Are Parliamentary Systems Better?
Comparative Political Studies
Volume 42 Number 3 March 2009 327-359 © 2009 Sage Publications 10.1177/0010414008325573
http:// cps. sagepub.com hosted at http://online.sagepub.com
John Gerring Strom C. Thacker Carola Moreno
Boston University
The institutional differences between presidential and parliamentary rule are well known, yet the practical effects of these divergent constitutional arrangements within democratic polities have received scant attention. This article employs a global data set to test the relationship between a historical measure of parliamentary rule and 14 indicators ranging across three policy areas: political development, economic development, and human development. The study revealed a strong relationship between parliamentarism and good governance, particularly in the latter two policy areas. To the extent that these institutions influence the quality of governance, parliamentary systems may offer advantages over presidential systems of democratic rule.
Keywords:  presidentialism; parliamentarism/parliamentary; governance; development; separation of powers
The constitutional structure of the executive has been a major research question in political science since the birth of the discipline. Beginning with Walter Bagehot and Woodrow Wilson, scholars of American and European politics have pondered the effect of unified and separated powers on campaigns and elections, on voter behavior, on the behavior of party elites, on the workings of the bureaucracy and the courts, and more generally, on the policy-making process. More recently, the structure of the executive has emerged as an important question within the field of comparative politics, a consequence of the rise of new democratic polities around the world. Thus, many of the same questions that arose in comparisons between the United States and the United Kingdom are now being raised in a global context.
Underlying much of this work within American and comparative politics is the implicit normative question, are parliamentary or presidential systems superior? Which method of structuring the executive leads to better governance?
327
328    Comparative Political Studies
It is difficult to read the literature on parliamentarism and presidentialism without catching a strong whiff of these underlying normative concerns (e.g., Haggard & McCubbins, 2001; Lijphart, 1992; Mainwaring & Shugart, 1997; Skach, 2005; B. Wilson & Schramm, 1994). And it is right that they should be there, for the practical question facing constitution makers around the world is one of institutional choice. Taking all possible causal effects into account, the question then becomes, which system is preferable? We live in an age of constitutional reform, where core elements of the polity are subjected to continual criticism and not infrequent amendment (Reynolds, 2002; Sartori, 1994). If political science cannot offer guidance on this matter, then the discipline is, in essence, passing the buck. Moreover, this fundamental normative question—if answerable—may lead to significant theoretical gains for the discipline. It is not simply a question of exigency.
Of course, it may be that neither system is better overall. Conceivably, the only consequential difference between these two constitutional molds is that one system (presidentialism), by virtue of greater institutional fragmentation, offers greater resistance to change (Tsebelis, 2000). There may also be as many differences within each constitutional type as there are between multiple types. Indeed, presidentialism and parliamentarism are unified labels for variegated realities (Haggard & McCubbins, 2001; Mainwaring, 1993; Shugart & Carey, 1992; Tsebelis, 1995). Alternatively, parliamentarism and presidentialism each may demonstrate varying strengths and weaknesses along different policy dimensions (Weaver & Rockman, 1993). Or it could be that the performance of the executive is contingent on cultural, socioeconomic, and historical factors that vary from country to country and from period to period. If so, parliamentarism may be more appropriate in some contexts and presidentialism, in others. The modal position among political scientists appears to be that the issue is complicated and that no clear and consistent advantage can be found for either executive type.
In any case, questions regarding the overall governance impact of these two fundamental constitutional forms seem well worth posing. Without taking a broad cut at these matters, we cannot hope to answer the question of whether there might be distinct strengths and weaknesses along different policy dimensions, among different regions of the world, or across different periods or whether there is in fact some consistent result that holds across dimensions, measures, regions, and eras.
It is striking that the question of executive organization and its effect on the quality of governance has received so little systematic crossnational empirical analysis, given its theoretical and normative importance. There is,
Gerring et al. / Are Parliamentary Systems Better? 329
of course, plenty of speculation on these matters, a bit of formal modeling, and quite a number of case studies. In addition, many large-sample cross-national studies have been conducted on what might be termed the proximate causal effects of presidentialism and parliamentarism—for example, on electoral behavior, legislative behavior, interbranch relationships, cabinet formation, the operation of the bureaucracy, the policy-making process, the organization of interests within civil society, and the size of the welfare state (e.g., Samuels, 2007). Yet, none of these outcomes has unambiguous ramifications for the quality of governance. Indeed, the only outcome of obvious normative importance that has been subjected to extensive cross-national empirical analysis is the purported affinity between parliamentarism and democracy, a matter about which there seems to be little scholarly agreement (contrast Cheibub, 2007 and Stepan & Skach, 1993).
Our concern here is with issues of governance internal to democratic polities. That is, given a modicum of multiparty competition, what are the policy effects of different executive structures? This question has scarcely been broached.1 Granted, a number of writers have tested aggregated models of democratic governance, which bring together a number of political institutions into a single variable—for example, the consensus model (Lijphart, 1999), the checks index (T. Beck, Demirguc-Kunt & Levine, 2000), the political constraints index (Henisz, 2000), and the centripetalism index (Gerring, Thacker, & Moreno, 2005). Our interest here is more narrowly targeted on the construction of the executive and its possible role in achieving good governance.
In attempting to judge this matter, we observe three broad policy areas— political development, economic development, and human development— composing a total of 14 specific outcomes understood as indicators of good or bad governance across these various dimensions. The first section of the article presents the theoretical debates surrounding the efficacy of presidential and parliamentary systems. These debates generate the competing hypotheses tested in the empirical section, and the article concludes with a discussion of the results and implications of the research.
Theoretical Debates
The debate over presidentialism and parliamentarism has a long history, and the terms of this debate have changed little over the past century. The debate persists despite the fact that the proximate effects of different executive structures are generally accepted. Ceteris paribus, it is understood that
330    Comparative Political Studies
parliamentary systems foster stronger political parties, more centralized and party-aligned interest groups, a more centralized decision-making process, and more centralized and hierarchical administrative structures. These differences are aptly, if somewhat ambiguously, summarized by the concept of veto points (Tsebelis, 1995). Separate powers introduce an additional veto point into the political process, fragmenting politics at all levels. The effect of this concentration or dispersion of power is to create a contrast between systems that are decisive and those that are resolute (Cox & McCubbins, 2001). Parliamentary systems are decisive; presidential systems, resolute. In addition, parliamentarism fosters a style of politics and policy making that is probably more institutionalized, centered as it is on political parties, whereas presidentialism fosters a more personalized and free-floating style of leadership centered on individual politicians and smaller, less established organizational entities.
Yet, none of these proximate causal effects has clear-cut implications for the quality of governance in a polity. Partisans of separate powers prize the diffusion of power, whereas parliamentarists prize its unification. Thus, most of the arguments that we review here may be understood as being derivative of a single centralist-decentralist dimension.
Consider, first, the question of party organization. There is general agreement that the existence of an independently elected executive with strong policy-making prerogatives weakens party cohesion (Shugart, 1998). Members of the same party are less likely to vote together in the legislature, and party allegiances are weaker in the electorate. But there is intense disagreement over whether strong (cohesive) political parties foster good governance. Supporters of the responsible party government model view strong political parties as the linchpin of democratic accountability and effective governance (American Political Science Association, 1950). Critics of this ideal point out that undue partisanship may blind voters and legislators to the public interest and may prevent them from reaching compromise. Good public policy, not the interests of a party, should guide behavior in the ballot box and on the floor of the assembly (Herring, 1940).
Consider, next, the question of interest organization. A separate powers system is generally thought to encourage the formation of a highly fragmented, non-party-aligned ("independent") interest group community. The concept of separate powers is thus a key feature of interest group pluralism, whereas parliamentarism is a key feature of corporatism (G. Wilson, 1992). Pluralism may have positive ramifications for the quality of governance insofar as (a) a greater diversity of views will be expressed, (b) each group will retain a high degree of independence from political parties and the state,
Gerring et al. / Are Parliamentary Systems Better? 331
and (c) important groups will exercise an informal veto power over questionable legislation (Truman, 1951). It may have negative ramifications for governance to the extent that groups represent narrow ("special") interests, are able to veto legislation that serves the general interest, and defect rather than cooperate with other political actors (McConnell, 1953; Olson, 1982).
Consider, third, the question of bureaucratic organization. Lines of bureaucratic accountability are more complicated in a separate powers system because bureaucrats have two masters: their nominal chief executive and the controllers of the budget (Huber & Shipan, 2002; Moe & Caldwell, 1994). However, the effects of this dispersion of authority are disputed. Defenders argue that bureaucratic accountability is ensured through multiple principals, semi-independent agencies carrying strict and highly specific mandates, and overlapping jurisdictions such that bureaucrats check and balance each other and offer comparative benchmarks that make the task of policy effectiveness easier to assess (Breton, 1996). For example, the price of bribes may be driven down by the opportunity to forum-shop among various agencies (Rose-Ackerman, 1978). Opponents of bureaucratic fragmentation point out that it hinders cooperation among agencies, is less efficient, leads to problems of micro-management, and, by blurring lines of responsibility, erodes principal-agent accountability (Moe & Caldwell, 1994).
Consider, fourth, the question of transparency, presumably a key element of democratic accountability. Advocates of separate powers underline the fact that a direct vote for the chief executive leads to a transparent relationship between elections and electoral outcomes. Advocates of parliamentarism point out that postelection coalition negotiations affect only those circumstances in which no single party is able to gain a majority of the legislature. They also point out that these coalition negotiations are usually quite predictable (it is rare for ideologically distant parties to coalesce); often, they are explicitly advertised such that two coalitions in waiting face off in the general election. In either instance, they scarcely attenuate the level of accountability that exists between the voter and his or her chosen party.
