Solow Model

Jinfeng Ge

Fudan University

2 19th, 2014

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 1 / 74

Outline

Overview of stylized facts of economic growth.

Solow growth model: Theory.
Solow growth model: Empirics.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 2 / 74

Long Run Economics Growth I
Economic growth over the very long run.
Figure 2: Economic Growth over the Very Long Run
INDEX (1.0 IN INITIAL YEAR)

45

40 Per capita GDP

35

30

25

20

15

10
Population
5

0 200 400 600 800 1000 1200 1400 1600 1800 2000
YEAR

Note: Data are from Maddison (2008) for the “West,” i.e. Western Europe plus the United
States. A similar pattern holds using the “world” numbers from Maddison.
Sourse: Maddison (2008)
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 3 / 74
Long Run Economics Growth II
Introduction to Modern Economic Growth

10 9 Western Offshoots
log gdp per capita
8

Western Europe
7

Latin
America
Asia

Africa
6

1000 1200 1400 1600 1800 2000

year

Figure 1.11. The evolution of average GDP per capita in Western Offshoots,
Western Europe, Latin America, Asia and Africa, 1000-2000.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 4 / 74

The Magic of Compounding

A 2% growth rate means your income doubles in 35 years.

Rule of thumb: the time to double the income is 70 dividing growth
rate.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 5 / 74

 Growth and Development: The Questions Cross-Country Income Di¤erences
Cross-country Income Difference
Cross-Country Income Di¤erences
Huge income differences.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 6 / 74

 Growth and Development: The Questions Origins of Income Di¤erences and World Economic Growth

Origins of Income Differences

Origins of Income Di¤erences and World Growth

10

USA
9
log gdp per capita

Spain

China
8

Britain

Brazil
7

India
Ghana
6

1800 1850 1900 1950 2000

year

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 7 / 74

 41
Post-war Economic THE FACTS OF ECONOMIC GROWTH
Growth

Figure 24: GDP per Person, 1960 and 2011

GDP PER PERSON (US=1) IN 2011
Luxembourg
Singapore Norway
Ireland U.S.
1 Japan
Netherlands Switzerland
Taiwan H.K. Australia
South Korea Spain
New Zealand
1/2 Malta
Barbados Trinidad/Tobago
Zimbabwe
Romania Argentina Mexico Gabon
Botswana
1/4 Malaysia
China Thailand Venezuela
Ecuador South Africa
Tunisia
1/8 Egypt
India
Jamaica
Cape Verde Honduras
1/16 Pakistan
Ghana
Mauritania Congo
Cameroon Nigeria
Lesotho Nepal Zambia
1/32 Kenya
Senegal

Ethiopia Madagascar Guinea

Malawi
1/64 C. Afr. Republic Niger

1/128
1/64 1/32 1/16 1/8 1/4 1/2 1
GDP PER PERSON (US=1) IN 1960

Source: The Penn World Tables 8.0.

growth in the 1980s (and before), Japan peaked at an income relative to the U.S. of 85
percent
Jinfeng in 1995.
Ge (Fudan Since
University) 1995, though, Japan has fallen back to around 75 percent
Solow Model of the
2 19th, 2014 8 / 74
Growth miracles (Catching up)
Growth Miracles I
Log GDP per capita for several countries
10.5
China
Japan
10 Korea
US

9.5

9
log dollars of 2000

8.5

7.5

6.5

5.5
1500 1600 1700 1800 1900 2000 2100
year

FrançoisGeGourio
Jinfeng (Fudan(BU)
University) Ec 502
Solow Model 2011
2 19th, 2014 96 // 74
12
Growth Miracles Figure
II 2. Output per worker: 1960 versus 2000.
Table 2
Fifteen growth miracles, 1960–2000

Country Growth 1960–2000 Factor increase

Taiwan 6.25 11.3

Botswana 6.07 10.6
Hong Kong 5.67 9.09
Korea, Republic of 5.41 8.24
Singapore 5.09 7.29
Thailand 4.50 5.83
Cyprus 4.30 5.39
Japan 4.13 5.04
Ireland 4.10 5.00
China 3.99 4.77
Romania 3.91 4.63
Mauritius 3.88 4.58
Malaysia 3.82 4.48
Portugal 3.48 3.93
Indonesia 3.34 3.72

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 10 / 74

 h. 8: Growth Econometrics 56

Growth Disasters Table 3

Fifteen growth disasters, 1960–2000

Country Growth 1960–2000 Ratio

Peru 0.00 1.00

Mauritania −0.11 0.96
Senegal −0.26 0.90
Chad −0.43 0.84
Mozambique −0.50 0.82
Madagascar −0.60 0.79
Zambia −0.61 0.78
Mali −0.77 0.74
Venezuela −0.88 0.70
Niger −1.03 0.66
Nigeria −1.21 0.62
Nicaragua −1.30 0.59
Central African Republic −1.56 0.53
Angola −2.04 0.44
Congo, Democratic Rep. −4.00 0.20

mainlyJinfeng
located in East
Ge (Fudan Asia and Southeast
University) SolowAsia.
Model These countries have sustained
2 19th, 2014 11excep
/ 74
Argentina vs. Switzerland
Growth Stagnation
Log GDP per capita for several countries
10.5
Argentina
Switzerland

