John Carey – Presidentialism 25 Years After Linz

From the archives

This is the consolidated version of three guest posts by Professor John Carey. The posts are based on the keynote address that he gave to the Conference on Coalitional Presidentialism at St. Anthony’s College, Oxford, on May 2, 2014.

John Carey

Presidentialism 25 Years After Linz

Does constitutional regime type affect outcomes we care about? One of the most influential answers to that question was offered by Juan Linz about 25 years ago. At the time, most of Latin America was emerging from long stretches of military authoritarian rule. Politicians, activists, and academics were asking whether anything could be done to minimize the risk of repeating the region’s longstanding pattern of democratic breakdowns. I first encountered Linz’s paper on the Perils of Presidentialism, which was circulating in samizdat form in the late 1980s, just as I entered grad school.

I would summarize Linz’s central claim as, “If Latin America had had parliamentarism instead of presidentialism in the mid-20th Century, it might have avoided Pinochet’s regime in Chile, 20 years of military dictatorship in Brazil, Argentina’s Dirty War, Operation Condor (which pioneered the practice of ‘extraordinary rendition’ before that term was ever dreamed up), and a host of other catastrophes.” As a new grad student, this struck me as an incredibly exciting proposition – that if we could just get the formal rules right, we could avoid incalculable injustice, violence, and suffering. What more important challenge could political science take on than to figure out whether this was actually right? Could we figure out how to engineer constitutions to minimize the risk, even on the margin, of democratic breakdowns, and the parade of horrors that can follow?

With the wave of democratizations cresting in Latin America, and building in sub-Saharan Africa and throughout Central and Eastern Europe and Central Asia, it was a no-brainer that the survival of democratic regimes mattered. The effects of non-democracy in all these contexts were so apparent. And Linz made a compelling case that presidentialism undermined democracy. What’s more, of all the ways presidentialism did this, in Linz’s account, stifling the development of strong parties and stable party coalitions was the most important. Democracy didn’t endure on a national level without competition among viable political parties, and the central pillar of Linz’s critique was that parliamentarism fostered strong parties and collective accountability whereas presidentialism undermined them.

Linz’s concern, amplified by Mainwaring’s seminal article on the “difficult combination” of presidentialism and multi-partism, had a huge impact on how a whole generation of scholars studied presidentialism. An avalanche of research followed into whether and how presidentialism affects democratic outcomes, much of which directly challenged Linz’s claims.

What are the central post-Linzian lessons on presidentialism? One is that, Linzian skepticism notwithstanding, presidentialism, and its hybrid cousins that combine popularly elected presidents with some measure of cabinet dependence on parliament, have multiplied far more rapidly than pure parliamentarism in recent decades, to the point where there are now about equal numbers of regimes in each category. The research also showed that presidentialism is not necessarily a recipe for democratic regime collapse, or for the impossibility of stable party coalitions in support of presidents. Along similar lines was the evidence that strong formal authorities for presidents don’t necessarily doom presidential democracy. They might even provide tools for the maintenance of coalition-based presidentialism that is distinct from a Madisonian vision based on conflict between the elected branches.

A lot of the most influential scholarship came out of Brazil, a country that commands our attention for a variety of reasons. One is, of course, is that Brazil is a giant. Another is that it produced a remarkable cohort of scholars, both Brazilian and others, who have focused on Brazilian presidentialism. But as important as any of this is the drama of the story they had to tell. Remember that Brazil was often singled out in the early presidentialism scholarship as an institutional basket case – fragmented, personalistic, clientelistic parties, untrammeled presidential powers. Twenty years ago, it was seen as the perfect presidential storm, yet it transformed into everyone’s darling. For classic liberals, the former dependency theorist who used the presidency to deliver economic stabilization without ignoring poverty was irresistible. For the left, successive Worker’s Party presidents who extended social welfare programs and oversaw measurable reductions in economic inequality looked even better.

But if the evolving accounts of Brazilian presidentialism presented challenges to Linz, the post-Linzian comparative scholarship also provided support for a number of his original claims. Presidential democracies do operate differently from parliamentary ones on a number of counts. Legislative parties and coalitions exhibit lower voting unity under presidentialism than under parliamentarism. Presidents have lower “batting averages” than do parliamentary executives in getting their policy initiatives approved as law. There are more broken campaign promises and less ‘mandate accountability’ under presidentialism than under parliamentarism.

