Solow

The Solow Model
Siddhartha Chattopadhyay
Department of HSS
IIT KGP
Spring, 2022-23
Chattopadhyay (IIT KGP) Solow Spring, 2022-23 1 / 71

Solow Model with Population Growth
Motivation
Considerable di¤erence in the income level among countries

across the world. Why?

Motivation

Motivation
Considerable Di¤erence in the growth rate among countries.

Why?

Motivation

Motivation

Motivation

Motivation
Conditional Convergence: Growth rate of the rich countries are

lower than the poor countries. Why?

Motivation
Income and Welfare: Positive relationship between income and

consumption

Motivation
Income and Welfare: Positive relationship between income and life

expectancy at birth

Kaldor Facts (1963): Long-run
Percapita income grows over time and its growth rate does not falls
Physical capital grows over time
The rate of return to capital, share of capital and labour in national
income and capital to output ratio nearly constant
Growth rate of percapita output di¤ers substantially across countries

Objective
Solow (1956) develops an analytical model to explain stylized facts

discussed above

Assumptions
Time: Continuos
Market: Perfectly Competitive
Technology: Constant Returns to Scale (CRS) with two factors of
Production - capital (K ) and labour (N ). Output is denoted by Y .
The production function follows diminishing marginal productivity
with respect to capital and labour. We will con…ne our analysis to the
Cobb-Douglas production function given below.
Y (t ) = F (K (t ), N (t )) = K (t )α N (t )1 α
,0 < α < 1 (1)
Share of capital and labour in national income is constant and

denoted by α and (1 α) respectively.

Assumptions
Economic Agents: Firms and Households

Firms: maximize pro…t by choosing capital and labour. Price of
output is normalized to unity. Labour and capital are paid a real wage
(w ) and real rental (r ) ) they are paid in terms of …nal goods.
Households: a non-optimizing household saves a constant fraction,
0 < s < 1 of its total income. Household owns the …rm

Assumptions
Population: growing exponentially at an exogenously given rate

0<n<1
N (t )
= n
N (t )
N (t ) = N (0) e nt
Capital depreciates at the rate 0 < δ < 1

Percapita output and percapita capital are denoted by y and k and
de…ned as,
Y K
y = ,k =
N N

Pro…t Maximization
Production function in percapita form: obtained by dividing both

sides of equation (1) by N (t )
y (t ) = f (k (t )) = k (t )α
Pro…t Maximization: CRS technology and perfect competition )
factors are paid in terms of their marginal products ) product
exhausts ) zero pro…t
Π (t ) = F (K (t ), N (t )) w (t ) N (t ) r (t ) K (t )
π (t ) = f (k (t )) w (t ) r (t ) k (t )
α
= k (t ) w (t ) r (t ) k (t )
0 1
k (t ) : r (t ) = f (k (t )) = αk (t )α
0
π (t ) = 0 ) w (t ) = f (k (t )) k (t ) f (k (t ))
= (1 α ) k (t ) α
The Fundamental Equation
Total income of household is spend in consumption and savings.

Percapita consumption and percapita savings are denoted by c (t )
and e
s (t )
y (t ) = c (t ) + e
s (t )
Investment:
. dK (t )
I (t ) = K (t ) + δK (t ) , K (t ) =
dt
.
K (t )
i (t ) = + δk (t )
N (t )
e
s (t ) = sy (t ) = sf (k (t ))
= sk (t )α

Goods market equilibrium: investment equals to savings
K (t )
+ δk (t ) = sf (k (t )) (2)
N (t )
K (t )
Note, k (t ) = N (t ) . Taking log both sides and di¤erentiating with
respect to time gives,
k (t ) K (t ) N (t ) K (t )
= = n
k (t ) K (t ) N (t ) K (t )
.
K (t )
k (t ) = nk (t ) (3)
N (t )

