You are on page 1of 7

COURSE 006 (EDPI)

February 2020
 The Neoclassical (Solow-Swan) Model of
economic growth predicted that countries having
the same technology, population growth rate and
savings rate would converge to the same steady-
state level of per capita GDP in the long run.
 During this process, poorer countries would grow
faster and eventually catch up with richer
countries. (Reason: diminishing returns to K.)
 If countries differ in technology etc, they will
converge to different steady states.
 The tendency to converge should be stronger
for states/regions within a country, because:
◦ Intra-national differences in technology, savings
and population growth rates should be less than
international differences;
◦ Relatively freer movement of factors within a
country should allow reallocation of K from capital-
abundant to capital-scarce regions with higher
productivity and rewards for K; vice versa for L.
◦ Relatively freer movement of goods within a country
should reinforce this (FPE theorem).
 σ-convergence: calculate a measure of
dispersion of the distribution of the
national/regional per capita GDPs for each
year and check if there is an increasing or
decreasing trend over time.
◦ BUT do not use the standard deviation or variance:
they fail the property of ‘scale invariance’.
◦ Use one of the following:
 the max/min ratio (very crude)
 the coefficient of variation
 the standard deviation of the logs of the pcGDP
 the Gini coefficient
 β-convergence: fit a regression equation of the
following form, where y is pcGDP, i indexes
states, and is the initial year:
ln ln
, ≡

A negative and significant estimated β coefficient is


evidence of (absolute/unconditional) convergence.

 Conditional convergence regressions include


other variables on the RHS to control for other
determinants of transitional growth, like human
K, infrastructure, institutions. These variables are
taken at their values in the initial year .
 Tests for states of the U.S., countries and
regions of Europe, prefectures of Japan, and
provinces of China all show strong evidence
of absolute convergence.
 But Indian states show the opposite (β > 0,
=> absolute divergence), esp in the post-
reform period.
 See the charts in ch.10 of the Economic
Survey 2016-17.
 Since we have only 28 states, we cannot add
more explanatory variables on the RHS of a
state-level convergence regression because of
inadequate degrees of freedom.
 Some authors have done so with district-level
data, which gives many more observations,
although district-level data are not very reliable.
 These studies show evidence of conditional
convergence of districts, and the importance of
infrastructure and literacy in promoting growth.

You might also like