What the evidence shows
We turn to more formal evidence on the pattern of lending of the World Bank and the
IMF. We focus on the 1984- 2001 period, for which a complete data set could be
assembled. We distinguish five developing regions: Africa, Asia, developing Europe,
Latin America, and the Middle East16. Our discussion so far has highlighted that IFI
disbursements may reflect the outburst of a financial crisis, regional income patterns
(with poor countries likely to receive a disproportionate amount of funds relatively to
their GDP) and the direct interests of influential shareholders. In an attempt to capture
this set of factors we relate the flows of annual World Bank and IMF disbursements
toward a given region to the following variables:
• A set of dummy variables taking a value of one if a balance of payments crisis did
take place in that year, and zero otherwise. While financial crises would be better
measured at the country level, most of such events, at least in the 1980s and even
more so during the 1990s, occurred at the regional level. We feel, therefore,
confident that our regional balance of payments crisis variable adequately
captures periods of financial stress. Table 2 provides more detailed information on
such variable.
• Regional fixed effects, which are intended to capture both the regions’ ranking in
per capita incomes and other time invariant factors. To allow for changes in the
regional ranking of per capita incomes we also tested whether regional fixed
effects showed a break between the 1980s and the 1990s.
• The trade and financial exposure of influential shareholders toward a given
region17. Trade exposure is simply measured as the ratio of exports toward a
certain region of destination to total shareholder’s exports. Alternatively, we
could have use total trade (rather than exports), but this would not have changed
much the results, being the two highly correlated. As noticed before, financial
exposure is measured as the ratio to GDP of consolidated banking claims of an
influential shareholder toward a capital-recipient region.
As dependent variable we take the ratio of World Bank and IMF disbursements to
regional GDP. This is to uncover the link between the trade and the financial exposure of
the main shareholders and the patterns of IFI disbursements. Otherwise, the small weight
of say Africa in the total exports of Europe and the United States would seem to bode ill
for the ability of the region to benefit from IFI loans. However, while it is obviously true
that large regions, or countries, should have a larger clout in accessing the resources of
IFI’s, their needs will be proportionately larger, thereby placing a significant burden on
the limited resources of such institutions and crowding out the lending possibilities open
for other countries. In this sense, therefore, Africa could still benefit from the lobbying
16 We plan to extend our analysis to a data set that relies on individual countries rather than on regional
aggregates.
17 In a closely related paper, Rodrik (1995) measures political influence through a set of dummy variables
that take a value of one if a given country is deemed to be a close friend of the US (France, OPEC).
13