Consider, fifth, the question of information. Advocates of presidential-ism rightly point out that the division of power between executive and legislature creates an information-rich environment. Each branch has an incentive, and the requisite constitutional authority, to investigate the other branch; each also has an incentive to publicize (either by formal proclamation or informal leaks) information favorable to the achievement of its political power and policy preferences. Thus, wherever there is disagreement between the two branches or some important actors within these rather fragmented bodies, we can expect that the public will learn a good deal
332    Comparative Political Studies
about policies and processes of government (Persson, Roland, & Tabellini, 1997). Advocates of parliamentarism counter that more information does not always lead to better governance. There is such a thing as too much information or, worse, misleading information. If members of the two branches are constantly attacking each other, engaged in vituperative "smear" campaigns, then the resulting information will not serve as a useful check against bad policies. It will instead enhance citizen apathy and alienation (Cox & Kernell, 1991). Parliamentarists point out that plenty of information is publicly available in the normal workings of a democratic polity with a free press. The more important issue is how that information is framed and disseminated.
Consider, sixth, the vexed question of electoral accountability. Advocates of separate powers see greater accountability at the local level—between constituents and their individual representatives, whose positions and legislative record are distinct from their parties and the presidential candidates of the major party. Advocates of parliamentarism see greater accountability at the national level—between constituents and their political party.
In the process of governing, advocates of separate powers see accountability arising from the diffusion of authority among multiple independent bodies, who are thereby able to effectively monitor and, if necessary, block one another's actions. For them, the process of accountability is continuous—as exercised through multiple bodies, each "checking and balancing" the other (Laffont & Meleu, 2001; Persson et al., 1997). Advocates of parliamentarism, however, see accountability arising from the concentration of authority in a single set of hands (the ruling party or coalition and its leadership). For them, accountability is a retrospective process of reward or punishment by which principals (voters) exact accountability from agents (elected officials). Whether the principal-agent relationship concerns voters and representatives or representatives and bureaucrats, the chief point at issue is the relative clarity of responsibility and the prospect of ex post facto reward and punishment. From this perspective, accountability is rendered more tenuous—and the tendency to play "blame games," more tempting—whenever responsibility for governance is diffused across multiple bodies (Fiorina, 1980; Linz, 1994).
However, strict accountability between elected officials and voters may not lead to better governance outcomes. Insofar as voters' demands are unrealistic, short-sighted, or simply wrong-headed, it may be useful for government officials to find ways to mask their responsibility for unpopular measures. Thus, measures that impose costs on powerful constituencies with less visible benefits to the general public or to other constituencies— for example, the closure of unneeded military bases, the lifting of trade
Gerring et al. / Are Parliamentary Systems Better? 333
barriers on protected sectors, and budget-balancing initiatives (raising revenue or reducing expenditures)—may benefit from the plausible deniabil-ity of responsibility by government actors. Because authority is more divided in a separate powers regime, it may be easier to engineer this gerrymandering for the public good (Pierson & Weaver, 1993).
Consider, seventh, the question of political change. Advocates of separate powers often emphasize the virtues of political stability. The primary task of politics is to ensure a stable political environment so that the behavior of political actors is predictable from one election to the next. And only where institutions are constitutionally separate can the government attain credible commitment to a set of policies (Henisz, 2000; Keefer & Stasavage, 2003). By contrast, advocates of parliamentary rule emphasize the problem of the status quo. For them, the chief political problem is to adapt to changing demands and changing circumstances. Political and economic sclerosis is the problem; flexible government, capable of overcoming strong vested interests and exercising leadership, is the solution (American Political Science Association, 1950; Olson, 1982).
The same contrasts also apply to the short term. Presidential systems are rigid, in the sense that, absent an impeachment, there is no way to remove a sitting president in between elections, which are usually held at fixed intervals (Linz, 1990; Shugart & Carey, 1992). This lends the separate powers system a high degree of short-term stability. Parliamentary systems are more flexible because the prime minister (and the ruling coalition at large) may be removed at any time by parliamentary vote and because elections may be called at any time. Stability in the executive is a virtue insofar as it allows leaders to credibly commit to policies and stay the course; it is a vice insofar as presidents may lose legitimacy or pursue wrongheaded policies (Linz, 1990).
The question of political change is framed differently if viewed from the lens of policy innovation, the proposal and consideration of new policy ideas. Arguably, a separate powers system, because it ensures more points of access and greater independence on the part of politicians (each of whom seeks to carve out his or her own turf on the national stage), provides a breeding ground for new ideas. Every interest group, as well as every candidate, is a potential policy entrepreneur. Yet, there is no guarantee that any of these ideas is worthwhile or politically viable. Many new ideas are called forth, but few are chosen.
Consider, eighth, the question of institutionalization. Parliamentarism fosters a highly predictable, institutionalized form of politics and policy making in which participants are part of the establishment. By contrast, presidentialism
334    Comparative Political Studies
fosters a more personalized form of political behavior in which presidents, legislators, interest group leaders, and even bureaucrats all enjoy a degree of independence from the institutions of which they are nominally members (Linz, 1990; Lowi, 1985). Each may play the role of policy entrepreneur, as we have noted. Each may also play an oppositional role. Under the influence of the media, this political structure may lead to greater political conflict than what would be expected in a parliamentary system, where negotiated settlements (often behind closed doors) are the norm. Although the institutionalization of conflict is generally a good thing, some may question whether parliamentarism fosters too much of a good thing. Achieving good governance is often a matter of achieving a proper balance between various attributes.
Consider, ninth, the question of contestation and consensus. In a separate powers system, conflict is endemic and continual. Each branch is assumed to represent a somewhat different constituency or the same constituency in different ways. Yet, because a higher threshold of consensus is necessary for agreement on any policy measure, it might be said that consensus is mandated by a separate powers constitution. By contrast, power in a parliamentary system is temporarily monopolized by a single party or coalition. Other groups may voice their opposition, but they have no formal mechanism by which they might affect policy outcomes. Consequential conflict is thus episodic, occurring during elections but not in between. Thus, perhaps a separate powers system is one where all parties engage in policy making at the same time (with greater or lesser influence on policy results), whereas a parliamentary system is one in which parties take turns (depending on electoral results and postelection coalition agreements).
In which sort of system is the public interest more likely to be served? Cox and McCubbins (2001; see also Linz, 1994) stipulate that "the greater the number of effective vetoes, the more private regarding will be the policies enacted" (p. 28). This, they argue, is a consequence of the bargaining process among veto holders,
where each veto player will be able to demand, and receive, side payments in the form of narrowly targeted policies. Thus, when the effective number of vetoes is great, even broad public policy will be packaged as a set of individual projects, or it will be packaged with narrowly targeted programs, (p. 28)
However, another sort of outcome is possible. If a single policy can be found that offers a Pareto improvement for all veto holders, then no bargaining is necessary (Buchanan & Tullock, 1962; Persson et al., 1997). Indeed, the multiplication of veto holders grants vetoes to persons who may wield them
Gerring et al. / Are Parliamentary Systems Better? 335
against special interest legislation or unwise policies. Veto power may prevent a variety of pathologies, including the tyranny of the majority, populist-inspired legislation and socially inefficient redistributive coalitions (Riker, 1982).
Consider, finally, the question of strength or energy ("decisiveness") in the executive. To some, this is a sign of danger, of corruption: an invitation to the abuse of power. To others, energy in the executive is a sign of state capacity. From this perspective, government must be empowered to do the right thing; otherwise, it will do nothing at all or, worse, different things at the same time. Cox and McCubbins (2001) point to the serious problem of indecisiveness encountered in many presidential regimes, particularly when operating under conditions of divided government. These problems include '"institutional warfare' of varying intensity, unilateralism (where the executive and the legislature attempt to circumvent each other in implementing policy), various forms of gridlock, greater fiscal pork and rents, and a tendency toward larger budget deficits" (p. 29). Those concerned about problems of indecisiveness look more favorably on parliament's concentration of political power in the hands of a single party or coalition.
In sum, academics and policy makers who are working in this area seem to agree on one thing and one thing only: Constitutional structures matter. They disagree on how constitutional structures, such as the separation of powers, affect the quality of governance in a democracy. On this central question there is no consensus, even though the proximate effects of presidentialism and parliamentarism are well understood and generally agreed on. Disagreements hinge on basic assumptions about the state, the citizenry, the workings of the marketplace, and the nature of politics. Not surprisingly, these opposing theoretical suppositions are long-standing. Arguments by Montesquieu, Madison, and their contemporaries (Vile, 1967/1998) are echoed by advocates of separate powers in the present era. Arguments by 19th-century parliamentarists such as Walter Bagehot, Woodrow Wilson, and Henry Jones Ford are reiterated by advocates of responsible party government in the 20th century (American Political Science Association, 1950).
It is a testament to the recalcitrance of a theoretical debate that it fails to progress. Each side remains entrenched; each argument retains plausibility. Indeed, there are strong theoretical grounds justifying both sides of the argument. Each virtue of presidentialism/parliamentarism can also be seen as a vice, depending on one's theoretical lens. They are mirror images of each other. Alternatively, arguments for each side may cancel each other out—that is, the null hypothesis. This is all that we can safely conclude on the basis of our brief review of the literature.