10

9.5
log dollars of 2000

8.5

7.5

7
1840 1860 1880 1900 1920 1940 1960 1980 2000 2020
year

FrançoisGeGourio
Jinfeng (Fudan(BU)
University) Ec 502
Solow Model 2011 12 /12
2 19th, 2014 74 /
Correlates of Economic Growth I
Introduction to Modern Economic Growth

.08
Average growth rate of GDP per capita 1960−2000

TWN
.06

KOR
CHN

THA MYS
JPN
.04

LKA PRT IRL

ESP
LUX
NORPAN
ISR AUT
ITA GRC
INDISL FIN MUS
DOM BEL EGY
PAK FRA
BRA GHA MARNLD
TTO USA DNK
TUR CAN CHL GBR
.02

AUS SWE MEX

COL IRN
CRI ECU
PRY CHE
URY PHL ETH NZL
MWI
ZAF
ARG GTM PER SLV
ZWE BEN JAM
NGA UGA BFA BOL HND
VENKEN ZMB
0

GIN JOR
NIC

−.04 −.02 0 .02 .04

Average growth of investment ratio 1960−2000

Figure 1.15. The relationship between average growth of GDP per capita
and average growth of investments to GDP ratio, 1960-2000.
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 13 / 74
Correlates of Economic Growth II
Introduction to Modern Economic Growth

.06
KOR
HKG
average growth gdp per capita 1960−2000

THA
CHN JPN
.04

IRL
BRB
PRT MUS MYS
COG
IDN ESP
GRC
PAK ITA FIN ISR NOR
AUT
IND
SYR DOM
BRA ISL BEL
EGY FRA
TTO
CHL
PAN NLD CAN USA
TUR LKA DNK AUS
.02

IRN GBR SWE

ZWE LSO
COL MEX
MWI PRY
NPL DZA CHE
UGAGTM JOR ECU
CRI PHL
GHA
BGD KEN URY NZL
ZAF ARG
PER
SLV JAM
GMB
CMR HND BOL
BEN
BDI
0

MLI TGO
RWA
SEN
VEN
ZMB
MOZ
NIC
NER
−.02

0 2 4 6 8 10
average schooling 1960−2000

Figure 1.16
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 14 / 74
Facts and Questions

What determines both the level and the growth of income?

Is there something that a government can do to accelerate growth?
Why do some countries remain stubbornly poor? Bad policies? Bad
institutions? Bad endowments, such as geography? Bad luck?

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 15 / 74

Production Function

The production function can be written as

Yt = F (Kt , At Lt ).

We assume that the economy produces a single good Yt by means of

two factors of production: capital, Kt and labor, Lt .
Capital is the same as the final good of the economy, and used in the
production process.
The productivity of these inputs also depends on the level of
technology At . We assume that At is labor augmenting: At improves
the effecctiveness of workers in production.
Most commonly used production function-the ”Cobb-Douglas”
production function
Yt = Ktα (At Lt )1−α .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 16 / 74

Key Assumptions I

Continuity, Differentiability, Positive and Diminishing Marginal Products,

and Constant Returns to Scale.
∂F (·) ∂F (·)
FK (Kt , At Lt ) = > 0, FL (Kt , At Lt ) = > 0,
∂K ∂L
∂2 F (·) ∂2 F (·)
FKK (Kt , At Lt ) = < 0, F LL ( K t , A L
t t ) = < 0,
∂K 2 ∂L2
F (λKt , λAt Lt ) = λF (Kt , At Lt ), λ > 0

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 17 / 74

The Intensive Form of the Production Function

This section introduces a technique that is useful to simplify the

analysis.
Given the aggregate production function, we divide all inputs by At Lt ,
we obtain
Yt Kt At Lt
= F( , ).
At Lt At Lt At Lt
Let
Kt Yt
kt = , yt =
At Lt At Lt
we obtain
yt = F (kt , 1) = f (kt ) .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 18 / 74

Key Assumptions II

Inada Condition: First, Each input is strictly necessary:

F (0, At Lt ) = 0,

F (Kt , 0) = 0.
Second, f (k ) is strictly concave and

lim f 0 (k ) = ∞,
k →0

lim f 0 (k ) = 0.
k →∞

When capital is very low, having an additional bit of capital is very

valuable, but when capital is very high, an additional bit of capital is
almost worthless. These conditions guarantee a positive level of per-capita
output in the long-run.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 19 / 74

Evolution of Technology and Labor

Population (labor supply) Lt grows at a constant rate n over time

Lt +1 = (1 + n ) Lt ,

where n is the percentage change in L in any period.

Technology At also grows at a constant rate g over time

At +1 = (1 + g ) At ,

where g is the percentage change in A in any period.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 20 / 74

Evolution of Capital I

The capital stock increases owing to new investment and decreases

owing to depreciation. Capital depreciates at rate δ where 0 < δ < 1.
The capital accumulation equation is :

Kt +1 = It + (1 − δ) Kt ,

where It represents the amount of output devoted to capital

accumulation (gross investment).
We assume that a constant fraction of output in the economy is
saved each period:

St = sYt ,
where s denotes the savings rate with 0 < s < 1. This assumption is
quite restrictive: the saving decision is exogenous determined by the
parameter s. This assumption will be relaxed later.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 21 / 74

Evolution of Capital II

In the closed economy, goods market equilibrium demands

Ct + It = Yt ,

And investment equals saving

It = St .

