effects vary across regions. While inflation has been a deterrent factor to FDI inflows to
Latin American countries, inflation has had a smaller marginal effect in Asian and African
countries. Also, high economic growth rates have been relatively more important for Asian
countries than African and Latin American countries.
A number of observations from Section 3 are confirmed by the general-to-specific analy-
sis. International openness (trade) is important in all regions except Asia; the stability of
the exchange rate regime is important for Asian and Latin American countries; financial
and political stability (external debt, current account balance, corruption and rule of law)
are important in Latin American countries; while low wages have been a comparative
advantage in Asian countries. The results also indicate that the focus on economic risk in
studies of FDI into Asia is misleading since political and commercial risk (political rights
as well as voice and accountability) are equally important for this region.
International openness has typically been proxied by total trade as a share of GDP,
the import share or the export share. Since trade is a linear combination of imports and
exports it is not possible to include all three of them at the same time. Table 7 reports the
results when we use the import and export shares instead of total trade as our openness
proxy. The results from Table 6 are confirmed and we see that the positive effect of trade
was driven by import, which was also the case in Janicki and Wunnava (2004) and Ferris
et al. (1997).
Also, we find that some variables are region-specific: GDP per capita, land area, roads,
international reserves and government expenditure for Africa; wage earnings, political
rights and the Kaufmann voice and accountability index for Asia; and telephone, external
debt, corruption and ores export for Latin America. The remaining six variables lie
somewhere in between where four variables turn out significant in both Asia and Latin
America (urban population, current account, change in the exchange rate, variance of
the exchange rate) and two enter both the specification in Africa and Latin America
(trade openness and rule of law). This suggests that Africa and Asia do not seem to be
well described by the same set of variables and one should exercise caution when pooling
the two regions. Latin American and Asian countries can more readily be pooled but
interaction terms should still be incorporated to adjust for heterogeneity in the impact of
common explanatory variables.
5 Summary and Conclusion
This paper provides a comprehensive and structured review of the part of the empirical
literature that has analysed the regional differences in FDI inflows. A number of observa-
tions are worth highlighting. First, regional dummy variables rarely turn out to be robust
once structural characteristics of the host country are properly accounted for. Second,
the large number of significant interaction terms between regional dummies and selected