ABSTRACT
Guided by the above econometric strategy, we undertake the estimation of several randomeffects discrete-choice models using a large sample of around 3600 annual observations for 101 countries in the period 1970-2010, the longest for which consistent data are available. We estimate these panel data probit regressions using all right-hand side variables lagged one period to reduce potential biases arising from simultaneity. This bias, nevertheless, is not expected to be important since most of the variables are of institutional nature or move slowly in time, being usually less affected by conflicts contemporaneously or in the very short run. A detailed discussion of the econometric results follows in Tables III and IV. We report the marginal effects as they comprise sign, size, and statistical significance.
