Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

This book examines semi-presidentialism in the Caucasus and Central Asia. Semi-presidentialism is the situation where a country’s constitution makes provision for both a directly elected fixed-term president and a prime minister (PM) and cabinet who are collectively responsible to the legislature. Currently, there are four countries with semi-presidential constitutions in this region: Armenia, Azerbaijan, Georgia, and Kyrgyzstan, plus Kazakhstan, which is on the constitutional cusp of semi-presidentialism and presidentialism but which is presidential. We are interested in the extent to which presidential institutions in these countries have had an impact on the practice of politics there since statehood in the early 1990s.

There is now a large literature identifying the consequential effect of institutions and, more specifically, a literature identifying both the effect of different forms of executive/legislative relations on political outcomes (Samuels and Shugart 2010) and the effect of institutional variation within semi-presidentialism (Samuels and Shugart 2010; Elgie 2011). This work gives us reason to think that institutions should have shaped outcomes in the five countries under investigation. At the same time, though, there is also a literature suggesting that the effect of formal institutional structures matters only in relatively democratic countries. In the five countries we examine, the degree of democracy has varied greatly both generally from one country to the next and over time within a number of countries. This leads us to expect that the effect of institutions will vary accordingly. What is more, there is a further area-centric body of work indicating that outcomes in countries of the former Soviet Union are shaped more by informal vertical power relations than formal institutions. The countries under investigation in this book are all part of the former Soviet Union and achieved statehood in relatively similar circumstances at around the same point in time. This suggests they are likely to share certain post-Soviet political practices that may go so far as to negate the standard impact of formal constitutional arrangements.

In this context, we ask the following questions: firstly, what is the relationship between semi-presidentialism in the formal constitution and the verticality of actual power relations in these five countries; secondly, to what extent, if any, have presidential institutions been responsible for the relative performance of democracy in these countries; and thirdly, in what ways have particular institutional arrangements shaped the relationships both within the executive between the president, the prime minister (PM), and ministers, and between the executive and the legislature? At the risk of a major spoiler alert, we find that in some cases, vertical power relations have made formal constitutional structures almost entirely redundant. In other cases, though, institutions have mattered, and semi-presidential arrangements have had consequential effects.

This book has eight chapters. There are chapters on each of the five countries under investigation. These chapters are preceded by a chapter on the general features of post-Soviet politics in this region. There is also this introduction to the themes of the book and a conclusion that draws together the findings. In this introductory chapter, we set the scene by identifying the concept of semi-presidentialism and reviewing the existing literature about the supposed effects of this constitutional form. In particular, we address the importance of institutional variation within semi-presidentialism and discuss some common misconceptions about semi-presidentialism. We also provide an institutional overview of the countries to be examined and the effects we might expect to identify in those countries on the basis of the existing literature. Finally, the chapter outlines the structure of the book and the specific organization of the country case studies. To begin, we outline the concept of semi-presidentialism and how it has changed over time.

Semi-presidentialism: What Is It and Where Is It Found?

Semi-presidentialism was first conceptualized by Maurice Duverger (1970) to describe a type of regime that was different from both presidential and parliamentary systems. The introduction of the direct, popular election of the president in France in 1962 was the institutional change that prompted Duverger to identify semi-presidentialism as a separate regime type. Duverger’s original definition was as follows:

[A] political regime is considered as semi-presidential if the constitution which established it combines three elements: (1) the president of the republic is elected by universal suffrage; (2) he possesses quite considerable powers; (3) he has opposite him, however, a prime minister and ministers who possess executive and governmental power and can stay in office only if the parliament does not show its opposition to them. (Duverger 1980, p. 166)

Duverger’s conceptualization of semi-presidentialism was a welcome innovation and addition to the study of comparative governmental systems. For more than a century, attention had been focused solely on presidentialism and parliamentarism, even though the first countries with semi-presidential constitutions had appeared as far back as 1919. Duverger’s definition of semi-presidentialism quickly took hold and became the standard way of thinking about semi-presidentialism. More than that, given the context in which the concept was formulated, Duverger’s own writing on France and his position within the French academy, France became the default semi-presidential reference point. Largely thanks to Duverger, when scholars thought about semi-presidentialism, they thought about the French case.

Gradually, though, Duverger’s definition was contested. In this regard, a recurring problem not only with Duverger’s definition but also with subsequent attempts at regime classification was the inclusion of clauses referring to the actual powers of either president or PM in order for a regime to be considered semi-presidential. As seen above, Duverger’s original definition states that the president must have “quite considerable powers” for a country to be considered semi-presidential. In his alternative formulation, Sartori (1997) focused on the effective powers of the PM. As a condition for a country to be classified as semi-presidential (ibid., 131), he believed that there should be a “dual authority structure” where the president shared power with a PM. However, definitions of semi-presidentialism that include reference to essentially subjective criteria such as the presence of “quite considerable” presidential powers hamper our ability to agree on a common set of countries for comparison. How “considerable” do the president’s powers have to be for a country to fall in the semi-presidential category? Subjective classification criteria mean that the universe of country cases is likely to shift from one researcher to the next. In addition, focusing on the effective powers of the president and the PM (e.g. power sharing between the president and the PM) as part of the definition of semi-presidentialism conflates operation with constitutional form. This hinders our ability to infer about the effects of institutions, given such effects are already part of the definition of the institutional variable we are studying. For these reasons, there was dissatisfaction with Duverger’s definition of semi-presidentialism and with equivalent reformulations.

