our
approach allows the possibility of unstable adjustments does not predict the
ultimate explosion or complete collapse of market systems. Rather, it suggests
that we must search beyond the price and quantity adjustment processes empha-
sized in the neoclassical synthesis to find a complete explanation for the general
stability exhibited by postwar market economies. It is likely that other aspects of
the economy not considered in our approach, or in the mainstream theory, limit
the potentially unstable dynamics that we have identified. For example, govern-
ment interventions, planned or accidental, may have provided barriers that have
contained unstable paths.
Of course, another possible explanation for the general stability of market
economies over the last several decades is that the empirical parameters of the
system are such that the stable case prevails, and the long-run implications of the
neoclassical synthesis remain relevant. After all, we do not argue that the aggre-
gate demand curve necessarily slopes upward, but only point out that the unstable
case cannot be ruled out theoretically. But simply claiming that ex post results rule
out instability on empirical grounds is not a satisfactory response to the issues
raised in this chapter. The relative importance of the cash flow effect is not
constant through time but will vary with changes in financial conditions and
institutions. Given the secular trend toward increasing private debt burdens in the
U.S., for example, one can make a strong case for the proposition that the
monetary authorities should design policy and institutional structures to contain
endogenous instability even though the recent behavior of the system may appear
generally acceptable. 8
These results emphasize the importance of the link between financial institu-
tions and macroeconomic performance. This crucial relationship is prominent in
the work of Keynes (1936) as well as Davidson (1972) and Minsky (1975), for
example. Yet, it has not received much attention within the mainstream theories
of the neoclassical synthesis. As discussed in the last section, the adjustment path
of the system when it is out of long-run equilibrium depends on how quickly cash
payment commitments can be brought into line with changing cash flows. This
speed of adjustment will depend on the term structure of outstanding debt. The
dynamic properties of the adjustment will also depend on the level of indebted-
ness in the economy as well as the nature of the payment obligations set up in debt
contracts. As these institutional features of the economy evolve, the dynamic
properties of the aggregate market system may change in significant ways.
Macroeconomic theory cannot be separated from the study of changes in financial
practices and institutions. 9
This conclusion is particularly relevant for policy analysis. Much of the litera-
ture in macroeconomics proposes universal policy rules, applicable to all market
economies at all times. The results of our alternative analysis suggest that this
quest is destined to fail. The macroeconomic impact of policy changes will
depend on the particular institutional environment in which they occur. In a
system with low levels of indebtedness and simple financial relationships, a tight
money policy may be effective against inflation and may have low costs in terms
of foregone real output. In an economy with extensive and complex debt con-
tracts, however, the same policy can cause persistent changes in output and can
lead to instability.
Let us consider this point in more detail. In the previous section, we demon-
strated that the net effect of falling prices on aggregate demand is a key determi-
nant of the economy's dynamic behavior. The conventional wisdom of the neo-
classical synthesis generally holds when the demand curve is downward sloping,
but an upward sloping curve leads to a system in which the deflationary pressures
generated by unemployment are destabilizing. Now suppose that firms and
households have sufficient financial resources so that a small decline in prices or
the inflation rate increases aggregate demand through the conventional channels.
If prices fall farther from their previously anticipated levels and financial prob-
lems become more severe, however, the destabilizing cash flow effect dominates
the stabilizing Keynes and Pigou effects and demand begins to fall as prices fall.
A demand curve that has these characteristics will be nonlinear; an example is
graphed as D0 in Figure 4a.
This kind of demand curve can cause the system to exhibit ''corridor effects. ''
When small disturbances occur, the endogenous response of the system will be
stabilizing, but large disturbances push the economy outside its stable corridor
and unstable dynamics result. Consider the effect of a small contraction in
demand represented by the backward shift of the demand curve from 0 0 to D 1 in
Figure 4a. The new equilibrium point lies on the downward sloping portion of the
demand curve. Therefore, the deflationary pressure caused by the downward
shift of supply pushes output toward its full employment level as Y2 moves
toward Y*. In Figure 4b, however, the same system receives a larger initial
Figure 4a
Price
Level
D~ s~"' s,
~I
I : Y*
y,-Y2 Output
Figure 4b
Price
Level
D~ s~"' s,
Y 2 -y, Y* Output
demand shock. The equilibrium after the shock now lies on the upward sloping
portion of the demand curve, and deflation will push Y2 further away from full
employment. 10
This nonlinear analysis provides a basis for "gradualism" in policy imple-
mentation. If the monetary authorities tighten money growth to fight inflation,
for example, then the policy must be implemented slowly. If the system receives
too large a shock, so that the expectations embodied in payment commitments
differ significantly from actual circumstances, the stabilizing adjustment mecha-
nisms in the system may break down. This also suggests that policies designed to
contain instability in the financial sector, such as lender of last resort policies,
may be very important when the system receives large shocks.
This result fundamentally differs from the "new classical" macroeconomics
view of monetary policy. In the neoclassical synthesis model, changes in mone-
tary policy only affect nominal variables in the long-run steady-state; real output
converges to the full capacity level determined by tastes and technology. When
one incorporates rational expectations into this model, agents are assumed to fully
understand the long-run consequences of changing monetary policy. Therefore,
they instantaneously arbitrage away any differences between current nominal
values and the steady-state levels as soon as information about monetary changes
becomes available.
In our model, this kind of process cannot occur. Price expectations established
prior to a monetary change are more than subjective forecasts that can be costless-
ly abandoned when new information emerges. The expectations become em-
bodied in objective nominal debt commitments. The legacy of past expectations
cannot be erased immediately when a policy change occurs, even ifthe monetary
authorities are fully credible. 11 The real expenditure effects caused by squeezing
nominal cash flow relative to payment commitments implies that the nominal
level of economic variables can have an important impact on the adjustment
dynamics of the system. Thus, whether expectations are "rational" or not,
monetary policy must not interfere with agents' ability to validate debt contracts.
Otherwise, our analysis shows that real instability can result.
* * *
Ever since the publication of The General Theory, mainstream macroeconom-
ics has attempted to overturn Keynes's conclusion that endogenous market
mechanisms do not necessarily lead to a stable, full employment equilibrium. The
neoclassical synthesis emerged out of this effort. But the synthesis rests on some
fragile theoretical propositions. A reasonable and seemingly small change in one
of these propositions can cause a large change in the implications of the theory.
In this chapter, we have analyzed the consequences of changing one of the
underlying assumptions of the neoclassical synthesis model: the proposition that
deflation will stimulate aggregate demand. When payment commitments in
nominal terms are extensive, deflation has a negative cash flow effect on aggre-
gate demand because the reduction of nominal cash flows relative to nominal
payment obligations increases the likelihood of insolvency. We have indicated
how this can reduce the expenditure of both debtors and creditors. If this effect is
large enough, deflation causes aggregate demand to fall, and the market response
to the system may be unstable. The faster that wages and prices fall, the more
severe an economic contraction can be. This overturns a central result of the
neoclassical synthesis that attributes the real costs of aggregate demand contrac-
tions to the failure of these nominal variables to adjust quickly enough.
It is important to recognize that our results do not lead to the prediction that
market economies will be unstable at all times. We claim only that the conver-
gence to general, full employment equilibrium cannot be taken for granted. A
market economy can behave in qualitatively different ways based on the institu-
tional settings. The neoclassical synthesis is not sufficiently detailed to provide a
basis for macroeconomic analysis in contemporary, financially complex econo-

mies. We need a theoretical framework that explicitly recognizes the links be-
tween financial structure and macroeconomic performance. This chapter pro-
vides a step in that direction.
Notes
1. Of course, there is considerable latitude within the broad bounds of the neoclassi-
cal synthesis for different policy perspectives. Some neo-Keynesians assume that the
endogenous adjustment of wages and prices is sufficiently sluggish that the convergence to
full employment can be quite protracted. Government demand management policies are
required to make the system perform acceptably. Monetarists generally believe that the
lags involved in the adjustment process are so complex and variable that stabilization
policy will most likely fail. Finally, the "new classical" macroeconomists argue that the
endogenous stabilizers work very fast due to •'rational'' expectations, and that macroeco-
nomic policy has no systematic effect on real variables.
2. The idea that deflation would stimulate aggregate demand contradicts Keynes's
conclusion; see chapter 19 of The General Theory. Similar views are also expressed by
Fisher (1933), Davidson (1972), Minsky (1975) and Tobin (1975). A critique of the
supply-side of the neoclassical synthesis can be found in Fazzari (1986).
3. A notable exception can be found in Tobin (1975) where the distributional effects
of a deflation affect aggregate consumption because debtors and creditors are assumed to
have different propensities to consume.
4. Mishkin (1976, 1978) presents empirical evidence that financial conditions affect
consumption behavior. Fazzari and Mott ( 1986) and Fazzari and Athey ( 1987) find signifi-
cant empirical effects of financial variables on firms' investment spending.
5. This analysis becomes more complicated when existing debt contracts carry vari-
able rate financing terms. Then interest rate changes affect outstanding as well as prospec-
tive payment commitments. This issue is considered further in Caskey and Fazzari ( 1986).
6. See Stiglitz and Weiss (1981); Bernanke (1981); Greenwald, Stiglitz, and Weiss
(1984); and Calomiris and Hubbard (1985).
7. This issue is discussed more fully and modeled formally in Caskey and Fazzari
(1987). Similar points are made by DeLong and Summers (1986). In their model, antici-
pated future deflation reduces aggregate demand by increasing current real interest rates.
8. The chapter by Wolfson in this volume and Wolfson (1986) provide empirical
support for increasing secular debt burdens and declining margins of safety for debt.
Similar evidence from a cyclical perspective is presented in the chapter by Niggle in this
volume. In the framework considered here, these observations increase the likelihood of
instability and make the need for containing policy intervention more critical.
9. These issues are analyzed in greater depth in Caskey and Fazzari ( 1986).
10. Of course, a full dynamic analysis of this kind of model can lead to many different
kinds of qualitative system behavior, as the chapters in the first part of this volume
demonstrate.
11. In this sense, our nominal debt contracts play a role similar to nominal wage
contracts in models such as Fischer (1977).
References
Bernanke, Ben. 1981. "Bankruptcy, Liquidity and Recession." American Economic Re-
view 71(2).
Calomiris, Charles, and R. Glenn Hubbard. 1985. "Price Flexibility, Credit Rationing
and Economic Fluctuations: Evidence from the U.S., 1879-1914," NBER Working
Paper No. 1767.
Caskey, John, and Steven Fazzari. 1986. "Macroeconomics and Credit Markets." Jour-
nal of Economic Issues 20(2).
- - - . 1987. "Monetary Contractions with Nominal Debt Commitments: Is Wage
Flexibility Stabilizing?" Economic Inquiry, 25(4): 583-97.
Davidson, Paul. 1972. Money and the Real World. New York: John Wiley and Sons.
DeLong, J. Bradford, and Lawrence Summers. 1986. "Is Increased Price Flexibility
Stabilizing?" American Economic Review 26(5): 1031-44.
Fazzari, Steven. 1986. "Sales Expectations and Output Constrained Firms: A New View
of the Microfoundations of Aggregate Supply." Washington University, St. Louis.
Mimeo.
Fazzari, Steven, and Michael Athey. 1987. "Asymmetric Information, Financing Con-
straints, and Investment." Review of Economics and Statistics, 69(3): 481-87.
Fazzari, Steven, and Tracy Mott. 1986. ''The Investment Theories of Kalecki and Keynes:
An Empirical Study of Firm Data 1970-1982." Journal of Post Keynesian Economics
9(2): 171-87.
Fisher, Irving. 1933. "The Debt Deflation Theory of Great Depressions." Econometrica
1: 337-57.
Fisher, Stanley. 1977. "Long-Term Contracts, Rational Expectations, and the Optimal
Money Supply Rule." Journal of Political Economy 85(1): 191-205.
Greenwald, Bruce, Joseph Stiglitz, and Andrew Weiss. 1984. "Informational Imperfec-
tions in the Capital Market and Macro-Economic Fluctuations." American Economic
Review 74(2).
Keynes, John M. 1936. The General Theory of Employment, Interest, and Money. Lon-
don: Harcourt, Brace and World.
Mishkin, Fredric. 1976. "Illiquidity, Consumer Durable Expenditure, and Monetary
Policy." American Economic Review 64(4): 642-53.
- - - . 1978. "The Household Balance Sheet and the Great Depression." Journal of
Economic History 38(4): 918-37.
Stiglitz, Joseph, and Andrew Weiss. 1981. "Credit Rationing in Markets with Imperfect
Information." American Economic Review 71.
Tobin, James. 1975. "Keynesian Models of Recession and Depression." American Eco-
nomic Review 65(2).
Wolfson, Martin. 1986. Financial Crises: Understanding the Postwar U.S. Experience.
Armonk, New York: M.E. Sharpe, Inc.
Ill
Empirical Evidence on Debt
and Financial Instability
The Cyclical Behavior of
Corporate Financial Ratios and
Minsky's Financial Instability Hypothesis
Christopher J. Niggle
This chapter presents an empirical evaluation of several aspects of Hyman

