This is a guest post by Pedro C. Magalhães, Researcher at the Institute of Social Sciences of the University of Lisbon, Portugal
If you are a Portuguese political scientist like myself, it is almost certain that, regardless of sub-field, at some point, you have been interested in “semi-presidentialism”. In my case, I am afraid it has happened again.
You see, semi-presidential regimes, where heads of state are popularly elected but where there are also prime ministers responsible before the legislature (in Elgie’s impeccable “dispositional” clarification of Duverger’s inicial concept), are very intriguing. While the “semi-“ part suggests something “weak” or “incomplete” about the presidency, presidents in these regimes can be enormously powerful, sometimes arguably more than presidents in presidential regimes, depending on the prerogatives constitutions assign them, given the lack of separation between legislature and executive, and their potential role as actual de facto majority leaders (think the French president). However, they can also be non-partisan former archaeologists and museum curators. So there’s some puzzlement to be enjoyed here.
The article I have written with Jorge Fernandes, now available for early view in the European Journal of Political Research, deals with a particular puzzle that much better political scientists than me have faced before when addressing a particular question about semi-presidentialism. It is a question about government stability. On the one hand, many excellent works have suggested, with the help of sheer logic but also deeply researched case studies, that two things tend to be particularly inimical of government stability under semi-presidential regimes: whether presidents can dismiss governments as will (what Shugart and Carey famously coined as a “president-parliamentary” regime, in contrast with “premier-presidentialism”) and whether the president’s party is absent from the cabinet (what has been called “cohabitation”). However, once one uses the statistical tools normally employed to systematically examine the determinants of government survival, neither presidential powers (including cabinet dismissal but also parliamentary dissolution) nor cohabitation seem to make much of a difference in that respect (see here, here or here, for example). So what’s going on?
Jorge and I examined the post-War political history of twelve European semi-presidential regimes, using the invaluable European Representative Democracy dataset, which contains data about the duration and several other features of cabinets in those countries. We then added a few variables collected on our own, particularly concerning presidential powers of discretionary dismissal of cabinets and dissolution of parliaments, and tracking how those powers have changed with time. Furthermore, we looked at the party affiliation of presidents (if any) and identified the contexts where the president’s party was in the cabinet, when it was not, and when the president was non-partisan.
Having done this, we then used survival analysis to determine how presidential powers, cohabitation, and their interaction affected the risks of cabinets falling, controlling for a bunch of other things that a very large body of research has shown to make a difference in this respect (minority v. majority cabinets, single-party vs. coalition, time until end of regular legislative term, investiture rules, which we found to have impacts similar to those found in other studies). And we added a small twist in relation to the previous literature: besides looking at the competing risks for a cabinet of ending through a parliamentary dissolution or through a cabinet replacement (without elections), we also look at two types of replacement: one where the new cabinet involved a change in the PM’s party, and one where it did not. This seemed important to us for the same reasons that it also seemed important to Cheibub and Chernykh: you can have a lot of apparent “instability” (successive and frequent cabinet replacements) which hides, nonetheless, a basic continuity in the cabinets’ make-up and control (an extreme – non semi-presidential – example is Italy’s First Republic).
So what do we find? First, in semi-presidential systems where presidents enjoy the discretionary power to dissolve parliament, the risk of a government being terminated through a parliamentary dissolution increases massively. Although this does not really seem earth shattering, it is nice to find it empirically confirmed. Furthermore, once we employ a refined (we hope) taxonomy of cabinet termination through replacement, the effects of “president-parliamentarism” (presidents than can dismiss cabinets) finally emerges with clarity: although such power does not affect the risk of “generic” cabinet replacement, it does increase the risk of replacements that involve a change in the partisan control of the government. And to top it off – and this emerged out of a reviewer’s suggestion – we find that “continuity replacements” (the cabinet is replaced without elections, but the partisan control remains the same) becomes more likely in the combination of premier-presidentialism and unified government, a finding that fits nicely with several arguments presented by Samuels and Shugart on how premier-presidential regimes work when the same party controls the presidency and the cabinet.
Interestingly, though, cohabitation emerges as mostly irrelevant in our story about cabinet stability. Although we tried to be as thorough as we could (and as the reviewers helped us to be), it may be the case that we are missing something, either in terms of the measurement of the phenomenon or model specification. But we do speculate a bit in the end about why this might be the case. Having a president whose principal is the electorate is a tricky thing. Will she remain a purely partisan actor, intent on bringing her party to the cabinet and throwing out opponents using her powers (when available)? Or will she consider median voter preferences, public opinion, and the state of the economy, i.e., everything she needs to consider in order to preserve the support of the majority of voters that brought her to power? Even when presidents are facing a non-friendly cabinet, their partisan concerns and their own office and electoral concerns may collide in such a way as to make “cohabitation” less of a threat to cabinets than previously thought. Jorge and I (and Luís) are now in the middle of a paper looking at how presidential powers, cohabitation, and the state of the economy interact in order to examine this a bit further.
Take a look if this interested you. Comments most welcome.