The solution to these definitional problems was to define semi-presidentialism and regime types in general without reference to the actual powers of the presidents and the prime ministers (Elgie 1999). This way of understanding semi-presidentialism has now been adopted by the majority of scholars (Schleiter and Morgan-Jones 2009). In line with this way of thinking, we define semi-presidentialism as the situation where a direct or popularly elected fixed-term president exists alongside a PM and cabinet who are collectively responsible to the legislature. By responsible to the legislature, we mean that the legislature has the constitutional power to refuse to approve the cabinet in an investiture vote and/or to topple the cabinet through a vote of confidence and/or no confidence. We stress that the responsibility must be collective. If the legislature can remove the head of government without that necessarily resulting in the collapse of the cabinet as a whole, then we do not class the constitution as semi-presidential. Equally, if the legislature can vote to recommend the removal of both the PM and cabinet but if the final decision still lies with the president, then we do not class the constitution as semi-presidential. By contrast, if the legislature can dismiss the PM and cabinet collectively, but the dismissal automatically leads to the dissolution of the legislature, then we still classify the constitution as semi-presidential.

While it is necessary to specify some of these rules, defining semi-presidentialism in this way provides a much more reliable way of identifying a set of semi-presidential countries than any definition, including Duverger’s and Sartori’s definitions, that requires an assessment of the actual powers of the president and the PM in practice. This is because the definition presented here requires no expert knowledge of any given country, knowledge that in any event is bound to be contested among the set of experts working on that country. Instead, all that is required is access to the wording of a country’s constitution. Moreover, in the age of the Internet, such access is now freely available for all countries of the world. What is more, including only references to the constitutional text in the definition of semi-presidentialism avoids problems of endogeneity and self-fulfilling prophesies about the functioning and effects of semi-presidentialism. In short, it leaves us better placed to identify the effects of semi-presidential institutions.

Whereas this constitutional mix was relatively rare until the early 1990s, since this time, it has become the most common constitutional choice for new democracies. There are now over 50 countries with a semi-presidential constitution across the world. (For a list of countries with semi-presidential constitutions, see Fig. 1.1.) While this constitutional form has been predominantly popular in Western Europe (including countries such as France, Portugal, and Finland), in the post-communist world (e.g. Bulgaria, Russia, and Lithuania), and across Africa (e.g. Cape Verde, Mali, and Tunisia), semi-presidentialism is also found in Asia (Mongolia, Sri Lanka, Timor-Leste, and Taiwan) and even in Latin America and the Caribbean (Haiti and Peru). New democracies with a colonial history have often followed in the constitutional path of their former colonial power—which explains why semi-presidentialism is the predominant constitutional form in Francophone and Lusophone Africa. In other cases, semi-presidentialism offered itself as a convenient compromise constitutional choice in debates between supporters of presidentialism on the one hand and the introduction of a parliamentary system on the other. This was the case in Mongolia, for example (Moestrup and Ganzorig 2007).

Fig. 1.1
figure 1

List of semi-presidential countries and the date when a semi-presidential constitutional was adopted or reinstated after the collapse of democracy

Full size image

Some Common Misconceptions about Semi-presidentialism

As noted above, the definition of semi-presidentialism that we employ in this book has been adopted by the majority of scholars (Schleiter and Morgan-Jones 2009). Indeed, there is now a considerable body of work both on the concept of semi-presidentialism as it is understood here and on the politics of countries with semi-presidential constitutions (Elgie 1999, 2011; Elgie and Moestrup 2007, 2008; Elgie et al. 2011; Sedelius and Mashtaler 2013). In fact, the study of semi-presidentialism has been a burgeoning research area in the last decade or more (Elgie 2004). However, semi-presidentialism remains a somewhat controversial and poorly understood topic in some quarters. There are some scholars who simply refuse to recognize the validity of the concept, or at least who use the term but who deliberately refuse to employ it in the way that it is understood here. There are also other scholars who are perfectly willing to accept the concept and who use the term quite freely, but who unwittingly refer to it in a way that is different from how it is understood in this project, usually defaulting to Duverger’s original definition and thinking of France as the archetypal semi-presidential case. The problems surrounding the use of the term are further complicated by the fact that the concept has very practical political application. Countries debate whether or not they should adopt semi-presidentialism. However, this means that the concept can become laden with party political connotations. These vary according to the context of the particular situation. For example, in Italy, the introduction of semi-presidentialism has been debated for many years. However, the party politics of the debate have meant that semi-presidentialism is automatically associated with the introduction of a powerful president who will provide decisive leadership in a fragmented system. Immediately, this generates a set of people who are in favor of semi-presidentialism for their own party political reasons and a set of people who are opposed to it for their own reasons. In such a context, carefully crafted academic definitions with their accompanying semantic niceties soon go out of the window, all of which adds to the conceptual and empirical confusion surrounding the term generally.

Against this background, we address some of the most commonly held misconceptions about semi-presidentialism. We will not be able to convince everyone of the validity of the concept, nor do we aim to do so. We will certainly not be able to change the terms of the political debate in some countries. However, we can help to clarify the way in which we believe the concept should be understood, how it should be applied, and what we can reasonably say about it.

  • Misconception no. 1—Semi-presidentialism is a regime type midway between presidentialism and parliamentarism

It is not uncommon to hear people referring to semi-presidentialism as a hybrid regime, a mixed regime (Cheibub 2010), or worse, a “bastard” regime (Bahro and Veser 1995). In one sense, there is nothing unproblematic with such a description, or at least perhaps with the first two. After all, if we define presidentialism as the situation where there is a directly elected, or popularly elected, fixed-term president and where the government is not collectively responsible to the legislature, and if we define parliamentarism as the situation where there is either a monarch or an indirectly elected president and where the PM and cabinet are collectively responsible to the legislature, then semi-presidentialism does indeed exhibit one characteristic of each of the other two main regime types. In this sense, it can indeed be described as mixed. In another sense, though, we should be a little wary of this terminology. If we think about regimes in this way, then semi-presidentialism is not the only mixed type. We can think of what Shugart and Carey (1992) call assembly-independent regimes, that is, Switzerland, as being mixed regimes too. They are characterized by the situation where there is an indirectly elected president and where the government is not collectively responsible to the legislature. Again, such regimes exhibit one characteristic of each of the other two main regime types. In these terms, then, assembly-independent regimes are equally as mixed as semi-presidential regimes. And yet, we rarely, if ever, hear them being referred to in this way. Thus, labeling semi-presidential regimes as hybrid or mixed singles them out for unnecessary and perhaps misleading attention.