Minsky's theory of systemic financial instability and fragility. Minsky argues that
mature capitalist economies are financially unstable in expansions: certain key
financial ratios change systematically as such economies grow. 1 Nonfinancial
firms' profit expectations and risk evaluations encourage them to increasingly
leverage their capital as expansions proceed and seek external sources of capital
to finance positions in real and financial assets. Financial intermediaries and
other lenders are willing to supply funds to these borrowers (allowing the non-
financial firms' leverage ratios to increase beyond levels previously considered
prudent) as their own expectations of profit and risk are revised; a collective
"euphoria" sets in and debt grows rapidly as a result. Simultaneously, the firms'
perceptions of diminished risk, coupled with the usual expansion-induced in-
crease in interest rates, encourage the nonfinancial corporations (NFCs) to re-
duce their holdings of liquid financial assets (''cash kickers'' in Minsky's evoca-
tive language). As a consequence, the financial balance sheets of firms in the
nonfinancial corporate sector deteriorate in expansions: debt/asset and borrow-
ing/investment (leverage) ratios rise, the short-term composition of their debt
increases, and they become increasingly nonliquid.
Minsky argues that in contractions these processes are reversed. Borrowing,
investment, and purchases of financial assets decline. Because borrowing de-
clines more rapidly than investment expenditures, leverage ratios also shrink,
Thanks to Martin Wolfson, Willi Semmler, Gary Evans, and Bob Pollin for their
comments on earlier versions of this paper; of course, all errors and omissions are my
own.
203
204 CHRISTOPHER J. NIGGLE
while perceptions of increasing risk lead firms to attempt to increase the liquidity
of their asset portfolios. In an economy in which relatively high and stable levels
of employment and profits are increasingly seen as normal (because of a long
period of relative prosperity, such as during the first three postwar decades) and
recessions as temporary aberrations, perceptions of risk decline from cycle to
cycle. Consequently, an upward ratchet effect in leverage and nonliquidity takes
place from cycle to cycle. This deterioration in the balance sheets of the NFCs has
implications for both the stability of the economy and countercyclical policy. The
economy may be financially fragile and deflationary macroeconomic policies
consequently dangerous. It faces the threat of debt deflation (a collapse of asset
values) which may precipitate a severe depression. 2
In order to determine the extent to which corporations' behavior coincides
with that postulated in Minsky's model, the cyclical behavior of the sources and
uses of funds for the U.S. nonfinancial corporate sector over six complete
business cycles (1953 to 1982) was studied, using the National Bureau of Eco-
nomic Research's Reference Cycle Program. 3 This program, based upon the
business cycle analysis begun by Wesley C. Mitchell and continued by Arthur F.
Burns and others at the NBER, is known as the Standard Business Cycle Analysis
of Time Series. The program translates a time series into reference cycle units
which normalize the values of a series with respect to its average value over the
cycle (thereby correcting for intracycle trend), defines turning points for the time
series, calculates the amplitude of the rise and fall of the series, and computes an
"average cycle" on the basis of the cycles analyzed. The business cycle is
divided into expansion and contraction phases. They include nine stages: trough;
early, mid, and late expansion; peak; early, mid, and late contraction; trough. 4
Nonfinancial Corporations' Sources

and Uses of Funds and
Federal Reserve Board Accounting Concepts
In order to clarify some aspects of the analysis of the NFCs' financial behavior
presented below, this section will review the sources and uses of funds for
nonfinancial firms. It will also discuss the relationship between those abstract
concepts and the somewhat different accounting techniques underlying the statis-
tics found in the Federal Reserve Board's Flow of Funds Accounts (FOFA),
which furnish the data base for this study. 5 A corporation (and the entire NFC
sector) can run a deficit or surplus on capital account (or match capital expendi-
tures with internal funds). The internal funds available for capital expenditures
come from two source: retained earnings (profits net of taxes and dividends) and
depreciation allowances. Capital expenditures include expenditures on plant and
equipment, construction of residential units, and inventory investment. The NFC
sector usually runs deficits on capital account, but occasionally runs surpluses
during contractions as investment expenditures fall below the level of internal
CYCLICAL BEHAVIOR OF CORPORATE FINANCIAL RATIOS 205
funds (which usually increase throughout contractions due to the strong upward
trend in the stock of capital and depreciation allowances).
The deficit (D) or surplus (S) on capital account is defined as the difference
between internal funds available for investment (N) and actual investment (I):
(1) D(S) =I - N
These variables correspond to the Federal Reserve Board FOFA statistics "Fi-
nance Gap," "Capital Expenditures," and "Internal Funds." Expenditures on
capital account are the sum of fixed investment (F) and inventory investment (V):
(2) l=F+V
Internal funds (N) are retained profits net of corporate income tax liabilities (P)
and depreciation allowances (A):
(3) N=P+A
Internal funds are reported by the FRB's Flow of Funds Division at both book
value and with an inventory valuation adjustment (IVA) which adjusts reported
book profits for the effect of price changes. Price inflation raises reported profits
as the market value of firms' inventories increase; internal funds with the IVA are
lower than the book value of internal funds during inflationary periods. Although
this adjusted value of internal funds would be the most appropriate statistic for
many purposes, the actual book value of internal funds is most relevant for
purposes of considering the ability of the NFCs to internally finance the nominal
levels of investment expenditures which they undertake to finance. The deficit
concept most appropriate for our purposes is the difference between the nominal
value of investment and the book value of internal funds (see Poll in, [1986] for an
interesting discussion of the importance and implications of this issue).
The NFCs' uses of funds on capital account as defined by the FOFA are not,
however, restricted to purchases of real assets. Corporations also use funds to
acquire financial assets (FA); in the FOFA, these financial assets are divided into
liquid financial assets (LFA) which includes cash and other short term assets,
trade credit (TC), consumer credit (CC), and miscellaneous assets (MA) which
consists primarily of U.S. NF Cs' direct foreign investment (all of these statistics
are net). The identity for net acquisitions of financial assets (NFA) is:
(4) NFA = LFA + TC + CC + MA
The total net uses of funds for capital expenditures (TNU) by the NFCs then
includes their deficit on capital account (I - N) plus net acquisitions of financial
assets:
206 CHRISTOPHER 1. NIGGLE
(5) TNU = D + NFA
Total sources offends (TS) for the NFCs include external funds (X) and internal
funds (N):
(6) TS= X + N
The FOFA divides the NFCs external funds into net funds raised in credit
markets (NF) which are sales of debt (B) and equity (E) securities both net of
retirements, trade debt (TD), and foreign direct investment in the U.S. corporate
sector (Fl).
(7) NF=B+E
(8) X = NF + TD + FI = B + E + TD + FI
Note that both components of net funds, credit market borrowing (B) and equity
issues (E), are net of retirements of the NFCs' own issues; borrowing (B) is
equivalent to "net issue of debt instruments" in the FOFA tables and consists of
all borrowing by corporations which involves the issue of some kind of a debt
security for which a secondary market exists (bonds, mortgages, commercial
paper, bank loans, acceptances, finance company loans and government loans).
Another potential source of funds for capital expenditures is the liquidation of
financial assets in the NFCs' portfolios (especially money and other liquid assets),
and including this in a model of the NFC sector's sources and uses of funds has
much appeal. (See Earley, Parsons, and Thompson [1976] for a discussion of this
approach and its application to an analysis of expenditures by economic units.)
This study utilizes an alternative approach in order to be consistent with the Fed's
FOFA approach and data. Net acquisitions of financial assets (purchases net of
sales) is treated as a use of funds-net of the source of funds (for some firms in the
sector) of disinvestment in their financial assets. Total uses of funds on capital
account (TU) must equal total sources of funds; this identity may be written:
(9) TU= TS
(10) I + NFA = X + N
(11) I + LFA + TC + CC + MA = B + E + TD + FI + N
In order to conform to the FOFA approach which defines a deficit on capital

account net of internal funds, total net uses of funds (TNU) is defined and is
equivalent to Equation (5):
(12) TNU = D + NFA
TNU must be financed externally. Since total net uses must equal total external
funds:
(13) TNU = D + NFA = X
Although these sources and uses of funds as defined are equal by definition, in the
real world, the FRB FOFA estimates of the NFCs' sources and uses of funds
result in large statistical discrepancies. ' 'Accounted for sources of funds'' almost
always exceeds "the accounted for uses of funds." The FRB FOFA statisticians
consider the raw data from which they estimate the sources of funds statistics as
more reliable than the data underlying the uses estimates. The "sources" data is
taken primarily from IRS and SEC reports, while some of the "uses" data is
culled from sources such as reports in specialist journals and business publica-
tions. These discrepancies can be interpreted plausibly as reflecting unaccounted
for uses of funds-outflows from the NFC sector-which may have some inter-
esting analytic significance which will be discussed below. The basic identity for
sources and uses of funds which conforms with FOFA techniques but explicitly
includes this statistical discrepancy (SD) and interprets TNU as total reported
uses then becomes:
(14) TNU + SD =D + NFA + SD =X

The Cyclical Behavior of the NFCs'
Sources and Uses of Funds and
Minsky's Financial Instability Hypothesis
Minsky's financial instability hypothesis may be interpreted as a series of propo-

sitions regarding the behavior of certain corporate financial variables and ratios;
the hypothesis is specified here as the following cyclical behavior:
First, Minsky argues that upward revisions of profit expectations and other
factors cause investment expenditures (I) to increase more rapidly than internal
funds (N) in expansions: d(l)ldt > d(N)ldt.
Second, since perceptions of both borrowers' and lenders' risk are revised
downward in expansions, the use of external funds (X) and credit market borrow-
ing (B) both increase in expansions: d(X)ldt > 0 and d(B)ldt > 0.
Third, since firms borrow to purchase financial assets and existing real assets
as well as to finance capital formation, the ratio of borrowing to investment (a
"flow" measure of leverage) increases in expansions: d(BII)/dt > 0 where B/I
represents the flow measure of leverage. Note: ceteris paribus if B/I increases,
the traditional measures of leverage (ratios of total debt to total capital or total
assets, however represented) would increase.
Fourth, Minsky argues that firms reduce their holdings of liquid financial
assets (LFA) in expansions as their perceptions of risk diminish: d(LFA)ldt < 0.
Fifth, these processes should all reverse themselves in contractions, except for
borrowing which may increase through the early stages of the contraction because
of forced borrowing by firms which have contractual cash obligations which they
cannot meet out of their (declining) sales receipts. Fazzari and Caskey discuss
forced borrowing in this volume: d(B)ldt > 0 in early stages of contractions, but
d(B)ldt < 0 in late stages.
This specification appears consistent with the descriptions of the expected
cyclical behavior of these variables as found in many of Minsky's articles treating
financial instability, although Minksy does not formally model this aspect of his
theory; see Minsky (1974, 1980) for concise expositions. For comparison, see the
detailed discussions of Minsky's views on the behavior of financial variables
around the peak of the business cycle which are found in Wolfson (1986) and in
Taylor and O'Connell (1985, reprinted in this volume). Because of a lack of data
in the appropriate format, the cyclical behavior of the maturity composition of
corporate data was not studied.
Corporate Financial Variables and

Ratios in Expansions
Since it is the increase in debt and the resultant increase in leverage ratios which
Minsky identifies as the most important factor in undermining the financial
strength of the nonfinancial corporate sector and, ultimately, in determining the
degree of financial fragility of the economy, the behavior of those ratios is the
focus of this study. And since the stock measures of leverage (debt/capital,
debt/assets or debt/equity) move slowly, adjusting to changes in net borrowing
and purchases of assets sluggishly, the cyclical behavior of flow measures of
leverage are emphasized. Four such measures of leverage for the NFC sector
were studied: the ratios of both credit market borrowing (B) and total external
funds (X) to gross investment (I) and to fixed investment (F). All four measures
of leverage (X/I, X/F, B/I and B/F) behaved consistently with point three of the
specification of Minsky's hypothesis: all increased in the expansions, peaking in
the late expansion. Figure 1 illustrates the behavior of B/I which peaks in stage 4
of the average cycle. More complete versions of this study and the behavior of
other ratios can be found in Niggle (1984, 1986). Stock measures ofleverage fall
through the early expansion, then rise to peak late in the contraction. See Figure 2
for the behavior of one stock measure ofleverage, Credit Market Debt Outstand-
ing/Market Value of Corporate Equity (CMD/MVE).
Consider the individual behavior of investment expenditures, internal funds,
and the deficit on capital account. All measures of investment-plant and equip-
ment expenditures, fixed investment and capital expenditures-rise in the expan-
sion more than either measure of internal funds (book value or with the IVA), and
Figure 1. Leverage: Credit Market Borrowing/Capital Expenditures (Bii).
160
Cycle 1
100 1954.4-1958.4
40 160
100 Cycle 2
1958.4-1961.1
160 ------- ... 40
Cycle 3
100
1961.1-1970.1
.l!l
·c:
:::J
Cl)
0 40 160
>-
0
--------·
Cl)
0
c:
~
100 Cycle 4
'*
1970.1-1975.1
a:
160 40
Cycle 5
1975.1-1980.7
40 ---- 160
100 Average cycle
- - - - - - - - · - - - - - · - - - - - 40
II Ill IV V VI VII VIII IX
T p T
Cycle stages
Figure 2. Leverage: Outstanding Credit Market Debt/Market Yalue of Corporate

Equity (CMD/MVE).
160
100 ~
- ~
_...... - Cycle 1
1954.4-1958.4
40 160
r---...
- - 100 Cycle 2
1958.4-1961.1
160 40
100 ,_,,-
~ Cycle 3
~ 1961.1-1970.1
c:
::I
Q)
0
>.
/"-.. ...... 160
J
0
Q)
40
0
c:
~
~
-----------
100 Cycle 4
a: 1970.1-1975.1
160 40
100
40
....._
--- -- ....__
- 160
Cycle 5
1975.1-1980.7
-- 100
~
Average cycle
40
II Ill IV V VI VII VIII IX
T p T
Cycle stages
Table 1
"Average Cycle" Values for Trough-Peak and Peak-Trough Changes in

Financial Variables and Ratios (U.S. Nonfinancial Corporations, 1953-81)
Variable or Ratio
B x F N D NFA SD Bii B/F
Trough-Peak Increase (in Reference Cycle Units)
( +) 71.3 69.1 44.3 55.6 45.8 246.9 70.9 47.0 20.1 32.8
Peak-Trough Decrease (in Reference Cycle Units)
(-) 25.9 18.6 3.4 17.9 4.8 170.6 37.9 + 11.8 7.0 19.6
B: Net Credit Market Borrowing