336    Comparative Political Studies
Is it possible to move forward toward a coherent—that is, unified— theory postulating why presidentialism or parliamentarism might lead to stronger governance? This is, of course, the ultimate objective. However, consider that the set of possible and plausible causal mechanisms listed in the foregoing pages is quite large and equally diverse. A single conceptual schema may not be able to capture these realities. Even if it could, it would likely become so overburdened with caveats and conjectures that it would serve as little more than a laundry list of possibilities. Alternatively, it is always possible to construct a coherent theory by the simple expedient of reductionism—that is, by ignoring issues that do not fit neatly into the specified theory. But this is hardly convincing, given the strength of these manifold claims in the present instance. It is also possible to create the appearance of coherence by maintaining a sufficiently high level of abstraction. Yet, such models tend to resist empirical testing: They may be true, or they may be false; we can never know for sure (Green & Shapiro, 1994).
In the absence of a genuinely coherent, operational, and plausible theory of how democratic institutions work, we propose here to reverse the conventional sequence of scientific inquiry. Rather than theorize first, then test, we propose to test first, in the hopes that fruitful theorizing will follow. It is a well-accepted feature of scientific progress that new theories often arise only when faced with a set of empirical findings that are striking enough to warrant theoretical exploration (Lakatos, 1978). Findings thus pave the way for theory. And it is a procedure followed in some recent work. For example, Persson and Tabellini's wide-ranging study (2003) of the economic effects of constitutions offers only a thin patina of theory, followed by an extensive empirical tour. Their influential book is aptly subtitled What Do the Data Say? We adopt a similar approach in the present context.
Fortuitously, the question of executive organization and governance is somewhat more tractable in empirical terms than it is in theoretical ones. We can measure inputs and outputs in the governance equation, even if we cannot test specific causal mechanisms. We therefore view the present study as a first step toward the formulation of a new theoretical framework. In the concluding section, we offer some thoughts on how the multiple threads of a potentially unifying theory might be knit together.
Classifying Executive Types
With the spread of democracy around the world in the postwar era, com-parativists face daunting challenges of definition and classification. What is a
Gerring et al. / Are Parliamentary Systems Better? 337
"parliamentary" system? How does it differ from a "presidential" one? How might we classify the various intermediate types? Although we must impose a single definition on this complicated matter, it is important not to exaggerate the differences among these various classifications, which usually affect only a handful of borderline cases (mostly falling in the "semipresidential" zone). Thus, any results, though to some extent contingent on matters of classification, are likely to be robust across alternative classifications.
We understand parliamentarism as a system of government in which the executive (the prime minister and cabinet: collectively, "the government") is chosen by and responsible to an elective body (the legislature), thus creating a single locus of sovereignty at the national level. Presidentialism, its contrary, is a system where policy-making power is divided between two separately elected bodies: the legislature and the president. The president's selection is usually by direct popular election, though it may be filtered through an electoral college (as in the United States); futhermore, the rules pertaining to victory (i.e., by relative or absolute majority) vary from country to country. His or her tenure cannot be foreshortened by parliament except in cases of gross malfeasance. She or he is actively engaged in the making of public policy and, in this sense plays a political (i.e., partisan) role.
Between these polar types we find various admixtures, known gener-ically as semipresidential polities (Bahro, Bayerlein, & Veser, 1998; Elgie, 1999; Roper, 2002; Skach, 2005). Thus, we conceptualize the parliamentary-presidential distinction as a continuum with two dimensions: first, the degree of separation (independence) between president and parliament (unity = parliamentary, separation = presidential) and whether there is any separation at all; second, the relative power of the two players (the more power the president possesses, the more presidential is the resulting system). We capture this complex reality with a three-part coding scheme (0 = presidential, 1 = semipresidential, 2 = parliamentary). As a general name for this three-part coding scheme, we employ the term parliamentarism.
However, because our interest lies in the causal effect of parliamen-tarism/presidentialism (hereafter, parliamentarism), we must concern ourselves with not only a country's current status but also its institutional history. How many years has it retained a parliamentary or presidential system? We suppose that it takes time for institutions to exert an appreciable effect on governance outcomes. A country that is switching from a presidential to a parliamentary system (or establishing a parliamentary system in a newly democratic or independent setting) should not expect to see immediate, dramatic changes in the quality of governance. Instead, these effects
338    Comparative Political Studies
are likely to cumulate as the new institutional rules begin to affect actions and expectations (Gerring, Thacker, et al., 2005).
To empirically represent this historical component, we created a new variable that draws on the annual ("raw") scores for parliamentarism described above. We assume that history matters but that recent history matters more. Thus, we calculate a moving, weighted sum of the annual scores of each of our underlying explanatory factors, beginning in 1901 and ending in the observation year. Weights capture long-term historical patterns while giving greater weight to more recent years. For instance, a country's score in 1996 is the weighted sum of its scores from 1901 to 1996; its score in 1998 is the weighted sum of its scores from 1901 to 1998; and so on.
Country years figure in this weighted summation process so long as a country surpasses a minimum threshold of democracy during that year. (Recall that we are concerned with democratic governance; parliamentarism has quite a different meaning in an authoritarian setting, if it has meaning at all.) We employ a relatively low threshold of democracy because we wish to include as many plausible cases as possible in our analysis and because we believe that the distinct dynamics instilled by parliamentarism or presidentialism should be in evidence in any situation where there is a modicum of multiparty competition. Thus, we include a country year in the following analysis so long as it obtains a score greater than zero, on a scale ranging from -10 to 10, on the Polity2 measure of democracy (drawn from the Polity IV dataset [Marshall & Jaggers, 2002]).
Dependent Variables
To test the possible causal effects of parliamentarism on the quality of governance, we must identify a set of outcome indicators that are measurable, valid cross-nationally, and reasonably clear in their normative implications (good or bad relative to the public interest). Length limitations preclude an extensive discussion of our choice of outcome variables (see Gerring & Thacker, 2008). Suffice to say, we include as wide an array of governance outcomes as possible: those that are survey based ("soft") and those that are directly measurable ("hard"), those that measure processes (e.g., the workings of the bureaucracy) and those that measure policy effort or policy outcomes. There is no best way to measure the quality of governance; our recourse is to extend the empirical purview to a range of plausible measures and to discern lessons from the broader patterns observed.
Gerring et al. / Are Parliamentary Systems Better? 339
It is especially important to take a wide cut at this material for the simple reason that any individual result may be spurious. Regardless of the care that one exercises in the choice of control variables and tests for robustness, it is still possible that empirical correlations with a given outcome variable may be driven by unmeasured factors (a problem of simultaneity or country-specific effects) or by measurement error. Cross-country regression is a sensitive empirical tool, and few results are robust across all specifications. However, if one observes a fairly consistent pattern across a range of associated variables, greater confidence is warranted.
Outcome measures number 14 in total and fall loosely into three policy areas: political development, economic development, and human development. Indicators of political development include two measures of corruption control: one from Political Risk Services (Howell, 1998) and another from the World Bank's governance indicators (Kaufmann, Kraay, & Mastruzzi, 2007). We also employ a measure of bureaucratic quality (from Political Risk Services). Finally, we include three additional World Bank indicators: government effectiveness, political stability, and rule of law (Kaufmann et al., 2007). In each case, higher scores suggest better governance in the area of political development.
We include five indicators of economic development: four based on the World Bank's World Development Indicators (2003) and one based on Euromoney (2004). The number of telephone mainlines per 1,000 inhabitants captures an important dimension of economic infrastructure and communication. Import duties as a percentage of imports provide a measure of trade protection, a negative governance outcome. Trade openness, as measured by the value of total trade divided by gross domestic product (GDP), is an outcome-based measure of a country's trade policy. To measure the overall investment climate, we use Euromoney's country risk ratings, which are based on polls of economists and political analysts and supplemented by quantitative data, such as debt ratios and access to capital markets. Finally, our broadest economic measure captures the level of prosperity within a country, as measured by real per capita GDP. Except in the case of import duties, higher values indicate better economic governance.
Three variables capture the level of human development in societies around the world, all based on the World Development Indicators (World Bank, 2003). Infant mortality rate (per 1,000 live births, age 0 to 1) is a critical health and quality-of-life indicator. Encapsulating the entire life cycle are estimates of life expectancy. Illiteracy measures the percentage of the population that is unable to read and write a sentence about daily life, in its
340    Comparative Political Studies
native language. Lower scores for infant mortality and illiteracy and higher scores for life expectancy indicate better governance performance.
Empirical Strategy
Our method of analysis is largely contingent on the nature of the evidence at hand. Where data are not comparable over time, as with the World Bank governance indicators, we employ a between-effects estimator. This approach averages the results of a series of cross sections, with data drawn from four adjacent time periods (measured every 2 years from 1996 to 2002).
Where comparable time-series data are available, we employ a time-series-cross-section format. Although this sort of analysis offers distinct advantages over simple cross-sections, it is subject to simultaneous spatial and temporal difficulties. We cannot employ a unit-based fixed-effect research design to address spatial issues, such as unobserved heterogeneity, because our causal variable, parliamentarism, does not sufficiently vary from year to year. Its movement through time is "sluggish" (S. Wilson & Butler, 2003; see also, N. Beck & Katz, 2001; T. Beck et al., 2000). We do, however, employ a set of regional fixed effects and a geographically weighted version of the dependent variable (see below) to help mitigate sample heterogeneity. With respect to temporal issues, we employ a statistical correction for first-order autocorrelation and a time-trend variable to control for possibly spurious correlations between heavily trended dependent and independent variables.