There is, though, another way of referring to semi-presidentialism that is more problematic. This is when it is classed as a regime that is midway between presidentialism and parliamentarism, or, as Shugart and Carey (1992, 23) put it, “a regime type that is located midway along some continuum running from presidential to parliamentary.” We agree with Shugart and Carey that the temptation to think of semi-presidentialism in this way comes from the use of the prefix “semi,” meaning half of something. While it is perhaps true that a semi-detached house is only half as freestanding as its detached neighbor down the road, and that the World Cup final includes only half the number of teams as the semifinals, this does not mean that semi-presidentialism is necessarily either half as presidential as presidentialism or half as parliamentary as parliamentarism. What scale is being used? Does an invitation to a semi-formal dinner really mean that half of your clothes are expected to be formal and the other half informal? If so, which half? In other words, while the root of the term “semi” does mean half, the term is also used in cases where it no longer has this specific numerical implication. Thus, we agree with Shugart and Carey that this way of thinking about semi-presidentialism is mistaken. To do so is, in political science terms, to treat regime types as a continuous or, at least, discrete variable. Yet, regime types do not possess the values that allow us to think about them in that way. Perhaps, though, we can think about semi-presidentialism, presidentialism, and parliamentarism as nominal or ordinal categorical variables. This sounds more plausible but it raises the problem of institutional variation within semi-presidentialism.

  • Misconception no. 2—Semi-presidential regimes have quite powerful presidents but never very powerful or very weak presidents

As we have noted, the immediate source of the conceptualization of semi-presidentialism was the 1962 constitutional reform in France. This reform kept the basic parliamentary system of the 1958 constitution intact, including the collective responsibility of the government to the National Assembly, but introduced the direct election of the president. This reform had long been sought after by President Charles de Gaulle, and particular events in 1962 gave him the opportunity to propose a referendum to bring about such a change. The change was subsequently approved in the popular vote. President de Gaulle was returned to power in the first direct election in 1965, and he remained in office for another 4 years. De Gaulle’s presidential successors lacked his personal authority, but by then, the expectation of presidential leadership had become the norm. As we have also noted, France came to be the ideational default case of semi-presidentialism. Therefore, when scholars pictured a semi-presidential regime, they pictured a post-1962 Gaullist-type system with a president with “quite considerable” powers who was in a position to exercise decisive leadership, though in conjunction with a parliamentary-style PM who was fundamentally loyal but who also enjoyed at least some independent political authority by virtue of having the confidence of parliament.

Against this background, the list of countries with semi-presidential constitutions in Fig. 1.1 is immediately striking. France is included. So, too, are French-style semi-presidential systems, such as Mongolia, Romania, and more recently, Ukraine, where the president is the senior political player but where the PM is also a major political figure. However, countries such as Austria, Croatia, Iceland, Ireland, and Slovenia are also included in the list of semi-presidential countries. These countries have figurehead presidents whose main job might be thought of, like de Gaulle’s characterization of the weak and indirectly elected president of France prior to 1958, as being merely to open flower shows. What is more, countries such as Belarus, Egypt, Mozambique, and Peru are also included in the list. These countries have very powerful presidents. Here, the PM is merely a figurehead, an adviser, or a chief of staff. What does a set of countries ranging from Iceland to Mongolia to Mozambique have in common? The answer is nothing, apart from the fact that they share a basic constitutional structure. They all have a directly elected president and a government that is collectively responsible to the legislature. In other words, they share nothing in common apart from their constitutional semi-presidentialism. The result is that while semi-presidentialism is a category that captures common constitutional features of certain countries, it is also a category that includes countries with very different power relations in practice. To put it another way, the set of semi-presidential countries are constitutionally homogenous but politically heterogeneous. (Arguably, the same point applies to presidentialism and parliamentarism too, but we leave this point aside for the purposes of the following discussion.)

The political heterogeneity of semi-presidentialism means that for explanatory purposes, we should be very wary of treating semi-presidentialism as a nominal or ordinal categorical variable relative to presidentialism and parliamentarism. For sure, in constitutional terms, we could consider semi-presidentialism in this way, but it is not entirely clear what conclusions we would be able to draw from it. Let us assume that we want to estimate the effect of semi-presidentialism relative to these other regime types (Hicken and Stoll 2008). What would we expect the overall effect of constitutional semi-presidentialism to be, given the extreme political heterogeneity of the set of countries in this category? Maybe we might think that the parliamentarizing effect of the countries with weak presidents will balance out the presidentializing effect of countries with strong presidents, again leaving semi-presidentialism somewhere in between full presidentialism on the one hand and full parliamentarism on the other. Well, maybe we might, but this would be a heroic assumption. Rather, we should avoid operationalizing semi-presidentialism as a discrete explanatory variable at all. Instead, we should think of the distinction between semi-presidentialism, presidentialism, and parliamentarism as an exercise in constitutional taxonomy. These three regimes, and indeed others, such as assembly-independent regimes, have discrete constitutional forms. Taxonomically, they are different. However, in itself, this taxonomical classification generates no empirical expectations.

