Transcript The changing policy framework

Latin America’s infrastructure gap:
a macroeconomic perspective
Luis Servén
The World Bank
ECLAC
January 2005
Plan
1. The changing policy framework
2. The infrastructure gap
3. The output cost
4. The lessons
The changing policy framework
• Until the 1970s, the public sector dominated infrastructure
provision in both industrial and developing countries.
• Since the 1980s (earlier in Chile and the UK) Latin America
led the worldwide drive towards opening up of infrastructure
to private initiative – in various forms and extents.
• The drive was propitiated by a hardening of fiscal discipline
in response to financial instability and macroeconomic crises
• In most countries, the fiscal retrenchment led to a sharp
contraction of public infrastructure investment (similarly to
the post-Maastritch fiscal adjustment in the EU)
The changing policy framework
Latin America: Public Investment in Infrastructure
(weighted average of 7 countries, percent of GDP)
4.0
3.5
3.0
2.5
% 2.0
1.5
1.0
0.5
Total
Roads plus Rails
Note: 7 Latin Am erica countries, ARG, BOL, BRA, CHL, COL, MEX, PER.
Power
Water
Telecommunications
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
0.0
The changing policy framework
Latin America’s fiscal adjustment:
Contribution of consumption and investment
Changes between 1980-84 and 19992001 (percent of GDP)
Public
infrastructure
investment
(1)
Public
consumption
(2)
Argentina
-2.87
8.22
Bolivia
Brazil
Chile
Colombia
Mexico
Peru
-2.48
-2.57
-1.38
-0.59
-2.20
-1.43
2.38
9.97
-2.51
11.30
1.31
0.53
Source: Calderón and Servén (2004b).
(a)
Contributions to fiscal
adjustment
Primary
surplus
(3)
Public
infrastructure
investment
-(1)/(3)
Public
consumption
-(2)/(3)
6.23
0.46
-1.32
0.40
0.62
1.89
0.17
0.42
2.10
-0.39
-2.42
3.45
-3.23
-0.25
-0.78
6.15
4.12
0.73
3.50
5.24
0.68
(b)
The changing policy framework
Latin America: Private Investment in Infrastructure
(weighted average of 7 countries, percent of GDP)
1.6
1.4
1.2
1.0
% 0.8
0.6
0.4
0.2
Total
Roads plus Rails
Note: 7 Latin Am erica countries, ARG, BOL, BRA, CHL, COL, MEX, PER.
Power
Water
Telecommunications
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
0.0
The changing policy framework
Latin America: Total investment in Infrastructure
(weighted average of 7 countries, percent of GDP)
The changing policy framework
Latin America: Total investment in Infrastructure
(6 major countries, percent of GDP)
The changing policy framework
Latin America: Investment in Infrastructure (public + private)
(weighted average of 7 countries, percent of GDP)
4.0
3.5
3.0
2.5
% 2.0
1.5
1.0
0.5
Total
Roads plus Rails
Note: 7 Latin America countries, ARG, BOL, BRA, CHL, COL, MEX, PER.
Power
Water
Telecommunications
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
0.0
The changing policy framework
The private sector response
•
Private initiative surged in the 1990s -- but with diversity
across industries (and countries)
•
Strong response in telecommunications, much less in transport.
•
Evidence of public-private complementarity, not only substitution:
countries maintaining higher public investment attracted more private
investment (Chile, Bolivia, Colombia)
•
The rise in private investment was not enough for asset
accumulation to keep up with other world regions
•
The investment fall contributed to widen Latin America’s
infrastructure gap – in terms of quantity and quality -widened over the 1980s and 1990s
The changing policy framework
Brazil: the power sector
Investment
Capacity change
The infrastructure gap
Power Generation Capacity
(megawatts per 1,000 workers, Medians by Region)
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1980
1985
LAC (19)
1990
EAP7 (7)
1995
MIDDLE (64)
IND (21)
2001
The infrastructure gap
Road plus Railway Length
(km per area, Medians by Region )
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
1980
1985
LAC (19)
1990
EAP5 (5)
1995
MIDDLE (53)
IND (21)
2001
The infrastructure gap
Total Telephone Lines
(lines per 1,000 workers, Medians by Region)
3,000
2,500
2,000
1,500
1,000
500
0
1980
1985
LAC (19)
1990
EAP7 (7)
1995
MIDDLE (64)
IND (21)
2001
The infrastructure gap
Perceived
infrastructure quality
Figure 2.17. Overall Infrastructure Quality
(Medians by region, 2000)
Medians by region and income level, 2000
6
5
4
3
2
1
0
LAC (11)
EAP7 (7)
MIDDLE (27)
Question: The quality of the infrastructure is among the best in the world (1=strongly disagree; 7 =strongly agree).
Source: World Competitiveness Report.
IND (24)
The output cost
Why do we care about infrastructure ?
• The availability and quality of infrastructure services is
key for productivity and profitability
• Robust association between infrastructure availability and
aggregate output / growth within and across countries
• Partly driven by reverse causality (growth encourages demand
for infrastructure services)
• But there is broad agreement that infrastructure development has
a strong causal effect of on economic development.