X: External Funds (Credit Market Borrowing Plus Other
External Sources of Funds, Including Equity Issues)
F: Expenditures on Fixed Investment
I: Capital Expenditures (Fixed Investment Plus Inventory
Investment)
N: Internal Funds
D: Deficit on Capital Account (I - N)
NFA: Net Acquisitions of Financial Assets
SD: Statistical Discrepancy
Bil: Flow Measure of Leverage
B/F: Flow Measure of Leverage
Source: Federal Reserve Board Flow of Funds Accounts and Niggle (1984).
they also rise more rapidly than do internal funds (see Table 1 for this and other
data). The corporate sector begins most expansions with a surplus on capital
account which becomes a deficit as the expansions proceed. Capital expenditures
rise 55.6 reference cycle units in the average expansion (this means that the
variable's increase from cycle trough to peak is equal to 55.6% of its average
absolute value over the cycle), while book value of internal funds rises 45. 8 points
and internal funds with the IVA rises 40.5 points. This behavior is consistent with
point one of the specification of Minsky's· hypothesis.
As a consequence of the cyclical behavior of internal funds and investment
expenditures, the FOFA deficit on capital account (the gap between investment
and internal funds) grows dramatically in expansions, increasing by 246.9 refer-
ence points from the initial trough to the cycle peak (Figure 4 and Table 1). This
investment gap can be financed in two ways: the corporations can reduce their
holdings of financial assets in their portfolios and/or they can resort to external
sources of funds. The corporations adopt the latter solution; rather than dissave,
the NFCs actually increase their net acquisitions of financial assets by 70.9
reference points in the average expansion (NFA peaks in stage 4-see Figure 3
and Table 1). Consequently, the NFCs must issue securities and raise the neces-
sary funds to finance their growing deficits on capital account externally. Credit
market borrowing (B) for the NFC sector rises by 71.3 reference units in the
average expansion, peaking in stage 5, the cycle peak; and net funds raised in
markets (NF-borrowing plus net new equity issues) rises by 64.1 points, also
peaking in stage 5. Total external funds (X) also rises in expansions (Table 1 and
Niggle [ 1986]). This behavior is consistent with point two of the specification of
Minsky's analysis.
Thus, some of the increase in external finance which occurs in expansions can
be explained as due to differences in the timing or rhythm of the cyclical changes
in the flows of internal funds and investment expenditures. Over the course of
entire cycles, internal funds are approximately adequate for capital formation
(Niggle 1984), but because internal funds increase more slowly than the funds
necessary for the desired amount of investment in expansions, firms are forced to
seek external sources of funds adequate for their targeted investment expendi-
tures (some of the external funds are necessary to finance the deficit on capital
account). And since the flow leverage ratios of external finance and borrowing to
capital formation increase in expansions (B/I by 20.1 points and B/F by 32.8
points), borrowing must increase absolutely and more rapidly than either invest-
ment expenditures or the finance gap itself. This means that a portion of these
external funds must be used for something other than capital formation. Two
broad possibilities exist: they may be used to finance increases in working capital
(which would appear in the FOFA as acquisitions ofliquid financial assets) or for
other purposes not tracked by the FOFA, such as speculative purchases of debt
and equity securities (including those attendant on mergers and acquisitions).
Consider the identity for total net uses and external sources of funds (Equation
14); dividing this identity by the level of investment (I) gives an expression
including one of the leverage concepts (X/I):
(15) XII = (D + NFA + SD)/I
This identity shows that the behavior of the flow leverage ratio is determined by
the collective behavior of the total net uses of funds (net of internal funds and
including the unreported uses reflected in the statistical discrepancy, SD) and
investment expenditures; leverage ratios will increase if any of the three variables
in the numerator increases. But unless internal funds are declining (increasing D
without increasing I), an increase in leverage (X/I) must be the result of an
Figure 3. Net Acquisitions of Financial Assets (NFA).
160
Cycle 1
100
1954.4-1958.4
40 160
100
Cycle 2
1958.4-1961.1
160 40
100 Cycle 3
~ 1961.1-1970.1
c
::s
CD
-g. 40 160
CD
0
c
I!!
J!! 100
CD
a: Cycle 4
1970.1-1975.1
160 40
100
Cycle 5
1975.1-1980.7
40 160
100 Average cycle
40
II Ill IV v VI VII VIII IX
T p T
Cycle stages
Figure 4. Deficit on Capital Account (D).
160
Cycle 1
100 1954.4-1958.4
40 160
100
Cycle 2
1958.4-1961.1
160 40
100 Cycle 3
~ 1961.1-1970.1
c
:::>
CD
0
>-
0 40 160
CD
0
c
I!?
.!! 100
CD Cycle 4
a: 1970.1-1975.1
160 40
100
Cycle 5
1975.1-1980.7
40 160
100 Average cycle
40
T p T
Cycle stages
increase in one or both of the other two uses of funds. Since internal funds
increase in expansions, the increase in flow leverage ratios such as X/I which
occurs in expansions must be due to increases in the other uses of funds.
Consistent with point three of our specification of Minsky's financial instabil-
ity theory, both acquisitions of financial assets and other, unreported uses of funds
(reflected in the statistical discrepancy) do increase in expansions. Although
firms reduce their holdings of liquid financial assets-point four of the specifica-
tion of Minsky's theory-in stages 3, 4, and 5 of the expansion (they initially
increase their holdings of liquid assets in stage 2, the first one-third of the
expansion), their holdings of other financial assets (trade credit, consumer credit
and direct foreign investment) rise rapidly and swamp out the fall in their hold-
ings of liquid assets. These financial uses of funds are large with respect to the
external funds raised (the ratio of net acquisitions of financial assets to the deficit
was 3/1 in the cycles of the 1970's) and account for much of the growing
difference between external funds raised and the deficit on capital account; funds
flow out from the NFC sector to finance their customers' purchases and in search
of profitable financial investment opportunities. Nonfinancial corporations bor-
row to lend in expansions, acting like financial intermediaries.
Other financial uses of funds which increase in the expansion may be reflected
in the statistical discrepancy between reported sources and uses of funds in the
FOFA. Debt-financed purchases of equity securities to effect mergers, acquisi-
tions and buyouts, and the capitalization of interest (borrowing to finance interest
payments in periods of financial stringency) are other possible explanations for
the increase in leverage ratios which occur in expansions; the original data
sources for the FOFA do not track these activities, and casual observation of
corporate behavior in recent decades indicates that these activities may absorb
significant amounts of borrowed funds. 6 The statistical discrepancy is large both
absolutely and with respect to the levels of borrowing and reported expenditures
by the NFCs, occasionally approaching the size of total credit market borrowing
itself. The statistical discrepancy rises in most expansions (all but one, in which it
fell 4.8 points) for an average trough-peak increase of 47 .0 reference points. To
the extent that its behavior does reflect the speculative financial uses suggested
here, it offers further evidence in support of point three of Minsky's theory.
To recapitulate, the increase in external financing of corporate economic
activity which occurs in the expansion phase of the typical business cycle is
partially explained by the differential rates of increase in capital expenditures and
internal funds, in differences in the timing of these uses of funds and internal
sources of funds. But increases in other (reported and unreported) uses of funds
must be called upon to give a complete account of the increase in external finance
as well as in the leverage ratios of the nonfinancial corporations. The NFC sector
increases its leverage, changing its capital structure toward a heavier debt and
external sources of funds weight not just to finance positions in new real capital
assets, but also to finance its sales, its investment abroad, and holdings of
financial assets. In addition, the increased leverage might also refinance its
outstanding debt, service that debt, and finance restructuring and combinations of
firms. As expansions mature, all of these uses of funds outstrip the growth of
internal funds, forcing the firms to seek external funds.
This observed behavior in the sources and uses of funds and in leverage ratios
is exactly the pattern that Minsky's model of upward financial instability in
expansions would predict. Borrowing and leverage both increase in expansions
because firms' demand for capital assets (real and financial) outstrip the growth
rate of internal funds, and their holdings ofliquid financial assets decline, further
weakening their financial positions. (Wolfson [1986, and in this volume] also
interprets the behavior of financial variables from the perspective of Minsky's
theory; although critical of certain aspects of Minsky's model, Wolfson finds that
the model gives a plausible account of the behavior of the NFC sector during
financial crises, which occur around cycle peaks.)
Corporate Financial Variables

and Ratios in Contractions
In expansions, flow leverage ratios increase because external finance rises more
rapidly than expenditures on investment, which in turn rise more rapidly than
internal funds. This increase in external finance above that necessary to finance
the growing deficit on capital account is necessary to finance other activities of
the NFCs such as acquisitions of financial assets, corporate restructurings and
combinations of firms. In the contraction phase of the cycle this process is
reversed: investment, acquisitions of financial assets, borrowing and leverage
ratios all decline, although not back to their levels at the beginning of previous
expansions; this behavior is consistent with point five of Minsky's hypothesis.
Flow leverage ratios tend to peak late in the expansion (stage 4 in the average
cycle) and then to decline slowly throughout the contraction, often exhibiting a
second (lower) peak in the mid-to-late contraction. These contraction peaks in
leverage ratios, often higher than the expansion peaks, appear to be a combina-
tion of two factors: first, investment declines steadily throughout contractions,
reducing the denominator of leverage ratios, and second, borrowing increases in
the mid-to-late contraction, increasing the numerator in the ratios. In the cycles in
which investment declines a great deal, a relatively small increase in borrowing
results in a significant increase in leverage ratios for the sector. The increase in
borrowing in the contraction appears to be the result of several factors: one is the
increase in the extension of credit by the NFCs to their customers (consumer
credit rises in late contractions), and another is direct foreign investment, which
also rises in contractions. A third factor, which is consistent with Minsky's
financial fragility and instability hypotheses, is forced borrowing by firms to
cover operating losses (to pay their fixed costs of administrative overhead, etc.),
Table 2
Average Values of Various Leverage Ratios for U.S. Nonflnanclal

Corporations During Trough-Peak-Trough and Peak-Trough-Peak Cycles,
1953-81
B/F Bil CMD/MVE

TPT Cycles (five) % % %
1 (1954.4-1958.4) 29.0 27.3 36.2

2 (1958.4-1961.1) 29.2 28.1 33.4
3 (1961.1-1970.10) 38.6 34.6 33.4
4 (1970.10-1975.1) 55.2 48.9 50.4
5 (1975.1-1980. 7) 45.7 42.0 74.3
PTP Cycles (six)
1 (1953.7-1957.7) 26.0 23.9 37.9

2 (1957.7-1960.4) 29.7 28.8 39.9
3 (1960.4-1969.10) 36.9 32.9 32.5
4 (1969.10-1973.10) 50.3 46.1 41.5
5 (1973.10-1980.1) 49.6 44.5 73.7
6 (1980.1-1981.7) 43.3 41.5 71.1
B: Net Credit Market Borrowing

F: Fixed Investment
I: Capital Expenditures (Fixed Investment Plus Inventory Investment)
CMD: Credit Market Debt Outstanding
MVE: Market Value of Corporate Equity
Source: Federal Reserve Board Flow of Funds Accounts and Balance Sheets for the US Economy
1945-81; Niggle 1984.
to make interest payments, and to refinance outstanding debt. (See Minsky's

discussion of forced borrowing in credit crunches and their aftermaths [Minsky,
1982b]; forced borrowing is also discussed by Fazzari and Caskey, and Wolfson,
both in this volume.)
The leverage ratios are almost always higher at the end of the trough-peak-
trough (T-P-T) cycle (stage 9) than at the beginning (stage 1), producing an
upward trend. Table 2 illustrates this upward ratchet effect in the flow measures
of leverage which occurred in the first four of the five T-P-T cycles; the stock
measures of leverage (such as credit market debt outstanding to the market value
of corporate equity) reflect that upward ratchet effect.
The long-run intercycle behavior of these leverage ratios is also consistent
with Minsky's model, which is a synthesis of cyclical and secular analysis (Pollin
1982, 1986). The flow measures of leverage peaked in cycle 4 (1970-75); the
expansion phase of this cycle succeeded the long expansion of the 1960s (105
months) and the brief recession of 1969-70 ( 12 months). One could argue from a
Minskian perspective that the brief and shallow recession following the 1960s'
boom was not severe enough to dissipate expectations of profit and perceptions of
risk which, if no longer "euphoric," were still strongly optimistic for the repre-
sentative firm and its bankers. The greatest relative increases in leverage (in
reference cycle units) occurred in cycle 3 (the long 1960s expansion) and cycle 4
(Table 2).
The upward drift in leverage during the period analyzed paused after the
fourth cycle (1970-75). Although leverage increased in the expansion of the
subsequent cycle (1975-80), it did not rise as rapidly as in the previous cycles,
and it fell back farther than in those cycles. This appears to be due to a combina-
tion of two factors. First, the sector's deficit on capital account was lower during
this cycle than during previous expansions, damping the increase in external
funds raised (see Figures 1 and 4); second, the sector's acquisition of financial
assets (especially consumer and trade credit) plunged precipitously in the 1980
contraction, lessening the demand for funds by the NFCs (Figure 3).
This reduction in flow leverage over the last cycle may imply a return toward a
lower "normal" leverage ratio, the attainment of a new plateau as forseen by
Gurley and Shaw (1956) for financially mature and developed economies, or
simply a temporary pause in the secular pattern. Notice (Table 2) that the stock
measures of leverage reflect the reduction in flow leverage very slightly:
CMD/MVE only fell from 73% to 71 % between cycles 5 and 6 (P-T-P cycles),
and the decade averages for the stock measures are much higher for the 1970s
than for the earlier period. This behavior can also be interpreted as consistent
with Minsky's model: the more frequent and severe recessions of the late 1970s
and early 1980s, coupled with the deteriorating asset portfolio quality of financial
institutions, reduced speculative expectations, effecting a decline in borrowing
and leverage. And although the current expansion (which began late in 1982)
continues, making cyclical analysis inappropriate, leverage began increasing
again in the mid 1980s, so that B/I averaged 50.5% for 1986 (see FRB Statistical
Release Z7, June 5, 1987).
* * *
This analysis confirms that leverage and borrowing by nonfinancial corpora-
tions (taken as a sector) behave cyclically and secularly much as Minsky's theory
would have it (at least as that theory has been specified for this paper). Expendi-
tures on capital assets (real and financial, existing and new), borrowing and
leverage ratios all increase in expansions. The greatest increase in leverage
occurred during the longest expansion (that of cycle 3, 1961-70) and the absolute
level ofleverage peaked in cycle 4, after the brief 12 month contraction of 1969-
70 which followed the long expansion of the 1960s. And although leverage
temporarily declined in the late 1970s and early 1980s, that process reversed itself
as the economy recovered from the latest contraction of 1981-82; currently
leverage ratios are the highest ever recorded for U.S. nonfinancial corporations.
Notes
1. Minsky has discussed financial instability and financial fragility in a series of books
and articles. His John Maynard Keynes (1975) and Can "It" Happen Again? (1983)
contain extended discussions of his theory. "A Theory of Systemic Fragility" (1977)
contains a concise statement of his views.
2. See Minsky (1977); Minsky (1978), pp. 17-20; Minsky (1980), pp. 508-11; and all
of his 1981 paper for discussions of the financial crises and possible ensuing deep depres-
sion which the state of financial fragility implies. Minsky (1977) explains the role ofliquid
reserves in preventing economic collapse. Minsky attributes the argument that debt defla-
tions and ensuing deep depressions occur when economic units are financially fragile to
Irving Fisher's pioneering work in the 1930s (Fisher, 1933).
3. Another attempt at comparing Minsky's theory with the empirical record is Martin
Wolfson's Financial Crises: Understanding the Postwar U.S. Experience (1986); also see
his chapter in this volume.
4. Wesley C. Mitchell's What Happens During Business Cycles (1951), and W. C.
Mitchell and Arthur F. Burns, Measuring Business Cycles describe the techniques they
developed for business cycle analysis. Gerhard Bry and Charlotte Boschan describe the
NBER programs in a thorough but non-technical manner in their Cyclical Analysis of Time
Series, Technical Paper 20 (1971), NBER. For a detailed discussion of an application of
the NBER program to the nonfinancial corporations' sources and uses of funds which
discusses the strengths and weaknesses of the NBER approach, see Niggle (1984, chap-
ter 3).
5. The Federal Reserve Board's Flow of Funds Accounts attempt to represent the
interaction between financial and nonfinancial economic activity on the macroeconomic
and sectoral levels. The FRB began developing the FOFA in 1947; Introduction to the
Flow of Funds (FRB 1980) describes the accounting concepts utilized, the organization of
the accounts, the sources of data, and the various publications which report the time series.
Morris Copeland's pioneering work forms the conceptual basis for the accounts (see
Copeland 1952). Cohen (1972) offers an assessment of Copeland's contribution to flow of
funds accounting techniques and their usefulness for economic analysis.
6. See the statement by Preston Martin, then Vice President of the Board of Governors
of the Federal Reserve Board, in the Federal Reserve Bulletin 71(July1985): 514-17 for
the Fed's views on the extent of the nonfinancial corporations' use of funds for corporate
combinations and restructurings.
References
Bry, Gerhard, and Charlotte Boschan. 1971. Cyclical Analysis of Time Series. Technical
Paper No. 20. New York: National Bureau of Economic Research.
Cohen, Jacob. 1972. "Copeland's Moneyflows After 1\venty-five Years: A Survey."
Journal of Economic Literature 10:1-25.
Copeland, Morris. 1952. A StudyofMoneyflows in the United States. New York: National
Bureau of Economic Research.
Earley, James, Robert J. Parsons, and Fred A. Thompson. 1976. "Money, Credit, and
Expenditure: A Sources and Uses Approach." The Bulletin. Center for the Study of
Financial Institutions, Graduate School of Business Administration, New York Univer-
sity, No. 3.
Fazzari, Steven, and John Caskey. 1988. "Debt Commitments and Aggregate Demand: A
Critique of the Neoclassical Synthesis." In this volume.
Federal Reserve Board. 1980. Introduction to the Flow of Funds. Washington, D.C.:
Board of Governors of the Federal Reserve.
Fisher, Irving. 1933. "The Debt Deflation Theory of the Great Depression." Econome-
trica 1:337-57.
Gurley, John G., and Edward S. Shaw. 1956. "Financial Intermediaries and the Saving-
Investment Process." Journal of Finance (May):257-66.
Minsky, Hyman P. 1974. "The Modelling of Financial Instability: An Introduction."
Modelling and Simulation Volume 5: Proceedings of the Fifth Annual Pittsburgh
Conference, Instrument Society of America, pp. 267-73.
- - - . 1975. John Maynard Keynes. New York: Columbia University Press.
- - - . 1977. "A Theory of Systemic Fragility." Financial Crises: Institutions and
Markets in a Fragile Environment. E. D. Altman and A. W. Sametz, eds., New York:
John Wiley and Sons, pp. 138-52.
- - - . 1978. "The Financial Instability Hypothesis: A Restatement." Thames Papers
in Political Economy (Autumn):5-10.
- - - . 1980. "Finance and Profits: The Changing Nature of American Business
Cycles" The Business Cycle and Public Policy. Washington, D.C.: U.S. Government
Printing Office, pp. 209-44.
- - - . 1982. Can "It" Happen Again?. Armonk, New York: M.E. Sharpe, Inc.
Mitchell, Wesley C., and Arthur F. Burns. 1946. Measuring Business Cycles. New York:
National Bureau of Economic Research.
Mitchell, Wesley C. 1951. What Happens During Business Cycles. New York: National
Bureau of Economic Research.
Niggle, Christopher J. 1984. The Cyclical and Secular Behavior of Leverage for the U.S.
Nonfinancial Corporate Sector 1953-81: A Sources and Uses Approach. Ph.D. diss.,
University of California, Riverside.
- - - . 1985. The Cyclical Behavior of U.S. Nonfinancial Corporations' Financial
Ratios and Minsky's Financial Instability Hypothesis. University of Redlands. Mimeo.
Pollin, Robert. 1982. Corporate Financial Structure and the Crisis of U.S. Capitalism.
Ph.D. diss., The Graduate Faculty, New School for Social Research, New York.
- - - . 1986. "Alternate Perspectives on the Rise of Corporate Debt Dependency:
The U.S. Postwar Experience." Review of Radical Political Economics 18:205-35.
Taylor, Lance, and Stephen A. O'Connell. 1985. "A Minsky Crisis." Quarterly Journal
of Economics 100:871-85.
Wolfson, Martin. 1988. "Theories of Financial Crises." In this volume.
- - - . 1986. Financial Crises: Understanding the Postwar U.S. Experience.
Armonk, New York: M. E. Sharpe, Inc.
Theories of Financial Crises
Martin H. Wolfson
With the increase in financial crises in recent years, there has also been an
increase in theoretical analysis concerning the causes of this phenomenon. Sever-
al theorists writing on this topic recently have approached the problem from a
relatively similar perspective. For want of a better term, this perspective could be
termed the "business cycle, credit market" perspective.
Although there is broad agreement on a general approach, there are important
differences in the actual analyses of various writers. The purpose of this chapter is
to discuss the major theories of financial crises within this general perspective, to
indicate how they differ, and to note some ways to compare the theories with the
experience of financial crises in the postwar U.S. economy.
A Perspective on Financial Crises
The business cycle perspective sees financial crises as the result of the normal
functioning of the economic and financial systems over the course of the business
cycle. Endogenous developments near the peak of the cycle create the conditions
that make financial crises likely.
According to the credit market perspective, the use of credit is essential to the
business cycle expansion. Near the peak of the expansion, the demand for credit
becomes increasingly inelastic. However, limitations on the supply of credit
imply that all demands for credit cannot be accommodated. The financial crisis is
associated with the end of the expansion of credit.
This general perspective is an important framework of analysis, but nonethe-
less it is only a framework, not a theory of financial crises. Important and
significant differences in analysis are contained within the overall perspective. In
order to gain a better understanding of these differences, several contemporary
The views expressed in this chapter are those of the author and do not necessarily
represent those of the Board of Governors or of any other staff member.
221
222 MARTIN H. WOLFSON
theories of financial crises will be briefly highlighted, followed by a discussion of