The extreme political heterogeneity of semi-presidential countries and the problems with operationalizing constitutional semi-presidentialism as a discrete explanatory variable are precisely the reasons why some scholars refuse to accept the term. We believe that they should consider thinking about regime taxonomies in a way that makes semi-presidentialism a valid and reliable concept. However, we acknowledge that if we wish to move beyond taxonomies and examine the effect of institutional variation on political outcomes, then we need to go further than merely identifying a set of countries with semi-presidential constitutions. In this regard, the standard move is to adopt Shugart and Carey’s (1992) distinction between two basic forms of semi-presidentialism: president-parliamentarism, where the president can dismiss the PM and cabinet on his or her own initiative, and premier-presidentialism, where the PM and cabinet are responsible only to the legislature. (For a list of president-parliamentary and premier-presidential countries, see Fig. 1.2.) Shugart and Carey’s distinction is seductive because they apply an additional constitutional rule to produce a distinction between these two forms of semi-presidentialism. Therefore, we can reliably identify a set of president-parliamentary and premier-presidential regimes. However, it has also been widely adopted because of its empirical implications. At bottom, Shugart and Carey are engaging in a taxonomic exercise. However, the taxonomic categories they arrive at generate much less within-category variation in political practice than occurs within the category of semi-presidentialism alone. Generally speaking, president-parliamentary countries have stronger presidents than premier-presidential countries. If we assume, reasonably enough, that in practice presidential countries have stronger presidents than president-parliamentary countries and also that premier-presidential countries have stronger presidents than parliamentary countries, then we generate an ordinal variable. Moreover, we can have reasonable expectations that these institutional differences may bring about differential empirical effects. Indeed, when Samuels and Shugart (2010) reasoned in this way, they found that these regimes were associated with their expected outcomes or at least that presidential and president-parliamentary regimes were associated with different outcomes from premier-presidential regimes, which, in turn, were associated with different outcomes from parliamentarism.

Fig. 1.2
figure 2

List of president-parliamentary and premier-presidential semi-presidential regimes

Full size image

There are problems with the president-parliamentary/premier-presidential distinction. Most notably, while the within-category variation is reduced relative to semi-presidentialism, there is still variation in political practice within both president-parliamentarism and premier-presidentialism. For example, the president-parliamentary category includes Austria and Iceland, both of which have figurehead presidents. Indeed, Samuels and Shugart (2010) excluded Austria from their empirical analysis, precisely because the gap between constitutional form and actual political practice was so great. What is more, even though the list of premier-presidential countries includes classic semi-presidential cases, such as France and Romania, it also includes countries such as Ireland and Slovenia that operate in a purely parliamentary-like manner, even though they have directly elected presidents. So, even though the president-parliamentary/premier-presidential distinction reduces the within-category heterogeneity of semi-presidentialism, it does not eliminate it. For this reason, some scholars prefer to understand this distinction as another purely taxonomic one. When it comes to exploring the impact of variation in presidential practice, such scholars choose to estimate the effects of presidential power (Doyle and Elgie 2015) rather than regime types. For example, both Shugart and Carey (1992) and Siaroff (2003) provide a set of presidential power scores for different countries using a particular set of indicators. The former is based on purely constitutional powers. The latter is based at least partly on the practice of political power. The use of such composite indices to measure presidential power in individual countries has its limits, as argued forcefully by Fortin (2013). For the purposes of the current discussion, though, the key point is that these measures do not estimate the effect of semi-presidentialism or the effect of Shugart-and-Carey-like varieties of semi-presidentialism. They are estimating the effect of presidential power on a country-by-country basis. When making this move, scholars are no longer considering the effect of institutional variation based on any form of taxonomic classification. They are engaging in an exercise based on a different way of thinking about presidential politics.

The bottom line is that it is fundamentally mistaken to think that the set of semi-presidential countries includes only those cases where the president has Duverger-like “quite considerable” powers. When we distinguish between presidential, semi-presidential, and parliamentary regimes on a purely taxonomical basis, then we find that the set of countries with semi-presidential constitutions is empirically heterogeneous. It includes countries with very strong presidents, very weak presidents, as well as presidents with “quite considerable” powers. If we add an additional taxonomic criterion and distinguish within semi-presidentialism between president-parliamentary and premier-presidential countries, then the within-category degree of political heterogeneity is certainly reduced, and this can have important and beneficial empirical applications. However, even if we do so, while the set of premier-presidential countries does include those cases where presidents do have “quite considerable” powers like Romania, it also includes plenty of cases where the president is a figurehead like Slovenia. So, the category of premier-presidential countries is not restricted to those whose presidents have “quite considerable” powers. The same point applies to president-parliamentary countries. Finally, if we wish to identify the effect of more fine-grained institutional differences, then we can employ a measure of presidential power. In doing so, though, we are no longer estimating the effect of presidential power in semi-presidential countries or even a set of semi-presidential countries. Instead, we are pooling presidents, including those in presidential and indeed parliamentary regimes understood taxonomically, to see whether these individualized country-by-country institutional differences have an empirical effect across the whole set of countries with the presidency. Nothing, therefore, leads us to the conclusion that the term “semi-presidentialism” implies a category that is restricted to countries whose presidents have “quite considerable” powers.

  • Misconception no. 3—Autocracies cannot be semi-presidential

There is now a large body of work on the effect of institutional variation in democracies. This work includes the standard debate as to whether regime types have a good or bad effect on democracy as a whole. For example, is presidentialism perilous for democracy (Linz 1990; Cheibub 2010)? However, it also includes debates relating to a much broader set of outcomes, including the effect of regime types on public policy within consolidated democracies (Weaver and Rockman 1993). For instance, is presidentialism associated with the supply of more public goods than parliamentarism (Shugart 1999)? At the same time, there is also now an almost equally large body of work on non-democracies. Some of this work aims to identify different types of non-democracies. For example, we might wish to distinguish between on the one hand authoritarian regimes where political competition is outlawed and on the other competitive authoritarian regimes where competition takes place but where it is managed in a way such that it does not threaten the survival of the regime (Levitsky and Way 2010). Why is that some non-democracies remain purely authoritarian, whereas others have competitive elements to them? A feature of this work on non-democracies, though, is that it rarely, if ever, investigates the effect of constitutional regime types. This point applies to presidentialism, parliamentarism, and semi-presidentialism. The difference between the work on democracies and non-democracies in this regard seems to be based on a simple assumption; whereas institutions can have a profound effect on outcomes in democracies, they do not matter at all in non-democracies. The ultimate logic of this argument is that it makes no sense to label non-democracies according to the institutional forms that can be found in democracies. To put it another way, countries that are not democratic cannot be semi-presidential.