With this equilibrium condition, the capital accumulation equation

becomes:
Kt +1 = sYt + (1 − δ) Kt .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 22 / 74

Detrending Variables I

Dividing the capital accumulation equation through by At Lt we obtain

Kt + 1 sYt (1 − δ) Kt
= + .
At Lt At Lt At Lt

Given that Lt +1 = (1 + n )Lt and At +1 = (1 + g )At we obtain

Kt +1
= (1 + n)(1 + g )kt +1 .
At Lt
The normalized capital accumulation equation is expressed as:

s (1 − δ )
kt +1 = yt + kt .
(1 + n)(1 + g ) (1 + n)(1 + g )

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 23 / 74

Detrending Variables II

If we assume F (Kt , At Lt ) is the Cobb-Douglas production function, it

leads to
yt = f (kt ) = ktα .
Then we obtain the key equation of the Solow model:

s (1 − δ )
kt + 1 = kα + kt .
(1 + n)(1 + g ) t (1 + n)(1 + g )
This difference equation characterize the dynamics of Solow model.
Given an initial condition k0 , this equation describes how kt evolves
over time. Once we know kt ; we can find the other variables easily.
(In technical jargon, kt is the state variable of this economy.)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 24 / 74

Steady-state Equilibrium I

At this point, we can also define a steady-state equilibrium for this

model. A steady-state equilibrium is an equilibrium path in which
kt = k ∗ for all t.
Consider the basic Solow growth model and suppose that
Assumptions 1 and 2 hold. Then there exists a unique steady state
equilibrium where k ∗ is given by

s (1 − δ )
k∗ = k ∗α + k ∗,
(1 + n)(1 + g ) (1 + n)(1 + g )
Then   1−1 α
∗ s
k = .
(1 + n)(1 + g ) − (1 − δ)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 25 / 74

Steady-state Equilibrium II
Introduction to Modern Economic Growth

k(t+1)
45°

sf(k(t))+(1–δ)k(t)
k*

k(t)
0 k*

Figure 2.2. Determination of the steady-state capital-labor ratio in the

Jinfeng Ge (Fudan
SolowUniversity)populationSolow
model without Modeland technological change. 2 19th, 2014
growth 26 / 74
Existence and Uniqueness of Steady State

The examples of nonexistence and non uniqueness of interior steady states

when assumptions 1 Introduction
and 2 aretonot satisfied.
Modern Economic Growth

sf(k(t))+(1–δ)k(t) 45°
k(t+1) 45° k(t+1) k(t+1)
45° sf(k(t))+(1–δ)k(t)

sf(k(t))+(1–δ)k(t)

k(t) k(t) k(t)

0 0 0
Panel A Panel B Panel C

Figure 2.5. Examples of nonexistence and nonuniqueness of interior steady

states when Assumptions 1 and 2 are not satisfied.

simple way, and assume that

f (k) = af˜ (k) ,
where a > 0, so that a is a shift parameter, with greater values corresponding to greater
productivity of factors. This type of productivity is referred to as “Hicks-neutral” as we will
Jinfeng Gesee
(Fudan
below,University) Solow
but for now it is just a convenient Model
way 2 19th, 2014
of looking at the impact of productivity 27 / 74
Convergence and Stability
By the phase-diagram, we can see that the Solow model converges to the
steady-state k ∗ . In technical terms, we say that the steady-state k is
Introduction to Modern Economic Growth
(globally) stable.

k(t+1)
45°

k*

k(t)
0 k(0) k* k’(0)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 28 / 74

The Kaldor facts I

By balanced growth, we mean a path of the economy consistent with the

Kaldor facts (Kaldor, 1963), that is, a path where, while output per capita
increases, the capital-output ratio, the interest rate, and the distribution of
income between capital and labor remain roughly constant. The following
graph shows the evolution of the shares of capital and labor in the US
national income.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 29 / 74

The Kaldor facts II
Introduction to Modern Economic Growth

100%

90%
Labor and capital share in total value added

80%

70%

60%

50%
Labor
Capital
40%

30%

20%

10%

0%
1929

1934

1939

1944

1949

1954

1959

1964

1969

1974

1979

1984

1989

1994
Figure 2.11. Capital and Labor Share in the U.S. GDP.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 30 / 74

 THE FACTS OF ECONOMIC GROWTH 15
The Kaldor facts? Recently?
Figure 6: Capital and Labor Shares of Factor Payments, United States
PERCENT

80

Labor share
70

60

50

Capital share
40

30

20
1950 1960 1970 1980 1990 2000 2010
YEAR

Note: The series starting in 1975 are from Karabarbounis and Neiman (2014) and measure
the factor shares for the corporate sector, which the authors argue is helpful in eliminating
Source: Karabarbounis and Neiman (2014)
issues related to self-employment. The series starting in 1948 is from the Bureau of Labor
Statistics Multifactor Productivity Trends, August 21, 2014, for the private business sector.
Jinfeng Ge The factor
(Fudan shares add to 100 percent. Solow Model
University) 2 19th, 2014 31 / 74
Balanced Growth Path I

If the steady state equilibrium satisfies the balanced growth path, we

need the Cobb-Douglas production function.
Constant factor income share. Assume that the factor price is
determined by the marginal product:

rt = MPKt = FK (Kt , At Lt ) = αKtα−1 (At Lt )1−α ,

wt = MPLt = FL (Kt , At Lt ) = (1 − α) Ktα A1t −α Lt−α .