Substituting equation (2) to equation (3) yields the

fundamental equation of Solow model
k (t ) = sf (k (t )) (n + δ ) k (t )
α
= sk (t ) (n + δ )k (t ) (4)
We can solve for k (t ) = k (t ) from the fundamental equation and

consequently can solve
α
y (t ) = f k (t ) = k (t )
α
c (t ) = (1 s ) y (t ) = (1 s )f k (t ) = (1 s ) k (t )
α
e
s (t ) = sy (t ) = sf k (t ) = s k (t )
0 α
w (t ) = f k (t ) k (t ) f k (t ) = (1 α ) k (t )
0 α 1
r (t ) = f k (t ) = α k (t )
The Long-run
The long-run is technically known as the Balanced Growth Path

(BGP) in the literature where, capital and output grow at identical
rate. Note the steady state where, percapita capital stock is neither
growing nor decresing is also a BGP where,
0 1 ! 0 1 0 1
@ k (t ) A = y (t ) =0)@
K (t ) A
=@
Y (t ) A
=
N (t )
=n
k (t ) y (t ) K (t ) Y (t ) N (t )
ss ss ss ss

The Long-run
At steady state:
k (t ) = sf (k (t )) (n + δ ) k (t ) = 0
1
sf (kss ) s 1 α
) kss = = (5)
n+δ n+δ
α
s 1 α
) yss = f (kss ) = kssα = (6)
n+δ
α
s 1 α
) css = (1 s )f (kss ) = (1 s)
n+δ
α
s 1 α
) esss = sf (kss ) = s
n+δ
0
) wss = f (kss ) kss f (kss ) = (1 α) (kss )α
0 1
) rss = f (kss ) = α (kss )α
The Long-run

The Long-run

The Long-run

The Long-run

The Long-run

The Phase Diagram

The Long-run
Stability: strict concavity of the production function yields unique

and stable steady state.
k (0)1 < kss ) savings is greater than investment ) percapita capital
stock grows and movings towards steady state
k (0)2 > kss ) savings is less than investment ) percapita capital
stock shrinks and movings towards steady state

Savings rate rise

The Long-run
Higher savings rate increases long-run capital stock.

It in turn increases
long-run output, investment (savings) and consumption.
factor prices (price of capital and labour) in the long-run.

Population growth rises

The Long-run
Higher population growth rate reduces long-run capital stock.

It in turn reduces
long-run output, investment (savings) and consumption.
factor prices (price of capital and labour) in the long-run.

The Short-run Implication
Equation (4) gives,
k (t ) f (k (t ))
γk (t ) = =s (n + δ ) (7)
k (t ) k (t )
1
= s (k (t ))α (n + δ ) (8)
" #
1 α
kss
= (n + δ ) 1
k (t )
" 1 α #
y (t ) yss α
γy (t ) = = α (n + δ ) 1
y (t ) y (t )
Equation (7) shows

short-run growth rate depends on the level of capital stock and (n + δ).
short-run growth rate depends of the distance from the steady state
The Phase Diagram
f (k (t ))
s k (t )
is measured along the vertical axis

The Convergence
Solow (1956) predicts that the poor countries should grow faster than
the rich countries if they converge to the same steady state )
conditional convergence and not an absolute convergence.

No Absolute Convergence
Barro and Sala-i-Martin (2004): average annual growth rate of 114

countries across the world from 1960 to 2000.

Conditional Convergence

Barro and Sala-i-Martin (2004): average annual growth rate of 18

original OECD countries across (formed in 1961) from 1960 to 2000.

Madisson (2010):

Achievements and Problems
Data shows that some countries in the world are rich with
higher percapita income and some are poor with lower
percapita income.
Equation (6) shows that countries with higher savings rate and/or
lower population growth rate and/or lower rate of capital depreciation
rate have higher percapita income than others. Therefore, Solow
(1956) model gives answer to our …rst question - why some countries
are rich and some are poor in the world.
Conditional convergence with poor countries growing faster than rich
countries can be explained by Solow (1956)
Equation (6) shows no growth of percapita output in the
long-run. But data shows percapita income in the long-run is
not constant. Therefore, a modi…cation of the model is
required.

