In his book, Robert Elgie makes a significant contribution to studies of semi-presidential government that will be essential reading for scholars interested in the connection between a regime’s subtype and democratic performance. Since Maurice Duverger’s identification of the regime type, scholars have debated how to conceptualize this constitutional format and how to theorize its impact on interactions within the executive of president, prime minister, and cabinet; on relations between the dual executive and legislature; and on democratic politics more broadly. Elgie’s study contributes to an important recent shift in semi-presidentialism studies from a focus on conceptual and theoretical development toward empirically testing of hypotheses about the consequences of semi-presidentialism and variations within the regime type for political outcomes. Most of the book is concerned with testing his argument about the consequence of variation within semi-presidential regimes for democratic performance through a combination of cross-national statistical analysis and detailed case study narratives.

In his previous work, Elgie’s major conceptual contribution to the study of semi-presidentialism has been to insist on a definition of the regime type based on strictly institutional rather than behavioral criteria. He identifies a constitution as semi-presidential when there is both a directly elected fixed-term president and a prime minister and government accountable to the legislature. As he acknowledges, this definition delineates a rather heterogeneous set of regimes, in which the formal constitutional powers allocated to the president, government, and legislature can vary widely, as can their observed ability to use these powers. Semi-Presidentialism develops and tests an argument about important aspects of this variation for democratic performance. This argument builds on a distinction between two subtypes of semi-presidentialism first noted by Matthew Shugart and John Cary in 1992. They characterize president-parliamentary regimes as those in which the prime minister and cabinet are accountable to both president and legislature, and premier-presidential regimes in which the prime minister and cabinet are responsible only to the assembly and the president has no formal ability to remove the government. The central argument of the book is that the adoption of president-parliamentary regimes leads to political instability with detrimental impacts on democratic survival and the quality of democracy.

Chapter 1 locates the argument within the broad debates about the impact of constitutional formats on democracy initiated by Linz’s comments on the perils of presidentialism. Juan Linz’s criticisms of semi-presidentialism’s dual legitimacies, Elgie suggests, have contributed to a consensus view that emerging democracies should avoid choosing this regime type. Chapter 2 defines semi-presidentialism and its two subtypes and classifies the universe of semi-presidential countries. This chapter also outlines Elgie’s theory linking semi-presidentialism subtypes to democratic performance, which, he suggests, identifies a similar but somewhat different causal mechanism to Shugart and Cary’s 1992 account. Elgie argues that because governments can be removed only by the legislature under premier-presidentialism, this provides incentives for cooperation between president and assembly to facilitate stability. In contrast, under president-parliamentary constitutions, where both president and assembly can dismiss the government, each institutional actor may seek to maximize influence over the government by working against each other, inducing political conflict and instability damaging to democratic stability and performance.

Chapter 3 moves the book firmly toward empirical analysis. Through both descriptive statistics and the use of multivariate duration models, the author tests the effects of semi-presidentialism’s subtypes on the survival of democracy. The chapter establishes that the president-parliamentary government subtype significantly raises the hazards of democratic failure. Chapter 4 changes the focus of the analysis from democratic survival to democratic performance and provides a series of controlled statistical analyses that probe the link between semi-presidentialism’s subtypes and the level of democracy as measured by Polity IV and Freedom House. This chapter argues that consistent with the theory, there is a statistically significant link between premier-presidentialism and improvements in democracy scores across a range of statistical models and measures of democracy.

Chapter 5 tests the argument through paired comparative case studies of Guinea-Bussau and Mali. Elgie argues that tracing the causal mechanisms in these most similar cases demonstrates how the president-parliamentary format determined the collapse of democracy in Guinea-Bussau compared to the comparatively stabilizing influence of premier-presidentialism on democratic politics in Mali. Chapter 6 focus on two countries that changed constitutional formats, which are treated as natural experiments. The chapter argues that evidence of decline in democratic performance in Sengal after a switch from premier-presidentialism to president-parliamentism government and of improvement in democratic performance in Portugal after a switch from president-parliamartism to premier-presidentailism is consistent with the book’s central arguments.

The final chapter discusses a number of potentially confounding cases (Taiwan, Austria, Iceland, and Namibia) where president-parliamentary regimes performed better than expected and other cases (Congo-Brazzaville, Haiti, and Niger) where premier-presidential regimes performed worse than expected. It argues that either these regimes are unusual in ways that do not undermine the theory or they display the expected dynamics even if the expected effect on democracy has not been observed. A conclusion summarizes and points toward policy implications and avenues for further research.

Semi-Presidentialism employs a rich range of evidence and different modes of causal argument to focus on the effects of president-parliamentarism and premier-presidentialism. The book makes a convincing case that that there is an association between semi-presidential regime subtypes and democratic performance and makes progress in explaining the character of this association. The central challenge faced by this study, like other investigations into the effect of institutions, is in addressing the question of endogeneity. In this case, to what extent are countries that are more likely to perform poorly choosing president-parliamentary regimes rather than president-parliamentary regimes? Elgie explicitly acknowledges this problem, and in addition to addressing it in the case studies, develops a series of logit models that include, alongside indicator variables for semi-presidential subtypes, a potential proxy for a preexisting desire for autocratic rule and the percentage of votes cast for the winning candidate at the presidential election prior to the constitutional choice. He finds that including this variable does not influence the negative relationship between president-parliamentarism and democratic survival, and argues that this analysis demonstrates that semi-presidentialism subtypes exert their effect on democracy even when controlling for the endogeneity problem.

The proxy identified by Elgie is though capturing only one of a number of preexisting views in the literature about how the choice of regime types may be driven by factors that undermined the potential for democracy: For example, Shugart’s 1999 claim that large underdeveloped countries with weak fragmented party systems are more likely to choose formats in which the presidents have more power. Extensions could usefully engage with these and further arguments in the literature on institutional choice and also employ statistical techniques for addressing endogeneity, such as instrumental variables, to increase confidence that the casual arrow points in the expected direction.

Further studies could also reduce the distance between measures and theory. The theory focuses on conflicts between executive and legislature, measures of the quality of democracy. Focusing the statistical analysis on the extent to which semi-presidential subtypes create different patterns of cabinets and policy instability and then exploring the links between cabinet instability and policy conflict to democratic stability and performance would give more confidence that the links between the different elements of the theory are tight. A further refinement might be to more closely integrate case studies with the statistical analysis. In particular, by selecting cases that are predicted well by the statistical analysis and investigating if the expected mechanisms pertained in these cases would improve confidence that statistical models are capturing theoretically relevant relationships and preclude concerns about selection bias and the overdetermination of outcomes in freestanding case studies.

Theoretically, Semi-Presidentialism outlines a theory based on a strategic logic in which president and assembly anticipate each other’s reactions under different constitutions rules. This theory is close to, but distinct from, Shugart and Cary’s argument, and further work could present the theory’s assumptions more formally and outline more clearly how it is distinct and why this matters. As should be clear, though, these points only highlight the significance of Elgie’s study as an important contribution to our understanding of semi-presidential government, which will stimulate further research on this important set of regimes.