Factor income shares are
rK wL
αK = = α, αL = = 1 − α,
Y Y
αK + αL = 1.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 32 / 74

Balanced Growth Path II

In the steady state equilibrium, kt = k ∗ and yt = y ∗ . From the following

conditions
Yt Kt
Yt = yt At Lt , = yt At , Kt = kt At Lt , = kt At , wt = (1 − α) At yt .
Lt Lt
We obtain that the aggregate output, output per capita, capital, capital
per capita and wage rate grows at a constant rate

Yt +1 yt +1 At +1 Lt +1 y ∗ At +1 Lt +1
lim = = = (1 + n)(1 + g ),
t →∞ Yt yt At Lt y ∗ At Lt
Yt +1 /Lt +1 Kt + 1
lim = (1 + g ), lim = (1 + n)(1 + g ),
t →∞ Yt /Lt t → ∞ Kt

Kt +1 /Lt +1 wt +1
lim = (1 + g ), lim = (1 + g ).
t →∞ Kt /Lt t → ∞ wt

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 33 / 74

Balanced Growth Path III

From the following conditions

Kt kt Yt yt
= , rt = α =α .
Yt yt Kt kt
We obtain that the capital-output ratio and interest rate are constant
Kt
lim = k ∗1− α ,
t →∞ Yt

lim rt = αk ∗α−1 .
t →∞

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 34 / 74

From Discrete Time Model to Continuous Model I

Start with a simple difference equation

xt +1 − xt = g (xt ) ,

Now consider the following approximation for any 4t ∈ [0, 1],

xt +4t − xt ' 4tg (xt ) ,

When 4t = 0, this equation is just an identity. When 4t = 1, it

gives the first equation.
In-between it is a linear approximation. The approximaiton is not too
bad if
g (x ) ' g (xt ) for all x ∈ [xt , xt +1 ] .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 35 / 74

From Discrete Time Model to Continuous Model II

Divide both sides of this equation by 4t, and take limits

xt +4t − xt
lim = ẋ (t ) ' g (x (t ))
4t →0 4t
where
dx (t )
ẋ (t ) =
dt
Then the differential equation represents the difference equation for
the case in which 4t is small.
Nothing has been changed on the production side.
Saving and consumption are

S (t ) = sY (t ) , C (t ) = (1 − s ) Y (t ) .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 36 / 74

From Discrete Time Model to Continuous Model III

Introduce population and technology growth

L̇ (t )
L (t ) = exp (nt ) L (0) , = n,
L (t )

Ȧ (t )
A (t ) = exp (gt ) A (0) , = g.
A (t )
Recall that
K (t )
k (t ) = ,
A (t ) L (t )
we obtain that

k̇ (t ) K̇ (t ) Ȧ (t ) L̇ (t ) K̇ (t )
= − − = − g − n.
k (t ) K (t ) A (t ) L (t ) K (t )

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 37 / 74

From Discrete Time Model to Continuous Model IV

The law of motion of capital is

K̇ (t ) = sF (K (t ) , A (t ) L (t )) − δK (t ) .

Then we can derive the key differential equation

k̇ (t ) = sf (k (t )) − (n + g + δ) k (t ) .

Let k̇ = 0 be the steady state equilibrium, we know that

  1−1 α
∗ s
k = .
n+g +δ

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 38 / 74

 We still have a fraction δ of the capital stock depreciating. In addition, the capital stock of
The Steady
the economyState
also has to increase as population grows in order to maintain the capital-labor
ratio constant. The amount of capital that needs to be replenished is therefore (n + δ) k.

output
(δ+n)k(t)

f(k(t))
f(k*)
consumption

sf(k(t))
sf(k*)

investment

k(t)
0 k*

Figure
Jinfeng Ge (Fudan 2.8. Investment and consumption
University) in the state-state equilibrium with
Solow Model 2 19th, 2014 39 / 74
Convergence and Uniqueness
Introduction to Modern Economic Growth

k(t)
k(t)

0 k(t)
k*

f(k(t))
s –(δ+g+n)
k(t)

Figure 2.9. Dynamics of the capital-labor ratio in the basic Solow model.
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 40 / 74
Comparative Dynamics I

Comparative dynamics: dynamic response of an economy to a change

in its parameters or to shocks.
Let us analyze the effect of an upward shift of saving rate.
What is the effect on the steady state?

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 41 / 74

After this point, it follows the dashed arrows on the horizontal axis.
Comparative Dynamics II

k(t)
k(t)

0 k(t)
k* k**

f(k(t))
s’ –(δ+g+n)
k(t)
f(k(t))
s –(δ+g+n)
k(t)

See also
Jinfeng Ge figure
Figure 1.5 in
2.13.
(Fudan University) the textbook.
Dynamics following an increase in the savings rate2 from
Solow Model
s to
19th, 2014 42 / 74
The Golden Rule

The Gold rule level of the capital stock is proposed by Phelps (1961).
(Phelps: Nobel Laureate in 2006).
The Gold rule level of the capital stock is defined to maximize the
consumption per capita in the steady state:

k̂ = arg max
∗
c ∗ (k ∗ )
k

c ∗ (k ∗ ) = f (k ∗ ) − (n + g + δ ) k ∗ .
Then we can see that
  1−1 α
α
k̂ = .
n+g +δ

To achieve that level of capital stock, we need the saving rate s = α.