the important points of difference.
Particular Theories of Financial Crises
The view of Albert M. Wojnilower (1980, 1985) is that the demand for credit near
the peak of the business cycle expansion is not only interest inelastic; it is
essentially unbounded. Corporations seek to expand production and to finance
this expansion with as much credit as they can obtain. According to Wojnilower,
the banks are willing and able to supply the credit required. The expansion of
credit is halted, however, by sudden shocks that make the banks either unwilling
or unable to continue lending. The abrupt limitation of credit puts an end to the
business cycle expansion and initiates a recession.
The perspective of Allen Sinai (1976, 1977; Eckstein and Sinai, 1986) is quite
different. He does not see the economy as out of control near business cycle
peaks; rather, his view is that limitations on the supply of credit are slowly and
surely bringing the expansion to an end. Tight monetary policy over a prolonged
period and reduced deposit inflows make the banks increasingly unable to meet
the demands for credit. Production is gradually curtailed, and the economy
slowly moves into recession.
The analysis of Hyman P. Minsky (1975, 1982, 1986) is focused on the
endogenous development of financial fragility, and the implications of fragility
for the development of financial crises. An important aspect of a fragile financial
environment is the increasing inability of units to meet payment commitments,
especially commitments due to debt. According to Minsky, an important reason
for an inelastic demand for credit near the business cycle peak is the need to
borrow to pay these debts. Another important reason is the need to finance
ongoing investment projects that had been initiated at an earlier time.
Like earlier theorists of financial crises, such as Thorstein Veblen (1904) and
Wesley Clair Mitchell (1971), Minsky notes the unwillingness of banks to debt-
finance investment because of the decreasing creditworthiness of business bor-
rowers. These earlier theorists emphasized a decline in profits as the key variable
in changing the assessments of lenders. Minsky notes the role of profits, but his
main emphasis seems to be on rising interest rates. He stresses their negative
effect on the ability of borrowers to meet debt commitments, and on the outlook
for future profitability. His emphasis is also different from that of Karl Marx
(1967, 1968), who stressed the key role of a decline of profits in the inability to
meet fixed payment commitments.
Differences Among the Theories
As can be seen from the above discussion, there are important differences among
the theorists even though there are similarities in the overall perspective. These
THEORIES OF FINANCIAL CRISES 223
differences generally concern (a) the demand for credit, (b) the supply of credit,
and (c) the nature of the financial crisis.
An important difference concerns the reason that the demand for credit be-
comes inelastic near the peak of the expansion phase of the business cycle. One
perspective is that business firms are actively seeking to expand production, and
higher interest rates are not a sufficient deterrent to these plans. An alternative
perspective is that the demand is inelastic because it is necessitous. Firms need
money either to finish investment projects that had been started in the past (and
whose abandonment would result in unnecessary loss), or to meet payment
commitments on outstanding debt.
This difference of opinion is related to another one, concerning the relation-
ship between the financial crisis and the business cycle recession. Those who
view the demand for credit as a demand for funds to finance the growth of
production consider the interruption of the supply of credit to be responsible for
halting this growth and initiating the recession. On the other hand, the view of the
demand for credit as a necessitous one, involving funds to meet debt payment
requirements and to finish partly completed investment projects, sees the finan-
cial crisis as occurring either during the recession or after the conditions for
recession have already been established (because plans for new investment have
been curtailed).
The distinction is important, not only for pinpointing the causal relationship
between the financial crisis and the recession, but also for understanding the
potential threat of the crisis itself. If the crisis involves a denial of funds to
borrowers who cannot meet debt payments, then a cascading downward spiral of
defaults, bankruptcies, and credit liquidations, typical of the debt-deflation pro-
cess (Fisher 1933), is set to occur. However, if production can be curtailed
without requiring debt repudiation, then perhaps the economy can avoid the
threat of these severe repercussions.
There is a further difference of opinion among those who emphasize the
inability of borrowers to meet debt payment commitments. It is whether this
situation is due primarily to an increase in interest rates, or whether a decline in
profits is mainly responsible.
A second major issue is the reason for the limitations on the supply of credit.
The differences of opinion revolve around the ability and willingness of creditors,
especially the banks, to extend credit. One point of view is that the main reason
for the limitation is the voluntary restriction of credit by the banks, due to the
decreased creditworthiness of business borrowers. The alternative perspective is
that the banks try to meet the credit demands of their borrowers but are prevented
from doing so either by the slow squeeze on reserves and deposits or by a sudden
shock to the system.
A third issue is the nature of the financial crisis. One dimension of difference
is whether the crisis occurs slowly or abruptly. Another is whether it is only a
squeeze on the supply of credit, or if it also involves a sudden jolt to confidence
and a disruption of normal financing patterns. In other words, the controversy is

over whether the financial crisis involves a slow winding down of credit exten-
sions or a sudden cutoff of credit and a desperate scramble for money to meet
payment commitments.
Theories and the U.S. Experience
It is not easy to reconcile these differences, because they involve attitudes and
motivations as well as questions of fact. Although it is not possible to get
definitive answers, some light can be shed on the relevant issues by examining the
recent experience of financial crises in the United States. Timing relationships
and surveys of behavior will be particularly useful.
First of all, it would be useful to briefly identify the timing of the postwar
crises. The first crisis following World War II in the United States was called the
Credit Crunch of 1966, which took place in August, 1966. It was followed by a
crisis surrounding the bankruptcy of the Penn Central Railroad in June of 1970.
In May of 1974, the troubles of the Franklin National Bank precipitated another
crisis and was followed in March, 1980 by the events surrounding the Hunt
brothers' unsuccessful silver speculation. During the summer of 1982 (June
through August), a series of events associated with the collapse of Drysdale
Government Securities, Inc., the failure of Penn Square Bank, and the threatened
default of Mexico on its international obligations created a crisis of confidence in
the nation's commercial banks.
If we count the growth recession of 1966-67, then a financial crisis has taken
place near the peak of each business cycle expansion since 1966. Each of the
above mentioned crises, in fact, has taken place shortly after the recession has
begun. All of them took place within six months after the peak of the expansion
(except for the crisis in 1982, which took place approximately one year after-
wards). Thus it appears that this timing evidence is incompatible with the view
that sees the recession caused by the liquidation of credit during the financial
crisis.
There are also the most recent crises: the effective failure of Continental
Illinois National Bank, the runs on deposits of state-insured thrifts in Ohio and
Maryland, and the stock market crash. These took place during the business cycle
expansion, before the peak had been reached. Their timing is related to long-term
structural and institutional changes which will not be discussed further here. 1
Let us now consider the other issues raised in the previous section. The first
issue concerns the reasons for the inelastic demand for credit near the business
cycle peak. There is some basis to think that this is a necessitous demand for funds
to meet payment commitments.
The financial condition of the nonfinancial corporate sector has shown a
strong cyclical deterioration as the peak of the expansion has approached. The
traditional measures of balance-sheet strength have worsened: debt/equity and
debt/maturity ratios have increased, and liquidity ratios have declined. 2 In addi-
tion, the interest coverage ratio has declined near the peak of the expansion. This
measure, which is the ratio of gross profits before tax plus interest paid to interest
paid, indicates the ability to pay interest costs out of capital income.
The peaks of the interest coverage ratio closely coincide with the peaks in the
profit ratio. This closeness in timing suggests that the decline in profits near the
expansion peak is responsible for the debt payment difficulties observed. Interest
rates, on the other hand, have typically begun to increase from six months to a
year before the interest coverage ratio begins to fall. This timing pattern suggests
that corporations can use debt to exploit profitable investment opportunities, and
these actions can increase interest rates without necessarily creating debt repay-
ment problems. It is only when the means to repay the debt-profits-begin to
decline that repayment problems become more severe. 3
The alternative view concerning the demand for credit.is that corporations are
seeking credit in order to finance increases in production. However, the empirical
record indicates that real investment in plant and equipment has generally peaked
before the financial crisis, although it has remained at a relatively high level. In
addition, the series on contracts and orders for plant and equipment, which
indicates plans for future investment, has consistently peaked approximately one
to two quarters before the investment series itself.
The most likely interpretation of these relationships is that a form of "involun-
tary" investment is responsible for at least part of the demand for credit in the
period immediately preceding the financial crisis. During the time in which plans
for investment have been curtailed but investment spending is still relatively
high, it is likely that corporations are seeking to fund not new investment proj-
ects, but those which had been initiated in the past and which require continued
funding in the present. This "involuntary" investment takes the form of a
payment commitment similar in some respects to debt payment requirements.
The second major issue concerns the reasons for the limitations on the supply
of credit. In examining the conditions affecting lenders, we find that monetary
policy has consistently tightened near the postwar business cycle peaks. In addi-
tion, losses on loans of commercial banks have risen as the peak of the expansion
has approached. Thus banks have had an incentive to cut back on their lending
and have also had to deal with pressures from the monetary authorities to limit
their lending. How they have reacted to this situation, however, has differed for
different types of customers.
Generally, banks have restricted credit near the business cycle peaks, but at the
same time have tried to meet the necessitous loan demands of their long-standing
corporate customers. The pressures on their reserves have led banks to limit the
growth of their investments and to slow the growth of nonbusiness loans in favor
of business loans. Evidence from the Federal Reserve Board's Survey of Bank
Lending Practices indicates that during these times, banks have tightened lending
policies much more for new business borrowers than for established customers.
They have been able to meet the needs of their best customers by using liability
management and "purchasing" relatively expensive, volatile, and uninsured
funds in the money market, e.g., large negotiable certificates of deposit, Euro-
dollar and commercial paper borrowings, and fed funds.
The final issue concerns the nature of the financial crisis. The difference of
opinion centers around the issue of whether the financial crisis is a slow liquida-
tion of credit or a more abrupt reaction. In fact, the pressures on the banks
referred to above have led to a slow reduction of credit for many borrowers prior
to the business cycle peaks. However, the banks' attempts to meet the needs of
their best customers have also led to a lending situation which has been disrupted
suddenly by surprise events.
There were two types of surprise events in the postwar experience in the United
States. The first was an institutional constraint imposed by government authori-
ties; in 1966 Regulation Q interest rate ceilings were kept in place, and in March,
1980 explicit credit controls were instituted. The second type of surprise event
was either a default or failure (or the threat of one) which shocked investor and/ or
depositor confidence. In 1970, it was the Penn Central bankruptcy, for example,
and in 1974, the threatened failure of Franklin National Bank, etc.
The significance of these surprise events is that they have abruptly cut off the
supply of credit to important borrowers. In the case of the institutional con-
straints, banks were suddenly prevented from further lending by government
authority. The surprise defaults and bankruptcies typically caused a rapid with-
drawal by institutional investors from the commercial paper market and the
markets for bank purchased funds. In both cases, lending suddenly stopped.
It was at this point that the Federal Reserve Board quickly intervened in the
financial markets as a lender of last resort. By making money available to those
who were desperately in need of it and were abruptly cut off from sources of
credit, the Federal Reserve prevented the financial crisis from spreading and
deepening. This intervention was usually followed by easier monetary and fiscal
policies which, for the time being, moved the economy away from crisis.
Thus the essence of the financial crisis is an intense demand for money,
brought about by the sudden cessation of credit. Although much of the economy
is subject to a gradual credit squeeze as the business cycle expansion is ending,
this alone does not capture the full extent of the threat that financial crises pose to
the financial stability of the U.S. economy.
Notes
1. For a discussion of these changes and their significance for the business cycle
perspective, see Wolfson (1986, 1987).
2. The empirical record is discussed in detail in Eckstein and Sinai (1986) and Wolfson
(1986).
3. Neither is it the case that interest costs, which are a deduction from profits, were
responsible for the decline in profits; the profit rate with interest costs added back to profits
shows the same cyclical pattern.
References
Eckstein, Otto, and Allen Sinai. 1986. "The Mechanisms of the Business Cycle in the
Postwar Era." The American Business Cycle: Continuity and Change, pp. 39-105.
Robert J. Gordon, ed. NBER: University of Chicago Press.
Fisher, Irving. 1933. "The Debt-Deflation Theory of Great Depressions." Econometrica
1:337-57.
Marx, Karl. Capital. 3 vols. New York: International Publishers. 1967.
- - - - . 1968. Theories of Surplus Value. 3 parts. Moscow: Progress Publishers.
- - - - . 1982. Can "It" Happen Again? Essays on Instability and Finance. Armonk,
New York: M. E. Sharpe, Inc.
----.1986. Stabilizing an Unstable Economy. New Haven: Yale University Press.
Mitchell, Wesley Clair. 1971. Business Cycles and Their Causes. Berkeley, California:
University of California Press.
Sinai, Allen. 1976. "Credit Crunches-An Analysis of the Postwar Experience." Param-
eters and Policies in the U.S. Economy, pp. 244-74. Otto Eckstein, ed. Amsterdam:
North Holland.
- - - - . Discussion of Papers on the American Financial Environment. Financial
Crises: Institutions and Markets in a Fragile Environment, pp. 187-203. Edward I.
Altman and Arnold W. Sametz, eds. New York: John Wiley & Sons, 1977.
Veblen, Thorstein. 1904. The Theory of Business Enterprise. New York: Charles Scrib-
ner's Sons.
Wojnilower, Albert M. 1980. "The Central Role of Credit Crunches in Recent Financial
History." Brookings Papers on Economic Activity, No. 2, pp. 277-326.
- - - - . 1985. "Private Credit Demand, Supply, and Crunches: How Different are the
1980s?" American Economic Review 75:351-56.
Wolfson, Martin H. 1986. Financial Crises: Understanding the Postwar U.S. Experience.
Armonk, New York: M. E. Sharpe, Inc.
- - - - . 1987. Financial Instability, the Business Cycle, and Macroeconomic Policy.
Paper presented at the Meeting of the Eastern Economic Association. (March 6).
The Political Economy of
the External Debt and Growth:
The Case of Peru
Felix Jimenez and Edward J. Nell
The Baker Plan signaled the U.S. Administration's desire to reverse the slow-
down in lending to the Third World from 1986 to 1988. Accordingly, first, debtor
countries must adopt free-market policies; second, banks should increase new
lending to the debt-ridden nations; and third, the IMF and the World Bank should
intensify their monitoring of debt-plagued countries and speed up their own
lending programs. Underlying the Plan are the unchallenged postulates of Ameri-
can policy makers: the free market is the best engine of economic growth, and
foreign capital always contributes to the long-run economic growth of recipients.
American economic policy has affected Third World countries by implement-
ing two measures: one monetarist (or bastard Keynesian) and the other Keynes-
ian. The first is a dollar devaluation, intended to change both the U.S. level of
trade and the terms of trade of the U.S. economy. 1 Since Reaganomics has
reduced the already damaged purchasing power of the Third World countries, the
second measure is to pump fresh money into the biggest debtor countries when-
ever they adopt free-market policies; this fresh money, in turn, can be expected to
help create demand for the manufactured goods of the U.S. and other industrial-
ized economies. 2
The general object of this paper is to show that the free-market approach is, at
best, useless, since it does not deal with the basic problems which prevent Latin
American countries from streamlining their economies, and, at worst, danger-
ous, since it can destroy the fragile industrial base these economies have devel-
oped. The point will be illustrated with the case of Peru, which is reasonably
representative. Free-market policies are part of the problem, not part of the
solution.
228
EXTERNAL DEBT & GROWTH: PERU 229
Table 1
Total External Debt
Debt-GDP Debt-Ex~ort
Year1 Public Private Total2 Ratio3 Ratio
1966 604.7 1153.1 1757.8 30.3 194.7

