suggests the existence of a link between secure property rights and reliable contract enforcement, on the one hand, and democratic political regimes, on the other. It also suggests that differences in regime type may be critical to successful adjustment only in the context of the constituent elements of their political capacity -- to the extent that their institutions are credible, that government policies are explicit, debated and negotiated, and their relative political extraction remains close to the norm given their level of development.
Tables 4.8 and 4.9 present the results of a logit analysis showing the differential impact of current and previous regime type on the probability of a change in political regime. The inclusion of additional variables to control for differences in regime longevity, current levels of development (measured by the log of current nominal GDP per capita and the value of agricultural production as a percentage of GDP), and the size of government consumption with respect to the private sector also test the hypothesis that type of political regime is endogenous, not exogenous, to domestic political and economic forces. Systematic extraneous effects of sample selection are controlled for by inserting a floating "country" constant. Therefore the impact that unknown countryspecific attributes can exert on the probability of regime change is built into the model.
Odds ratios are reported in place of maximum likelihood coefficients. The comparison group is the "Continuous Democratic" category. The values in the "Odds Ratio" column are the amounts by which the odds favoring Y = 1 are multiplied per one-unit increase in a particular X variable (assuming other X variables' values stay the same). 25 The z ratios and ρ > |z| are analogous to t-ratios and are interpreted in the same way. The magnitude of the odds ratios and zratios for the regime variables show that the survivability odds for continuous democratic or authoritarian regimes are the same, but for transitioning regimes they are significantly different. If a country is a transitioning democracy, the probability of it experiencing a regime change is 76 percent greater than that of a continuous democracy. If it is a transitioning authoritarian system, the probability of it experiencing a regime change is 99 percent greater than that of a continuous democracy. Every additional year that a regime remains in existence decreases the probability of regime change by 71 percent. Similarly the odds of the regime surviving are not significantly different if the previous regime was continuously authoritarian or democratic. The