The literature on governance suggests an array of political, economic, geographic, demographic, and cultural factors as plausible control variables. Although we cannot measure all these influences easily in a global setting, we include all the standard controls plus some not-so-standard ones that, we surmise, may also play an important role.
A country's regime history is likely to influence the quality of governance (Gerring, Bond, Barndt, & Moreno, 2005). Thus, we include a variable that captures the cumulative democratic history of each country over the course of the 20th century. A country's level of economic development is also likely to be strongly associated with good governance. We control for this in standard fashion, by including GDP per capita (in logarithmic form). We measure this once, in 1960, rather than on an annual basis, to alleviate concerns about endogeneity between GDP and our various outcomes (World Bank, 2003). (Results obtained using contemporaneous annual scores for per capita GDP, rather than the base year, yield the same substantive conclusions as those
Gerring et al. / Are Parliamentary Systems Better? 341
presented here.) We include regional controls for Africa, Asia and Latin America / Caribbean to capture whatever cultural, demographic, geographic, historical, and socioeconomic factors might be common to those regions. Dummy variables for Africa and Latin America / Caribbean are expected to reflect lower levels of governance in those regions, when compared to others, whereas expectations for Asia are somewhat mixed but generally more favorable. We anticipate that a significant period of socialist rule (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 1999) will have positive effects on human development and negative effects on our political and economic development indicators (with the exception of telephone mainlines, with which we anticipate a positive relationship). Having an English legal origin is often thought to promote good governance (LaPorta et al., 1999). Other colonial systems are thought to have been uniformly bad for governance outcomes in the developing world and are hence encapsulated in this variable. To the extent that countries farther from the equator have better governance, latitude (absolute value, scaled to 0-1, logarithm; LaPorta et al., 1999) should correlate with better outcomes. Expectations for ethnic (and linguistic) fractional-ization (Alesina, Devleeschauwer, Easterly, Kurlat, & Wacziarg, 2003) are more tentative, but heterogeneity is generally expected to hamper the quality of governance in a country. To the extent that having a large population (total population, logarithm, World Bank, 2003) makes certain governmental tasks more difficult, population in general might be expected to diminish governance capacity (except in the case of investment rating, where expectations are ambiguous because population may serve as a proxy for market size). Distance (in thousands of kilometers) from the nearest financial center (Tokyo, New York, or London) is intended to capture the negative impact of geographic distance from the cores of the international economy. Production levels of oil (millions of barrels per day, per capita) and diamonds (rescaled to billions of metric carats per year, per capita) capture the resource curse (Humphreys, 2005); yet, these resources also provide sources of revenue and wealth. As such, expectations are mixed.
We also include a control variable that measures the average value of the dependent variable across all countries, weighted by the inverse of the geographic distance (in kilometers) of each country from the country in question. (In the case of GDP per capita, we weight the average value of the dependent variable by each country's share of trade with the observed country rather than by the inverse of the geographic distance between the countries.) Countries lying close to one another may display similar values for extraneous reasons (culture, geography, diffusion, etc.). Thus, we anticipate a positive sign for this variable. The inclusion of this variable in all
342    Comparative Political Studies
regressions should help minimize possible spatial autocorrelation in the sample.
We employ additional variables in selective regressions, as appropriate. Thus, a measure of Protestantism is included in the analysis of political development and Islam in the analysis of human development outcomes (both measured as a percentage of the total population). Prior research suggests that a Protestant heritage may improve state capacity (Gerring & Thacker, 2004), whereas having a large Muslim population may impede human development (Moon, 1991). Linguistic fractionalization (Alesina et al., 2003) substitutes for ethnic fractionalization in our analysis of illiteracy.
Because there is no standard benchmark model for any of these dependent variables, we conducted two tests for each dependent variable. The first is a full model, including all variables discussed above. The second is a reduced-form model, omitting most controls that do not pass the threshold of statistical significance (p < .10, two-tailed tests) in the expected direction. We retain the geography-weighted control and the time-trend (for time-series-cross-section models), democracy stock, and GDP-per-capita variables in all models, regardless of statistical significance, because of our strong expectation that these variables will capture important and otherwise unobserved effects.2
Results
Tables 1-3 present the results of a series of tests of the empirical relationship between parliamentarism and various governance outcomes. In each case, the model fit is quite good, with F values significant at better than the .0001 level and with R21 pseudo R2 ranging from .47 to .90.
Parliamentarism appears to be associated with some aspects of political development, though the relationship is not robust across all dependent variables and specifications. Model 4 shows parliamentarism to be correlated with the World Bank measure of corruption control but only in the reduced-form model. Parliamentarism is strongly related to bureaucratic quality in both models (5 and 6) and with rule of law in the reduced-form model (12). Results for the Political Risk Services measure of corruption control, government effectiveness, and political stability are inconclusive. On balance, parliamentarism appears to do no harm to political development (as measured here), and it appears to be associated positively with certain aspects.
Clearer patterns emerge in the area of economic development. Results suggest that parliamentarism is associated with better telecommunications
Table 1
Estimation Results: Political Development
Corruption Corruption Bureaucratic Government Political Rule of
Control Control Quality Effectiveness Stability Law
Outcome (PRS) (WB) (PRS) (WB) (WB) (WB)
Model	1	2	3	4	5	6	7	8	9	10	11	12
Estimator	NW	NW	BE	BE	NW	NW	BE	BE	BE	BE	BE	BE
Parliamentarism	0.001	0.001	0.003	0.004**	0.002**	0.003***	0.0002	0.002	0.0003	0.0003	0.003	0.005***
(Historical)	(0.002)	(0.002)	(0.002)	(0.002)	(0.001)	(0.001)	(0.002)	(0.002)	(0.003)	(0.002)	(0.002)	(0.002)
Geography weighted	0.002	0.011*	0.002	0.026*	0.027***	0.028***	0.008	0.034"	0.020	0.017	0.026	0.042***
Dependent variable	(0.009)	(0.006)	(0.018)	(0.015)	(0.008)	(0.005)	(0.018)	(0.015)	(0.032)	(0.029)	(0.020)	(0.016)
Trend	-0.031*" (0.007)	-0.026*** (0.007)			0.011" (0.005)	0.010" (0.005)						
Democracy	0.637***	0.337"	1.083***	0.888***	0.598***	0.507***	0.994***	0.876***	0.671***	0.587***	0.869***	0.680***
(Historical sum, In)	(0.184)	(0.149)	(0.189)	(0.153)	(0.121)	(0.098)	(0.190)	(0.150)	(0.248)	(0.191)	(0.190)	(0.151)
GDP per capita, In	0.480***	0.459***	0.158***	0.189***	0.475***	0.509***	0.201***	0.221***	0.155***	0.149***	0.207***	0.184***
(1960)	(0.044)	(0.032)	(0.042)	(0.037)	(0.036)	(0.027)	(0.043)	(0.037)	(0.059)	(0.048)	(0.042)	(0.037)
Africa	-0.125 (0.169)		-0.175 (0.175)		-0.366*** (0.134)		-0.167 (0.177)		0.286 (0.254)		0.027 (0.174)	
Asia	0.263*		0.010		0.407***	0.630***	0.054		0.106		0.285*	
	(0.147)		(0.163)		(0.114)	(0.084)	(0.164)		(0.231)		(0.163)	
Latin America /	-0.902***	-0.814***	-0.197	-0.174*	-0.510***	-0.479***	-0.245*	-0.225"	0.020		-0.075	
Caribbean												
	(0.120)	(0.098)	(0.134)	(0.100)	(0.100)	(0.078)	(0.134)	(0.100)	(0.206)		(0.132)	
Socialism	0.484*** (0.177)		0.131 (0.169)		0.260" (0.120)		0.047 (0.170)		0.159 (0.223)		0.177 (0.169)	
English legal origin	-0.391***		-0.141		0.436***	0.393***	-0.080		-0.129		-0.037	
	(0.085)		(0.107)		(0.063)	(0.059)	(0.109)		(0.145)		(0.107)	
(Continued)
Table 1 (continued)
Corruption Corruption Bureaucratic Government Political Rule of
Control Control Quality Effectiveness Stability Law
Outcome (PRS) (WB) (PRS) (WB) (WB) (WB)
Model	1	2	3	4	5	6	7	8	9	10	11	12
Estimator	NW	NW	BE	BE	NW	NW	BE	BE	BE	BE	BE	BE
Latitude, In	-0.068		0.297***	0.280***	-0.095*		0.267***	0.275***	0.339***	0.416***	0.249***	0.281***
	(0.049)		(0.073)	(0.059)	(0.052)		(0.074)	(0.060)	(0.110)	(0.090)	(0.076)	(0.060)
Ethnic fractionalization	0.130 (0.180)		-0.012 (0.205)		-0.057 (0.126)		0.159 (0.207)		-0.396 (0.307)		0.036 (0.204)	
Population, In (1960)	-0.079*"	-0.053***	-0.018		0.071***		0.016		-0.113***	-0.112***	0.003	
	(0.020)	(0.019)	(0.024)		(0.015)		(0.024)		(0.037)	(0.032)	(0.024)	
Distance from	0.033*		-0.009		0.003		-0.021		-0.013		-0.030	
financial center	(0.018)		(0.022)		(0.012)		(0.022)		(0.030)		(0.022)	
Oil production,	-0.776***	-0.607"	-0.397		-0.470"	-0.298*	-0.305		0.058		-0.291	
per capita												
	(0.280)	(0.291)	(0.609)		(0.239)	(0.175)	(0.614)		(0.765)		(0.609)	
Diamond production,	0.038*		0.056*		0.070***		0.059*		0.026		0.075"	
per capita	(0.020)		(0.033)		(0.018)		(0.034)		(0.044)		(0.033)	
Protestant	0.009***	0.008***	0.007***	0.005***	0.001		0.003				0.006***	0.004"
	(0.001)	(0.001)	(0.002)	(0.002)	(0.001)		(0.002)				(0.002)	(0.002)
Constant	-0.393	1.160	-7.759***	-7.040***	-7.887***	-6.213***	-7.900***	-7.042***	-3.192*	-2.621"	-7.070***	-5.620***
	(1.410)	(1.315)	(1.296)	(0.941)	(0.919)	(0.786)	(1.302)	(0.916)	(1.736)	(1.268)	(1.300)	(0.925)
Observations	1,218	1,244	311	323	1,218	1,251	328	341	294	297	321	333
Countries	97	99	118	123	97	100	118	123	106	107	119	124
Sample period	1984-	-2001	1996-	-2000	1984-	2001	1996-	-2000	1996-2000		1996-2000	
R2/PseudoR2	.64	.64	.84	.82	.73	.72	.80	.78	.62	.60	.82	.80
Prob > F	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000
Note: PRS = Political Risk Services; WB = World Bank; NW = Newey-West standard errors, including correction for first-order autocorrelation; BE = between-effects ordinary least squares regression; GDP = gross domestic product. All samples limited to country years that are minimally democratic (Polity2 > 0). 'p< .10. "p< .05. "'p< .01.