What is the implication of this saving rate?

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 43 / 74

Summary: Implications of the Solow Growth Model

Solow model shows us that if there is no technological progress, there

will be no sustained growth.
Generate per capita output growth, but only exogenously:
technological progress is a blackbox.
Convergence: If all countries have access to the same technology A
and the same savings rate s, the Solow model implies convergence in
per-capital incomes Y /L, i.e. everything else equal, the Solow model
implies that a poor country should grow faster than a rich country
and catch up. (Could you prove this result?)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 44 / 74

Solow Growth Model and the Data

Now we will use Solow model or extensions to interpret both

economic growth over time and cross-country output differences.
Focus on proximate causes of economic growth.
We will discuss growth accounting and growth econometrics.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 45 / 74

Solow Model: Empirics

In the following lecture, we will discuss the empirical implication of

Solow growth model.
I will start with a simple example proposed by Lucus (1990) (Lucas:
Nobel Laureate in 1995).

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 46 / 74

Income Differences and Investment Return Differences I

Lucas assume that the US technology can be exported to other

countries. Then the production function, both domestically and
abroad, would be
Y = AK α L1−α .
Then imagine a less developed country whose annual (per capita)
output is a seventh of the US output:
yLDC 1
= .
yUS 7

Using per capita variables (LUS = LLDC = 1), the marginal return on
capital investment in the US is calculated as:
α −1
RUS = αAKUS − δ,

where the parameters α and δ take values of 1/3 and 0.1, respectively.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 47 / 74

Income Differences and Investment Return Differences II

Then we obtain  α−1 1

RUS + δ

KUS
= .
RLDC + δ KLDC
By the production function, we also obtain
α
AKLDC
yLDC 1
= α = ,
yUS AKUS 7

Then
KUS
= 73 .
KLDC
The net rate of return on capital in the US can be estimated to be
6.5% per annum, i.e RUS = 0.065.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 48 / 74

Income Differences and Investment Return Differences III

Therefore, the net rate of return on capital in the less developed

country should be
RLDC = 7.985.
This is saying that if the US production techniques could be exactly
replicated in less developed countries, the net return on investment
would be 798.5%. This result is striking since if this is the case, then
capital should be massively moving out of the US and into less
developed countries.
Of course, this riddle will disappear if we let ALDC < AUS .
This example highlight the importance of technology differences to
understand income differences across countries!

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 49 / 74

Growth Accounting I

Solow (1957) introduce the growth accounting framework. It is most

commonly applied for decomposing the sources of growth over time.
Conceptually, growth accounting can be thought of as quantifying the
relationship

Income = F (Factor Inputs, Efficiency).

The consensus view is that Efficiency plays a very large role.

Differences in efficiency account for at least 50% of differences in per
capita income.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 50 / 74

Growth Accounting II

Given the aggregate production function

Y (t ) = F [K (t ) , L (t ) , A (t )] = K (t )α (A (t ) L (t ))1−α .

Taking logs and differentiating it we obtain

ln (Y (t )) = α ln (K (t )) + (1 − α) ln L (t ) + (1 − α) ln A (t ) ,

Ẏ (t ) K̇ (t ) L̇ (t ) Ȧ (t )
=α + (1 − α ) + (1 − α ) ,
Y (t ) K (t ) L (t ) A (t )
gY = αgK + (1 − α) gL + (1 − α) gA .
Thus, the sources of growth in output can be attributed to growth in
factor inputs, weighted by their factor share, and the effect of growth
in efficiency.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 51 / 74

Growth Accounting III

For the 1947-1973 period, the U.S experienced GDP growth of

4.02%. Growth from capital accounted for 1.71% Growth from labor
accounted for 0.95%. TFP growth was 1.35% which accounted for
33.6% of growth in GDP per capita over this period. In contrast,
Germany experienced 6.61% GDP growth over this period, 56% of
this growth was accounted for by TFP.
Young (1995) argues that, contrary to past beliefs, the economic
miracles of South-East Asia had more to do with accumulation of
physical and human capital, and less to do with technological change.
But Hsieh(2002) use the different method to reach the different
conclusion.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 52 / 74

Growth Accounting IV

Sourse:
Jinfeng Ge Hsieh (2002)
(Fudan University) Solow Model 2 19th, 2014 53 / 74
Dual Growth Accounting

Given
Y (t ) = r (t ) K (t ) + w (t ) L (t ) .
Taking logs and differentiating it we obtain

ln (Y (t )) = ln (r (t ) K (t ) + w (t ) L (t )) ,
   
Ẏ (t ) K̇ (t ) ṙ (t ) L̇ (t ) ẇ (t )
=α + + (1 − α ) + ,
Y (t ) K (t ) r (t ) L (t ) w (t )
Ẏ (t ) K̇ (t ) L̇ (t ) ṙ (t ) ẇ (t )
−α − (1 − α ) =α + (1 − α ) .
Y (t ) K (t ) L (t ) r (t ) w (t )
Then
Ȧ (t ) ṙ (t ) ẇ (t )
(1 − α ) =α + (1 − α ) .
A (t ) r (t ) w (t )

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 54 / 74

Key behind Dual Growth Accounting

134 AEA PAPERS AND PROCEEDINGS MAY 1999

S 0 -- - - - - - - ----- ---- -- -- ---- ---- ---- --- -- ---- ---- ---- ---- - -- ---- ... ... . . .. ... 2 0 - --- --- --- ---- --- --- ---- --- - -- ---- --- --- . -- ------ --- . ..... .

