The Institutions and Growth Debate: Part II

The Institutions and
Development Debate: Part II
Institutions, Inequality and Growth
Paul Dower
NES
Inequality in transition countries
Income GINI, Gini
transition countries
50
Russian Federation
45
Turkmenistan
40
Armenia
Georgia
Moldova Azerbaijan
Tajikistan
35
30
Kyrgyz Republic
Estonia
Latvia Lithuania
Bulgaria
Poland
Kazakhstan
Belarus
Romania
Croatia
Ukraine
Albania Macedonia, FYR
Uzbekistan
Slovak Republic Czech Republic
Hungary
25
20
0
2000
4000
6000
8000
10000
12000
14000
16000
GDP per capita, PPP, 2002
Before transition, it used to be from 20-28 (depending on a country)
18000
Outline of talk
1. A theory of institutional persistence.
2. Structural inequality and
underdevelopment.
3. Simply inequality or institutional
inequality?
Why do Institutions Persist?
• Institutions and poverty traps:
– Too poor to afford good institutions; bad institutions
make poor.
• Self-reinforcing institutions:
– Ex. Factor endowments favor land inequality; initial
land inequality leads to elite capture of political
institutions; elites set up institutions that suppress
majority; suppression of the majority stabilizes land
distribution. North America (family farms) vs. South
America (plantations). (Engerman and Sokoloff 1997)
– Ex. Resource Curse.
Easterly (2007)
• Institutional persistence and structural inequality
– Structural vs. Market Inequality
– Structural inequality should correspond to average
inequality over longer periods.
• Test Engerman and Sokoloff hypothesis:
– Factor endowments affect initial inequality;
– Initial inequality persists because of poor institutions;
– Poor institutions leads to underdevelopment.
Another Geographical IV
• Suitability of land for growing wheat or
sugar:
– Sugar farms lead to higher inequality through
plantations.
– Wheat farms lead to more equal land
distribution through smaller farms.
• Use wheat to sugar ratio as instrument for
inequality.
What should we expect to see?
• Structural inequality as predicted by initial
factor endowments should positively
correlate with underdevelopment.
• Higher structural inequality should also
predict lower investments in institutions
that empower the majority such as private
property and education.
Overidentification Test
• Test whether no direct effect on
development is present.
• Idea:
– Both are exogenous and both should be
excluded from the main regression.
– Adding one as a control should have no
explanatory power once we properly
instrument for endogeneity.
Summing up
• Findings robust to alternative measures and
samples and a variety of controls
–
–
–
–
Ethnic fractionalization
Legal origin
Commodity exporter
Share of tropical land
• Evidence supports initial hypothesis that
inequality affects development through
education and institutions.
• Over-identification test fails to reject exclusion
restriction (in all but one specification)
Income inequality or institutional
inequality?
• Easterly (2007) can not distinguish
between income inequality creating bad
institutions and institutions that are bad
because there is inequality in access.
• Both interpretations are plausible since
institutions index does not vary within
countries.
Observations from Banerjee-Duflo
(2005)
• Within country differences are comparable to across
country differences in returns to capital and technology
adoption.
– Fafchamps (2000): Trade credit at 2.5%/month for dominant
trading group. Double that for the minority trading group.
• Investment rates in developing countries do not differ
substantially lower than they should be.
– Duflo et al. (2003): only 15% of farmers take up fertilizer despite
over 100% return.
• Human capital externalities do not appear to be high.
– Rauch (1993): modest positive externalities in US cities (3%5%). Would have to be 25% to explain cross-country differences
• Average rates of return are too high given the TFP ratio
implied by the productivity gap.
Understanding growth by understanding
inequality within countries
• If the returns to investment in physical and human capital
are different within poor countries, using aggregate
models and cross-country studies may miss the main
issue.
• Hypothesis: the different rates of return are driven by
poor institutions and understanding how these
differences affect the growth process is key to
understanding the great divergence.
• Alternative: excessively different rates of return are
driven by market failure.
Land Inequality and Industrialization
Galor, Moav and Vollrath (2009)
• As in ES, high land inequality leads to low levels of
public investment in human capital.
– Tsarist Russia: Provincial councils dominated by wealthy
landowners; in 1896, rural literacy rate was 21%. After Stolypin
reforms, share of education in provincial council budget
increased by 50%.
• Without a sufficient level of human capital,
industrialization in a particular region could not occur.
• In contrast to ES, this effect only delays industrialization
due to increasing pressure from industrial elite to invest
in human capital.
Empirical Strategy
• US States vary by land inequality and public investment
in human capital, especially before 1950.
• Resistance to Industrialization:
– In 1900, Alabama spent $2.58 (1929 dollars) per child
on education. Massachusetts spent $36.45 per child!
– In 1950, Alabama spent $63.50 (1929 dollars) per
child on education. Massachusetts spent $107.55 per
child.
Location of poverty areas in the US
Property Rights and Finance
Johnson et al. (2002)
Question: Are property rights sufficient for
investment?
• Survey of firms in transition countries:
– Firms give subjective perception of security of
property.
– Detailed information on investments and
assets.
• Possible to test between wealth inequality
and institutional inequality.
Observations
• Entrepreneurs invest less of own funds if
perceive property rights are insecure.
• Absence of bank finance does not
preclude investment.
Caveats
• Barriers to entry
• Firms that have survived
• Small-scale
Conclusions
• Structural inequality reflects institutional
persistence.
• Primacy of property institutions in
contributing to structural inequality.