So is it possible to size up the progress in the study of presidentialism? What approaches have provided the most traction? I think the advances beyond Linz followed from two key shifts in how scholars of comparative institutions approached the study of presidential systems, but that each of those shifts has brought its own set of limitations and challenges.

One shift was the increasing appreciation of the problems that omitted variables, and endogeneity, present for the comparative study of institutions. Jose Cheibub’s focus on presidentialism in the aftermath of military dictatorships was a huge advance here, demonstrating that, planted on similar soil to parliamentary regimes, presidential democracies were not more likely to collapse – but that the soil conditions where presidentialism takes root are systematically less hospitable to democracy. Robert Elgie’s efforts to parse the impact of variants of semi-presidentialism on regime collapse were similarly exemplary. (And I, for one, am sympathetic to the argument that divided control over the cabinet encourages regime crises.)

Lurking behind the whole question of whether and how regime type might matter to democratic performance was a question Adam Przeworski posed in a paper about ten years ago called “Is the science of comparative politics possible?” The point – not really new then, and even more familiar by now – was that institutions are the products of context, and the array of factors encompassed by the word “context” here inevitably shapes the outcomes we care about – stability, democracy, prosperity, equality, justice, security, public goods provision, etc. Przeworski’s concern, of course, just foreshadowed the “identificationist” wave that was about to wash over our discipline.

We have to ask: Can presidentialism research respond to changing expectations and standards for inference that the identificationists demand? I’m not asking this question in a kind of rhetorical build-up to a big reveal. I honestly cannot think of a research agenda in political science that presents bigger challenges for identification and inference than the study of how constitutional regime type at the national level affects the quality of democracy. I don’t have a solution, but it would require an ostrich-like capacity for denial not to acknowledge the problem in a review of this sort.

The other key post-Linzian shift was the influence in research on comparative presidentialism of theories of legislative politics initially developed in studies of parliamentary democracy and of the US Congress. The list of names here is long – and includes Tsebelis, Laver, Shepsle, Schofield, Strom, Huber, Krehbiel, Cox, McCubbins, Feddersen, Diermeier, and others. I won’t risk tedium by rehearsing the long list of studies that have applied or adapted their theories by testing them against evidence from presidential regimes. What I want to do, instead, is to raise a warning flag that we have, I think, occasionally prioritized the theories as objects of our research over more basic questions about the quality of democracy – the sort of questions Linz would not have lost sight of.

In the last decade, I have reviewed more manuscripts than I care to recall that stated their central goals as “filling gaps” in the empirical examination of existing theories. Let me suggest that when we find ourselves describing our motivation this way, it’s time for a little introspection. I’m not without blame here. I’ve spent plenty of time deep in the weeds of legislative roll calls or district-level election returns, parsing data for evidence to support hypotheses the fascination of which escapes most of my colleagues – and reviewers – never mind my parents, siblings, or children.

I’m not saying all of our research needs to reveal how to prevent the next Dirty War. Shedding light on the factors that tip the balance of influence between executives and legislatures in one direction or another is a time-honored vocation. But Madison and Montesquieu and Locke all made the case that the balance of powers mattered because it affected the likelihood of tyranny. They persuaded their audiences that the separation of powers could affect outcomes that everyone recognized as important. And remember that, for Linz, too, tyranny was front and center. My worry is that, for all the theoretical advances in the study of presidentialism since Linz, we have too often lost sight of why we, or our audience, should care.

In an article published just recently in Democratization, Paul Chaisty, Nic Cheeseman, and Timothy Power emphasized why we should care, even if we now know (or are reasonably confident) that presidentialism does not necessarily pave a straight road to tyranny. They write: “Twenty years of research have shown presidentialism to be remarkably durable, and in particular its multiparty variant has vastly over-performed relative to early predictions … [However, the authors go on to wonder whether] … The very same presidential tools that enhance governability may also undermine accountability.” I agree that the governability—accountability trade-off is what scholars of presidentialism should be studying, and I want to highlight some recent studies of presidentialism that imagine new ways to think about accountability – ways that I think would be relatively easy to explain to your aunt at Thanksgiving, or if you got interviewed by Terry Gross or Melvyn Bragg.