1968 762.0 1199.6 1961.6 36.1 201.3
1971 1000. 7 1210.5 2211.2 29.0 207.0
1973 1491.0 1201.0 2692.0 26.5 199.8
1975 3066.0 1286.0 4352.0 28.0 261.6
1978 5135.0 1340.0 6475..0 55.0 272.8
1981 6210.0 1507.0 7717.0 30.7 191.0
1984 9775.0 1464.2 11239.2 53.6 312.5
Notes:
'Cycle-peaks: 1966, 1971, 1975 and 1981
Cycle-troughs: 1968, 1973, 1978 and 1984.
2 Millions of U.S. dollars.
'Variables of this ratio were measured by the same currency at current prices.
Sources:
Banco Central de Reserva de! Peru; Instituto Nacional de Estadistica; IMF, International Financial
Statistics; World Bank, Peru: Long-term Development Issues, Vol. III, 1979.
First, the external and internal factors underlying the evolution of the Peruvian
debt will be identified. The debt crisis arises not only from the inadequacies of the
international financial system, 3 e.g., deterioration in the terms of trade and world
recession. It is also a result of internal economic factors, since internal policies to
counteract international shocks are ultimately limited by the requirements of
domestic capital accumulation.
Second, the contribution of foreign finance to economic growth will be quanti-
fied and the macroeconomic consequences of both external debt and foreign
direct investment will be identified. We want to stress that the need for external
finance mostly reflects the structural dependency of the economic system on
imports of industrial inputs and capital goods due to the absence of an integrated
input-output system.
Sources of Growth of External Debt, 1973-1984
The figures in Table 1 show three important characteristics of the external debt
for the Peruvian economy. First, as can be seen by comparing column 2 and
column 4, the share of public debt increased rapidly from 34.4 percent in 1966 to
70.5 in 1975, rising to 87 percent in 1984. Second, the ratio of foreign debt to
GDP didn't change significantly during the cycle peaks: its mean was equal to
29.5; however, during the cycle troughs this ratio had a spectacular increase: its
230 FELIX JIMENEZ & EDWARD J. NELL
Table 2
Current and Capital Accounts of the Balance of Payments

(millions of current U.S. dollars)
1950-62 1963-68 1969-75 1976-80 1981 -84
Resource Balance -297.7 -254.4 -1507.8 1518.0 -825.1

Net factor
Service Income -529.8 -658.4 -1262.9 -3300.0 -4334.6
(Net Interest
Payments) (n.a.) (-235.0) (-843.8) (-2434.3) (-3774.2)
(Net repatriated
profits) (n.a.) ( -423.4) (-419.1) (-865.7) (-560.4)
Transfers 145.7 151.8 355.0 615.0 698.0
CURRENT ACCOUNT -681.8 - 761.0 -2415.0 -1167.0 -4461.0
Medium and Long
term Loans, Net 242.8 542.4 2243.9 2584.0 4342.2
Direct Foreign
Investment, Net 330.8 26.1 423.7 348.0 122.0
Short Term
Capital(*) 192.2 102.8 -267.4 -604.5 -176.0
CHANGE IN
RESERVES, Net 84.0 -89.7 -14.8 1160.5 -172.0
*This item includes Errors and Omissions.
Sources:
Banco Central de Reserva del Peru; World Bank, Peru: Long-term Development Issues, Vol III,
1979.
mean was equal to 42.8. A similar pattern is present in the debt/export ratio,
which had an average greater than 200 percent for the whole period. 4 Third, the
year 1973 constitutes the turning point in the long-run pattern of the external
debt; from that year through 1984, the debt increased almost exponentially.
The current and the capital account of the balance of payments clearly show the
existence of two radically different periods in the last 35 years (see Table 2).
From 1950 to 1968, the current account deficit represented only 18 percent of the
respective accumulated deficit in the last 16 years; repatriated profits constituted
the predominant item in the balance of factor service income during the same
period; and the short term capital flows were persistently positive.
The picture is radically different for the last 16 years, since financial capital
took over the economy. The outflow of interest payments from 1969 to 1975 was
3.5 times greater than between 1963 and 1969. Its amount tripled between 1976
and 1980 and became 4. 5 times greater from 1981 to 1984, compared to the levels
under the Velasco administration. Moreover, repatriated profits were half the
level of the net interest payments between 1969 and 1975, a third between 1976
and 1980, and only 15 percent between 1981 and 1984.
Nevertheless, it is worth pointing out that the average debt/GDP ratio during
Velasco's government was 12 points less than its mean for the 1976 to 1984
period. The corresponding values were 30 percent for 1968 to 1975, 42. l for
1976 to 1980, and 41.8 for 1981 to 1984. The cost of borrowing skyrocketed
between 1981 and 1984. In order to estimate the cost, the implicit nominal
interest rates are used-which may mean underestimating the contractual nomi-
nal interest rates. The average real rate, based on price-of-imports inflation, was
2.8 for the first Belaunde administration (1964-68), -0.2 during the first four
years ofVelasco's government, and -20.5 from 1973 to 1975. During the Mora-
les-Bermudez government, the average real rate became positive but stayed near
zero. But during the second Belaunde administration, the cost of credit increased
steeply: the real rate averaged 11. 7 percent. The cheap credit during Velasco' s
era and the low debt/GDP ratio (30 percent) contrasts with the high cost and the
high debt/GDP ratio (42 percent) during the last Belaunde administration. Are
the key factors which determine the differences in the behavior of the foreign debt
in these periods chiefly external?
When the debt service is related to exports of goods and services, the resulting
ratio for the later period is systematically higher than during the sixties. From
1950 to 1967, the debt service ratio averaged significantly less than 10 percent,
only reaching the neighborhood of 10 percent in 1966-67, which is when the
GDP rose furthest above its trend values for the entire period from 1950 to 1984
(see Table 3).
What follows is a detailed analysis of the internal and external factors underly-
ing the spectacular growth of foreign debt during the 1973-84 period. This
examination will help to evaluate the consequences of different economic policies
adopted by the last three governments: the administration of Velasco Alvarado
and the governments of Morales-Bermudez and Belaunde Terry. The first of these
is well known for its expansionary and interventionist stance, while the last two
welcomed subordination to the IMF's monetarist policies.
A methodology based on the balance-of-payments identity is used, which
makes it possible to decompose the difference between the actual and the expected
increase in debt into internal and external factors. 5 The accounting identity for
the net flow of total external debt is:
where: NB = net flow of total medium- and long-term debt;

M = imports of goods and non-factor services;
Pm = import prices;
X = exports of goods and non-factor services;
Px = export prices;
r = nominal interest rate on external debt;
D = total debt stock in current dollars;
Table 3
External Debt Flows and Debt Service Ratios
Net Flow1 Interest Interest Debt Cover-

Pay- Export Service age
Year Total Public ments2 Ratio3 Ratio4 Ratios
1963-88 542.4 474.7 235.0 4.8 9.7 99.2

1969-72 340.4 359.0 297.0 6.6 19.4 16.7
1973 333.5 313.5 101.0 7.5 32.5 71.6
1974 751.0 693.0 176.4 9.6 24.9 152.0
1975 819.0 793.0 269.4 16.2 28.5 167.2
1976 471.0 446.0 331. 1 21.0 30.7 92.0
1977 674.0 659.0 384.1 18. 1 29.4 105.9
1978 419.0 405.0 562.1 23.7 29.6 57.7
1979 585.0 617.0 540.8 12.6 19.3 74.8
1980 435.0 371.0 616.2 13.4 28.7 28.0
1981 523.0 388.0 767.8 19.0 43.5 22. 1
1982 1152.0 995.0 915.9 22.8 39.8 62.2
1983 1346.0 1431.0 971.2 27.9 50.7 80.9
19846 1321.2 1436.0 1119.3 31. 1 56.5 70.7
Notes:
'Medium- and long-term loans, in millions of U.S. dollars.
2 Total net interest payments, in millions of U.S. dollars.
'Ratio of total net interest payments to exports of goods and non-factor services.
4 Ratio of amortization plus interest payments on public loans to exports of goods and non-factor
services.
'Inverse ratio of amortization plus interest payments on public loans to net foreign public debt
flows.
•Values of debt service ratio and coverage ratio for 1984 are preliminary.
Sources:
Banco Central de Reserva del Peru; Instituto Nacional de Estadistica; World Banlc, Peru: Long-
term Development Issues, Vol Ill, 1979.
(Dt = NBt + Dt-1)

FI = net repatriated profits minus the net flow of foreign direct invest-
ment;
TR = net transfers;
SE = net flow of short term capital plus errors and omissions;
RA = changes in international reserves;
t = current year t
On the other hand, the expected net flow of total foreign debt can be defined:
(2)
where: NB* = expected net flow of total foreign debt;

Pm* = expected value of import prices;
Px* = expected value of export prices; 6
M* = amount of imports in year t given by applying the propensity to
import of year t-1 to the volume of domestic demand (apparent consumption) of
year t;
X* = amount of exports in year t given by applying the ratio of exports to
world demand in year t -1 to the level of world demand of the same year;
r* = forecasted value of the implicit interest rate estimated by regressing
the natural log of the implicit interest rate on time for 1966-72.
D*t = is equal to NB*t + D*t-l· The expected value of D* 0 (stock of
foreign debt for 1972) was estimated by regressing the natural log of the net
interest payments on time for 1966-72. The forecasted value for 1972 was
divided by r* of 1973 to calculate D* 0 •
By subtracting (2) from (1) and assuming:
FI*t = 0
TR*t = TRt
SE*t = 0
RA*t = 0
the additional external debt can be decomposed as follows: 7
NBt - NB*t =
(ii) (Pmt - Pm*t)M*t
(iii) (rt - r* t)Dt-1
(vii) Flt
(viii) r* t(Dt-1 - D* t-1)
(ix) SEt
(x) RAt
where: x is the share of exports in the world demand;

W is the real world demand proxied by imports of industrialized coun-
tries;
W* is the expected world demand; 8
m is the ratio of imports to domestic demand;
DD stands for the domestic demand, DD = GDP + M - X;
DD* is the expected domestic demand.
The term in row (i), the difference between expected and theoretical values of
exports calculated under the assumption that the share of exports in world de-
mand remains constant, accounts for the effect of deviation in world demand from
its trend values on the country's real exports. It was named World Recession
Effect and is measured at current prices. The two terms in row (ii) constitute the
Terms of Trade Effect. Changes in actual prices are measured against expected
prices only in order to capture the unexpected effect of inflation on the terms of
trade. The term in row (iii) indicates the effect of changes (with respect to
expected values) in the interest rate paid on the external debt. This Interest Rate
Effect is purely an external shock. All of these three first rows constitute the debt-
inducing effects of external disturbances. We named them External Factors.
The term in row (iv) accounts for the Import Penetration Effect. This effect
constitutes the increase in the country's imports explained only by the increase in
the propensity to import. Since this effect is also measured at current prices, it
incorporates the external shock due to unexpected inflation. The Austerity Policy
Effect is accounted for in row (v). Assuming a constant propensity to import, it is
the estimated increase in imports explained only by deviations of domestic de-
mand from its trend. Again, since this effect is measured at current prices, it
incorporates the external shock due to unexpected inflation. The term in row (vi)
quantifies the Export Reorientation Effect9 which is measured by the difference
between theoretical and expected values of exports under the assumption that the
share of exports in world demand remains constant. These last three effects are
basically associated with internal factors that further overloaded the debt prob-
lem.
The rows (vii) to (x) also constitute the debt-inducing effects of internal policy
reactions to internal or external shocks. Row (vii) accounts for the short-run and
the long-run impact of continuous foreign direct investment on the balance of
payments. It was named Foreign Investment Policy Effect. An economic policy
which permits profits to be repatriated in excess of net foreign direct investment
only exacerbates the financial distress in an economy with structural disequilibri-
um in its external sector. Row (viii) accounts for the Debt Stock Effect on Interest
Payments. The effect of policy decisions on speculators is accounted for in row
(iv). It was named Speculation or Smuggling Effect. Since it includes errors and
omissions it probably conceals smuggling, which itself is mainly a result of the
policies adopted by the government. Finally, row (x) accounts for the Reserve
Table 4
Sources of Growth of External Debt, 1973-1984

