Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.
Cheibub. 2002. Presidentialism and democratic performance. in The Architecture of Democracy: Constitutional Design, Conflict Management, and Democracy. Andrew Reynolds, ed. Oxford: Oxford University Press.
It has been argued in the lit that presidential democracies are less stable because (1) presidential democracies might have divided government, which (2) leads to deadlock, (3) which leads to bad things (in this study, democratic failure or electoral unaccountability). Cheibub examines each claim in turn.
Cheibub finds that divided government and deadlock do not affect democratic failure or accountability; when they do seem to matter, the effect is in the opposite direction. Thus, although we know that presidential regimes fail more often the parliamentary ones, those who would blame divided government and deadlock for the difference are incorrect. It's still a mystery, then, why presidentialism has a higher failure rate than parliamentarism. Cheibub speculates that reelection bans and interactions between party system and president's legislative powers may be playing key roles.
Cheibub shows, first, that presidentialism leads to divided government more frequently when (a) there is a higher effective number of parties, (b) legislative and executive elections are non-concurrent, and (c) electoral rules are more proportional than majoritarian. But it is not a given that divided government will happen in presidential systems.
Next, he shows that deadlocked government is possible in only some of the presidential democracies. First, there must be a veto institution, otherwise divided government doesn't matter. Second, the veto override mechanism must require more than a simple majority in the legislature, otherwise there is no deadlock (the legislature wins). And third, in those states that have both a veto institution and a supermajority override rule, the president's party must control (1) enough seats to sustain vetoes and (2) few enough seats that they aren't the majority. Based on this, Cheibub creates a variable to indicate the specific country-years in which deadlock is even possible.
Coups and Assassinations: It has been argued that presidentialism is bad because it allows deadlock, which leads to a search for extra-constitutional means (coups, assassinations, etc). But as Table 5.9 shows, deadlock conditions do not lead to an increased probability of transition from democracy (relative to non-deadlock conditions). Neither does divided government or almost anything else, with the "qualified exception" of multipartism.
Unaccountability: Some argue that divided government hurts accountability, since it makes it difficult to know whom to blame for the economy's performance. Cheibub argues that presidents are accountable if their "hazard rate" (probability of staying in office) correlates with the economy's performance. He tests this for presidents eligible for reelection, and again for all presidential parties (i.e. does the party retain the presidency or not). In both cases, the economy does not seem to influence the hazard rate at all. Moreover, divided government seems to help the president stay in office. (Cheibub doesn't say it, but this seems to suggest that divided government does hurt accountability.)
If the literature's arguments are incorrect, then how can we explain the fact that presidential regimes fail more frequently than parliamentary regimes? Cheibub proffers two speculations.
First, term limits: Many presidents are barred from seeking reelection. However, those that are allowed to run are far more likely to win than not--unlike prime ministers, who tend not to win reelection. If we assume that all presidents (even those who can't run for reelection) are this popular, then we have a problem: A popular president is forced by term limits to leave office, but he is popular enough that he might think he can get away with refusing to step down. Thus, it is the term limits that make democracy dangerous. A better way of reducing the president's incumbency advantage might be to use spending limits, free access to media (for opposition), and so on rather than term limits, since this allows the president to run but nonetheless not have too much advantage.
Second, we know that presidents frequently build coalitions with people out of their party. Thus, divided government might not be the problem. The larger problem might be that some presidents lack the resources necessary to influence legislative outcomes, so they resort to extra-constitutional means.
Research by the same authors
Research on similar subjects