Table 2
Estimation Results: Economic Development
Telephone Import Trade Investment GDP per
Mainlines Duties Openness Rating Capita
Outcome (In) (WDI) (WDI) (In) (WDI) (Euromoney) (In) (WDI)
Model	13	14	15	16	17	18	19	20	21	22
Estimator	NW	NW	NW	NW	NW	NW	NW	NW	NW	NW
Parliamentarism	0.006"*	0.003***	-0.067***	-0.058***	0.004***	0.004***	0.085***	0.089***	0.004***	0.003***
(historical)	(0.001)	(0.001)	(0.010)	(0.010)	(0.001)	(0.001)	(0.020)	(0.020)	(0.001)	(0.001)
Geography weighted	-0.001*	-0.001**	-0.030	-0.009	-0.0002	0.002*	0.008*	0.009*	0.037***	0.036***
Dependent variable	(0.001)	(0.0005)	(0.018)	(0.014)	(0.001)	(0.001)	(0.005)	(0.005)	(0.014)	(0.014)
Trend	0.049***	0.051***	-0.093***	-0.128***	0.008***	0.007***	0.296***	0.276***	0.014***	0.014***
	(0.003)	(0.002)	(0.032)	(0.029)	(0.001)	(0.001)	(0.085)	(0.086)	(0.001)	(0.001)
Democracy	0.174	0.212*	3.785***	6.288***	-0.073	-0.203***	12.786***	12.354***	0.348***	0.317***
(historical sum, In)	(0.136)	(0.128)	(1.303)	(0.786)	(0.050)	(0.041)	(2.347)	(2.268)	(0.102)	(0.095)
GDP per capita, In	0.709***	0.765***	-2.939***	-3.557***	-0.020	-0.043***	8.598***	8.893***	0.767***	0.767***
(1960)	(0.068)	(0.058)	(0.535)	(0.367)	(0.015)	(0.010)	(0.577)	(0.544)	(0.065)	(0.051)
Africa	-0.799***	-0.780***	2.980*		-0.144***		-4.034**		-0.186*	-0.273***
	(0.140)	(0.102)	(1.642)		(0.051)		(1.689)		(0.096)	(0.067)
Asia	-0.371"		2.635"		0.027		7.420***	8.744***	0.100	
	(0.146)		(1.098)		(0.051)		(1.805)	(1.752)	(0.097)	
Latin America /	0.294***		0.052		-0.248***	-0.175***	-11.035***	-9.470***	0.119"	
Caribbean										
	(0.076)		(0.900)		(0.033)	(0.025)	(1.442)	(1.197)	(0.049)	
Socialism	0.499***	0.406***	-3.909***		0.286***		_9_997»*	-9.317***	-0.281***	-0.341***
	(0.148)	(0.147)	(1.177)		(0.052)		(2.170)	(2.097)	(0.098)	(0.104)
English legal origin	0.038 (0.060)		1.133" (0.490)		0.015 (0.026)		-0.417 (0.980)		-0.048 (0.038)	
(Continued)
Table 2 (continued)
Telephone Import Trade Investment GDP per
Mainlines Duties Openness Rating Capita
Outcome (In) (WDI) (WDI) (In) (WDI) (Euwmoney) (In) (WDI)
Model	13	14	15	16	17	18	19	20	21	22
Estimator	NW	NW	NW	NW	NW	NW	NW	NW	NW	NW
Latitude, In	0.270"*	0.251***	0.502		-0.054***		1.900**	1.832"	0.192***	0.178***
	(0.053)	(0.050)	(0.562)		(0.019)		(0.852)	(0.854)	(0.042)	(0.045)
Ethnic fractionalization	-0.124		-0.901		0.372***		-2.307	-4.060**	-0.291***	-0.279***
	(0.115)		(1.200)		(0.049)		(1.995)	(1.926)	(0.071)	(0.074)
Population, In (1960)	-0.060***	-0.100***	-0.628**	*	-0.246***	-0.225***	2.841***	2.919***	-0.022***	-0.020***
	(0.012)	(0.011)	(0.194)		(0.006)	(0.006)	(0.236)	(0.225)	(0.007)	(0.007)
Distance from	-0.052***	-0.052***	0.487**	0.729***	-0.045***	-0.047***	-0.900***	-1.050***	-0.049***	-0.048***
financial center										
	(0.010)	(0.011)	(0.097)	(0.095)	(0.005)	(0.004)	(0.233)	(0.228)	(0.005)	(0.006)
Oil production,	-0.722***	-0.446**	0.594		-0.0003		-0.684		0.334**	0.402**
per capita										
	(0.142)	(0.184)	(1.465)		(0.086)		(2.392)		(0.150)	(0.162)
Diamond production,	0.062***		-0.237		0.018***		1.427***	1.295***	0.130***	0.130***
per capita										
	(0.015)		(0.176)		(0.007)		(0.211)	(0.206)	(0.007)	(0.007)
Constant	-5.128***	-5.338***	24.577**	3.310	7.753***	8.748***	-163.172***-	-162.295***	-0.830*	-0.650
	(0.769)	(0.730)	(8.399)	(5.797)	(0.375)	(0.317)	(17.637)	(17.709)	(0.445)	(0.426)
Observations	2,129	2,171	1,411	1,436	2,521	2,611	1,544        1,544       2,522 2,522			
Countries	126	128	97	98	126	132	122	122	124	124
Sample period	1965-	-2001	1970-	-2000	1961-2001		1982-2001		1961-2000	
7?2/Pseudo7?2	.79	.78	.50	.47	.68	.63	.81	.81	.90	.90
Prob > F	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000	.0000
Note: WDI = World Development Indicators; GDP = gross domestic product; NW = Newey-West standard errors, including correction for first-order autocorrelation; BE = between-effects ordinary least squares regression. All samples limited to country years that are minimally democratic (Polity2 > 0). *p<.10. *><.05. **><.01.
