Three Lectures on
Economic Efficiency
and Growth
Thorvaldur Gylfason
Outline and aims
Present a policy-oriented overview of
the theory and empirical evidence
of economic growth
Trace linkages between economic growth
and its main determinants: saving,
investment, and economic efficiency



Exogenous vs. endogenous growth
Liberalization, stabilization, privatization
Education, institutions, natural resources
Outline and aims
Lecture I
Saving, efficiency, and economic growth
Lecture II
Economic policy and growth
Lecture III
Education, natural resources, institutions,
and empirical evidence
Introduction
Growth theory
As old as economics itself
Smith, Marshall, Schumpeter, Keynes
Explicit growth theory started with Harrod
and Domar in 1940s
Why important?
Unfashionable in 1960s and 1970s
Limits to growth, etc.
Growth and development
Growing apart
Country B: 2% a year
GNP per capita
 Investment
 Efficiency
 Institutions
Threefold
difference after
60 years
 Policy
Country A: 0.4% a year
0
60
Years
National economic output
Economic growth:
The short run vs. the long run
Economic growth
in the long run
Potential output
Actual output
Upswing
Business cycles
in the short run
Downswing
Time
Other comparisons
1)
2)
3)
4)
5)
6)
West-Germany vs. East-Germany
Austria vs. Czech Republic
US vs. USSR
South Korea vs. North Korea
Taiwan vs. China
Finland vs. Estonia
See my Pictures of Growth


www.hi.is/~gylfason/pictures2.htm
Further comparisons
1)
2)
3)
4)
5)
6)
Thailand vs. Burma
Mauritius vs. Madagascar
Botswana vs. Nigeria
Tunisia vs. Morocco
Spain vs. Argentina
Dominican Republic vs. Haiti
Singapore and Malaysia:
GNP per capita 1962-2001
30000
Malaysia
25000
Singapore
20000
15000
Current US$,
Atlas method
10000
5000
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
0
Botswana and Nigeria:
GNP per capita 1962-2001
3500
Botswana
3000
Nigeria
2500
2000
Current US$,
Atlas method
1500
1000
500
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
0
Spain and Argentina :
GNP per capita 1962-2001
18000
16000
Argentina
14000
Spain
12000
10000
Current US$,
Atlas method
8000
6000
4000
2000
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
0
Mauritius and Madagascar:
GNP per capita 1962-2001
4500
4000
Mauritius
3500
Madagascar
3000
2500
Current US$,
Atlas method
2000
1500
1000
500
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
0
Ireland and Greece:
GNP per capita 1962-2001
25000
Ireland
20000
15000
Greece
Current US$,
Atlas method
10000
5000
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
0
Basic growth theory
A.
B.
C.
Harrod-Domar model
Solow model
Endogenous growth model
Harrod-Domar model
Two assumptions
S  sY
K  vY
Implications for growth
s
g  
v
0.21
g
 0.04  0.03
3
SI
S  sY  I  K  K
sY  vY  vY Y s



s  vY  vY
Y
v
Harrod-Domar model
Three propositions about growth
s
g  
v
dg
0
ds
dg
0
dv
dg
0
d
Solow model
gn
Four assumptions
Y Y
nt
L  L0e
F
A  const.
Y  ( AL) K
a
1 a
S  sY  I  K  K
 Y

g  an  (1  a )  s     an  (1  a ) g
 K

Solow model
Endogenous output/capital ratio
 y

g  an  (1  a )  s   
 k

Exogenous
Exogenous
Endogenous
Y  ( AL) K
a
1 a
y  Ak
1 a
y
a
 Ak
k
Solow model
Dynamic stability of output/capital ratio
K
k
L
k  K  L
y n 

k 0 
k
s
y



k  K  L  s    n  sAk  a    n
k
dk
 (1 a )
 asAk
0
dk
Solow model
Dynamic stability of output/capital ratio
dk
 asAk  (1 a )  0
dk
k
E
k
Solow model
Two equations in two unknowns, y and k
y  Ak
1 a
 n 
y
 s

k

 sA 
k 

 n  
 s 
y 

 n  
1 a
a
A
1
a
1
a
Solow model
Output per head
 n 
y
 s
E
y  Ak1 a
Output/capital ratio
Capital per worker

k

Solow model
 s 
y 

 n  
1 a
a
A
Four propositions about long-run growth
dy
0
ds
dy
0
dA
dy
0
dn
dy
0
d
1
a
Closed-form solution to
Solow model
Y  ( AL) K
a
1 a
yˆ  Y / AL
kˆ  K / AL
1 a
ˆ
yˆ  k
K  I  K  sY  K
L  L0e
nt
A  A0e
qt
AL  A0 L0e
( n  q )t
K
 syˆ  kˆ
AL
Closed-form solution to
Solow model
 K  K  K 
ˆ
k  
  2 2 ( AL)  syˆ  kˆ  nkˆ  qkˆ