50-20-~~~~~~~~~~~~~~~~~~~

30 ------------- t\;, --------- ' ----------- --------- ------------

-10 -- ------ ------- - - ------ --------- -------T T -t

-20 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992

1966 1969 1972 1975 1978 1981 1984 1987 1990 Year

Year . Return to equity Average lending rate . .. . EIP rate

Deposit rate . . Discount rate - . Curb loan rate

FIGURE 2. RETURN TO CAPITAL IN SINGAPORE

FIGURE 1. RETURN TO CAPITAL IN KOREA Note: The plots show annual interest rates.

Note: The plots show annual interest rates.

about the underlying technology. To see this,

This priced-based procedure measures the

Sourse: Hsieh (2002)

outward shift of the factor-price frontier- start with the basic national income accounting

essentially the dual representation of the identity that output is equal to the payments to

outward shift of the production possibilities the factors of production:2

frontier. The advantage of price-based esti-

mates is that they are based on market prices (1) Y = rK + wL.

(namely, wages and interest rates) paid by

Jinfeng Ge (Fudan University)

agents who have every incentive to get the Solow Model
This identity can be easily extended to allow 2 19th, 2014 55 / 74
 52 CHARLES I. JONES

Consensus
Table 6: Basic Development Accounting, 2010

GDP per Capital/GDP Human Share due

worker, y (K/Y )α/(1−α) capital, h TFP to TFP

U.S. 1.000 1.000 1.000 1.000 ...

Hong Kong 0.854 1.086 0.833 0.944 48.9%
Singapore 0.845 1.105 0.764 1.001 45.8%
France 0.790 1.184 0.840 0.795 55.6%
Germany 0.740 1.078 0.918 0.748 57.0%
U.K. 0.733 1.015 0.780 0.925 46.1%
Japan 0.683 1.218 0.903 0.620 63.9%
South Korea 0.598 1.146 0.925 0.564 65.3%
Argentina 0.376 1.109 0.779 0.435 66.5%
Mexico 0.338 0.931 0.760 0.477 59.7%
Botswana 0.236 1.034 0.786 0.291 73.7%
South Africa 0.225 0.877 0.731 0.351 64.6%
Brazil 0.183 1.084 0.676 0.250 74.5%
Thailand 0.154 1.125 0.667 0.206 78.5%
China 0.136 1.137 0.713 0.168 82.9%
Indonesia 0.096 1.014 0.575 0.165 77.9%
India 0.096 0.827 0.533 0.217 67.0%
Kenya 0.037 0.819 0.618 0.073 87.3%
Malawi 0.021 1.107 0.507 0.038 93.6%

Average 0.212 0.979 0.705 0.307 63.8%

1/Average 4.720 1.021 1.418 3.260 69.2%

Computed using the Penn World Tables 8.0 for the year 2010 assuming a common value of
α = 1/3. The product of the three input columns equals GDP per worker. The penultimate row,
“Average,” shows the geometric average of each column across 128 countries. The “Share due to
Sourse: Chad Jones (2016) TFP” column is computed as described in the text. The 69.2% share in the last row is computed
looking across the columns, i.e. as approximately 3.5/(3.5+1.5).
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 56 / 74
 Soviet
Growth Growth of Soviet Union
Accounting

1928–1987 annual growth rates:

gY/L gF gA

2.9% 3.3% -0.4%

Sourse: Easterly and Fischer (1994)

Source: Easterly and Fischer (1994)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 57 / 74

 Lost Decades of Japan
Figures 

Figure 1: Japan’s GDP Growth Rate: 1956-2000

16
14
12
56-73 Average (9.2%)
10
8
6
4
74-90 Average (3.8%)
2
91-2000 Average (1.3%)
0
-2
-4
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
Source: Japanese Cabinet Office, 1968 System of National Accounts
Sourse: Hoshi and Kashyap (2011)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 58 / 74

 in the growth in output per working-age-person.5 In spite of the low TFP growth in the 1973-83
Growth Accounting of Japan
period, output per adult increased at 2.2 percent. The reason that growth in output per adult was

higher in the 1973-83 period than in the 1991-2000 period is that in the earlier period there was

significantly more capital deepening and a smaller reduction in the labor input per working-age-

person.

Table I: Accounting for Japanese Growth per Person Aged 20-69

Factors
Period Growth rate
TFP Capital Workweek Employment
factor intensity length rate
1960-1973 7.2% 6.5% 2.3% -0.8% -0.7%
1973-1983 2.2% 0.8% 2.1% -0.4% -0.3%
1983-1991 3.6% 3.7% 0.2% -0.5% 0.1%
1991-2000 0.5% 0.3% 1.4% -0.9% -0.4%

Sourse: Hayashi and Prescott (2002)

Households
We model workweek length h as being exogenous prior to 1993 and endogenous thereafter.