Table 3
Estimation Results: Human Development						
	Infant Mortwality		Life Expectancy		Illiteracy	
Outcome	Rate (In) (WDI)		(In) (WDI)		(In) (WDI)	
Model	23	24	25	26	27	28
Estimator	NW	NW	NW	NW	NW	NW
Parliamentarism	-0.003"*	-0.004***	0.0003***	0.0003***	-0.001	-0.001
(historical)	(0.001)	(0.001)	(0.0001)	(0.0001)	(0.001)	(0.001)
Geography-weighted	0.002*	0.002*	0.00002	-0.0002*	0.055***	0.057***
dependent variable						
	(0.001)	(0.001)	(0.0001)	(0.0001)	(0.008)	(0.008)
Trend	-0.031***	-0.030***	0.003***	0.003***	-0.020***	-0.021***
	(0.001)	(0.001)	(0.0002)	(0.0002)	(0.002)	(0.002)
Democracy	-0.522***	-0.508***	0.027**	0.019"	-0.564***	-0.503***
(historical sum, In)	(0.066)	(0.061)	(0.011)	(0.008)	(0.119)	(0.116)
GDP per capita, In	-0.295***	-0.291***	0.046***	0.042***	-0.746***	-0.769***
(1960)	(0.036)	(0.024)	(0.006)	(0.005)	(0.031)	(0.025)
Africa	0.394***	0.389***	-0.191***	-0.206***	0.218***	0.257***
	(0.070)	(0.055)	(0.014)	(0.013)	(0.084)	(0.068)
Asia	-0.090		0.023**		-0.131	
	(0.086)		(0.012)		(0.097)	
Latin America / Caribbean	0.197***	0.175***	0.019**		0.517***	0.589***
	(0.051)	(0.043)	(0.008)		(0.094)	(0.077)
Socialism	-0.448***	-0.501***	0.019		-2.043***	-2.019***
	(0.078)	(0.080)	(0.012)		(0.158)	(0.149)
English legal origin	0.021		-0.016***	-0.017***	-0.311***	-0.315***
	(0.027)		(0.006)	(0.006)	(0.050)	(0.050)
Latitude, In	-0.040		0.016***	0.014***	-0.073*	
	(0.034)		(0.005)	(0.005)	(0.044)	
(Continued)
Table 3 (continued)
Infant Mortwality Life Expectancy Illiteracy
Outcome Rate (In) (WDI) (In) (WDI) (In) (WDI)
Model	23	24	25	26	27	28
Estimator	NW	NW	NW	NW	NW	NW
Ethnic fractionalization	0.501"*	0.521***	-0.066***	-0.060***		
	(0.059)	(0.049)	(0.011)	(0.012)		
Population, In (1960)	0.031***	0.017**	-0.007***	-0.006***	0.020*	
	(0.008)	(0.007)	(0.001)	(0.001)	(0.011)	
Distance from	0.042***	0.044***	0.001		0.060***	0.062***
financial center						
	(0.005)	(0.006)	(0.001)		(0.008)	(0.008)
Oil production, per capita	-0.313*** (0.108)		-0.001 (0.024)		-0.227 (0.276)	
Diamond production,	0.002		-0.003		0.011*	
per capita	(0.010)		(0.005)		(0.006)	
Muslim	0.005***	0.005***	-0.001***	-0.001***	0.009***	0.009***
	(0.001)	(0.001)	(0.0001)	(0.0001)	(0.001)	(0.001)
Linguistic fractionalization					0.715*** (0.088)	0.709*** (0.087)
Constant	10.851***	10.977***	3.543***	3.624***	11.865***	11.971***
	(0.402)	(0.388)	(0.065)	(0.046)	(0.852)	(0.767)
Observations	2,633	2,682	2,634	2,652	2,401	2,438
Countries	126	129	125	127	108	110
Sample period	1961-	-2001	1961-	-2001	1961-	-2001
7?2/Pseudo7?2	.82	.81	.75	.75	.85	.85
Prob > F	.0000	.0000	.0000	.0000	.0000	.0000
Note: WDI = World Development Indicators; NW = Newey-West standard errors, including correction for first-order autocorrelation; GDP = gross domestic product. All samples limited to country years that are minimally democratic (Polity2 > 0). *p<.10. *><.05. **><.01.
Gerring et al. / Are Parliamentary Systems Better? 349
Table 4 Summary Results
Specification
Outcome Full Reduced Models
Political development
Corruption control (PRS)	1,2
Corruption control (WB)	3,4
Bureaucratic quality (PRS)	5, 6
Government effectiveness (WB)	7,8
Political stability (WB)	9, 10
Rule of law (WB)	11, 12
Economic development	
Telephone mainlines (WDI)	13, 14
Import duties (WDI)	15, 16
Trade openness (WDI)	17, 18
Investment rating (Euromoney)	19, 20
GDP per capita	21, 22
Human development	
Infant mortality rate (WDI)	23, 24
Life expectancy (WDI)	25, 26
Illiteracy	27, 28
Note: PRS = Political Risk Services; WB = World Bank; WDI = World Development Indicators; GDP = gross domestic product. Probabilities fall in the expected direction—that is, models in Tables 1-3 that show parliamentarism to be positively correlated with political, economic, or human development. "p< .05. ">< .01.
infrastructure, lower import duties, greater trade openness, better investment ratings, and higher levels of per capita GDP, across both full and reduced-form models. These results are robust in both models and across a variety of alternative specifications, most of which we do not report here.
In two out of three cases, parliamentarism is associated with higher levels of human development. With all else being equal, longtime parliamentary systems have lower levels of infant mortality and longer life expectancy. Again, results are robust across various specifications and highly significant. No clear patterns emerge in the case of illiteracy.
Table 4 summarizes findings for all regression tests reported in Table 1. It appears that parliamentary systems hold distinct advantages over presidential ones across a range of indicators of political, economic, and human
350    Comparative Political Studies
development. In no case are parliamentary systems associated with bad governance, and in most cases they are strongly associated with good governance. The results are striking in economic and human development— where a strong positive relationship between parliamentarism and good governance can be seen in every case but one.
In addition to considering the statistical strength of the relationship between parliamentarism and good governance, it is worth considering the substantive import of this relationship. A hypothetical scenario is instructive here. Employing the coefficients for parliamentarism from the full models in Table 1 (and keeping all control variables constant), we find that 50 years of fully parliamentary rule (1951-2000, in this example) are associated with an improvement of 0.15 points on the 7-point scale of bureaucratic quality, 45% more telephone mainlines, a 5% reduction in import duties as a percentage of imports, 30% more trade openness, an increased investment rating of more than 6 points (on a scale of 100), a 30% higher per capita income, a nearly 23% lower infant mortality rate, and a greater life expectancy of more than 2%. Thus, whereas the substantive impact of parliamentarism appears quite modest in some areas (e.g., bureaucratic quality and investment rating), it is quite large in others (e.g., telephone mainlines, trade and GDP per capita). Perhaps most important—for we do not wish to invest too much importance in imprecise point estimates—parliamentarism shows a strong relationship with a variety of different outcomes, suggesting that we need to consider its overall cumulative effects across a broad spectrum of governance issues.
Robustness
Several robustness issues arise. Consider, first, the construction of the independent variable of interest: parliamentarism. In our coding, we recognize three levels: full presidentialism, semipresidentialism, and full parliamentarism. This imposes an interval measurement onto a set of ordinal distinctions, an imposition that could result in significant measurement error. However, it is important to keep in mind that there are relatively few semi-presidential systems and that they occupy the in-between position in this trichotomy. Under the circumstances, it is unlikely that another method of coding would bring results significantly different from those reported here.
Second, many of the countries with the longest history of parliamentary rule are located in Western Europe. These countries are also some of the best governed in the world. To control for this possibly confounding factor,
Gerring et al. / Are Parliamentary Systems Better? 351
we ran tests (not reported) including a dummy variable for Western Europe. Results for parliamentarism remained quite robust, whereas the Western Europe variable performed inconsistently.
Numerous additional specification tests were conducted, though we have no space to report them in Table 1. To our benchmark equations, we added additional control variables, such as the number of years that each country has enjoyed national sovereignty (logged), decade dummies (to further control for time effects), and alternative measures of fractionalization (ethnic, linguistic, religious). None of these alternate specifications appreciably disturbed the results for parliamentarism. We also experimented with the exclusion of various controls in the benchmark model. Again, we the results are reassuring. The relationship between parliamentarism and good governance does not appear to be conditional on the inclusion of particular control variables in the general model.
A fourth issue concerns the inclusion of observations during years when a country is only marginally democratic. Recall that we adopt a fairly low threshold for democracy (Polity2 > 0, on the Polity IV index, which stretches from -10 to 10). Our operating assumption is that a minimal level of multiparty competition is sufficient to set in motion whatever causal effects parliamentarism might have on the quality of governance. To the extent that the inclusion of less democratic cases weakens the potential impact of democratic institutions such as parliamentarism on governance, this liberal inclusion criterion should bias the results against our hypotheses. However, it is possible that the effects of parliamentarism in a marginally democratic society are different from the effects in a fully democratic society. Reassuringly, additional tests that set the bar of democracy higher, at 5 on the 21-point Polity2 scale, yield little substantive difference in the results reported here. Note that observations in all regressions consist of country years, not of country cases. This means, perforce, that greater weight is carried in the analysis by countries that have been democratic over a longer stretch of time; that is, the United States is a more influential case than recently democratized Namibia. As it happens, long-term democracies tend to have stronger democracies; thus, the results, as reported, reflect a sample weighted toward countries that are strongly democratic.
A fifth issue concerns possible interaction effects. It could be, for example, that parliamentary, as well as presidential, systems work differently in polities with different electoral system types (majoritarian, proportional representation, preferential vote, etc.) or across unitary and federal systems. This may be empirically probed through interaction terms. Although the empirical field of possible interaction terms is
352    Comparative Political Studies
virtually limitless, we did not discover significant interaction effects with the main variable of interest: parliamentarism.
A final issue concerns possible selection effects. The assignment of the "treatment" in observational research is not random. This is problematic if the process of selecting political institutions is correlated with the outcomes under investigation (good governance, broadly considered). Perhaps, that is, parliamentary systems are more likely to be adopted where prospects for good governance are more propitious. In this case, our key variable serves as a proxy for other, unmeasured factors. To neutralize this potential identification problem, we employed a series of instruments for parliamentarism in two-stage least squares estimations. Results so obtained are largely robust, thus providing at least some assurance that the effects reported here are not simply the product of a nonrandom choice of political institutions. Even so, we do not have a great deal of confidence in the two-stage models. All the instruments that we could construct violate at least one of the assumptions of instrumental-variable analysis (Reiss, 2003): They either poorly correlate with parliamentarism, or they correlate with the error term (i.e., they are probable causes of good/bad governance in the contemporary period). Thus, though the instrumental-variable results are supportive, we do not report them here.
In any case, we think it unlikely that the choice of constitutional institutions reflects, rather than shapes, a country's future prospects for good/bad governance. To be sure, whether a country becomes parliamentary or presidential (or somewhere in between) depends partly on a country's colonial heritage, on its size and demographics, and on patterns of government that obtain in a regional or historical context. However, these exogenous influences are relatively easy to model and so appear as controls in all our regression tests. Other factors influencing constitutional choice regarding the separation of powers do not seem to accord with a country's proclivity to good or bad governance and may therefore be regarded as being stochastic.