This line of argument is flawed. As we have seen, the distinction between presidential, semi-presidential, and parliamentary democracies is the result of a taxonomic exercise. These taxonomic classifications are derived from a reading of country constitutions. Almost all countries, including non-democracies, have a constitution. Certainly, some of these constitutions are unusual. For example, during Muammar Gaddafi’s autocratic rule in Libya, the country was formally run according to the highly idiosyncratic The Green Book. Most, though, take a form that is recognizable in terms of the presidential, parliamentary, and semi-presidential taxonomy. This means that almost all non-democracies can be classed as constitutionally presidential, parliamentary, or semi-presidential. Indeed, if they are constitutionally semi-presidential, then they can be further classed as either president-parliamentary or premier-presidential. Thus, in this sense, non-democracies can be taxonomically semi-presidential.

In fact, the argument, or at least the implicit assumption, that non-democracies cannot be semi-presidential is usually based on the same flawed logic that we identified in misconception no. 2. If we were to understand semi-presidentialism as the situation where a country has a president with quite considerable powers, but where there is also a PM with some political authority too, then, in practice, the set of semi-presidential countries in non-democratic regimes would indeed be empty. Typically, authoritarian and competitive authoritarian regimes are governed by a single, powerful political leader, often holding the office of the president. As we have seen, though, it is misleading to think of semi-presidentialism in this way. At root, the concept is taxonomic. There is no democracy criterion in the definition of constitutional regime types. In other words, the process of taxonomic classification does not have to be confined to democratic countries, not least because there is a bigger and even more contested debate as to what constitutes a democracy in the first place. Given the process of categorizing constitutional regime types is a taxonomic exercise and given that exercise does not include a democracy criterion, then it is perfectly reasonable, when the constitutional criteria apply, to classify certain authoritarian regimes as semi-presidential.

Even if it is conceded that non-democracies can be constitutionally semi-presidential, perhaps, the argument then reverts to the position that institutions simply do not matter in non-democracies and, therefore, are not worth studying. This may be true, but this is an empirical question that is at least worthy of investigation. The heart of the study of constitutional regime types is the idea that variation in the organization of the executive and executive-legislative relations has consequential effects. While it could be the case that in unequivocally authoritarian regimes, loyalty to the leader is so absolute that institutional arrangements are merely a façade, it is at least plausible to suggest that in certain competitive authoritarian regimes, institutional variation may make at least some difference (Gandhi 2008). For example, if the ruling party is factionalized and the president’s support base is weakening, perhaps, the threat of a breakaway group having enough support to lodge a no-confidence motion may be more than just idle talk. In other words, even when democracy is not the only game in town, there may be times when institutional arrangements can matter. If so, then it is worthwhile studying them. Overall, we can dismiss the idea that non-democracies cannot be semi-presidential, and we can at least entertain the idea that in some non-democracies under certain conditions, institutional arrangements may generate particular resources and/or constraints that may help to shape political outcomes.

  • Misconception no. 4—Semi-presidentialism is bad/good for democracy

As we saw, there is a long-standing debate about the effect of constitutional regime types of democratic outcomes. Typically, this debate has been expressed in terms of parliamentarism vs. presidentialism. Linz (1990) argued that presidentialism alone was perilous for democracy. Mainwaring (1993) argued that the combination of presidentialism and a multi-party system was dangerous for democracy. Cheibub (2010) countered by arguing that presidentialism was only dangerous because it was typically adopted in countries where there was a tradition of military involvement in politics. There is an equivalent debate as to whether semi-presidentialism is a good or bad constitutional choice for new democracies. Generally, semi-presidentialism is considered to be a problematic choice. Researchers looking at the impact of institutional design on democratic performance and survival have tended to ascribe many of the same evils of presidentialism to semi-presidentialism. Lijphart, for one, is critical of semi-presidentialism because “the zero-sum nature of presidential elections remains” (2004, 102). Moreover, in Lijphart’s view, the president in a semi-presidential system may become even more powerful than in a presidential one; he references the example of France as a country that has gone through “hyper-presidential” phases when the president and the PM have been from the same party with majority backing in the legislature (ibid.). The existence of a dual executive creates additional perils. There is the potential for intra-executive competition and tensions between the president and the PM, even when both executives are from the same majority party (Linz 1994). Such tensions can develop into destabilizing conflict, in the event of “cohabitation” between a president and a PM from opposing parties (Linz and Stepan 1996). There is also the potential for conflict between the president and the legislature, especially when there is no working majority in the legislature, and the president seeks to impose his or her choice of PM (Skach 2005). At the same time, there are those who argue that semi-presidentialism has advantages for young democracies. Most notably, the dual executive provides the opportunity for power sharing between competing political forces (Sartori 1997), giving both a stake in the system and encouraging the consolidation of democracy. In the event that the president is not backed by a majority in parliament, semi-presidentialism provides a cure for gridlock and stalemate not found in presidential systems—cohabitation. Should divided majorities occur, power effectively shifts to the PM. It is this “head shifting” solution that provides semi-presidential systems the flexibility presidentialism lacks, in Sartori’s view (ibid., 125). Semi-presidentialism may also facilitate coalition building through pre-election agreements on the sharing of posts among coalition partners (Issa Abdourahmane 1996, pp. 80–81). Thus, recent contributions to the literature on institutional design have focused on the importance of coalitional politics for regime survival in both presidential and semi-presidential countries (Chaisty et al. 2014).

