608 International Organization TABLE 7.Selection models of democracy and FDI Variables Standard OLS Selection OLS LAGGED FDI 0.308** 0.310** (3.962) (14.533) MARKET SIZE -0.608 -0.626 (-1278) (-1.387) DEVELOPMENT LEVEL 0.023** 0.023** (2.688) 3.101) GROWTH 0.939* 0.796 (2.055) (1.485) TRADE 0.007 0.007** (1.550) (2.350) BUDGET DEFICIT -0.025* -0.024** (-2.171) (-2.954) GOVERNMENT CONSUMPTION -0.045** -0.046*** (-2.440) (-3.298) FDI INFLOWS CONTROLS -0.037 -0.025 -0.267) (-0.129) DICTATORSHIP -0.369** -0.964*率* (-3.800) (-2.891) Time dummies Yes Yes Country dummies Yes Yes Observations 1568 1568 Countries 104 104 Rho 0.531** (7.923) Sigma 1.479*** (37.253) Lambda 0.786** (6.787) LR test,Chi-sq (probability) 11.28** (0.0008) Note:All regressions are ordinary least squares(OLS)regressions using annual net FDI inflows as a percentage of GDP as the dependent variable.LR test likelihood ratio test. *p<0l,*p<.05,*p<.10. GDP per capita and the number of past democratic breakdowns to generate probit estimates of the existence of democratic regimes,and then use this predicted re- sult in a standard OLS regression with country and time dummies.65 For this regression,all variables are the same as employed earlier,except that I have substituted the Polity III measure of democracy with a dichotomous measure of dictatorship from Alvarez et al.because of the need for a dichotomous measure 65.This is a similar empirical technique to that employed by Przeworski et al.2000.They find that the level of economic development and the number of transitions from authoritarian rule correctly predict 77.7 percent of the political regimes from 1950-90.GDP per capita and the number of past democratic breakdowns to generate probit estimates of the existence of democratic regimes, and then use this predicted re￾sult in a standard OLS regression with country and time dummies+ 65 For this regression, all variables are the same as employed earlier, except that I have substituted the Polity III measure of democracy with a dichotomous measure of dictatorship from Alvarez et al+ because of the need for a dichotomous measure 65+ This is a similar empirical technique to that employed by Przeworski et al+ 2000+ They find that the level of economic development and the number of transitions from authoritarian rule correctly predict 77+7 percent of the political regimes from 1950–90+ TABLE 7. Selection models of democracy and FDI Variables Standard OLS Selection OLS lagged fdi 0+308*** 0+310*** ~3+962! ~14+533! market size 20+608 20+626 ~21+278! ~21+387! development level 0+023*** 0+023*** ~2+688! ~3+101! growth 0+939** 0+796 ~2+055! ~1+485! trade 0+007 0+007** ~1+550! ~2+350! budget deficit 20+025** 20+024*** ~22+171! ~22+954! government consumption 20+045** 20+046*** ~22+440! ~23+298! fdi inflows controls 20+037 20+025 ~20+267! ~20+129! dictatorship 20+369*** 20+964*** ~23+800! ~22+891! Time dummies Yes Yes Country dummies Yes Yes Observations 1568 1568 Countries 104 104 Rho 0+531*** ~7+923! Sigma 1+479*** ~37+253! Lambda 0+786*** ~6+787! LR test, Chi-sq ~probability! 11+28*** ~0+0008! Note: All regressions are ordinary least squares ~OLS! regressions using annual net FDI inflows as a percentage of GDP as the dependent variable+ LR test 5 likelihood ratio test+ ***p , +01, **p , +05, *p , +10+ 608 International Organization