(Millions of current U.S. dollars)
Morales-
Velasco Bermudez Belaunde
Factor 1973-75 1976-80 1981-84
1. World Recession Effect 326.0 664.3 2747.5

2. Terms of Trade Effect -216.3 -461.9 4207.4
3. Interest Rate Effect 197.8 1092.3 1740.0
TOTAL EXTERNAL SHOCKS 307.5 1294.7 8694.9
4. Import Penetration Effect 54.6 -85.1 -664.7

5. Austerity Policy Effect 485.6 -2824.8 -1666.7
6. Export Reorientation Effect 687.8 -163.5 641.3
7. Foreign Investment Policy Effect -372.9 517.7 437.6
8. Debt Stock Effect on Interest Payments 39.9 26.0 996.2
9. Speculative or Smuggling Effect 160.7 604.5 176.0
10. Reserve Accumulation Effect -281.6 1160.5 -172.0
TOTAL INTERNAL SHOCKS 774.1 -764.7 -252.3
TOTAL ADDITIONAL EXTERNAL DEBT 1081.8 529.8 8442.7
Accumulation Effect, which basically reflects the financial pressure on monetary

authorities to maintain the external value of the currency. 10
The results are summarized in Table 4. As far as the additional external debt
(NB-NB*) is concerned-see the last row of Table 4-it amounted to $1,081.8
million in Velasco's period, decreased to $529.8 million in the Morales-Bermu-
dez administration and jumped up to reach the impressive value of $8,442.7
million in the second Belaunde administration.
One striking result concerns the role played by internal factors. These factors
explained a significant part of the additional debt only during the Velasco admin-
istration: 71.6 percent against -144.3 percent during the Morales-Bermudez's
government and -3.0 percent during the second government of Belaunde Terry.
Velasco's administration represents the last political effort to develop a nation-
al manufacturing sector based on the internal market, while biasing production
toward import-competing industries. The implementation of this strategy re-
quired increasing capital goods, technology, and food imports, which meant
financing these imports not only from export earnings but also by increasing
external debt. 11
While the average increase in the propensity to import accounted for only 5
percent of the total additional external debt, it is worth noting that the import
substitution policy didn't avoid import penetration. The Velasco expansionary
policy and the absence of a significant export reorientation accounted for 108.5
percent of the total additional net flow of foreign capital (see Table 4). The
foreign investment effect was negative in accord with the nationalist attribute of
this administration. By the same token, the smuggling effect was positive. As for
external shocks, it can be seen that their contribution to the additional debt only
amounted to about 28.4 percent. The favorable terms of trade effect more than
made up for the interest rate effect which, in turn, explained not more than 18
percent of the additional external debt. The associated process of import substitu-
tion was spurious; it didn't create an integrated national input-output system but
only intensified the necessity of importing capital and intermediate goods. 12
The industrialization process spurred by Velasco's government coincided with
the change in the form in which foreign capital has "migrated" to the less
developed countries. In the 1960s, foreign direct investment accounted for about
30 percent of all foreign capital inflows into Latin America, with bank loans and
bonds accounting for 10 percent. In the 1970s, the multinational share had
dropped to 21 percent, while the share of private international financiers had
risen to 59percent; 13 consequently, the interest costs overwhelmed not only the
outflow of repatriated profits, but also the inflow of foreign direct investment. 14
According to some analysts, the foreign financial borrowing, by facilitating the
trade of capital equipment, know-how, technologies, and the hiring of foreign
consultants to help set up the investment projects, has decreased the economic and
political risks associated with foreign direct investment.
In the subsequent two administrations, the role of internal factors decreased
primarily because of the draconian austerity policies. Without the effect of exter-
nal factors, the net flow of the total additional external debt would have been
negative. The positive additional debt was, therefore, explained basically by the
external shocks (see Table 4).
The interest rate effect was the worst external shock during the Morales-
Bermudez administration. But, the bulk of this effect was concentrated in the last
three years of that administration, just when the cost of the credit started to grow
in the international financial system. 15 On the other hand, the world recession
effect, discounting the favorable terms of trade effect, only amounted to about
15.6 percent of the total additional debt arising from external shocks.
As far as the internal factors are concerned, the government of that time
"succeeded" in driving down the net flow of foreign loans: the import bill was
cut by almost $3 billion through its severe austerity measures (see Table 4). Yet,
the export promotion policies were not significant in terms of relaxing the need
for foreign borrowing.
It is worth noting that without the draconian austerity policy, the accumulated
resource balance would have been negative and approximately equal to -$1,307
million (see Tables 2 and 4 for the 1976-80 period). Moreover, it is not surprising
that the open-door external policy-a well-known element of the IMF austerity
package-prevented a significant decrease in the average propensity to import in
the corresponding period. Finally, the most striking internal factor was the
reserve accumulation effect. Rather than draw down the reserves to finance the
current deficit, the Morales-Bermudez regime borrowed abroad to add to the
reserves; the increase in reserves was nearly $1,161 million between 1976 and
1980 (see Table 4).16
Trade liberalization started out in 1979 with the elimination of the Registro
Nacional de Manufacturas 17 introduced by Velasco's government to control im-
port competition. 18 The consequent significant increase in the propensity to
import and its negative impact on the consumer durables-producing sector inten-
sified the vulnerability of the state to pressures of international financial capital.
With foreign loans kept down, the state had to divert export earnings to finance
import penetration (and also increases in repatriated profits) and could accumu-
late reserves only in order to manage the dollarization of the economy on the local
currency. De-industrialization and the dollar driving the local currency out of the
internal market were the two main consequences of control over the state by
international financial capital.
Once the IMF and the World Bank paved the way for "restructuring" the
manufacturing industry, the international financial institutions increased the flow
ofloans to the new government ofBelaunde Terry. Hence, the additional external
debt skyrocketed between 1980 and 1984. The import bill decreased significantly,
again due to the austerity policies: -$1,667 million between 1981and1984 (see
Table 4). President Belaunde weakened the austerity prescriptions at the begin-
ning of his administration, perhaps because of enthusiasm generated by the recent
performance of the country's exports and the previously accumulated reserves.
This temporary alleviation occurred within a framework of free trade and conse-
quently resulted in a massive inflow of imports: the rise in the propensity to
import created a pressure on the balance of payments equivalent to 139 percent of
the same effect during the deepest import penetration period-1973 and 1974-of
Velasco's administration (See Table 4).
The upward cycle begun in 1979 peaked in 1981 simultaneously with the
current account deficit, which reached its highest value in the last 35 years. The
crisis of the external sector forced a reversal of the cycle and the story was
repeated-with an identical scenario imposed by the IMF. The government de-
creased the import bill by austerity measures; it fell -$1,766 million-which
might have been expected during 1983-84. The draconian austerity package was
imposed on an economy deeply dependent on imports, which also put a down-
ward pressure on the ratio of imports to domestic demand. As a result of this and
other restrictive external policies, the influence of changes in the propensity to
import on the additional external debt was severely negative in 1983 and 1984.
The policy of trade liberalization, begun in 1979, was halted in 1982. The
government decreed a general temporary tariff increase of 15 percent. But
this occurred at a point when the process of de-industrialization had al-
ready increased the income elasticity of demand for imports to more than twice
its value for 1968-75 (see Table 8). 19

On the other hand, the attempt to reduce the financial squeeze by export-
promoting policies was fruitless. The depressed international environment dur-
ing Belaunde's administration only showed, in a dramatic way, the unfeasibility
of the so-called export-led growth model, popularized by the apparent successes
in Asia. To try to penetrate international markets for nontraditional exports in an
unfavorable environment is likely to be a waste of effort.
Although the export reorientation effect during Morales-Bermudez's adminis-
tration had an opposite sign, the policies were basically the same as those of the
second administration of Belaunde Terry. The first was luckier than the second,
but both shared the free-market-conservative ideology. As a matter of fact during
both administrations, the sum of repatriated profits, discounting the net flow of
foreign direct investment, amounted to over $955 million. That was almost 34
percent of the outflow of international reserves due to the increase in the interest
rate in the international financial system (see Table 4). By the same token, the
outflow of short-term capital, which usually depends on confidence in internal
policy making, was approximately equal to $780 million. In absence of these two
effects, the net flow of foreign borrowing would have been $1,070 million less
than its expected value, calculated on the basis of the internal austerity and trade
policies adopted during 1981-84. Finally, the contrasting behavior of the reserve
accumulation effect during Belaunde's government doesn't mean a recovery of
monetary sovereignty, but shows a state unable to resist the pressure to devalue.
The highest rates of depreciation occurred between 1981 and 1984, and the
government was forced to change the monetary unit by creating the so-called
INTI.
In order to appreciate the foregoing analysis properly, it is important to
remember that the total effect of internal factors on the additional debt was
negative for both administrations. Therefore, the positive inflow of foreign
borrowing was basically connected to external shocks. As mentioned, during
Belaunde's administration the external shocks worsened. The world recession
effect was quadrupled; the deterioration in the terms of trade was of such a
magnitude that it created an additional need for international reserves approxi-
mately equal to $4,207 million; and the interest rate effect was around 60 percent
more than the value reached during the previous government (see Table 4).
To show the nature of the debt crisis during the last two governments, we need
to also consider the effects of the IMF policies on the internal economy as whole.
The average rate of growth during Morales-Bermudez's administration was only
about 1.9 percent, and it fell to -0.5 percent in 1981-84, to mention only one of
the more important economic indicators. Hence, the striking moral-for the
monetarists-that can be derived from the analysis so far is that free-market
policies cannot solve the persistent crisis of the balance-of-payments. The con-
tinuous dismantling of the machinery of state economic intervention only exacer-
bated the financial squeeze on the economy. This was reflected in the spectacular
rates of growth of external debt reached during the last two administrations.
Trade liberalization increased the vulnerability of the economy to import penetra-
tion; this brought about de-industrialization, which further increased the income
elasticity of demand for imports. The lack of control over the outflow of repatriat-
ed profits and over the inflow of foreign direct investment did not bring any relief
for the financial distress of the economy. Furthermore, the austerity measures
alleviated this distress in an irrational way-by deadening the economy and,
hence, worsening the financial misery of the country.
So far, it is clear that the Peruvian debt picture was overwhelmingly compli-
cated by external conditions in the last ten years. With world demand, terms of
trade, and interest rates growing at their trend rates, the net flow of foreign
capital would have been equal to -$4,100 million between 1981and1984. If this
hypothetical situation had occurred, could we argue that the free-market policies
would have resolved the current account crisis in a medium- or long-run perspec-
tive as Baker's plan assumes? In other words, could the economy reach a sus-
tained and stable rate of growth without incurring an increasing external debt, by
introducing free-market policies and assuming a favorable international scenar-
io? This question is answered in the following sections.
The Contribution of External Borrowing

to the Rate of Economic Growth
The discussion so far shows that for the Peruvian economy, the major constraint
on the rate of growth of output is the current account of its balance of payments.
The persistent tendency to deficit on current account sets the limit to the growth
of domestic demand, 20 a limit which certainly need not be compatible with the
needs of the economy. But this tendency is just a consequence of a mode of
accumulation which reproduces the non-integrated industrial structure. Hence,
there is a bias toward a diversified production of consumer-durables with its
concomitant income distribution. The absent, or scarcely developed, domestic
sector of capital goods and industrial inputs makes the process of accumulation
highly dependent on imports.
By the same token, the demand effect of public and private investment is
exported and consequently, the full internal operation of Verdoorn' s and Kaldor' s
law-growth of productivity and capital accumulation endogenous to the eco-
nomic system and dependent on final demand 21 -is weakened. The multiplier
effect of investment is lost at the same time that it generates an increase in imports
of capital goods and their corresponding inputs. Therefore, the tendency to
external disequilibrium is associated with the increasing import bill generated by
the basic process of accumulation in the domestic economy. So a rise in growth
has to be associated with an increase in pressure on the external balance. Without
public expenditure (not directed to generate capacity), the effect of investment on
profit will be nil although it enlarges production capacity. Therefore, the accumu-
lation process also creates a tendency to excess capacity, eventually creating

downward pressure on the rate of profit. 22
At the beginning of the import-substitution-led industrialization of the fifties,
the failure of investment to create additional effective demand for domestic
products was offset by the creation of new industries directed to produce goods
which were previously imported. This transformational growth ended as private
capital developed a bias toward the production of consumer durables between
1960 and 1966. Consequently, industrial development failed to integrate the
productive structure and to eliminate dependence on foreign technology and
equipment. Since the mid-sixties, the increasing public deficit offset the rising
tendency to export effective demand, but at the cost of permitting a growing
deficit on current account and thereby creating an increasing flow of foreign
borrowing. The root of the financial distress is, therefore, structural: the pres-
ence of a nonintegrated input-output system with a new dynamic sector-manu-
facturing industry-which is increasingly dependent on imported inputs and
capital goods. 23
Under these circumstances, expansionary policies are both self-defeating and
increasingly difficult as this spurious process of industrialization unfolds. The
relative contribution of foreign financial capital to economic growth will de-
crease partly because of austerity policies to overcome potential crises in the
balance of payments and partly because free-market trade policies will increase
the income elasticity of demand for imports. Open international competition will
force the economy to return to pre-industrial structures-as in Chile-but now
with higher elasticities due both to higher levels of consumption and to the more
diversified composition of consumption originated by the industrialization pro-
cess.
To test these propositions quantitatively, we divide the actual rate of growth as
follows: one part attributable to the rate of growth of exports, and the other
attributable to the financial flows of foreign capital. The relative price move-
ments (for example, the terms of trade effect) are dealt with as a residual factor
which also affects the rate of economic growth. A methodology based on a
balance of payments identity 24 is followed to analyze the contribution of foreign
capital to economic growth during the last three administrations. With an initial
current account deficit, the balance of payments can be expressed as:
where: Xt is the level of exports at constant prices in the year t