Solow Model with Population Growth and Technology
Growth
Labour Augmenting Technology
Production function with labour augmenting/Harrod neutral

technological progress
Y (t ) = K (t )α (A (t ) N (t ))1 α
(9)
Technology is growing exponentially at an exogenous rate
A (t )
0 < g < 1 ) A (t ) = g
Production function in terms of e¤ective labour force.
Y (t ) y (t )
yb (t ) = =
A (t ) N (t ) A (t )
α
= f kb (t ) = kb (t ) ,
K (t ) k (t )
kb(t ) = =
A (t ) N (t ) A (t )
Growth
Fundamental Equation
By equating savings with investment we get
K (t )
= sf kb (t ) δkb (t ) (10)
A (t ) N (t )
K (t )
Note, kb (t ) = A (t )N (t ) . Taking log both side and di¤erentiating with
respect to time gives,
.
.
K (t )
= kb (t ) + (g + n)kb (t ) (11)
A (t ) N (t )

Growth
Fundamental Equation
Substituting equation (11) to (10) gives,
kb (t ) = sf kb (t ) (n + δ + g ) kb (t )
α
= s kb (t ) (n + δ + g ) kb (t ) (12)
.
Steady State, kb (t ) = 0
sf kbss s
1
1 α
kbss = = (13)
n+δ+g n+δ+g
α
s 1 α
ybss = f kbss = (14)
n+δ+g

Growth
The Long-run
Equation (14) gives,

α
s 1 α
yss (t ) = A (t )
n+δ+g
! .
y (t ) A (t )
) = =g (15)
y (t ) A (t )
ss
Equation (15) shows that percapita income in the long-run is

not constant but grows at the rate of technological progress.

Solow Model with Population and Technology Growth
The Transitional Dynamics: Multiply kb (t ) α

to the both side on
equation (12)
kb (t ) kb (t ) α
=s (n + δ + g )kb (t )(1 α)
(16)
De…ne, x (t ) = kb (t )1 α
x (t ) kb(t )
= (1 α)
x (t ) kb(t )
x (t ) = (1 α) kb(t )kb(t ) α
(17)
Substituting equation (17) to (16) yields,

.
x (t ) = (1 α) s (1 α ) (n + δ + g )x (t ) (18)

Solution of equation (15): Integrating Factor

= e λt , λ = (1 α) (n + δ + g )
d
e λt x (t ) = (1 α) se λt
Z dt Z
d e λt x (t ) = (1 α) se λt dt
(1 α) s
x (t ) = + e λt m
λ
s s λt
x (t ) = + x (0) e (19)
(n + δ + g ) (n + δ + g )
s
m = x (0) (n + δ +g )

α
Substituting, x (t ) = kb (t )1 α
and yb(t ) = kb (t )α = x (t ) 1 α in
equation (19) gives,
s s (1 α)(n +δ+g )t
x (t ) = + x (0) e
n+δ+g n+δ+g
1
) + kb (0)1
1 α
kb (t ) = kbss1 α (1 e λt α
e λt
) + kb (0)1
1 α
k (t ) = kbss1 α (1 e λt α
e λt
A(t ) (20)
α
1 α 1 α 1 α
yb(t ) = ybssα (1 e λt
) + yb (0) α e λt
α
1 α 1 α 1 α
y (t ) = ybss (1
α
e λt
) + yb (0) α e λt
A(t ) (21)

Equation (21) shows percapita output and capital stocks grow

at the rate of technological progress along the BGP
After solving k (t ) and y (t ) from equations (20) and (21) we can
solve,
c (t ) = (1 s ) y (t ) , e
s (t ) = sy (t )
0
w (t ) = f (k (t )) k (t ) f (k (t )) = (1 α) (k (t ))α
0 1
r (t ) = f (k (t )) = α (k (t ))α

The Industrial Revolution (1760-1840)


The Growth Accounting
Solow (1957): equation (9) with B (t ) = A (t )1 α

gives,
Y (t ) K (t ) N (t ) B (t )
=α + (1 α) +
Y (t ) K (t ) N (t ) B (t )
B (t )
Solow Residual: B (t )
A (t ) 1 B (t )
Rate of growth of TFP: A (t )
= 1 α B (t )