We have to be very careful about how we think about the relations between institutions, including semi-presidential institutions, and democracy and, indeed, outcomes generally. There is still a tendency for some scholars to give the impression that they are putting forward monocausal, highly deterministic arguments. This type of argument states, or at least strongly implies, that if a particular institutional arrangement is adopted, then democracy is either inevitably doomed or bound to succeed, depending on the way in which the argument is pitched. However, we know that the determinants of democracy are very complex and are likely to be sensitive to local conditions. This does not mean that we cannot make general, probabilistic claims about the average effect of institutional forms, but we should be very wary of making strong claims that sound as if they can be applied to every case. In other words, we should be very wary of arguments suggesting that any constitutional form, including semi-presidentialism, is by itself either good or bad for democracy. The situation is likely to be much more complex, and at best, institutions are only likely to explain a proportion of the outcome under consideration.

There is, though, another reason why we should be wary of claims that semi-presidentialism is either good or bad for democracy. This is because, as we have seen, semi-presidentialism is a constitutional category that contains an extremely heterogeneous set of countries. Why would we expect this very varied set of countries to generate a monocausal effect on democracy? In this context, rather than making claims about the impact of semi-presidentialism as a whole, we need to explore whether or not some form of systematic variation within the set of semi-presidential countries is associated with better or worse democratic outcomes. Again, scholars have turned to the president-parliamentary/premier-presidential distinction. Shugart and Carey (1992) hypothesized that president-parliamentary regimes are likely to be more problematic for democracy than the latter because they generate an inherent conflict of legitimacy between the president and the legislature. Elgie (2011) has recently provided strong evidence to support this hypothesis, looking at the different incentive structures provided by the two subtypes. In premier-presidential systems, where the president has to work with the legislature to secure maximum influence over the government, he or she is more likely to engage in negotiations over government formation. In contrast, in president-parliamentary systems, where the president is at liberty to dismiss the PM and cabinet, there is less motivation for negotiation over the appointment of cabinet ministers with the legislature. The lack of negotiation in the cabinet formation phase is likely to carry over to other aspects of the political process as well. This could, for example, lead to a higher rate of vetoed legislation and to greater political instability overall. A large-n test of the correlation between democratic collapse and semi-presidential subtype has confirmed that president-parliamentary systems are much more likely to break down than their premier-presidential counterparts (ibid.). Overall, we should be wary of claims that semi-presidentialism is either good or bad for democracy. Instead, we should identify institutional variation within semi-presidentialism and examine the consequences of such variation.

To sum up, we understand semi-presidentialism purely taxonomically. This means that the set of countries with semi-presidential constitutions is very varied and is not confined to a small group of countries that have presidents with quite considerable powers. Given we think of semi-presidentialism taxonomically, this means that we can think of non-democracies as being semi-presidential. If their constitutions meet the formal definitional requirements of semi-presidentialism, then we can classify them as semi-presidential. However, again because we restrict ourselves to a taxonomical definition of semi-presidentialism, we make no claims that semi-presidentialism has any effect on outcomes of any sort either in democracies or non-democracies. The political variation within countries that meet the taxonomical requirements of semi-presidentialism is so great that it is ludicrous to think that there is likely to be a single semi-presidential effect of any sort. This point applies not least to the debate about the impact on democracy of countries with semi-presidential constitutions. In fact, we should be dismissive of monocausal deterministic-sounding claims about the impact, good or bad, of semi-presidentialism on democracy. Instead, we need to identify systematic variation within the set of countries with semi-presidential constitutions so as to reduce the within-category heterogeneity. To date, Shugart and Carey’s distinction between president-parliamentary and premier-presidential regimes is the favored way of doing so. This distinction allows president-parliamentary and premier-presidential countries to be identified reliably. This distinction is also fairly well correlated with stronger vs. weaker presidents. There may be good theoretical reasons that variation in presidential strength has a consequential effect on a range of political outcomes. Indeed, it is plausible that these effects may manifest themselves in both democracies and non-democracies, or at least in some competitive authoritarian regimes. Having established the basis of our thinking on semi-presidentialism and, hopefully, dispelled some common misconceptions about semi-presidentialism, we turn now to the way in which we are approaching the study of semi-presidentialism in the Caucasus and Central Asia.

Semi-presidentialism in the Caucasus and Central Asia

In the countries of the former Soviet Union, constitutional semi-presidentialism is particularly prevalent. As the Union of Soviet Socialist Republics (USSR) collapsed and former union republics acceded to statehood, most introduced a directly elected president, while maintaining parliamentary features of the Soviet constitution, following the constitutional example of Russia. Alexander Baturo discusses these developments across the region in more detail in the following chapter. That said, constitutional semi-presidentialism is not ubiquitous in the former USSR. In the Caucasus and Central Asia region, Uzbekistan, Tajikistan, and Turkmenistan have constitutions that do not meet the criteria for semi-presidentialism. See Fig. 1.3 that summarizes the classification of regimes in the subregion. In Uzbekistan, it is the president’s privilege to appoint and dismiss the PM and cabinet (article 93.9). While the cabinet must be approved by the legislature (article 98), the 1992 constitution does not include provisions enabling the Oliy Majlis (legislature) to initiate a vote of no confidence in the PM or cabinet or to dismiss the government by withholding approval of the government’s program. Constitutional provisions in Tajikistan are similar, giving the president the power to appoint and dismiss the PM and cabinet (article 69, 1994 constitution as revised in 2003). The two houses of the Majlisi Oli (legislature) may in joint session approve presidential decrees for the appointment or dismissal of the PM and other members of government (article 55), but the parliament does not have independent powers of dismissal of the government or PM. In Turkmenistan, the president forms and chairs the cabinet; the 2008 constitution (as the 1992 constitution before it) does not provide for a PM position (articles 72–74).