5
The average annual TFP growth rate over 1983-1991, for example, is calculated as

( A1991 / A1993 )1 /(1991−1983) − 1 .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 59 / 74
 0
20
40
60
0
20
40
60
Zombies

1981 1981
1982 1982
1983 1983
1984 1984
1985 1985
1986 1986
1987 1987
1988 1988

`
`
1989 1989
1990 1990
1991 1991

Jinfeng Ge (Fudan University)

1992 1992
1993 1993
All Firms

Real Estate
1994 1994
1995 1995
1996 1996
1997 1997
1998 1998
1999 1999
2000 2000
2001 2001
2002 2002

0
20
40
60
0
20
40
60
Figure 7: Cross-industry incidence of zombies

1981 1981
1982 1982
1983 1983
1984 1984
1985 1985
1986

Sourse: Hoshi and Kashyap (2011)

1986
1987 1987
1988 1988
`
`

1989 1989
1990 1990
Trade

1991 1991
1992 1992

48
1993 1993
1994
Manufacturing

1994

Solow Model
1995 1995
1996 1996
1997 1997
1998 1998
1999 1999
2000 2000
2001 2001
2002 2002
0
20
40
60
0
20
40
60

1981 1981
1982 1982
1983 1983
1984 1984
1985 1985
1986 1986
1987 1987
1988 1988
`
`

1989 1989
1990 1990
1991 1991
1992 1992
Services

1993 1993
Construction

1994 1994
1995 1995
1996 1996
1997 1997
1998 1998
1999 1999
2000 2000
2001 2001
2002 2002
2 19th, 2014
60 / 74
Zombies and Productivity Growth
Figure 8: Zombie Incidence and Productivity Growth

Zombies and TFP Growth

10%

8%

6%
TFP growth 1990-2000

4%

2%

0%

-2%
y = -0.3993x + 0.0336
-4%

-6%
0% 5% 10% 15% 20% 25%
Change in the zombie index: 81-92 average to 93-02 average

Note: Estimates for TFP Growth are from Tsutomu Miyagawa, Yukiko Ito, and Nobuyuki Harada (2004) “The IT Revolution and
Productivity Growth in Japan,” Journal of the Japanese and International Economies, 18(3), 362-389.
Sourse: Hoshi and Kashyap (2011)

Jinfeng Ge (Fudan University) 49

Solow Model 2 19th, 2014 61 / 74
Population Aging
Figure 5: The Proportion of the Elderly in Various Countries (65 years old and above) (%)

40

35

30

25

20

15

10

5
1960
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
2014
2017
2020
2023
2026
2029
2032
2035
2038
2041
2044
2047
2050
United States United Kingdom Japan Italy Germany France Canada

Sourse: Hoshi and Kashyap (2011)

Jinfeng Ge (Fudan University) Solow Model 46 2 19th, 2014 62 / 74
 Barry Bosworth and Susan M. Collins 47
Growth Accounting of China and India I
Figure 1
GDP per Capita
(constant 2000 international purchasing power parity dollars)

6000

5000

China
4000

3000

2000
India
1000

0
1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004
Source: World Bank’s 2006 World Development Indicators. This purchasing power parity measure of GDP
standardizes for differences in the prices of common products across countries and over time.

Sourse: Barry Bosworth and Susan M. Collins (2008)

This approach is based on a production function in which output is a function
Jinfeng Ge of capital,
(Fudan labor,
and a term for total
University) factor
Solow Modelproductivity. As discussed in more
2 19th, detail 63 / 74
2014
Growth Accounting of Accounting
China forand India
Growth: II China and India
Comparing 49

Table 1
Sources of Growth: China, India, and East Asia, 1978 –2004
(annual percentage rate of change)

Contribution to output per worker of

Output per Physical Total factor

Period/country Output Employment worker capital Education productivity

1978–2004 China 9.3 2.0 7.3 3.2 0.3 3.6

India 5.4 2.0 3.3 1.3 0.4 1.6
1978–1993 China 8.9 2.5 6.4 2.4 0.4 3.5
India 4.5 2.1 2.4 0.9 0.3 1.1
1993–2004 China 9.7 1.2 8.5 4.2 0.3 3.9
India 6.5 1.9 4.6 1.8 0.4 2.3
East Asia excluding China
1960–1980 7.0 3.0 4.0 2.2 0.5 1.2
1980–2003 6.1 2.4 3.7 2.2 0.5 0.9
1980–1993 7.3 2.7 4.6 2.6 0.6 1.4
1993–2003 4.5 2.0 2.5 1.8 0.5 0.3

Source: Authors’ estimates as described in text; Bosworth and Collins (2003).

Notes: The employment series is a census-comparable concept for both China and India. The East Asia
comparison includes Indonesia, South Korea, Malaysia, Phillipines, Singapore, Taiwan, and Thailand.
Sourse: Barry Bosworth and Susan M. Collins (2008)
Growth rates may not sum due to rounding.

countries are also notable for the extent

Jinfeng Ge (Fudan University)
to which their growth appears2to
Solow Model
have
19th, 2014 64 / 74
 Slow labor Productivity Figure 1
Labor productivity since 1973
Business Sector Labor Productivity
Cumulative growth since 1973Q2 Percent
80
1997Q2 2003Q4 Q2
70

60

50

40

30

20
Broken linear
trend
10

-10
1973 1977 1981 1985 1989 1993 1997 2001 2005 2009

Source: BLS and Fernald (2012).

