1

Putting Institutions into Perspective: Two Waves of Authoritarianism Studies and the Arab Spring1

Kevin Koehler and Jana Warkotsch European University Institute

Department of Political and Social Sciences kevin.koehler@eui.eu

jana.warkotsch@eui.eu

Paper prepared for the Panel Conceptualizing Autocracy at the 2011 General Conference of the European Consortium for Political Research, Reykjavik, August 2011

The mass protests that recently shook the Arab world and, in reminiscence of such earlier springs in the region as the Damascene Spring of 2000/2001 or the Cairo Spring in 2005 have come to be known as the Arab Spring (Anderson 2011), not only took the Arab and international public by surprise, but also challenged long-held assumptions within the academic community (Gause 2011). After all, the study of authoritarianism in recent years had mainly been focused on explaining why such regimes as those of the Arab world had been so remarkably (and maybe, one might add in hindsight, seemingly and superficially) stable and resilient. The events thus ushered in a period of self-reflection for many scholars of Middle Eastern politics and authoritarian rule. Plagued by the question of why we have been unable to foresee events of such magnitude and to predict the instability of these regimes, it is tempting to proclaim the failure of authoritarianism studies and to turn to other fields for conceptual salvation.

Instead of joining the chorus of those proclaiming the death of authoritarianism studies, we want to take this chance to thoroughly take stock of the discipline of what is left of it after the Arab Spring seemingly dealt it such a decisive blow. While

1 Parts of the paper are based on an earlier contribution to a workshop on authoritarian rule which is published

(in German) as Koehler & Warkotsch (2010). We would like to thank the participants in this workshop, and especially the editors of the conference volume, Holger Albrecht and Rolf Frankenberger, for their helpful feedback. We would also like to thank the participants in a workshop on authoritarianism in the MENA at the European University Institute, in particular the co-chairs Oliver Schlumberger and Philippe Schmitter, for their feedback. In addition, Adrienne Hritier and Peter Mair read and commented on this paper and gave us critical input and suggestions. Remaining shortcomings are nevertheless exclusively our responsibility.

2

the discipline has changed considerably over the years in terms of the questions asked and the approaches and methods employed, the problem of how to subdivide the space of authoritarian regimes into conceptually and empirically meaningful categories, as well as how to define the boundaries to neighboring areas, or more specifically to democratic regimes, remained at the heart of the subfield. While classificatory questions have always been central to the discipline, the uses to which they have been put varied quite significantly: Earlier studies of authoritarian rule in what we refer to as the first wave of authoritarianism studies in the 1970s and 1980s tended to see different forms of authoritarianism as political outcomes whose emergence was to be explained by specific constellations of political and social forces in developing nations, whereas second wave studies from the late 1990s onwards converged on seeing regime types as explanatory factors for regime stability and breakdown.

External events certainly drove the many changes in the way authoritarian regimes were studied and classified. These developments at the same time also mirror paradigmatic changes in the mother discipline of Comparative Politics itself. First wave approaches, broadly speaking, relied on the background of grand theorizing in the modernization theoretical tradition, focusing on how processes of socio-economic change during the transition from tradition to modernity affected political structures. Under the methodological influence of behavioralism, moreover, there was a strong tendency to explain macro-level outcomes as an aggregation of individual-level factors. Second wave approaches, on the other hand, adopted a (new) institutionalist focus on the role of institutions in shaping political behavior and outcomes via the reliance (in its rational choice variant) on formal models and regression analyses. Taking institutions as independent variables, second-wave approaches (in accordance with the general development in Comparative Politics), both reversed the logic of their first-wave predecessors and emphasized the independent influence of institutional factors.

However, the considerable academic attention authoritarian rule has received in the past decades notwithstanding, the questions of how to properly subdivide the space of authoritarian regimes as well as how to define them to begin with, remain fraught with problems. While there have been some attempts to do both in a comprehensive fashion, most notably by Juan Linz in 1975 (then republished in 2000), work in the first wave of authoritarianism studies barely bothered with exploring the range of authoritarian regimes and focused on specific subtypes instead; the second wave literature, by contrast, while producing some influential typologies generally relied on rather truncated definitions of authoritarianism and its bordering categories and tended to overemphasize the significance of institutional factors, thus reproducing conceptual difficulties that found their expression in the debate on hybrid regimes and new authoritarianisms that are still in the heart of debates on nondemocratic regimes.