Fig. 1.3
figure 3

Caucasus and Central Asia regime classification

Full size image

This book provides country cases of all four semi-presidential countries in the subregion—Armenia, Azerbaijan, Georgia, and Kyrgyzstan. In all four countries, the constitution provides for the position of the PM and gives the legislature the power to dismiss the cabinet. Among our country cases, Armenia, Georgia, and Kyrgyzstan fall in the premier-presidential category, since 2005, 2013, and 2010, respectively, while Azerbaijan is president-parliamentary. Armenia, Georgia, and Kyrgyzstan previously fell in this latter category, but subsequently transitioned to premier-presidentialism, as Alexander Markarov, Malkhaz Nakashidze, and Matteo Fumagalli discuss in their respective country chapters. In Armenia, the president must accept the resignation of the government if the National Assembly passes a vote of no confidence or fails to approve the program of the government, or if the PM resigns (article 55.4). The president does not have the independent initiative of dismissal of the PM or government. In Georgia, the 2010 amended constitution entered into force in 2013, upon the inauguration of the winner of the 2013 presidential poll, Giorgi Margvelashvili. The constitution now gives the PM nominated by the president the right to appoint and dismiss other members of the government (article 79.6). Parliament must vest the newly formed government with a confidence vote, or else the cabinet must be replaced (article 80). Parliament also has the power to initiate a vote of no confidence and nominate a candidate to replace the sitting premier (article 81). The president does not possess independent powers of dismissal of the PM or cabinet. The prerogatives of the Jogorku Kenesh (parliament) in Kyrgyzstan’s 2010 constitution are even more sweeping; it is the majority parliamentary group or majority coalition of groups that nominates a candidate for the PM position. This candidate then appoints a cabinet and submits the composition of the government and its program for approval by parliament. The president’s role is limited to issuing a decree appointing the PM and cabinet already validated by the Jogorku Kenesh (article 84). Only parliament has the power to oust a sitting cabinet through a vote of no confidence. The president is entitled to disagree with the no-confidence vote. However, if the Jogorku Kenesh “repeatedly” passes a no-confidence vote during a 3-month period, the president has to dismiss the government (article 85). In contrast, the constitution of Azerbaijan gives not only the legislature but also the president the right of initiative to dismiss the PM and/or cabinet—making the government accountable to two masters. In Azerbaijan, the 1995 constitution as amended in 2009 gives the president the power to appoint or remove the PM, albeit with the consent of the Milli Majlis (article 109.4). The president can furthermore, without legislative approval, appoint or remove other members of the cabinet.

The Kazakhstan case is more difficult and is dealt with in considerable detail in Chap. 6. In a nutshell, Kazakhstan had a short period of constitutional semi-presidentialism in the early 1990s, but we classify the current constitution as presidential, even if it is on the constitutional cusp of semi-presidentialism and presidentialism. Why so? The 1995 constitution establishes a directly elected fixed-term president (article 41). It also allows the parliament to express no confidence in the government (article 56.2) in a way that triggers the government’s collective resignation to the president (article 70.3). The president also has the unilateral right to dismiss the PM and government (articles 44 and 70.7). These constitutional clauses would seem to establish a classic example of president-parliamentarism. However, the 1995 constitution also states that if the legislature votes no confidence in the government, then the president has the right to accept or decline the government’s resignation (articles 70.4–70.6). If the president declines the resignation, then the government remains in office. In other words, it may look as if the legislature can dismiss the government, but according to the wording of the constitution, the final decision remains with the president. In our terms, this makes Kazakhstan constitutionally presidential, because the government is not responsible to the legislature. In this event, why are we including Kazakhstan in the volume? There are two reasons. The first is that there was a brief period of constitutional semi-presidentialism from December 1991 to January 1993. The second reason is that Kazakhstan acts as a neat counterpoint to address one of the important issues with which the book is dealing, namely do institutions matter in authoritarian and semi-authoritarian countries. If similar dynamics are at work in both constitutionally presidential and semi-presidential countries under authoritarianism and/or competitive authoritarianism, then presumably they do not. We now turn to the questions that drive this project.

Research Questions

Do Institutions Matter in Post-Soviet Countries?

To what extent, have semi-presidential institutions in Central Asia and the Caucasus had an impact on the practice of politics in the region since statehood in the early 1990s? Has the vertical power tradition of the Soviet system supervened on all institutional factors, or has there been sufficient openness and accountability for institutions to have an effect? Is it a waste of time to study the role of institutions in non-democracies or partial democracies? These are some of the questions the country chapters will collectively seek to address.

The countries of the Caucasus and Central Asia all achieved statehood/independence when the USSR collapsed in 1990. The unexpected and swift transition from Soviet rule had and continued to have considerable implications for governance in this region. In Kazakhstan, even though a new democratic-looking constitution was adopted in the years immediately after the collapse of the USSR, the elites remained largely unchanged; the practice of vertical power remained in place; and democracy never emerged. In Azerbaijan, pluralist politics emerged, and there was a short period of quasi-democratic politics, but soon the system was captured by elites who, controlling economic power, re-established an autocratic system. In the other semi-presidential countries in this region, democratic practice has varied. Armenia is a classic case of a stable quasi-democracy with an essentially pluralist system, but one in which the executive is able to distribute resources so as maintain a firm hold on power. In Kyrgyzstan, there have been repeated attempts by elites to capture the system, but there have also been periods of fragile but nonetheless pluralist democracy. In Georgia, the Rose Revolution swept away the façade democracy of the Soviet-era elite with recent elections bringing about a further alternation in power.

This variation in democratic practice across the region shows that despite a shared history as Soviet Republics, once independent, these five countries evolved quite differently. In Azerbaijan and Kazakhstan, the practice of vertical power seems to have prevailed, whereas in others, including Georgia, Kyrgyzstan, and even Armenia, politics has been much more competitive. Moreover, even in countries that have remained authoritarian or semi-authoritarian, constitutional tinkering has been a recurring theme. Ruling elites have found it necessary to govern with the veneer of constitutional legitimism. Indeed, in authoritarian Azerbaijan, as Jody LaPorte describes, the confluence of formal and informal sources of authority has served to enhance presidential powers.

If So, Do They Matter for Democratic Performance?

