602 International Organization tween democracy and a country's ability to attract FDI.This result is robust both under different model specifications and even a different measure of democracy.52 The substantive effects of different levels of democracy on FDI inflows are large. Countries that move from one standard deviation below the mean to the mean level of democracy-a change in democracy score from 3.03 to 10.9-increase FDI inflows an added 0.47 percent of GDP.A move to full democracy level mean would increase FDI as a percentage of GDP by 1.2 percent.The magnitude of these swings is quite remarkable,where the average level of FDI for the sample is 1.96 percent of GDP.A move from an authoritarian regime to a democratic regime increases FDI inflows by 60 percent. These positive results on the effects of democracy on FDI inflows remain ex- tremely robust under multiple specifications.To test the robustness of the democ- racy result,I have included a number of variables from the William Easterly Data set,including GOVERNMENT REPUTATION,EXPROPRIATION,CORRUPTION,RULE OF LAW,and BUREAUCRATIC QUALITY.33 The empirical results are reported in Table 3. None of these variables had any significant effect on the democracy variable's standard error or coefficient. Time-Series Cross-Sectional Results Cross-sectional empirical analysis is often criticized for its static nature.To ex- plore how domestic variables affect FDI inflows over time,I have constructed a time-series cross-sectional data set for 114 countries from 1970 to 1997.The meth- odology employed is an OLS regression with panel-corrected standard errors as recommended by Beck and Katz.34 All regressions were run with both random and fixed effects and with decade dummies.For presentational and theoretical rea- sons,I include only the fixed-effects regressions in the final tables.All estimates are consistent between the random and fixed-effects regressions unless noted.The unit of observation for the dependent variable is annual FDI inflows as a percent- age of GDP,as defined earlier.The independent variables GROWTH,DEVELOP- MENT LEVEL,MARKET SIZE,TRADE,GOVERNMENT CONSUMPTION,BUDGET DEFICITS, and DEMOCRACY are the same as used in earlier regressions.I have also included two measures of capital controls from Brune,Garrett,and Guisinger.35 Overall capital controls is a 9-point measure of the controls on inflows and outflows of capital,where a country is coded as a 9 if there are no controls on any capital flows.FDI INFLows is the same variable used in the earlier regressions.All 52.The empirical results are essentially unchanged under different measures of democracy.See Table 6. 53.Easterly 1999. 54.Beck and Katz 1995. 55.Brune,Garrett,and Guisinger 2001.tween democracy and a country’s ability to attract FDI+ This result is robust both under different model specifications and even a different measure of democracy+ 52 The substantive effects of different levels of democracy on FDI inflows are large+ Countries that move from one standard deviation below the mean to the mean level of democracy—a change in democracy score from 3+03 to 10+9—increase FDI inflows an added 0+47 percent of GDP+ A move to full democracy level mean would increase FDI as a percentage of GDP by 1+2 percent+ The magnitude of these swings is quite remarkable, where the average level of FDI for the sample is 1+96 percent of GDP+ A move from an authoritarian regime to a democratic regime increases FDI inflows by 60 percent+ These positive results on the effects of democracy on FDI inflows remain ex￾tremely robust under multiple specifications+ To test the robustness of the democ￾racy result, I have included a number of variables from the William Easterly Data set, including government reputation, expropriation, corruption, rule of law, and bureaucratic quality+ 53 The empirical results are reported in Table 3+ None of these variables had any significant effect on the democracy variable’s standard error or coefficient+ Time-Series Cross-Sectional Results Cross-sectional empirical analysis is often criticized for its static nature+ To ex￾plore how domestic variables affect FDI inflows over time, I have constructed a time-series cross-sectional data set for 114 countries from 1970 to 1997+ The meth￾odology employed is an OLS regression with panel-corrected standard errors as recommended by Beck and Katz+ 54 All regressions were run with both random and fixed effects and with decade dummies+ For presentational and theoretical rea￾sons, I include only the fixed-effects regressions in the final tables+ All estimates are consistent between the random and fixed-effects regressions unless noted+ The unit of observation for the dependent variable is annual FDI inflows as a percent￾age of GDP, as defined earlier+ The independent variables growth, develop￾ment level, market size, trade, government consumption, budget deficits, and democracy are the same as used in earlier regressions+ I have also included two measures of capital controls from Brune, Garrett, and Guisinger+ 55 Overall capital controls is a 9-point measure of the controls on inflows and outflows of capital, where a country is coded as a 9 if there are no controls on any capital flows+ fdi inflows is the same variable used in the earlier regressions+ All 52+ The empirical results are essentially unchanged under different measures of democracy+ See Table 6+ 53+ Easterly 1999+ 54+ Beck and Katz 1995+ 55+ Brune, Garrett, and Guisinger 2001+ 602 International Organization