• Evidence that infrastructure development helps reduce
income inequality – makes it easier for the poor to access
economic opportunities, jobs, health and education.
Infrastructure Stocks
and Economic Growth (1960-2000)
8%
Growth Rate of GDP per capita
6%
BWA
THA
CHN
-4
SGP
7%
y = 0.0056x + 0.0206
R2 = 0.2547
TWN
KOR
5%
HKG
JPN
CYP IRL
PRT
MUS
ROM
IDN
ESP
GRC
TUN
3%
AUTBEL
PAK
FINITANOR
DOM
ISR
IND
EGY
MAR SYRBRA
FRA
USA
PAN
NLD
CHL
CAN
TUR
TTO
LKA
HUN
DNK
AUS
SWE
GBR
IRN
DEU
2%
MEX
COL
ZWE
PRY
NPL
DZA ECU
CHE
PHL
JOR
CRI
UGA GNB
URY
NZL
BGD
KEN GHA GTM
ZAF
1%
ARG
PER SLV
JAM
PNG
BFA
MRT
ETH TZA
HND
BOL
CIV
0%
MLI
RWA GINSLE
SEN
VEN
-3
-2
-1 ZMB
0
1 POL
2
MDG NGA
-1%
NIC
NER
-2%
4% MYS
Index of Infrastructure Stocks (1st. Principal Component)
Source: Calderón and Servén (2004b)
3
Growth in GDP per worker
Infrastructure accumulation and growth
(1960-2000 country averages, percent)
7%
6%
5%
4%
3%
2%
1%
0%
-1%
-2%
-3%
-2%
y = 0.505x + 0.0006
R2 = 0.3253
0%
2%
4%
6%
8%
Growth in infrastructure stocks per worker
Rest
lac
Source: Calderón, Easterly and Servén (2003)
eap7
10%
Infrastructure Stocks
and Income Inequality (1960-2000)
0.7
0.6
ZWEBRA
KEN
HND
BWA
MEX
COL
PAN ZAF
ECU
CHL
BOL
GTM
0.5
MDG
PER
BFA
ZMB PHL TUR
DOM
SLV
THA
MYS
VEN CRI
NGA
TTO
PNG
IRN
PRY
URY JAM
MAR
TUN
TZA
ARG
CIV LKA
HKGSGP
0.4
ETH
JOR
MUS
UGA
EGY
AUS IRL
FRA
PRT
NZL
USA
GHA
GRC
ITA
JPN
CHN
KOR
NOR
BGD IDN
DNK
YSR
SWECHE
IND
PAK 0.3
ISR DEU
TWN CAN
RWA
NLD
FIN
GBR
AUT
ROMPOLESP
BEL
CYP
HUN
BGR
Gini Coefficient (0-1)
SEN
0.2
y = -0.0303x + 0.403
R2 = 0.2157
0.1
0.0
-4
-3
-2
-1
0
1
Index of Infrastructure Stocks (1st. Principal Component)
Source: Calderón and Servén (2004b)
2
3
The output cost
•
What is the contribution of infrastructure services to
aggregate output and/or its growth rate ?
Three main empirical approaches in the literature:
1. Empirical growth models
2. Augmented production (or cost) function
3. VARs
Caveats:
-- technical problems often severe (identification /
reverse causality, spurious regressions…)
-- all else equal: the costs of “getting there” are not
explored – large tax rises or cuts in other
expenditures that may have an output cost…
The output cost
The long-run growth approach:
• Adding infrastructure into a standard growth regression
• Infrastructure usually proxied by telecommunications
indicators (e.g., Easterly 2001, Loayza et al 2003)
Calderón and Servén 2004b: panel of 100+ countries, 40 years
 Consider both infrastructure quantity and quality
 Synthetic infrastructure indicator: first principal component of
{power, roads, telecom} – accounts for 80% of their variance.
 Endogeneity: identification via GMM-IV with (a) internal
instruments; (b) demographic variables
 Growth contribution of infrastructure quantity and quality is
statistically and economically significant.
The output cost
Additional growth in LAC countries due to increased infrastructure development
Source: Calderón and Servén 2004b
The output cost
The augmented production function approach:
• Unlike VARs and growth regressions, it is a structural approach
Y = F (K, H, Z); K = physical capital; H = human capital (often
omitted) ; Z = infrastructure capital (power, phone lines, roads)
• Productive services assumed proportional to asset stocks
• In actual data, Z often is already included in K: The coefficient on
Z captures the return differential on Z over K
• In addition to usual reverse causality problem, spurious correlation
problem when using time series: nonstationarity of Y, K, Z leads to
huge infrastructure coefficient estimates (Aschauer 1990)
The output cost
The augmented production function approach
Calderón and Servén 2005: panel time-series estimation for 90
countries, 40 years.