Pxt is the domestic price of exports
Mt is the level of import at constant prices
Pmt is the price of imports expressed in domestic currency
Ct is the value of capital flows measured in domestic currency
Taking rates of change of the variables in the previous equation we obtain:
(4) (E/R)(pxt + Xt) + (C/R)(ct) = pmt + mt

where the lower case letters represent rates of growth of the variables, and E/R
and CIR represent the share of exports and the capital flows in the total inflow of
international reserves or the total import bill financed by export earnings and
foreign borrowing.
Assuming a multiplicative import function and taking rates of change of its
variables,
(5)
where: mt is the rate of growth of imports

w is the price elasticity of demand for imports (w < 0)
pmt is the rate of growth of import prices
pxt is the rate of growth of export prices
n is the income elasticity of demand for imports (n > 0)
yt is the rate of growth of domestic income
Substituting equation (5) into (4), we get the balance of payments constrained
growth rate:
(6) ybt*= [w(pxt - pmt) + (pxt - pmt)

+ (E/R)xt + (C/R)(ct - pxt)]/n
where: [w(pxt - pmt)]/n measures the import volume effect of relative price
changes on income growth
[pxt - pmt]/n gives the terms of trade effect
[(E/R)xt]/n stands for the export growth effect
[(C/R)(ct - pxt)]/n gives the effect of the rate of growth of foreign
capital flows.
Assuming a residual price effect, the balance of payments constrained rate of

growth can be written as follows:
(7) ybt = [(E/R)xt + (C/R)(ct - pxt)]/n

Hence, the difference between the actual rate of growth of output and that
estimated by (7) becomes a measure of the terms of trade effect and any import
volume response from relative price changes.
In order to obtain ybt, first, the corresponding import functions were estimat-
ed-the results are shown in Table 8. Second, in the export variable, not only
Table 5
Relative Contribution to the Rate of Economic Growth
Actual Estimated Foreign

Growth Growth Financial
Regime Rate Rate Exports Capital
Velasco
1969-75 4.6 2.2 -2.3 4.5
(1973-75) (5.4) (3.7) (-5.5) (9.2)
Morales-
Bermudez
1976-80 1.9 -1.6 1.1 -2.7
Belaunde
1981-84 -0.5 0.5 -0.2 0.7
Note: The income elasticities of demand for imports of goods and non-factor services were 1.90 for
1969-75 and 5.98 for 1976-84.
exports of goods are included but also exports of non-factor services. Finally, the
variable C, which represents the foreign capital flows, includes the net inflow of
medium- and long-term loans plus the net foreign direct investment, minus the
net interest payments and repatriated profits adjusted for transfers.
Table 5 gives data on the actual growth rate and the growth rate predicted by
the model (7). The latter was decomposed into two parts: one associated with the
growth of export volume and the other connected with the growth of real foreign
capital.
Three important conclusions can be derived from the results of the estimated
model. First, the potential growth of GDP attributable to the growth of exports of
goods and non-factor services increased during the last two administrations, but
its significance was reduced by the spectacular increase in the income elasticity of
demand for imports. It is worth pointing out that the economy reached the highest
rate of growth during Velasco's government in spite of the sharpest drop in the
rate of growth of its exports. The Baker plan encourages export-led growth at the
expense of social programs, although adopting export promotion policies within
a free-trade framework implies an immense social cost, particularly when the tiny
contribution of exports to economic growth during the last two liberal administra-
tions is considered. Even in the best international scenario, from 1977 to 1980,
the contribution of exports to growth was significantly less than the rate of growth
of the labor force. It is not possible to consider the various measures of social cost
here; the point simply is that neither export promotion nor free-market policies,
let alone their combination (for they work at cross purposes), provide a path to
sustained growth.
The second important conclusion refers to the contribution of foreign capital.
Table 6
The Current-account-equilibrium Rate of Growth and Balance-of-payments

Constrained Rate of Growth
Actual Actual Estimated Estimated

Rate of Rate of Growth Rate Growth Rate
Growth Growth by Harrod Including
of of Trade Fore~n
Regime Output Export Multiplier Capital lows
Velasco
1969-75 4.6 -4.6 -2.4 2.2
(1973-75) (5.4) (-10.6) -5.6 3.7
Morales-
Bermudez
1976-80 1.9 6.9 1.2 -1.6
Belaunde
1981-84 -0.5 -1.3 -0.2 0.5
Note: The estimated rate of growth of output by Harrod Trade Multiplier is the ratio of exports
growth to the corresponding income elasticity of demand for imports.
External saving enabled the economy to grow rapidly during the expansionary
period of 1969 to 1975. However, during the free-market administrations, its
contribution decreased tremendously: from 1981 to 1984 when the largest in-
crease in the debt took place, the contribution of foreign capital was only 16
percent of its 1969-75 contribution. In Morales-Bermudez's period, the positive
rate of growth of output was exclusively linked to export growth.
The third conclusion concerns the impact of the terms of trade effect on the
rate of economic growth. The difference between the actual and the predicted rate
of growth was not significantly explained by the positive rate of growth of the
terms of trade during Velasco's administration. But when the open-door policy
was introduced, the difference between the two rates was almost perfectly corre-
lated to the changes in the terms of trade.
As may be noticed, the balance-of-payments constrained growth rates do not
provide a good prediction of the actual growth rates. As previously mentioned,
the difference between them is basically a measure of the extent to which the
"constrained" growth rates have been affected by relative price movements in
international trade. However, two qualifications are necessary. On the one hand,
only during Velasco's period did the balance-of-payments constrained growth
rates give a good prediction of the actual growth rates_ The foreign capital inflow
allowed growth to be faster than otherwise would have been the case. The
contribution of exports was significantly negative (see Table 6). On the other, the
relatively unimportant role played by foreign capital loans during the last two
"liberal" administrations is correlated with the emphasis that they put on equili-
Table 7
Components of the Difference between the Actual Growth Rate and the
Estimated Growth Rate by Harrod Trade Multiplier
Relative Price Effect

Regime Yt"Yht ybt·Yht Total Terms of Trade
Velasco
1969-75 7.0 4.6 2.4 1.0
(1973-75) (11.0) 9.3 1. 7 1.7
Morales-
Bermudez
1976-80 0.7 -2.8 3.5 2.0
Belaunde
1981-84 -0.3 0.7 -1.0 -1.0
Notes: y t is the actual rate of growth of output, yht is the rate of growth estimated by Harrod Trade
Multiplier, and ybt is the balance of paym..:nts constrained rate of growth which includes foreign
capital flows. The total price effect and the terms of trade effect are adjusted by the income elasticity
of demand for imports.
brating the current account of the balance-of-payments, one of the basic targets of
the IMF policy prescriptions. Only during these two governments did the rate of
growth estimated by Harrod trade multiplier give a good prediction of the actual
growth rate, as shown in Table 6. To better understand this, note that when the
current account equilibrium is a policy objective, then:
(8) .:iYt
.:i Xt m
where Y is the level of output, X is the level of exports, and m is the marginal
propensity to import. The inverse of this propensity is the so-called Harrod Trade
Multiplier. When this equation is made "dynamic" and it is assumed that exports
(X) are equal to imports (M), we obtain:
(9)
where yht is the estimated rate of growth by the Harrod Trade Multiplier, xt is the
rate of growth of exports, and n is the income elasticity of demand for
imports. 25
The previous two qualifications can be assessed by looking at Table 7. As
pointed out before, real capital inflows growing faster than exports (4.6 percent)
Table 8
Import Demand Functions

[log Mt =
log a + w log(Pmt/Pxt) + n log GDPtl
Con- Elasticities
Period stant Price(w) lncome(n) R F SSA
1968-82 -64.2 -0.302 5.770 0.83 15.8 0.102

(-3.31) (-1.955) (3.918)
1968-75 -13.5 -0.225 1.901 0.91 10.6 0.014
(-1.20) (-1.260) (2.184)
1976-82 -67.1 -0.433 5.981 0.87 6.50 0.046
(-1.56) (-1.043) (1.833)
Notes: The import variable incorporates goods and non-factor services. Bracketed terms under the
coefficients are t ratios. All the regressions were corrected for autocorrelation. SSR is the sum of
squared residuals which gives us a Chow's statistic equal to 1. 87. The end-year of each period is a
peak-cycle; this allows us to not understimate the import functions.
Source: Instituto Nacional de Estadistica.
only confirms that the balance of payments equilibrium was not a policy target
during the Velasco government. By the same token, the terms of trade effect,
though positive, played an insignificant role in the economic performance of that
period. Furthermore, reflecting the industrialist strategy based on foreign tech-
nology and capital goods, the import volume effect due to the price elasticity of
demand for imports was "perverse."
This pattern is drastically modified during the latter two administrations.
Between 1976 and 1980, the rate of growth of output was linked almost exclusive-
ly to the positive and significant rate of growth of exports and the favorable terms
of trade. The real capital inflows grew more slowly than exports ( -2.8 percent as
seen in Table 7).
During the Belaunde Administration, the rate of growth of output was again
associated with export growth and the terms of trade effect. In the first free-trade
administration, the export promotion program (intended to ease the financial
distress of the economy) succeeded, but only because of the existence of favorable
external markets. The gloomy international scenario of later years proved the
unfeasibility of that model conceived, as in the Baker plan, within a free-market
policy framework.
At this point it is appropriate to assess the Baker free-trade policy proposals.
Free trade has conclusively failed in the case of the Peruvian economy. As shown
in Table 8, the external structural disequilibrium brought on by the spurious
process of import substitution, 26 was exacerbated first by the strategy of export
promotion but also by the significant increase in the income elasticity of demand
for imports due to the open-door policy imposed by the IMF. With the increased
dependence of the economy on imports, the external disequilibrium worsened

between 1976 and 1984, contrary to what the official economists had expected.
An economy with a structural external disequilibrium requires a permanent
growth of real foreign capital flows to maintain a given positive rate of growth of
output. If the rate of growth of capital inflows becomes zero, the rate of growth of
output will be less than ybt. To maintain a rate of growth equal to what would be
possible by considering only the Harrod Trade Multiplier, the real capital flows
would have to increase at the same rate as exports. In general, to maintain a given
rate of growth of output, say ybr. (ct - pxt) would have to be equal to Xt. If xt
becomes zero or less than zero, (ct - pxt) would have to increase correspond-
ingly to compensate for the drop in exports. 27 If the implicit assumption of
constant terms of trade is dropped, an adverse international scenario to export
and of export prices would imply a greater increase in the rate of growth of real
foreign capital to maintain an initial positive rate of growth of output. For
example, to grow at the average 1969-75 rate of growth (i.e., 4.6 percent) during
Belaunde' s period, the rate of growth of real capital inflow would have had to be
equal to 34.7 percent, given an income elasticity of imports equal to 5.98, a
negative rate of growth of exports equal to -1. 3, and an average terms of trade
deterioration equal to -6.0 percent. However, as mentioned before, the weighted
average rate of growth of real capital inflows was only equal to 4.3 percent.
Therefore, it is not difficult to figure out that continuous borrowing within an
adverse international scenario has to cause an explosive increase in the debt/GDP
ratio and, by the same token, in the debt/service ratio. This will be so even with a
constant real rate of interest, provided it is greater than the rate of growth of
exports. The situation will be worse, of course, when the adverse international
scenario is accompanied by growing real rates of interest. 28 Moreover, in Peru,
the structural lack of financial capacity of the manufacturing sector created a
constant inflow of foreign financial capital. This, in turn, kept real national
income penrtanently below its potential level due to the constant outflow of
capital for interest payments and repatriated profits. 29
* * *
The facts analyzed and evaluated in this chapter show an economy entrapped
by international financial capital and without enough autonomy to pursue fiscal
and monetary policies. The IMF backed the commercial banks' demands for full
repayment by imposing free-market policies as a condition to get short-term
financial relief in the balance of payments. Its adjustment programs have indeed
contributed to the world recession by cutting the debtor countries' current ac-
count deficit. Baker's proposal to deregulate and free up the Third World econo-
mies is linked to this effect.
The difference between the IMF programs and the Baker plan does not center
on the nature of the policy prescriptions. The issue now is the direction of the new
loans. While the IMF made balance of payments loans on the condition that
countries accept the austerity package, Baker's plan contemplates the World Bank
making loans with the exclusive purpose of raising efficiency in trade and indus-
try. Efficiency in the free-market ideology means trade liberalization, a reduced
role of government, elimination of state-owned companies, and so on. So, the
new loans will be to assist these objectives. As with IMF conditionality, the
commercial banks would provide new financing after the countries had reached
lending agreements with the World Bank. But the borrowing terms will be
harder: only countries that commit themselves to free-market-oriented economic
reforms would get the help.
The Baker plan for a free-market economy doesn't contemplate, of course, an
economic restructuring based on the development of the internal market together
with the construction of a domestic capital goods sector. 30 A structural change of
this kind would mean developing an integrated national input-output system by
selectively controlling the external sector and reinforcing the economic role of
the state. Baker calls for just the opposite: divestment of public enterprises, lower
taxes and elimination of industrial subsidies, de-regulation, and trade liberaliza-
tion.
The Baker plan resolves the conflict between the Latin American states and
international capital in favor of the latter. It doesn't introduce structural changes
into the present organization of international financial capital. Belying its free
market ideology, the plan doesn't even allow the market to resolve the crisis.
From the market point of view, international commercial banks share responsibil-
ity for debt crises, since they seem to have made mistakes about the risks they
were taking when they financed the increasing expenditures of public sectors.
Thus, creditors should simply sell their claims in the free market and take the
losses. Why should the United States' government help them out? Free market
advocates have no business calling on the government to bail them out of their
mistakes. 31
Our analysis also suggests the existence of domestic conflicts associated with
the crises in the model of capital accumulation. As observed, continuous financial
capital inflow is a direct result of a manufacturing sector highly dependent on
imports. In this context, the external factors (sluggish demand for exports or
increases in real interest rates) only accelerate financial distress. The current
crisis has spurred internal conflicts between the export-led growth advocates and
the interests tied to the current import substitution regime, between the support-
ers of a national re-industrialization and capital linked to the manufacturing
''white line'' production, between the pro-agricultural strategy to re-industrialize
the economy and the native imitators of industrialization in the central countries,
between interventionists and free-market supporters, between the followers of
the statism and the participationists, etc. And finally, all the national economic
interests are confronted with international financial capital.
The development of the Peruvian external debt has revealed, over the last
fifteen years, some of the historical dynamics of these conflicts. The external debt
rocketed during the expansionary period of Velasco' s government. The "liberal"
(conservative) policies applied from 1976 to 1980 to deal with the consequent
financial crisis worsened the structural problem. The priority given to servicing
the debt decreased the rate of growth of the net capital inflow (medium- and long-
term debt) and hence put a brake on economic growth. The positive-but insig-
nificant-average rate of growth reached in this period was closely associated
with the export sectors which had the benefit of a favorable international environ-
ment. But the vulnerability of the economy to international competition and the
financial pressures on monetary authorities increased.
As occurred in 1976-80, the net inflow of the total debt due to internal factors
during the Belaunde administration did not increase significantly. But total debt
soared again, this time because of growing real interest rates and a depressed
international scenario which drastically undercut export earnings. All export-
promoting policies of this period proved fruitless, and the vulnerability of the
economy increased.
Notes
1. The Keynesian aproach to devaluation-the elasticity approach-can be trans-