In Fig. 1.4 below, the divergent democratic trajectories of the countries in the region are reflected by the Polity IV democracy scores and corresponding regime classifications. Polity IV scores measure qualities of governing authority on a scale running from −10 (institutionalized autocracy) to +10 (institutionalized democracy). These democracy scores are based on a composite index of six measures of executive recruitment, constraints on executive authority, and political competition (Marshall et al. 2014, pp. 20–28). The scores are converted into a three-part regime categorization: autocracies (−10 to −6), anocracies (−5 to +5), and democracies (+6 to +10). Anocracies are “incoherent” polities “in which odd combinations of democratic and autocratic authority patterns could be observed” (ibid., 9). The shaded bars in Fig. 1.4 provide a visual illustration of variation over time in Polity IV democracy scores in our country cases. Wider bars imply higher democracy scores. While Kazakhstan and particularly Azerbaijan have evolved into autocratic regimes, Kyrgyzstan and Georgia have developed predominantly democratic authority patterns. Finally, Armenia has moved from democracy through a short authoritarian period into a quasi-democracy.

Fig. 1.4
figure 4

Polity IV democracy scores 1991–2013.Source: Polity IV (2014). Adapted by the authors.Legend: DEM democracy, AUTO autocracy, ANO anocracy.Asterisk: This is a converted Polity2 score, converting the original Polity score of −88 for “political transition” to an annual score (Marshall et al. 2014, p. 17)

Full size image

So has semi-presidentialism, even in its superpresidentialized form, helped to shape political dynamics within the executive and/or between the executive and the legislature? In this event, has it had a positive or negative effect on democratic performance? To answer these questions, we need to place the institutional analysis in its broad historical, cultural, and regional context. For example, what is the legacy of Soviet-era practices and elites? What are the pertinent local factors in any given country? These may include the presence or absence of natural resources, the nature of ethnic divisions, and issues relating to territorial integrity, especially in countries such as Azerbaijan and Georgia. In addition, central to the dynamics of politics in this region since 1990 has been developments in Russia. Domestic developments in Russia—the difficult transition, the period of extreme party fragmentation in the Duma and countervailing presidential rule, and the emergence of a hegemonic party and super-presidentialism/prime ministerialism—are likely to have had a “demonstration effect” in countries of the Caucasus and Central Asia. Over and above the impact of such domestic factors, Russia’s geopolitical interests are also likely to have shaped the domestic politics of other countries in this region, generating pro- and anti-Russian camps, sometimes encouraging alliances, and sometimes provoking conflict. The story of democracy and the effect, however, large or small, of institutions on democratic performance cannot be told without integrating the impact of Russia into the narrative.

Whatever the Answer, Has the Organization of the Executive and Executive-Legislative Relations Had an Impact on Political Life?

This book provides an overview of variation in subtypes of semi-presidential regimes between countries in the region and over time within individual countries, tracing the political consequences, if any, of institutional variation in our country cases. The individual country chapters seek to unpack the institutional incentives provided by premier-presidential versus president-parliamentary constitutions as they relate to the exercise of presidential power, the relationship between the president and the PM, and the collaboration (or not) between the executive and the legislature. To sum up, we wish to identify whether or not semi-presidentialism, and in particular, institutional variation within semi-presidentialism broadly understood, has had an impact on governance in the Caucasus and Central Asia since 1990. If it has, then in relation to democracy, we wish to identify whether it has had a positive or negative effect. Moreover, even if semi-presidentialism in its various forms has had little or no effect on overall democratic performance, has it generated more specific institutional incentives to which political actors have reacted?

The Structure of the Book

The book includes a contextual chapter, five case study chapters, and a conclusion. In the contextual chapter, Alexander Baturo places the countries of the Caucasus and Central Asia in their broader historical and regional context. He discusses the legacy of the breakup of the USSR, introduces the notion of vertical power, and examines the impact of Russia in this region after 1990.

The five country chapters, each written by country experts, follow a similar template to ensure the coherence of the book. Each chapter is thus divided into three substantive sections, the first of which focuses on the origins of semi-presidentialism. This section outlines the conditions under which semi-presidentialism was chosen. This involves identifying the constitutional situation in the period immediately following the breakup of the USSR and the onset of statehood/democratization. It also involves identifying the debates (or non-debates) at the time when the first constitution was adopted. In some countries, there have been multiple constitutions and/or multiple constitutional amendments. If these constitutions and/or amendments affected the constitutional nature of the regime, then the contributors identify the context in which constitutional choices were made. Specifically, contributors identify the extent to which Soviet institutions were transposed in the post-Soviet context. The second section of each country chapter focuses on the formal constitutional situation as it currently stands, identifying the constitutional and legal powers or the president, the PM, cabinet, and the legislature. The authors also discuss how such powers have varied over time, through constitutional or other reform. Each country chapter concludes with a discussion of political practice, focusing on how the executive and legislative power has been exercised. This section focuses not solely on the president, but also on the role of the PM, the government collectively, and the legislature.

Finally, the concluding chapter of the book brings together the findings of the different case studies and reflects on the general questions that were posed earlier in this chapter.

Overall, the book aims to provide the first in-depth study of the constitutionally semi-presidential countries of Central Asia and the Caucasus. In so doing, we wish to reflect on whether institutions matter and, in particular, whether semi-presidential institutions matter. By matter, we mean whether semi-presidential institutions have affected the performance of democracy in these countries since 1990 and also whether they have shaped the conduct of day-to-day political practice more specifically. We would be surprised if institutions have had no effect anywhere. However, we have absolutely no expectations that they will have had the same effect everywhere. We have also a suspicion that in some places, semi-presidential institutions may have had very few practical consequences. In general, we expect vertical power to be very strong in our region, though stronger in some places more than others. Having identified the aim of the volume and the basic expectations with which we embark on the project, we now turn to a general overview of the post-Soviet space before zooming in on individual country case studies.