The Growth Accounting

The Rate of Convergence
Equation (20) and (21) imply rate of convergence of the percapita

capital stock and percapita output towards their steady state at the
rate λ = (1 α) (n + δ + g ) > 0
Equation (20) and (21) also imply rate of growth of percapita capital
stock and percapita output in the long-run is g

Linearizing equation (12) around the steady state by Taylor series

expansion gives,
kb (t ) α 1
= s kb (t ) (n + δ + g )
kb (t )
(1 α) log (kb (t ))
= se (n + δ + g )
α) log (kbss )
= (1 α) se (1
d log kb (t )
(α 1 )
= (1 α) s kbss d log kb (t )
= (1 α) (n + δ + g ) d log kb (t ) (22)

Continued from the previous slide,

d
log kb (t ) = λd log kb (t )
dt
d
log (yb (t )) = λd log (yb (t )) (23)
dt
log (yb (t )) = 1 e λt
log (ybss ) + log (yb(0)) e λt
(24)
Equation (23) used to estimate the convergence rate, λ by Mankiw,

Romer and Weil (MRW) (1992).

Introduction of Human Capital, MRW (1992)
Consider the Production function:

Y (t ) = K α (t ) (A(t )N (t ))1 α
(25)
Divide both sides of equation (25) by Y (t )α and then by N (t ) gives,
α α
K (t ) 1 α s 1 α
y (t ) = A(t ) = A(t )
Y (t ) n+δ+a
at steady state
α α
log (yss ,i ) = a + log (si ) log (ni + δ + g ) + e(26)
i
1 α 1 α
log (Ass ,i ) = a + ei at the steady state
a: drift ) level of technology common for all countries, ei : random
error ) technology level that di¤ers from all the countries
MRW estimated α from equation (26) using data of di¤erent
countries
Problem: MRW (1992)
si : average annual investment to output ratio from 1965-1980

(δ + g ) = 5%
ni : average annual population growth from 1965-1980

Problem: MRW (1992)
Data α = 13 . Estimated α not matching data. Here, g is the

technology growth rate.

Problem: MRW (1992)
Introduced human capital in the production function
Y (t ) = K α (t ) H (t ) β (A(t )N (t ))1 α β
, 0 < α < 1, 0 < β < 1 (27)
A fraction sk and a fraction sh of total income is spent on
accumulating physical capital and human capital respectively. But
type of capital depreciate at the δ
kb (t ) = sk f kb (t ) , b
h (t ) (n + δ + g ) kb (t )
β
= sk kb (t )α b
h (t ) (n + δ + g ) kb (t ) (28)
b
h ( t ) = sh f kb (t ) , b
h (t ) (n + δ + g ) b
h (t )
β
= sh kb (t )α b
h (t ) (n + δ + g ) b
h (t ) (29)
Problem: MRW (1992)
Equation (28): given kb (t ) a rise in b

h (t ) increases kb (t ) and
vice-versa. Therefore, kb (t ) rises in the region above the kb (t ) = 0
line and falls in the region below the kb (t ) = 0 line
Equation (29): given b h (t ) a rise in kb (t ) increases b
h (t ) and
b
vice-versa. Therefore, h (t ) rises in the region to the right side of
b
h (t ) = 0 line and falls in the region to the left side of b
h (t ) = 0 line
Phase diagram shows it is globally stable system

Problem: MRW (1992)

Problem: MRW (1992)
Equations (28) and (29) at steady state: kb (t ) = b

h (t ) = 0 give,
1 1
1 β β
!1 α β !1 α β
sk s h skα sh1 α
kbss = ,b
hss =
n+δ+g n+δ+g
β
!1 1
α β
skα sh
ybss = (30)
(n + δ + g ) α + β

Problem: MRW (1992)
Using identical manipulation as done to derive equation (26) from

equation (25), we can derive equation (31) using equations (30) and
(27)
α β
log (yss ,i ) = a + log (ski ) + log (shi ) (31)
1 α β 1 α β
α+β
log (ni + δ + g ) + ei
1 α β
MRW estimated α and β from equation (31)

Problem: MRW (1992)
MRW got α and β that match with the data. It shows that the
augmented Solow model with human capital in the production
function matches the stylized facts.