In an article forthcoming in Comparative Political Studies, “The Successor Factor: Electoral Accountability in Presidential Democracies,” Ignacio de Ferrari builds on the tradition (from Powell, to Stokes, to Samuels & Hellwig) of measuring accountability as the ability of voters to reward or punish an incumbent, governing party for economic performance. De Ferrari codes presidential candidates from governing parties as either incumbents (who were eligible to run again), successors (candidates anointed by outgoing incumbents), or non-successors (unconnected to incumbent presidents).

De Ferrari shows that the link between economic performance and the incumbent party candidate’s electoral fate varies systematically with incumbent/successor/non-successor status. This isn’t shocking – indeed, it would be pretty surprising if this were not the case. But then consider that de Ferrari finds no relationship between the economic performance of the incumbent government and the status of the candidates the president’s party nominated. That is, economically successful presidents were no more likely than unsuccessful ones to be able to anoint their successors. These are puzzles, and they suggest that the factors that determine whether we are in the world of high or low accountability (at least as measured by electoral rewards for economic growth) are opaque in many presidential systems.

Consider also a couple of papers by Ryan Carlin and Shane Singh, also based on data from Latin America. In one, “Executive Power and Economic Accountability,” the authors show that the stronger the constitutional and partisan powers of the president in a given country, the stronger is the relationship between a survey respondent’s evaluation of the economy and her evaluation of the president. Again, not surprising – maybe even reassuring – this suggests that citizens, in the aggregate, adjust their expectations for presidential performance according to the authorities their president wields. Yet in an article forthcoming in Political Research Quarterly, “Happy Medium, Happy Citizens: Presidential Power and Democratic Regime Support,” Singh and Carlin show that respondents’ satisfaction with the performance of democracy in their country is non-monotonically connected to those same formal authorities. The conditions that foster the tightest possible bond between economic performance and presidential approval are not the same ones that foster the greatest satisfaction with democracy more generally. So accountability in presidential democracies is not an easy animal to track and hunt.

What I’m suggesting is close attention to how we assess accountability, about what citizens want from their democracies, and what those democracies ought to deliver. If GDP growth is our indicator of good economic stewardship, then we should fear for accountability when a political system does not appear to reward growth. But maybe citizen satisfaction, measured in surveys, is a more appropriate measure of good performance. Or some measure of agreement between public support for policies and their rate of adoption? Or something else altogether?

n a spasm of curiosity, I collected data on regime type (presidential, parliamentary, or semi-presidential hybrid) from Robert Elgie’s website for 131 countries with populations above 1,000,000. Of these 41 are parliamentary, 43 are presidential, and 47 are hybrids. 97 of  the 131 (34, 33, and 30, respectively) had Polity scores of 5 or higher in the most recent year. Then I collected data from the World Bank, Polity IV, the United Nations, and Transparency International on the most recent annual measures on a wide range of regime performance and policy outputs that I think any sentient observer ought to care about: levels of democracy and stability, poverty, economic inequality, taxation, corruption, physical insecurity, and the rule of law.

Parliamentary regimes are, on the whole, wealthier than presidential ones and than hybrids.  This graph shows the distributions of Purchasing Power Parity across the regime types:

Figure 1

A lot of the performance indicators I’m going to look at here are correlated with national wealth, the distribution of which is skewed and the effects of which are likely subject to diminishing returns. So the graphs that follow will be scatterplots, and some fitted plots, of various outcomes we should care about against a log transformation of per capita wealth. Presidential regimes are marked by red Xs, parliamentary regimes by green dots, and hybrid regimes by blue triangles.  We can look quickly at the scatters and size up whether one regime type or another is over-performing or under-performing, relative to others at the same level of wealth.

Looking first at democracy levels, as measured by Polity IV. Wealthier countries are more democratic, but there’s no clear pattern of any of our three regime types systematically over- or under-performing.

Figure 2

The same is true for regime stability, as measured by the World Bank’s Stability Perceptions Index, which reflects“perceptions of the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means, including politically-motivated violence and terrorism.”

Figure 3

So – so far, my crude, cross-sectional snapshots are consistent with the more systematic evidence presented by Cheibub and others that presidentialism, per se, is not inconsistent with democratic stability. But when we look at some further indicators of regime performance – again, with my crude measures – the picture for presidentialism is less encouraging.