One must also consider the fact that constitution makers generally have notoriously short horizons of time. They are usually interested in installing a system that will benefit themselves, their parties, or their constituencies. In this respect, the type of constitution to which a country arrives is the product of a highly contingent political battle, with little or no bearing on a country's long-term governance potential. We think it reasonable to regard a country's choices among constitutional institutions as a largely stochastic phenomenon with respect to the outcomes of interest in this study: long-term patterns of good or bad governance.
Gerring et al. / Are Parliamentary Systems Better? 353
One type of selection effect deserves special mention. Some argue that democratic polities with presidential institutions are more liable to regime failure (e.g., Stepan & Skach, 1993). If true (the point has been strenuously debated), then the sample may be skewed toward parliamentary systems, given that we consider only democratic country years in our empirical tests. However, this selection effect concerns independent variables (democracy and parliamentarism), not dependent variables; as such, it should not introduce bias into our estimates. Indeed, insofar as descent into autocracy may be considered a bad governance outcome, our exclusion of such country cases from our analysis constitutes a conservative bias in the sample. We might expect stronger results if we did not drop these cases.
Conclusion
In an age of democratic transition, few questions have greater practical import than that of constitutional engineering. Yet, relatively little scholarly attention has been devoted to the impact of different types of constitutional design on policies and policy outcomes within democratic societies. The evidence presented here suggests that to the extent that the nature of the executive makes a difference, parliamentary systems offer significant advantages over presidential systems. In no case examined here does parliamentary rule seem to detract from good governance. In most policy areas, particularly in the areas of economic and human development, parliamentary systems are associated with superior governance.
What remains to be explored are the specific causal mechanisms by which the structure of the executive might influence various policy outcomes. To this point, we have merely reviewed some of the possibilities, as suggested by the voluminous literature on these questions. Plausible reasons why parliamentarism might lead to better governance include the following: stronger political parties, corporatist interest organization, tighter principal-agent relationships within the various arms of the bureaucracy, centralized (national-level) electoral accountability, the capacity for flexible policy making, a more institutionalized political sphere, and decisive leadership. The first section of the article reviews each of these mechanisms. Is it possible to arrive at a relatively concise and coherent general theory that ties and explains the findings of this otherwise largely empirical endeavor?
To this end, we put forth, provisionally and broadly, that what distinguishes parliamentarism from presidentialism—and what makes the former a more reliable vehicle for good public policy—is its capacity to function
354    Comparative Political Studies
as a coordination device. The state is often conceptualized as a solution to the multiple coordination problems that emanate from society. There are many variations on this theme—joint decision traps, shirking, underprovid-ing of public goods, overgrazing, the tragedy of the commons, common pool problems, collective action problems, free rider problems, prisoner's dilemmas, transaction cost dilemmas, and so forth. These are different ways of pointing out a central problem: Quite often, when individuals or groups pursue their own agendas, the result is not what society as a whole would prefer. Coordination problems thus involve a conflict between the part and the whole, between individual/group rationality and collective rationality.
When compared to presidentialism, parliamentarism arguably offers better tools for resolving these sorts of difficulties (within the framework of democracy) because parliamentarism integrates a diversity of views while providing greater incentives for actors to reach agreement. Consider a stylized comparison of decision-making processes within the two systems. Both feature a similar set of players: a legislature with majority and minority parties, committees and separate leadership hierarchies, a cabinet, a chief executive (prime minister or president), and various government agencies. Yet, the interaction of these players and the role that each institution assumes tend to be quite different. In parliamentary systems, debate occurs in a highly institutionalized fashion—within parties, within committees, within leadership groups, across parties, within the cabinet, and between cabinet-level ministers and high-level civil service appointees within the bureaucracy. By contrast, in presidential systems most of these units have greater independence, and those without independence (such as the cabinet) have very little power. The players that matter have the capacity (and often the incentive) to say no or to insist on side payments in exchange for support. Although successful coordination can occur in this highly fragmented institutional sphere, such resulting agreements may impose higher transaction costs than that anticipated within a parliamentary system, where the incentives of the key actors are generally to reach agreement. This, in turn, is a product of how political careers and electoral incentives align within these various constitutional systems.
The same sort of contrast operates at "mass" levels. In a parliamentary system, voters are encouraged to align their perspectives (interests, identity, ideals) with one of the national parties because these are the political institutions that matter. In a presidential system, voters may consider not only a candidate's party membership but also her or his fealty to the district, along with various personal characteristics and issue-specific positions that may or may not be consistent with her or his party's national strategy.
Gerring et al. / Are Parliamentary Systems Better? 355
Of course, we realize that the actual workings of parliamentarism/presidentialism within a polity depend on many additional factors—sociological, economic, political, and historical—exogenous to the theory. Parliamentarism within a proportional representation electoral system works differently from parliamentarism in a "first past the post" electoral system, for example. However, these additional considerations can, for the most part, be subsumed under the ceteris paribus caveat that accompanies all causal reasoning. Parliamentarism should be more successful than presidentialism in coordinating diverse views and interests, all other things being equal. This means that the persistent institutional conflicts that characterize political life in all democracies—for example, between legislature and executive, between backbenchers and party leaders, among parties, among diverse agencies, between national and subnational governments, among subnational governments, and among diverse constituencies—may be easier to solve in a parliamentary system than in a presidential system. This, in turn, should help to account for the higher quality of governance observed in parliamentary systems across many policy areas.
Notes
1. Foweraker and Landman (2002), whose primary concern is with the quality of democracy in parliamentary and presidential systems, also touch on the provision of property rights across regime types. Although they find no statistically significant difference between them along this dimension of economic policy, the regression analysis relies on a small sample (n = 40) that is potentially biased and subject to a host of possible specification errors (owing to the paucity of control variables included in the model and lack of robustness tests).
2. For descriptive statistics, as well as details about sources and coding procedures for all variables, see http://www.bu.edu/sthacker/data.htm.
References
Alesina, Alberto, Devleeschauwer, Arnaud, Easterly, William, Kurlat, Sergio, & Wacziarg, Romain. (2003). Fractionalization. Journal of Economic Growth, 8(2), 155-194.
American Political Science Association. (1950). Toward a more responsible two-party system. New York: Holt, Rinehart.
Bahro, Horst, Bayerlein, Bernhard H., & Veser, Ernst. (1998). Duverger's concept: Semi-presidential government revisited. European Journal of Political Research, 34, 201-24.
Beck, Nathaniel, & Katz, Jonathan. (2001). Throwing out the baby with the bathwater: A comment on Green, Kim and Yoon. International Organization, 55, 487-495.
Beck, Thorsten, Demirguc-Kunt, Aski, & Levine, Ross. (2000). Law, politics, and finance. Washington, DC: World Bank.
Breton, Albert. (1996). Competitive governments: An economic theory of politics and public finance. Cambridge, UK: Cambridge University Press.
356    Comparative Political Studies
Buchanan, James M., & Tullock, Gordon. (1962). The calculus of consent: Logical foundations of constitutional democracy. Ann Arbor: University of Michigan Press.
Cheibub, Jose Antonio. (2007). Presidentialism, parliamentarism, and democracy. Cambridge, UK: Cambridge University Press.
Cox, Gary W., & Kernell, Samuel. (Eds.). (1991). The politics of divided government. Boulder, CO: Westview.
Cox, Gary W., & McCubbins, Mathew D. (2001). Political structure and economic policy: The institutional determinants of policy outcomes. In Stephan Haggard & Mathew D. McCubbins (Eds.), Presidents, parliaments, and policy (pp. 21-63). Cambridge, UK: Cambridge University Press.
Elgie, Robert. (1999). Semi-presidentialism in Europe. Oxford, UK: Oxford University Press.
Euromoney. (2004, June 11). Country risk methodology. Downloaded from the Euromoney subscriber section: http://www.euromoney.com/includes/magazine/print.asp
Fiorina, Morris. (1980). The decline of collective responsibility in American politics. Daedalus, 109(3), 25-45.
Foweraker, Joe, & Landman, Todd. (2002). Constitutional design and democratic performance. Democratization, 9(2), 43-66.
Gerring, John, Bond, Philip, Barndt, William T, & Moreno, Carola. (2005). Democracy and economic growth: A historical perspective. World Politics, 57(3), 323-365.
Gerring, John, & Thacker, Strom C. (2004). Political institutions and corruption: The role of unitarism and parliamentarism. British Journal of Political Science, 34, 295-330.
Gerring, John, & Thacker, Strom C. (2008). Good government: A centripetal theory of democratic governance. Cambridge, UK: Cambridge University Press.
Gerring, John, Thacker, Strom, & Moreno, Carola. (2005). Centripetal democratic governance: A theory and global inquiry. American Political Science Review, 99(4), 567-581.
Green, Donald P., & Shapiro, Ian. (1994). Pathologies of rational choice theory: A critique of applications in political science. New Haven, CT: Yale University Press.
Haggard, Stephan, & McCubbins, Mathew D. (Eds.). (2001). Presidents, parliaments, and policy. Cambridge, UK: Cambridge University Press.
Henisz, Witold J. (2000). The institutional environment for economic growth. Economics and Politics, 12(1), 1-32.
Herring, Pendleton. (1940). The politics of democracy: American parties in action. New York: Norton.
Howell, Llewellyn D. (1998). The handbook of country and political risk analysis (2nd ed.).