Sourse: Fernald (2014)
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 65 / 74
Slow labor Productivity 45

Figure 2
Evolution of Key Growth-Accounting Variables

A. Total Factor Productivity B. Quality-Adjusted Capital-Labor Ratio

Cumulative growth rate since 1973Q2 Percent Cumulative growth since 1973Q2 Percent
35 100

Q2 30 Q2 90

25 80
Utilization-Adjusted
TFP 70
20
60
15
50
10
TFP 40
5
30
0
20

-5 10

-10 0
1973 1977 1981 1985 1989 1993 1997 2001 2005 2009 1973 1977 1981 1985 1989 1993 1997 2001 2005 2009

C. Labor Quality D. Utilization

Cumulative growth since 1973Q2 Percent Cumulative growth since 1973Q2 Percent
18 4
Q2

16 3

14 Q2 2

12 1

10 0

8 -1

6 -2

4 -3

2 -4

0 -5
1973 1977 1981 1985 1989 1993 1997 2001 2005 2009 1973 1977 1981 1985 1989 1993 1997 2001 2005 2009

Notes: Mean of utilization set to zero for full sample that starts in 1973:2.

Sourse: Fernald (2014)

Source: Fernald (2012). Also recent research by Steven Davis and John

Haltiwanger
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 66 / 74
Europe
164 QUARTERLY JOURNAL OF ECONOMICS

Downloaded from http://qje.oxfordjournals.org/ at Fudan University on F

FIGURE XII
Relative Aggregate Productivity—The Importance of Services
This counterfactual sets the productivity growth in services such that in the
last period in the sample relative productivity in services is the same as rela-
tive productivity in industry in each
Jinfeng Ge (Fudan University) country.
Solow ModelEach panel plots aggregate labor 2 19th, 2014 67 / 74
Comments on Growth Accounting

Mismeasurement! How to measure the variables?

Measure of labor input: what matters is not labor hours, but effective
labor hours. It is important (though difficult) to make adjustments for
changes in the human capital of workers.
Measurement of capital inputs: in the theoretical model, capital
corresponds to the final good used as input to produce more goods.
In practice, capital is machinery, need assumptions about how relative
prices of machinery change over time. Typical assumption was to use
capital expenditures but if machines become cheaper would severely
underestimate gK .

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 68 / 74

Does TFP equal to technology

Figure 2: Real GDP and technology usage per capita for four countries
48

Sourse: Diego Comin and Bart Hobijn (2011)

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 69 / 74
Convergence I

One important implication of the Solow model is that countries

should converge to their steady-state. If all countries had the same
steady-state, they should all converge to the same steady-state. This
is called absolute convergence. This can be tested by checking if the
poorer countries at date t have higher subsequent growth from t on.
Do a regression:

growthrate1960−1985 = a + bincome1960 + ε,

or can simply do a scatter plot.

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 70 / 74

Convergence II
THE FACTS OF ECONOMIC GROWTH 43
The data show that countries have not been converging in absolute terms
in the entire world.
Figure 26: The Lack of Convergence Worldwide
GROWTH RATE, 1960 - 2011

7%
South Korea

6% Botswana

Taiwan
5% Romania Malta
Japan Singapore
China Thailand Cyprus
4% Egypt Portugal
Ireland
Spain Hong Kong
Panama Argentina Italy
3% Cape Verde Brazil Israel Germany
India Tunisia Malaysia Turkey
France Luxembourg
Morocco Chile U.S.
2% Lesotho Paraguay
Australia Canada
Switzerland
Burkina Faso Philippines Norway
Nepal Guatemala New Zealand
1% Ethiopia
Mali Iran
Gabon
Malawi Namibia
Togo Trinidad/Tobago
0% Comoros Cote dIvoire Jamaica
Madagascar Nigeria Venezuela
-1% C. Afr. Republic Congo
Niger
Guinea
-2%
1/64 1/32 1/16 1/8 1/4 1/2 1
GDP PER PERSON (US=1) IN 1960

Source: The Penn World Tables 8.0.

Sourse: Chad Jones (2016)
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 71 / 74
Convergence
42
III CHARLES I. JONES

Within OECD. ”Club convergence”.

Figure 25: Convergence in the OECD
GROWTH RATE, 1960 - 2011

4.5% Japan

4%
Ireland

3.5% Portugal
Greece Spain

Italy
3%
Germany
Finland Austria
Turkey
2.5%
France
Netherlands Luxembourg
Belgium Sweden
Denmark U.S.
2% Iceland U.K.
Switzerland
Canada
Norway
1.5%
1/4 1/2 1
GDP PER PERSON (US=1) IN 1960

Source: The Penn World Tables 8.0. Countries in the OECD as of 1970 are shown.

Sourse: Chad Jones (2016)

below
Jinfeng Ge theUniversity)
(Fudan 45-degree line rather than above.
Solow Model 2 19th, 2014 72 / 74
Concluding Remarks

Solow mode predict the correlates of economic growth with physical

capital, human capital and technology, but these are only proximate
causes of economic growth and economic success. What is the
fundamental causes of economic growth?
Cross-country differences in income per capita cannot be understood
solely on the basis of differences in physical and human capital. What
determine TFP?

Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 73 / 74

Institution?
Figure 31: Korea at Night

Note: North Korea is the dark area in the center of the figure, between China to the north and
South Korea to the south. Pyongyang is the isolated cluster in the center of the picture. Source:
http://commons.wikimedia.org/wiki/File:North and South Korea at night.jpg
Sourse: Chad Jones (2016)
Jinfeng Ge (Fudan University) Solow Model 2 19th, 2014 74 / 74