formed easily into the monetarist. The monetarist approach incorporates the role of real
balance effect with money treated as a capital asset. See Dornbusch (1976).
2. An economist of Financial World comments, "Baker's approach reflects a more
sophisticated sense of macroeconomics than the administration's prior statements on the
debt crisis did. Having recently moved to devalue the dollar in concert with the West's
biggest central banks . . . the Treasury Secretary seems to be trying to ensure that nations
like Brazil the world's eighth largest economic entity, will be capable of importing some of
the United States' prosperity less expensive exports .... And an increase in American
exports might help forestall the drive for the sort of protectionist tariffs that also contribut-
ed to the Great Depression." See Peter Hall (1985).
3. For example, the average maturity of new medium- and long-term loans decreased
from IO years in 1973 to five years in 1975. See I. Mehdi Zaidi (1985).
4. In 1981, this ratio was equal to 271.5 for Latin-American countries and 77.0 for
East Asian countries. The higher ratio of debt to exports is, according to Jeffrey Sachs
( 1985), the critical factor in making Latin America so vulnerable to the external shocks of
the early 1980s.
5. The methodology used in this and in the following section as well as the results
were taken from Jimenez (1987). See also Jimenez and Schatan (1983).
6. The expected values of import and export prices are derived from their trend rates
of growth estimated by regressing their natural logs on time for 1960-72, 1960-75, and
1960-80. The rate of growth associated with the first regression was applied to obtain the
expected import and export prices for Velasco's administration; the second one to obtain
similar expected values for Morales-Bermudez's period; and the third one for the adminis-
tration of Belaunde Terry.
7. The assumption of expected values equal to zero is not an accurate reflection of
correct policy in relation to the corresponding economic variables. However, by assuming
FI* = 0, we are only hoping that we can avoid an outflow of capital greater than the net
inflow of foreign direct investment. On the other hand, by assuming SE* and RA* equal to
zero, we are presupposing the absence of speculative capital outflow and that the monetary
authorities are not vulnerable to international financial pressures.
8. The expected values of Wand DD were estimated by their corresponding trend
rates of growth observed during 1960-72, 1960-75, and 1960-80. The trend rates were
obtained by regressing the natural log of Wand DD on time for the same periods. The set
of expected values associated with the regressions for the first period was used to analyze
the Velasco administration; the set corresponding to the regressions for the second period
was used to examine Morales-Bermudez's government; and the third set of expected
values was used to analyze Belaunde's administration.
9. Kate and Wallace (1980) introduced this suggestive name, but they estimate it in a
different way.
IO. When private short-term capital movements become part of the balance of pay-
ments deficit or surplus, the related concept of equilibrium, as Thirlwall (1980) points out,
gives a measure of the financial pressure on the monetary authorities to maintain the
external value of the domestic currency.
11. See Angell and Thorp (1980), pp. 865-86.
12. The argument against industrialization based on the internal market doesn't take
into account the fact that the capacity of this market to drive and sustain economic growth
depends upon the social and institutional forces which are embodied in the specific
industrial strategy. For example, a growing purchasing power of the peasant economy can
indeed be important as a motive of industrialization. Gerschenkron (1966, 114) assessed
the importance of the internal market referring to the substitutes for the demand of the
peasantry. He says, "The internal market as represented by the peasantry can become
unimportant for industrialization . . . if someone else, say the state, is willing and able to
engender industrial production and at the same time constitute the market for the goods
produced. This is one of the substitution patterns that occur in the industrialization of
backward countries. Besides, the demand of the state is not the only possible substitution
of this kind .... Such substitutions often are very likely to take place, but there is no
assurance that they actually will. If they do not, the rate of industrial growth is likely to
suffer, but the rate that is maintained will surely depend on the existence of the demand
emanating from the internal market."
13. See lnteramerican Develoment Bank (1979, 1980, 1982).
14. "Throughout the 1970s, direct foreign investment flows to developing countries
were on the order of $IO billion annually; net external borrowing was six to eight times as
large. At current rates, interest obligations are more than $50 billion per year. It is
conceivable that direct investment could rise enough to meet these interest costs." See
Eaton and Taylor (1985).
15. The typical interest rate on developing country loans-LIBOR (London interbank
offered rate) plus a spread of l percent-increased from 7. 0 percent in 1977 to 9. 7 percent
in 1978, 13.0 percent in 1979, 15.4 percent in 1980, and it has fluctuated around this level
since then. "Borrowers became accustomed to low real interest rates in the 1970s. For
1961-70, LIBOR on U.S. dollar deposits minus the U.S. wholesale price increase pro-
duced an average real interest rate of 4.1 percent. But for 1971-80, this average was -0.8
percent; real interest rates were negative, on average, for the decade. By 1979 and 1980,
nominal interest rates were high (LIBOR averaged 13 .2 percent) and although U .S.infla-
tion was virtually equal to LIBOR, high nominal rates caused a cash flow squeeze for
borrowers. . . . By 1981-82, declining inflation without a corresponding decline in inter-
est rates meant high real interest rates (7 .5 percent in 1981 and 11.0 percent in 1982),
making matters worse." See Cline (1983), pp. 18-23.
16. This type of policy reaction to financial pressures seems to have been common in
the majority of developing countries just when the external debt was increasing exponen-
tially. In this regard, Dornbusch ( 1980) comments: '' ... the striking fact is that rather than
draw down their reserves to finance their deficits, the developing countries have borrowed
abroad to add to their reserves at a rate of nearly $12 billion over 1976-78," pp. 18-19.
17. See Banco Central de Reserva de! Peru (1983).
18. The World Bank emphasizes the fact that trade liberalization, the exchange rate
and the interest rate policies were in consonance with its position expressed in its report of
April, 1981. See World Bank (1981), p. 26.
19. To see the negative impact ofliberalization on the manufacturing sector, consider
some figures: in 1981, $52.6 million was spent on imported TVs, $370.6 million on
automobiles and automotive vehicles, $3.4 million on footwear products, and $5.5 million
on clothing. Of course, in all these industries the domestic product decreased due to
international competition. A dramatic example of de-industrialization is the case of the
public steel enterprise SIDERPERU, whose capacity utilization fell from 75 percent in
1980 to 50 percent in 1982, while the economy imported 240 tons of competitive products
in 1981. See lguiniz (1984).
20. See Jimenez (1984).
22. See Casar and Ros (1983).
23. See Jimenez (1985). For an interesting analysis of Latin American industrializa-
tion see Fajnzylber (1983).
24. We follow the methodology suggested by Thirlwall and Hussain (1983).
25. See Harrod (1933) and Thirlwall and Hussain (1983).
27. See A. P. Thirlwall and M. N. Hussain (1983), pp. 503-4.
28. In general, when income and export growth depend on the growth rate of world
demand, •• ... the slower is the growth of world demand, the greater the debt and the
higher the debt/GDP ratio; further worse terms of trade cause a higher import propensity
which raises the debt/GDP ratio." See Spaventa (1983), pp. 330-1.
29. For an explanation of the behavior of the real national income and import capacity,
see Jimenez (1987).
30. Thirty five years ago, the ECLA started out its theoretical work to show that the
de-regulated or free-market economy would only reproduce the structural problems and
limit the growth capacity of Latin American countries. Thirty years later, in 1979, Raul
Prebish (former conductor of ECLA), would say it is surprising to still find in our
continent advocates of the free-market ideology. They think, he said, that by opening up
the economy, the central countries, impressed by their devotion to the neo-classical
principles, will receive generously all of that they export or want to export. They have lost,
he added, the most elemental political realism. See Prebish (1980).
31. This solution was suggested by Wiesner ( 1985).
References
Angell, A., and R. Thorp. 1980. "Inflation, Stabilization and Attempted Redemocratiza-
tion in Peru 1975-1979." World Development 8.
Banco Central de Reserva de! Peru. 1983. El Proceso de Liberalization de lmportaciones:
Peru 1979-1982. Lima.
Casar, J. I., and J. Ros. 1983. "Trade and Capital Accumulation in a Process oflmport
Substitution." Cambridge Journal of Economics '7:257-67.
Cline, W. R. 1983. International Debt and the Stability of the World Economy. Washing-
ton: Institute for International Economics.
Dornbusch, R. 1980. Open Economy Macroeconomics. New York: Basic Books.
- - - - . 1976. "Devaluation, Money and Non-traded Goods." J.A. Frenkel and
H.G. Johnson, eds. The Monetary Approach to the Balance of Payments. London:
George Allen and Unwin Publishers, Ltd.

Eaton, J., and L. Taylor. 1985. Developing Country Finance and Debt. Mimeo.
Fajnzylber, F. 1983. La Industrializacion Trunca de America Latina. Mexico: Nueva
Imagen.
Gerschenkron, A. 1966. Economic Backwardness in Historical Perspective. Cambridge:
Harvard University Press.
Hall, P. 1985. "How Safe Are the Banks." Financial World, November.
Harrod, R. 1933. International Economics. New York: Cambridge University Press.
Iguiniz, J. 1984. La Crisis Peruana Actual: esquema para una Interpretation. Mimeo.
Interamerican Development Bank. 1979. Economic and Social Progress in Latin America.
Washington, D.C.
- - - - . 1980 and 1982. Economic and Social Progress in Latin America, the Exter-
nal Sector. Washington, D.C.
Jimenez, F. 1982. "Peru: la expansion de! sector manufacturero como generadora de
crecimiento economico y el papel de! sector externo." Socialismo y Participacion,
No. 18.
- - - - . 1984. "La balanza de pagos como factor limitativo de! crecimiento y el
desequilibrio estructural externo de la economia Peruana." Socialismo y Participa-
cion, No. 25.
- - - - . 1985. Peru: inflacion, deficit publico, desequilibrio externo y crecimiento
economico. Una critica al enfoque monetarista. Mimeo.
- - - - . 1987. Capital Accumulation, the State and Effective Demand: A Non-
Neoclassical Structuralist Approach to Peruvian Development 1950-1984. Ph.D. diss.
Jimenez, F., and C. Schatan. 1983. "Mexico: la nueva politica comercial y el incremento
de las importaciones de bienes manufacturados en el periodo 1977-1980." Aspectos
Metodologicos para el Analisis de/ Sector Externo, 2.
Kate, A. T., and R. B. Wallace. 1980. Protection and Economic Development in Mexico.
New York: St. Martin's Press.
Prebish, R. 1980. "Prologo." Octavio Rodriguez, La Teoria de/ Subdesarrollo de la
CEPAL, Siglo XX/.
Sachs, J. 1985. "External Debt and Macroeconomic Performance in Latin America and
East Asia." Brookings Papers on Economic Activity 2:523-73.
Spaventa, L. 1983. "Risks to the Stability of the International Financial System: Gloom
without Drama." International Lending in a Fragile World Economy. D.F. Fair and R.
Bertrand, eds., Hague: Martinus Nijhoff Publishers.
Thirlwall, A. P. 1980. Balance of Payments Theory and the United Kingdom Experience.
London: The Macmillan Press Ltd.
Thirlwall, A. P., and M. N. Hussain. 1983. "The Balance of Payments Constraint,
Capital Flows and Growth Rate Differences between Developing Countries." Oxford
Economic Papers 34 (3).
Wiesner, E. 1985. "Latin America Debt: Lessons and Pending Issues." The American
Economic Review. Papers and Proceedings (May).
World Bank. 1981. Peru: Principales Cuestiones y Recomendaciones en Materia de
Desarrollo. Washington, D.C.
Zaidi, I. M. 1985. "Saving, Investment, Fiscal Deficits, and the External Indebtedness of
Developing Countries." World Development(13)5: 573-88.

Book - Financial Dynamics and Business Cycles

Uploaded by

Copyright:

Available Formats

Book - Financial Dynamics and Business Cycles

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

Book - Financial Dynamics and Business Cycles

Uploaded by

Copyright:

Available Formats

What is the book about based on the introduction and table of contents?

What is the book about based on the introduction and table of contents?

What factors influence financial instability according to the chapter titles?

What factors influence financial instability according to the chapter titles?

Fi n a n ci al D y n a mi c s

LONDON AND NEW YORK

No part of this book may be reprinted or reproduced or utilised in any form or by

Product or corporate names may be trademarks or registered trademarks, and are

Library of Congress Cataloging-in-Publication Data

Financial dynamics and business cycles : new perspectives / edited by

HB3722.F56 1989 88-30857

ISBN 13: 978-0-87332-531-8 (hbk)

Finance, Instability, and Cycles

Debt-Financing of Firms, Stability, and Cycles in a

Accumulation, Finance, and Effective Demand

The Real and Financial Determinants of Stability:

Comparative Monetary and Fiscal Policy Dynamics

Monetary Stabilization Policy in a Keynes-Goodwin

Qualitative Effects of Monetary Policy

Debt Commitments and Aggregate Demand:

The Cyclical Behavior of Corporate Financial Ratios

Theories of Financial Crises

The Political Economy of the External Debt and Growth:

ity, the basic message of Keynes-that the proper subject of economics is a

The General Theme

a critique of neoclassical policy postulates. The chapter by Albin, commencing

Linear and Nonlinear Dynamical Models

their equilibrium solutions) are adequate representations of economic dynamics.

Dynamical Models of Macrofluctuations

studied in a framework of accelerator-multiplier models with money and finance

Economics and Mathematical Systems. Heidelberg and New York: Springer-Verlag.

(2) PX - wbX rwbX T X

(5) Investment demand = PI = [g 0 + h(r + e - i)]PK,

where 8o is a constant reflecting autonomous capital stock growth, and the

(6) Saving supply = srPK = srwbX.

(7) 80 + h(r +e- i) - sr = 0.

form for the capital stock growth rate g ( = II K) as

(8) s[go + h(e - i)]

A fall in the interest rate or an increase in anticipated profits leads to a higher

Simplified Balance Sheets for Firms and Rentiers

Their wealth increases from capital gains and financial saving.

(13) µ(i,r + e)W - M = O,

(15) - (3(i,r + e)W + B = 0,

where µ + ~ + (3 = 1. Only two of these three equations are independent. As

(17) Pe = W(l - ~)](FIE);

(18) µ,(i,r + e) = a[l - w,r + e)],

where a = Ml F is the share of fiscal debt issued as money.

(19) 71;di + 71,dr = -11,de + (1 - ~)da,

1/; = µ,; + a~;

Figure 1. Responses of the Interest Rate and Profit Rate to an Increase

so that for a fixed money growth rate M, ci falls as g increases.

Whenever profits decreased hedge units became speculative and speculative

Profits = Investment + Government deficit - Current account deficit.

1. Using Pk as a shadow price for investment decisions, of course, resembles Tobin's

Atkinson, A. B. 1975. The Economics of Inequality. Oxford: Clarendon Press.

An Economy with Concentrated Ownership of

The economy to be considered here is a stationary infinite horizon economy with

where Ct is consumption in period t and nt is labor supply. Here 'Y is a discount

(i) U I > 0, U II < 0, VI > 0, VII > 0,

(16) kt+2 - k] = [a/1 - /1E] rkt+l -kl

nv '(n) + n* 2v "(n*)

{3'(µ)E(µ,) + {3(µ)E'(µ,) =;; 0.

e; = oe(r + e, i)loi < oi:(>., e, r, i)loi.

e, = e,(r* + e' i)-reactions of households in their share holdings to

Lemma 7. With respect to>..= A., e = e = 0, r = R(>.., e,), i = J(>.., e),

Theorem5. SupposethatAssumptions 1-6holdandthatUA(A, e) > 0, Ue(A*,