• Spurious regression problem does not arise here (due to large N)
• Only one long-run relation found – resolves identification problem
• Pooled and country-specific estimates – permit assessing
heterogeneity across countries / regions
• Synthetic index and disaggregated infrastructure assets
• Results broadly similar to Calderón, Easterly and Servén 2003 –
in spite of very different approach (GMM-IV to deal with
identification; first-differencing to deal with nonstationarity)
The output cost
Estimated (log) infrastructure coefficients
(DOLS estimates, 1960-2001, synthetic index)
Coefficient
S.E.
0.091
0.013
All (89 countries)
0.130
0.019
Industrial (21)
0.080
0.027
Developing (68)
0.145
0.024
Pooled
Country-specific: mean by group
Source: Calderón and Servén 2005
The output cost
Country-specific estimates
[Synthetic Infrastructure Index, GLS--PIC (1,1)]
40
35
30
25
20
15
10
5
0
<0.20
-0.10
0.00
0.10
Source: Calderón and Servén 2005
0.20
0.30
The output cost
• The estimated return on infrastructure assets is significantly
higher than that on other physical capital in the vast majority of
countries.
• Infrastructure has significantly lower returns than other capital
only in 3 out of 89 countries [none in LAC]
• Across LAC countries, some heterogeneity too:
The differential return on overall infrastructure is significantly
higher than average in Peru, Mexico, Colombia…
Differences also across assets – e.g., the differential return
on power generation capacity is significantly lower than
average in Paraguay, but higher in Brazil
The output cost
Estimated (log) infrastructure coefficients
(DOLS estimates, 1960-2001)
Electricity
Generation
Capacity
Pooled DOLS
Roads
Main
Telephone
Lines
0.074 **
0.018
0.060 **
0.022
0.046 **
0.015
All (89 countries)
0.115 **
0.022
0.104 **
0.042
0.052 **
0.026
Industrial (21)
0.120 **
0.030
0.135 *
0.070
-0.016
0.038
Developing (68)
0.113 **
0.027
0.094 *
0.051
0.073 **
0.032
Country-specific: means by Group
Source: Calderón and Servén 2005
The output cost
The cost of the widening infrastructure gap: EAP vs LAC
Avg. 1991-00 vs.
Avg. 1981-90
Avg. 1996-00 vs.
Avg. 1981-85
1. Change in relative infrastructure endowments (%)
Main Phone Lines
Electricity Generating Capacity
Roads
27.6
37.9
30.3
41.1
58.0
50.3
2. Change in Relative GDP per worker (%)
31.6
41.9
3. Contribution of the infrastructure gap
9.3
14.5
4. Relative contribution (as % of [2])
29.2
34.7
Source: Based on Calderón and Servén 2005
The lessons
(1) Fiscal adjustment, as commonly measured and
enforced, tends to have an anti-investment bias
•
One (not the only) major factor is the use of inappropriate
fiscal rules targeting liquidity, the cash deficit and gross
public debt – rather than solvency and net worth, which are
key to fiscal sustainability.
•
Infrastructure projects have a negative short-run liquidity
effect -- it takes time to build the assets and get the returns.
•
The focus on fiscal liquidity discourages such projects –
even if they are consistent with good public economics; i.e.,
they enhance solvency.
The lessons
(2) Infrastructure investment cuts represent an
inefficient fiscal adjustment strategy
•
The direct effect of the spending cut is to raise liquidity
and public sector net worth
•
But there is an opposing indirect effect: less infrastructure
means less output and lower fiscal revenues tomorrow
•
The indirect effect offsets partly the direct effect – and can
even make fiscal adjustment self-defeating.
Summary
• Latin America’s infrastructure gap widened in the 1980s
and early 1990s, at a substantial cost in terms of output and
productivity.
• A major factor in the process was the investment slowdown
– caused by a public investment decline not offset (except in
telecom) by private sector participation.
• The public investment compression reflected a biased and
inefficient fiscal adjustment, encouraged by rules targeting
liquidity and debt rather than solvency and net worth.
• Ensuring adequate room for productive spending requires
fiscal rules that reconcile solvency and growth.
End
The changing policy framework
Fiscal discipline has led to a public investment fall not only
in developing countries – also in the EU
• The fiscal targets imposed in the Maastritch Treaty
contributed to a decline in public investment across
Europe:
• Out of 9 countries exceeding the Maastritch deficit limit in
1992, 8 met it in 1997. Public investment had fallen in all 8 !
• Infrastructure investment fell along with the total
The changing policy framework
Fiscal adjustment and public investment
(average of 9 EU countries, percent of GDP)
1.4
3
2
1.3
1
1.2
0
-1
1.0
-2
0.9
-3
Transport Investment, mean (left scale)
Primary Deficit, mean (right scale)
Sources: World Developm ent Indicators - World Bank; and provisional data from ECMT.
Notes: (a) Total = Roads + Rails + Airports.
(b) 9 EU countries: Austria, Finland, France, Netherlands, Norw ay, Portugal, Spain, Sw eden and United Kingdom .
2000
1999
1998
1997
1996
1995
1994
1993
1992
-6
1991
0.6
1990
-5
1989
0.7
1988
-4
1987
0.8
% GDP
% GDP
1.1