 AL  AL  A L 
kˆ
 yˆ 
g kˆ 
 s     n  q  skˆ  a    n  q
kˆ
 kˆ 
 kˆ
 ˆa
 ˆ    n  q  k  s
1
a
 k



s
ˆ

k



1

a

a
kˆ  kˆ
   n  q kˆ  s
  n  q
Closed-form solution to
Solow model
a
ˆ
vk
 kˆ 
v
 a 
ˆ
v
k
 
v  kˆ 
  v
a  kˆ 
1
a
ˆ
v  k   v    n  q v  s
a




s
s
a
 a   n  q t
ˆ
ˆ
e
vk 
 k0  
 nq 
   n  q 
a
Closed-form solution to
Solow model k̂


s
ˆ

lim k  
t 
  n  q 
1
a


s

lim yˆ  
t 
  n  q
Y 
lim y  lim   
t 
t  L
 
1 a
a


s

A
  n  q 
k̂0
time
1 a
a


s

 A0e 
  n  q
qt
1 a
a
Solow model:
Convergence
kˆ  an  q   kˆ
kˆ  0.70.01  0.01  0.04kˆ  0.7  0.06  k  0.04kˆ
e
e
0.04t
 0.5  t  17
0.04t
 0.8  t  40
Output per head
Solow model:
Convergence
 n 
y
 s
Rich country’s
initial income per head
y  Ak1 a
Poor country’s
initial income per head
Capital per worker

k

An Increase in the
Saving Rate
C
Output per head
C’
P
F
E
Capital per worker
An Increase in Static
Efficiency
C
Output per head
P’
F
P
E
Capital per worker
An Increase in Population
Growth or Depreciation
C’
Output per head
C
P
E
F
Capital per worker
An Increase in Dynamic
Efficiency (Technical Progress)
C’
Output per head
C
F
P’
P
E
Capital per worker
Solow model:
Conclusion
Three main points to note
Long-run growth is exogenous: g = n + q


No role for economic forces, policy or
institutions, just technology
But education is good for growth
Model implies convergence

Poor countries grow more rapidly than rich
The medium term can be quite long

Growth is endogenous for a long while
Solow model with
education
Y  ( AH ) K
a
H  Le
1 a
bt
AH grows at n + q + b
g nqb
dg
0
ds
Education stimulates long-run growth
Endogenous growth
Y  AL K
a
1 a
A  E ( K / L)
Y  EK
g  sE  
g  nq
a
Economic growth, g
Endogenous growth
Saving rate
times efficiency
minus depreciation
C
B
45°
O
Population growth
g  nq
g  sE  
A
dg
0
ds
dg
0
dE
Technological progress, q
A tax on education and
endogenous growth
Y  AGK
G  tK
Y  At K  EK
E  At
g  s At  
Inflation, money, and
endogenous growth
M 
Y   K
P
 M 
Y 
K
 PK 
M 
E 

 PK 
 1 
gs 
 
1  
M
1

PK 1  
Education and endogenous
growth, again
R&D model (Romer)
Y  A(1  b) L
A  BbLA
A
 BbL
A
dg
0
ds
g  BbL  n
dg
0
db
Education and endogenous
growth, once more
Human capital model (Lucas)
Y  AL K
a
Ac
1 a
g  n 1 h
a
Y  (cL) K
a
c
 1 h
c
1 a
dg
0
ds
How growth becomes
endogenous
Solow model: when s rises, E = Y/K falls due
to decreasing returns to capital, so g stays put
Y  ( AL) K
a
1 a
y  Ak
1 a
y
a
 Ak
k
Endogenous growth: when s rises, E stays put
due to constant returns to capital, so g rises
Y  EK
Y
y
  E  f (k )
K k
Empirical growth research
1)
Cross-country regressions

2)
3)
Cross sections vs. panels
Averages vs. initial values of
independent variables

