The Political Economy of FDI 611 of experts,who assign countries values in a number of economic and political categories and generate an aggregate measure of country risk using weighted av- erages.The Institutional Investor credit risk ratings come from a survey of roughly 100 international banks on the probability of default.Sticking to convention,I use the standard logarithmic transformation of both ratings.70 A small number of empirical studies have examined the economic determinants of country risk ratings.7!These studies have found that the level of economic de- velopment,the government's current account balance,and the level of country debt all are significant determinants of country risk.The baseline model in col- umn I reconstructs the economic determinants of country risk using data on the level of development(GDP per capita),debt(central government debt/GDP),and current account balance (current account/GDP),all taken from the World Bank's World Development Indicators. In Table 8,I present a simple OLS panel regression for seventy-nine countries from 1980-98 using the Institutional Investor and Euromoney risk ratings as the dependent variables.As expected,in all models the level of economic develop- ment has a positive and statistically significant effect on country risk ratings,while the level of country debt has a negative effect.The current account deficit does not have a statistically significant effect on country risk,while GDP growth has a positive,although not statistically significant,effect.All models were also tested using controls for levels of inflation or exchange-rate variations.72 In both of the random-effects models,democratic institutions are associated with higher country sovereign debt ratings.In the first fixed-effects model-the model using the Institutional Investor scores as the dependent variable-democracy is positive and highly statistically significant.In the final model-the fixed effect Euromoney regression-I find no relationship between Euromoney ratings and po- litical regimes.This last finding is not surprising,given the stability of Euromoney ratings over time.Essentially,the fixed-effects,country dummies do most of the work in this regression.Other important controls that do not vary much over time, such as the level of economic development,are also only weakly statistically significant. In summary,democratic institutions,when all other economic factors are con- trolled for,are generally associated with lower levels of political risk in terms of sovereign default risk.This result sheds some light on the earlier finding that dem- ocratic governments attract higher levels of FDI.As stated earlier,the political risks involved with multinationals'investment decisions are similar to those faced by multinational corporations investing in foreign markets.Although this is not a 70.This is the standard transformation used by Feder and Uy 1985,followed by Cosset and Roy 1990 and Lee 1993.The formula for the transformation is Dependent variable In[R/(1-R)],where R represents the Institutional Investor or Euromoney Rating divided by 100. 71.See Feder and Uy 1985;Cosset and Roy 1990;and Lee 1993. 72.I tested all models with controls for the average annual consumer price inflation and the real effect exchange rate using the World Bank's World Development Indicators 1999 data.of experts, who assign countries values in a number of economic and political categories and generate an aggregate measure of country risk using weighted av￾erages+ The Institutional Investor credit risk ratings come from a survey of roughly 100 international banks on the probability of default+ Sticking to convention, I use the standard logarithmic transformation of both ratings+ 70 A small number of empirical studies have examined the economic determinants of country risk ratings+ 71 These studies have found that the level of economic de￾velopment, the government’s current account balance, and the level of country debt all are significant determinants of country risk+ The baseline model in col￾umn 1 reconstructs the economic determinants of country risk using data on the level of development ~GDP per capita!, debt ~central government debt0GDP!, and current account balance ~current account0GDP!, all taken from the World Bank’s World Development Indicators+ In Table 8, I present a simple OLS panel regression for seventy-nine countries from 1980–98 using the Institutional Investor and Euromoney risk ratings as the dependent variables+ As expected, in all models the level of economic develop￾ment has a positive and statistically significant effect on country risk ratings, while the level of country debt has a negative effect+ The current account deficit does not have a statistically significant effect on country risk, while GDP growth has a positive, although not statistically significant, effect+ All models were also tested using controls for levels of inflation or exchange-rate variations+ 72 In both of the random-effects models, democratic institutions are associated with higher country sovereign debt ratings+ In the first fixed-effects model—the model using the Institutional Investor scores as the dependent variable—democracy is positive and highly statistically significant+ In the final model—the fixed effect Euromoney regression—I find no relationship between Euromoney ratings and po￾litical regimes+ This last finding is not surprising, given the stability of Euromoney ratings over time+ Essentially, the fixed-effects, country dummies do most of the work in this regression+ Other important controls that do not vary much over time, such as the level of economic development, are also only weakly statistically significant+ In summary, democratic institutions, when all other economic factors are con￾trolled for, are generally associated with lower levels of political risk in terms of sovereign default risk+ This result sheds some light on the earlier finding that dem￾ocratic governments attract higher levels of FDI+ As stated earlier, the political risks involved with multinationals’ investment decisions are similar to those faced by multinational corporations investing in foreign markets+ Although this is not a 70+ This is the standard transformation used by Feder and Uy 1985, followed by Cosset and Roy 1990 and Lee 1993+ The formula for the transformation is Dependent variable 5 ln@R0~1-R!#, where R represents the Institutional Investor or Euromoney Rating divided by 100+ 71+ See Feder and Uy 1985; Cosset and Roy 1990; and Lee 1993+ 72+ I tested all models with controls for the average annual consumer price inflation and the real effect exchange rate using the World Bank’s World Development Indicators 1999 data+ The Political Economy of FDI 611