Democracy, Dictatorship, and Term Limits

6. Endogenous Term Limits: Statistical Analyses

6 Endogenous Term Limits: Statistical Analyses In this chapter I bring together the previous discussion of the major obstacles that the presidents have to overcome when they desire political immortality (chapter 3) and of the value of political office (chapter 4). The previous chapter explored how the costs and benefits of holding office influenced presidents’ behavior in three cases. Here I turn to empirical estimations and attempt to map the costs and benefits of holding office to the observable outcomes of presidents extending their tenure or stepping down. I take several approaches to statistical analyses in this chapter. Following detailed discussion of the operationalization of dependent and independent variables, I estimate how the stakes of losing office increase the probability of attempts to prolong one’s stay in power. I examine whether a higher value of holding office manifested in larger size of Public sector, of Government procurement markets and of Natural resource rents, as well as a lower probability of retaining the spoils and immunity after expected exit from office (bad Previous ruler’s fate, low values of Contract-intensive money and the existence of personal Immunity concerns), for a given level of constraints, among which I distinguish elite-level constraints, collective action constraints, and international pressure, can explain whether leaders are more likely to take over their regimes and prolong their tenure. Next I turn to separate analyses of the effects of spoils and immunity concerns in democratic and partly democratic regimes, in partly democratic regimes only, and in nondemocratic regimes, in order to gauge whether the logic of costs and benefits of holding office holds in different political regimes. Results are supportive of the expectations and indicate that in substantive terms the most significant predictors are the size of public sector, protection of property rights, the fate of the previous ruler, 146 Endogenous Term Limits 147 personal control over president’s own party, and democratic constraints. The results are robust to different model specifications, including instrumental-variable probit regression, different measurements of the magnitude of spoils and of immunity, and a different specification of dependent variable. For ease of discussion, the majority of the complementary analyses that gauge the robustness of results are reported in the appendix to this chapter; the variables are described in detail in an online appendix to this book. In the final section, building on the preceding discussion of various institutional and international constraints in chapter 3, I additionally disaggregate the composite measure of constraints and examine which institutional factors are the most important in deterring the sitting presidents. Results indicate that presidents that manage to extend their stay in office usually face a compliant judiciary, a weaker legislative opposition and a larger number of effective legislative parties, which suggests that these leaders rule in a more institutionally malleable regime. Likewise, leaders are responsive to their international environment, and I find that presidents facing lower linkages with the West, fewer democratic neighbors in the same region, and those that have witnessed numerous examples of successful tenure extensions by their peers, again, in the same region, are more likely to subvert constitutions. Extending and Stepping Down: The Dependent Variable In chapter 3 earlier I explained the possible ways that presidents turn to in order to remain in power. In the empirical analyses I simplify the choice of available strategies to two: to extend tenure by whatever means or step down, and construct a binary dependent variable. There are two potential challenges to such a specification. First, it is possible that a small number of presidents that extended for one additional term only were driven by different considerations than their more dictatorial peers, as argued previously. If this is the case the model can be mis-specified, and two separate models or the model with multiple outcomes would be more suitable. I address this possibility in the section on robustness in the appendix to this chapter by distinguishing between short-term — those retaining formal limits — and long-term — those that can be extended for unlimited periods — prolongation. Second, the binary-outcome variable will not distinguish between cases where presidents stepped down without a fight and when they fought to extend their terms and failed. But how can we define an 148 Democracy, Dictatorship, and Term Limits Figure 6.1: Selection of Presidents for Analyses yes died in office, deposed, still in power, etc. limits at entry? yes yes no completed terms and stepped...