East Syracuse, NY: PRS Group. Huber, John D., & Shipan, Charles R. (2002). Deliberate discretion? The institutional foundations of bureaucratic autonomy. Cambridge, UK: Cambridge University Press. Humphreys,  Macartan.  (2005).  Natural resources,  conflict,  and conflict resolution:
Uncovering the mechanisms. Journal of Conflict Resolution, 49(4), 508-537. Kaufmann, Daniel, Kraay, Aart, & Mastruzzi, Massimo. (2007). Governance matters TV:
Governance indicators for 1996-2006. Washington, DC: World Bank. Keefer, Philip, & Stasavage, David. (2003). The limits of delegation: Veto players, central
bank independence, and the credibility of monetary policy. American Political Science
Review, 97(3), 407-423. Laffont, Jean-Jacques, & Meleu, Mathieu. (2001). Separation of powers and development.
Journal of Development Economics, 64(1), 129-145. Lakatos, Imre. (1978). The methodology of scientific research programmes. Cambridge, UK:
Cambridge University Press.
Gerring et al. / Are Parliamentary Systems Better? 357
La Porta, Rafael, Lopez-de-Silanes, Florencio, Shleifer, Andrei, & Vishny, Robert W. (1999).
The quality of government. Journal of Economics, Law and Organization, 15(1), 222-279. Lijphart, Arend. (Ed.). (1992). Parliamentary versus presidential government. New York:
Oxford University Press. Lijphart, Arend. (1999). Patterns of democracy: Government forms and performance in thirty-six countries. New Haven, CT: Yale University Press. Linz, Juan. (1990). The perils of presidentialism. Journal of Democracy, 1, 51-69. Linz, Juan. (1994). Presidential or parliamentary democracy: Does it make a difference? In
Juan J. Linz & Arturo Valenzuela (Eds.), The failure of presidential democracy (pp. 3-87).
Baltimore: The Johns Hopkins University Press. Lowi, Theodore. (1985). The personal president: Power invested, promise unfulfilled. Ithaca,
NY: Cornell University Press. Mainwaring, Scott P. (1993). Presidentialism, multipartism, and democracy: The difficult
combination. Comparative Political Studies, 26, 198-228. Mainwaring, Scott, & Shugart, Matthew J. (Eds.). (1997). Presidentialism and democracy in
Latin America. Cambridge, UK: Cambridge University Press. Marshall, Monty G., & Jaggers, Keith. (2002). Polity IV Dataset Project: Political regime
characteristics and transitions, 1800-1999. College Park: University of Maryland, Center
for International Development and Conflict Management, Integrate Network for Societal
Conflict Research Program. McConnell, Grant. (1953). The decline of agrarian democracy. Berkeley: University of
California Press.
Moe, Terry M., & Caldwell, Michael. (1994). The institutional foundations of democratic government: A comparison of presidential and parliamentary systems. Journal of Institutional and Theoretical Economics, 150, 171-195.
Moon, Bruce E. (1991). The political economy of basic human needs. Ithca, NY: Cornell University Press.
Olson, Mancur. (1982). The rise and decline of nations. New Haven, CT: Yale University Press.
Persson, Torsten, Roland, Gerard, & Tabellini, Guido. (1997). Separation of powers and political accountability Quarterly Journal of Economics, 112, 1163-1202.
Persson, Torsten, & Tabellini, Guido. (2003). The economic effects of constitutions: What do the data say? Cambridge, MA: MIT Press.
Pierson, Paul D., & Weaver, R. Kent. (1993). Imposing losses in pension policy In R. Kent Weaver & Bert A. Rockman (Eds.), Do institutions matter? Government capabilities in the United States and abroad. Washington, DC: Brookings Institution.
Reiss, Julian. (2003). Practice ahead of theory: Instrumental variables, natural experiments and inductivism in econometrics. Unpublished manuscript.
Reynolds, Andrew. (Ed.). (2002). The architecture of democracy. New York: Oxford University Press.
Riker, William H. (1982). Liberalism against populism: A confrontation between the theory of democracy and the theory of social choice. San Francisco: Freeman.
Roper, Steven D. (2002). Are all semipresidential regimes the same? A comparison of premier-presidential regimes. Comparative Politics, 34(3), 253-272.
Rose-Ackerman, Susan. (1978). Corruption: A study in political economy. New York: Academic Press.
Samuels, David S. (2007). Separation of powers. In Susan Stokes & Carles Boix (Eds.), The Oxford handbook of comparative politics (pp. 703-726). Oxford, UK: Oxford University Press.
358    Comparative Political Studies
Sartori, Giovanni. (1994). Comparative constitutional engineering: An inquiry into structures, incentives and outcomes. New York: New York University Press.
Shugart, Matthew Soberg. (1998). The inverse relationship between party strength and executive strength: A theory of politicians' constitutional choices. British Journal of Political Science, 28, 1-29.
Shugart, Matthew Soberg, & Carey, John M.. (1992). Presidents and assemblies: Constitutional design and electoral dynamics. New York: Cambridge University Press.
Skach, Cindy. (2005). Borrowing constitutional designs: Constitutional law in Weimar Germany and the French Fifth Republic. Princeton, NJ: Princeton University Press.
Stepan, Alfred, & Skach, Cindy. (1993). Constitutional frameworks and democratic consolidation: Parliamentarianism versus presidentialism. World Politics, 46, 1-22.
Truman, David B. (1951). The governmental process. New York: Knopf.
Tsebelis, George. (1995). Decision making in political systems: Veto players in presidentialism, parliamentarism, multicameralism and multipartism. British Journal of Political Science, 25(3), 289-326.
Tsebelis, George. (2000). Veto players in institutional analysis. Governance 13(4), 441-474.
Vile, M. J. C. (1998). Constitutionalism and the separation of powers. Indianapolis, IN: Liberty Fund. (Originally published 1967)
Weaver, R. Kent, & Rockman, Bert A. (Eds.). (1993). Do institutions matter? Government capabilities in the United States and abroad. Washington, DC: Brookings Institution.
Wilson, Bradford P., & Schramm, Peter W. (Eds.). (1994). Separation of powers and good government. Lanham, MD: Rowman & Littlefield.
Wilson, Graham K. (1992). Interest groups and the constitution. In Peter F. Nardulli (Ed.), The constitution and American political development: An institutional perspective (pp. 207-233). Urbana: University of Illinois Press.
Wilson, Sven E., & Butler, Daniel M. (2003). Too good to be true? The promise and peril of panel data in political science. Unpublished manuscript, Brigham Young University, Provo, UT.
World Bank. (2003). World Development Indicators 2003 [CD-ROM]. Washington, DC: World Bank.
John Gerring received his doctorate from the University of California at Berkeley in 1993. He is a professor of political science at Boston University, where he teaches courses on methodology and comparative politics. His books include Party Ideologies in America, 1828—1996 (1998), Social Science Methodology: A Criterial Framework (2001), Case Study Research: Principles and Practices (2007), A Centripetal Theory of Democratic Governance (with Strom Thacker, 2008), Concepts and Method: Giovanni Sartori and His Legacy (edited with David Collier; forthcoming), Global Justice: A Prioritarian Manifesto (in press), and Democracy and Development: A Historical Perspective (with Strom Thacker, in press). His articles have appeared in a variety of political science journals. He served as a fellow of the School of Social Science at the Institute for Advanced Study (2002-2003) and a member of the National Academy of Sciences' Committee on the Evaluation of US AID Programs to Support the Development of Democracy (2006-2007). He is the recipient of a grant from the National Science Foundation to investigate the global impact of colonialism (2007-2010), and he serves as president of the American Political Science Association's Organized Section on Qualitative and Multi-Method Research (2007-2009).
Gerring et al. / Are Parliamentary Systems Better? 359
Strom C. Thacker received his doctorate in political science from the University of North Carolina, Chapel Hill, in 1996. He is currently associate professor of international relations and director of Latin American studies at Boston University. His research and teaching focus broadly on questions of political economy, governance, and development, with a regional focus on Mexico and Latin America. His books include Big Business, the State, and Free Trade: Constructing Coalitions in Mexico (2000), A Centripetal Theory of Democratic Governance (with John Gerring, 2008), and Democracy and Development: A Historical Perspective (with John Gerring, in press). He is also working on a project on the politics of human development. He has published articles in the American Political Science Review, the British Journal of Political Science, Business and Politics, International Organization, the Journal of Interamerican Studies and World Affairs, and World Politics. He also has an ongoing interest in the politics of foreign aid and lending and the International Monetary Fund. He is a faculty affiliate of the David Rockefeller Center for Latin American Studies at Harvard University and a fellow at the Frederick S. Pardee Center for the Study of the Longer-Range Future at Boston University. He has been a visiting associate professor of government at Harvard University, a Susan Louise Dyer Peace Fellow at Stanford University's Hoover Institution, and a Fulbright scholar.
Carola Moreno is a doctoral candidate in economics at Boston University. She received her master of arts in political economics in 2006 from Boston University and her bachelor of arts from Universidad de Chile in 1998. She is currently a senior economist at the Global Trends Unit of the Economics Research Department at BBVA since August 2006 and an economics instructor at Universidad Los Andes in Santiago, Chile. She is a macroeconomist whose research interests include economic forecasting, international finance, and development. She previously served as an advisor to the Ministry of Finance (1998-2001) in the International Finance and Capital Markets Division. She has published articles in the American Political Science Review and World Politics. She was a visiting researcher at the University of Toronto in 1998 and received a scholarship (Beca Presidente de la Republica) from the government of Chile to pursue graduate studies (2001-2005).