Taxation is the cornerstone of government capacity to deliver public goods. We know that wealthier states tend to tax at higher rates, and of course parliamentary regimes are better represented at that end of the scale. Nevertheless, if we look at the distribution of regimes above and below the best linear fit line, parliamentary regimes are about twice as likely to be over-performers than under-performers, and for presidential regimes, the reverse is true.

Figure 4

In this plot, the green line shows the linear fit for the relationship between per capita wealth and taxation for presidential regimes, and the blue line shows parliamentary and hybrids pooled, and we can see that the positive relationship is driven by the latter set.

Figure 5

We might ask whether the distinct patterns for presidential and other regimes are driven by the inclusion of non-democratic cases, but dropping all regimes with Polity scores below 5 only strengthens it. Wealthier presidential democracies actually tax marginally less as a share of GDP than do poorer ones; the reverse is true among parliamentary and hybrid regimes.

Figure 6

Maybe the tax share of GDP is not an ideal measure of government accountability. Let’s consider some other things that are affected by government policies in pursuit of public welfare. The next graph shows Gini indices of economic inequality plotted against per capita wealth.

Figure 7

There’s substantial dispersion but, on the whole, wealthier societies are slightly less unequal than poorer ones. But again, look at the relative distribution of presidential, as opposed to parliamentary regimes above and below the best fit line. Or easier, here are the linear fits for presidential regimes and for the pooled set of parliamentary and hybrids.

Figure 8

Economic inequality rises with wealth among presidential regimes whereas it declines in the others. In an article entitled “Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens,” forthcoming in Perspective on Politics, but already lighting up the wonkosphre, Gilens and Page make the case that there is massive elite bias in influence over public policy outcomes in the United States. This graph raises the question: Does the Gilens and Page result generalize beyond the United States to other presidential systems?

If we look at poverty rather than economic inequality, we don’t get as dramatic a difference between regime types – richer countries tend to have lower poverty rates across the categories – but presidential regimes again under-perform on poverty mitigation.

Figure 9

There is a discernible difference between presidential regimes and others, with poverty rates about 10% higher across the range of income levels.Figure 10

Another conventional indicator of government accountability is corruption, so let’s take a look at Transparency International’s Corruption Perceptions Index. (Remember, higher TI scores represent less perceived corruption among its survey respondents.)

Figure 11

The pattern here is a little less stark, but the fitted plots suggest that, as countries increase in per capita wealth, the rate of improvement on corruption is flatter among presidential regimes than among parliamentary systems and hybrids.

Figure 12

We get a similar kind of pattern if we consider another key government function – guaranteeing individual physical security. The data here show homicide rates from the comprehensive United Nations report released last month. Again, there’s a general pattern of greater security in richer countries, but the rate of improvement with wealth is flatter among presidential regimes than the others.

Figure 13

Finally, we can look at the less concrete, but more catholic conceptions of Rule of Law, or of Accountability, compiled by the World Bank as Governance Indicators. These are based on a combination of survey responses and expert assessments. On both their Rule of Law index, and their Accountability index, we see the same familiar pattern, with improvement across wealth levels in presidential regimes lagging that in parliamentarism and the hybrids.

Figure 14

Figure 15

I want to emphasize that these scatterplots are just suggestive. I collected the data because, as I thought about what we’ve learned about presidentialism since Linz, I went back to Linz’s essay, and then reviewed much of the literature on presidentialism that followed. In part, I found myself conducting the inevitable scorekeeping exercise. Linz appears to have been more right about some things than others. (By the way, when I’m done playing, I’d be happy to have a record even close to his.) But when you read Linz, why he cared about regime type is never in doubt. So much outstanding research has followed Linz. My words of encouragement as we continue this work is that our scholarship should be as clear as Linz’s was with regard to why we care about the phenomena we study.

John M. Carey is the Wentworth Professor in the Social Sciences, and the chair of the Government Department, at Dartmouth College.  He is a member of the American Academy of Arts and Sciences, and the author of 5 books and of over 50 academic articles on democratic institutions.  Research, datasets, and further information about his work are available on his website at http://sites.dartmouth.edu/jcarey/

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.