In 2000, public spending on infrastructure in developing countries averaged 9% of government
spending, or 1.4 % of GDP (IMF Statistics). Despite the magnitude of such spending,
and a widespread belief that infrastructure is integral to development, evidence on how
investment in physical infrastructure aects productivity and individual well-being remains
limited (Bank 1994). There is also very little evidence on the distributional consequences of
such investment.
    In this paper we have examined these questions in the context of large dam construction
in India. We have argued that any credible evaluation of large dams must address the
endogeneity of dam placement, which depends both on the wealth of dierent regions and the
expected returns from dam construction. This problem of endogenous placement is central
to the evaluation of any large infrastructure project. Placement will reflect both regional
need and a complicated decision-making process (Gramlich (1994)). While a growing cross-
country literature finds that productive government spending enhances growth, most studies
are unable to convincingly control for unobserved heterogeneity (see, for instance, Canning,
Fay, and Perotti (1994) and Esfahani and Ramirez (2003)). By taking account of geographic
suitability for infrastructure we provide a potentially generalizable way of estimating returns
to investment. Our estimation strategy also allows us to examine the relative importance
of dierent factors in driving infrastructure decisions. For instance, we find that our 2SLS
results are, with a few exceptions, very similar to OLS results when we restrict the comparison
to districts within a state, with a few exceptions. This suggests that while the overall number
of dams a state builds is endogenous, dam placement within a state has possibly been driven
by cost considerations rather than political economy or rate of returns concerns.
    We have also emphasized the importance of identifying who the beneficiaries of invest-
ment projects are, and who bears the costs. To return to the example of the World Bank, in
Spring 2005 the Bank announced 270 million dollars in grants and guarantees for the Nam
Theun 2 dam in Laos. The New York Times (June 5, 2005) quotes a bank ocial justifying
the return to dam lending as driven by the need to support infrastructure development in a

‘practical’ way since “You’re never ever going to do one of these in which every single person
is going to say, ’This is good for me”’ (Fountain 2005). Implicit in this statement is the belief
that projects with an average positive return should be undertaken, as it will be possible to
compensate the losers. We, however, find a strikingly unequal sharing of costs and benefits of
large dam construction in India. The results on poverty suggest that the government failed
to compensate people living near the dam for the inherent inequities in the gains and losses
associated with dam construction. Given that dams are at best marginally cost-eective the
absence of redistribution may well have been required to generate a political constituency
for large dams. We provide suggestive evidence that one reason for the lack of compensation
relates to the institutional framework within which policy decisions are made. In areas where
the institutional structure favors the politically and economically advantaged dams cause a
greater increase in poverty. We conclude that distributional implications of public policies
should be central to any evaluation, and that more attention should focus on understanding
the institutions, and power structures, that lead to the implementation of marginally cost
eective projects with significant negative distributional implications.

by Esther Duflo and Rohini Pande

Artikel Terkait

| 0 comments | Labels: ,


Poskan Komentar