Golden Rule Level of Capital
Is steady state capital stock maximizes the long-run consumption? A

social planner’s problem is to choose that golden rule level of capital
stock that maximizes consumption.
b
css = (1 s) f kbss = f kbss sf kbss (32)
= f kbss (n + δ + g )kbss
α
= kbss (n + δ + g ) kbss

Di¤erentiating Equation (32) with respect to kbss gives a golden rule

level of capital stock kbssg that maximizes long-run consumption
0
kbssg : f kbssg = (n + δ + g )
1
α 1 α
kbssg =
n+δ+g


Problem: Dynamic Ine¢ ciency
We have got,
1 1
s 1 α α 1 α
kbss = , kbssg =
n+δ+g n+δ+g
Steady state capital stock equals to golden rule level of capital stock
kbss = kbssg when s = α. In this case the allocation of the Solow
model is called dynamically e¢ cient. However, the savings rate (s ) in
Solow model is not optimally chosen. As a result, there is no
guarantee to achieve e¢ cient allocation in the Solow model.

Over accumulation of capital: kbss > kbssg ) s > α ) steady state

allocation of the Solow model is dynamically ine¢ cient. Reduction in
savings not only increases short-run consumption but it also increases
long-run consumption. In fact, any allocation belonging to Region B
of the next …gure is dynamically ine¢ cient.
Under accumulation of capital: kbss < kbssg ) s < α ) steady state
allocation of the Solow model is neither dynamically e¢ cient nor
dynamically ine¢ cient. Increase in savings increases long-run
consumption but it reduces short-run consumption.
The problem of dynamic ine¢ ciency arises because the savings
rate in the Solow model is exogenously given and not optimally
chosen.

Region A: Inconclusive ; Region B: Dynamically Ine¢ cient Region

Thank You

Solow

Uploaded by

Copyright:

Available Formats

Solow

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

Solow

Uploaded by

Copyright:

Available Formats

The Solow Model

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 1 / 71

Considerable di¤erence in the income level among countries

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 2 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 3 / 71

Considerable Di¤erence in the growth rate among countries.

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 4 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 5 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 6 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 7 / 71

Conditional Convergence: Growth rate of the rich countries are

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 8 / 71

Income and Welfare: Positive relationship between income and

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 9 / 71

Income and Welfare: Positive relationship between income and life

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 10 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 11 / 71

Solow (1956) develops an analytical model to explain stylized facts

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 12 / 71

Share of capital and labour in national income is constant and

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 13 / 71

Economic Agents: Firms and Households

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 14 / 71

Population: growing exponentially at an exogenously given rate

Capital depreciates at the rate 0 < δ < 1

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 15 / 71

Production function in percapita form: obtained by dividing both

Total income of household is spend in consumption and savings.

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 17 / 71

Goods market equilibrium: investment equals to savings

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 18 / 71

Substituting equation (2) to equation (3) yields the

We can solve for k (t ) = k (t ) from the fundamental equation and

The long-run is technically known as the Balanced Growth Path

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 20 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 22 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 23 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 24 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 25 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 26 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 27 / 71

Stability: strict concavity of the production function yields unique

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 28 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 29 / 71

Higher savings rate increases long-run capital stock.

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 30 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 31 / 71

Higher population growth rate reduces long-run capital stock.

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 32 / 71

Equation (4) gives,

Equation (7) shows

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 34 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 35 / 71

Barro and Sala-i-Martin (2004): average annual growth rate of 114

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 36 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 37 / 71

Barro and Sala-i-Martin (2004): average annual growth rate of 18

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 38 / 71

Chattopadhyay (IIT KGP) Solow Spring, 2022-23 39 / 71