4)
5)
Large samples, beginning in 1960 or 1970
Cost of simultaneity bias vs. cost of
discarding available data
Recursive modeling vs. instruments
Levels of income vs. rates of growth
Recursive modeling
Growth regression
g = a0 – a1y0 + a2x + a3z
(1)
where x is exogenous and z is endogenous
z = b0 + b1y0 – b2x
(2)
where z is, say, education and x is natural
resource reliance
Eq. (2) makes z exogenous, so (1) and (2)
can be estimated by OLS
TSLS calls for instruments that help explain z
without being correlated with g: Not easy
Levels of income vs.
rates of growth
g  a  by0  cx
Levels of income vs.
rates of growth
g  a  by0  cx
g
g
y  y0  T ln( 1 
)  y0  T (
)
100
100
Levels of income vs.
rates of growth
g  a  by0  cx
g
g
y  y0  T ln( 1 
)  y0  T (
)
100
100
T
y  y0 
(a  by0  cx)
100
Levels of income vs.
rates of growth
g  a  by0  cx
g
g
y  y0  T ln( 1 
)  y0  T (
)
100
100
T
y  y0 
(a  by0  cx)
100
y    y0  x
T
 a
100
T
 c
100
Levels of income vs.
rates of growth
g  a  by0  cx
g
g
y  y0  T ln( 1 
)  y0  T (
)
100
100
T
y  y0 
(a  by0  cx)
100
y    y0  x
T
 a
100
T
 1 b
100
T
 c
100
Absolute convergence:
Growth rates
Growth of GDP per capita 1960-2000 (%)
15
y = -0,864x + 8,3057
R2 = 0,1821
10
5
0
4
6
8
10
-5
-10
Log of GNP per capita 1960
12
Absolute convergence:
Levels of income
12
Log of GNP per capita 2000
y = 0,6575x + 3,2827
R2 = 0,4523
10
8
4
6
45º
4
4
5
6
7
8
9
10
Log of GNP per capita 1960
11
12
From efficiency to growth
Basic result
If it – anything! – increases economic
efficiency, it is also good for growth
Follows from Harrod-Domar model as well
as from endogenous-growth theory
and also, as a proposition about the
medium run, from the Solow model
In practice, Solow model and endogenous
growth are hard to distinguish
So, let’s look more closely at efficiency
Liberalization Increases
Economic Efficiency
D
H
Domestic,
distorted
price ratio
E
Traditional
output
C
O
G
Modern output
Liberalization Increases
Economic Efficiency
World price ratio
D
H
If output gain = E and
price distortion = c,
then E = mc2
Traditional
Output
output
gain
OC = modern output
CA = traditional output
OA = total output
O
Domestic,
distorted
price ratio
E
Price
distortion
F
C
A
G
Modern output
B
Liberalization Increases
Economic Efficiency
World price ratio
Welfare
gain
D
H
Exports
J
K
E
Traditional
output
Price
distortion
Output
gain
Domestic,
distorted
price ratio
Imports
F
C
O
A
G
Modern output
B
Liberalization Increases
Economic Efficiency
Welfare
gain
D
H
J
E
M
Traditional
output
N
Transition
takes time:
From E to F
via M, N,
and Q
Price
Q distortion
F
C
O
A
G
Modern output
B
Liberalization Increases
Economic Efficiency
Welfare
gain
D
World price
ratio
J
K
E
Productivity
gain
Price
AB = static gain
BC = dynamic gain distortion
AC = AB + BC = total gain
F
A
G
Traditional
output
O
Modern output
H
B
C
Stabilization Increases
Economic Efficiency
Distorted
price ratio
D
C
Inflation
distortion
E
F
Real
capital
If output gain = E and
inflation distortion = c,
then, again, E = mc2
Output
gain
Undistorted
price ratio
G
H
O
Output after
Output before
45°
Financial capital
A
B
Privatization Increases
Economic Efficiency
Undistorted
price ratio
D
E
Distorted
price ratio
H
Public
output
Price and
quality
distortion
F
J
Output
gain
45°
O
N Private output
A K
G
B
From efficiency to growth:
Same story time and again
Free trade is good for growth
– Reduces the inefficiency that results from
restrictions on trade
Price stability is good for growth
– Reduces inefficiency resulting from inflation
Privatization is good for growth
– Reduces inefficiency resulting from SOEs
Education is good for growth
– Reduces the inefficiency that results from
inadequate education
Investment and growth,
1965-1998
Growth of GNP per capita 1965-1998, adjusted for initial
income (% per year)
6
An increase in investment by
4% of GDP is associated with an
increase in per capita growth by
1% per year
4
2
Botswana
Thailand
0
0
5
10
15
20
25
30
-2
1%
-4
Jordan
4%
-6
r = 0.65
Nicaragua
-8
Gross domestic investment 1965-1998 (% of GDP)
35
Education and growth,
25 point increase in secondary1965-1998 Aschool
enrolment goes along with
an increase in per capita growth by
1% per year
Per capita economic growth 1965-98,
adjusted for initial income (% per year)
6
4
2
Japan
Thailand
Finland
0
0
50
100
-2
-4
Ghana
-6
Diminishing returns: The additional
benefit from education becomes
smaller as enrolment increases
r = 0.72
-8
Secondary-school enrolm ent 1980-97 (%)
Natural resources and
growth, 1965-1998
Growth of GNP per capita 1965-98, adjusted for initial
income (%)
6
An increase in the
natural capital share by
8% goes along with a
decrease in per capita
growth by 1% per year
4
Mauritius
2
0
0
10
20
30
40
50
Cameroon
-2
Mali
-4
Madagascar
-6
r = -0.64
-8
Share of natural capital in national wealth 1994 (%)
60
Growth of GDP per capita 1960-2000, adjusted
for initial income (% per year)
Democracy and growth,
1965-1998
8
Equatorial
Guinea
Botswana
6
Singapore
Korea
4
Saudi-Arabia
China
Cyprus
Malaysia
2
0
-10
-5
0
5
-2
-4
-6
r = 0.48
-8
Index of dem ocracy 1960-2000
10
Growth regressions
Based on World Bank data
World Development Indicators, published
each year on CD
Wide coverage: 208 countries, 42 years
Could also use Penn data (compiled by
Summers and Heston), but they cover
fewer countries
Here, we report cross-sectional evidence,
representing each country by a single
observation for each variable
Conclusion
Saving and efficiency
are good for growth
Efficiency gains take
many different forms
Liberalization, stabilization, privatization
Conversion of inputs into output is not
solely a matter of technology, but also
efficiency, so economic policy matters