This paper thus serves several purposes at once. On the one hand we will present the state of the art in authoritarianism studies by reviewing the two distinct waves of scholarship on authoritarian rule mentioned above and by situating the development of the subfield in the broader context of theorizing in Comparative Politics. On the other hand, we aim to show how the current state of the art suffers from empirical

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

3

and conceptual problems arising out of second wave scholarship. The recent turn towards a stability debate that has dominated authoritarianism studies in the 2000s and focused on authoritarian institutions began to address some of the conceptual issues but left others unresolved and hence reproduced the main theoretical problems. We will first review the development of the discipline, then address these methodological and theoretical issues, and finally try to point to ways of moving the field forward by drawing on illustrations from regime trajectories in the Arab Spring.

Two Waves of Authoritarianism Studies

Authoritarianism studies developed in two distinct waves that differed in terms of the main focus of the respective debates. One of the most fundamental differences between these two waves is the degree to which the different authors take into account factors transcending more strictly institutional aspects of political regimes. Whereas the first wave of the 1960s and 1970s analyzed nondemocratic political orders primarily in their interaction with broader socio-economic conditions, thus focusing on the interrelations between the political, social, and economic subsystems, second-wave approaches from the late 1990s onwards mainly restricted their analyses to features of the political regime proper and tended to focus on formal institutional structures. In the next sections, we provide a schematic (and necessarily incomplete) overview over some central contributions to both debates that together constitute the state of the art in conceptual thinking about authoritarian rule.

Political Order, Development, and Authoritarian Rule: First-Wave Approaches to Authoritarianism

In order to understand the emergence of first-wave approaches to the study of authoritarianism, it is imperative to locate them within the broader development of Comparative Politics. Developing mainly from the 1970s onwards, first-wave scholarship was deeply embedded in the dominant paradigm of that time modernization theory (see Almond & Coleman 1960; Apter 1965; Huntington 1968; Huntington & Dominguez 1975) which arguably shaped the discipline in conceptual as well as methodological regards. First-wave approaches generally remained committed to this conceptual tradition, even though many authors more or less strongly rejected some of its theoretical assumptions or conclusions.

Thus the classical studies of authoritarianism we review here in general shared a common interest in the socio-economic conditions shaping authoritarian rule, although many authors explicitly rejected the supposedly uniform relationship between economic and political development implicit in modernization theory (e.g. Huntington 1968; ODonnell 1973; Schmitter 1971). According to classical modernization theory, the traditional societies of the developing world were expected to gradually develop more complex economic, political, and social structures in a process of modernization that would ultimately result in the emergence of modern

FarkMac

FarkMac

FarkMac

FarkMac

4

political and economic systems modeled after the Western example (Apter 1965; Binder et al. 1971; Rustow 1967). The original impetus behind models of modernization and political development was thus to understand political processes in the rapidly expanding universe of independent countries of the 1950s and 1960s by focusing on how they tackled the supposedly universal challenges and crises of modernization and political development.

With modernization the prospects of economic wealth and political stability appeared on the horizon of underdeveloped nations that could, it was hoped, draw on the earlier experiences of modernization and political development in the west and thus avoid some of the more painful by-products of the process.2 On the other hand, however, such developments were also seen as posing significant dangers and challenges to societies undergoing modernization. Increasing levels of economic development, industrialization, the expansion of education and social mobility, the emergence of new social roles, urbanization, and associated processes would ultimately, it was assumed, lead to attitudinal and behavioral changes that were bound to exert adaptive pressures on the political system. Since the fundamental process of modernization was thought to be universal (see especially Apter 1965; Bendix 1977; Binder 1971; Rustow 1967), analysts were mainly preoccupied with understanding the conditions under which political instability could be avoided and political order maintained under such circumstances. The link between the process of modernization and challenges to political stability was provided by the theory of relative deprivation (see especially Gurr 1970; also see Huntington & Dominguez 1975, 8): The social and attitudinal changes associated with modernization would lead to increasing aspirations among different social groups; if these increasing aspirations were not met by increasing opportunities to achieve their fulfillment, the likelihood of civil violence was thought to increase.

Far from simply describing the supposed social mobilization and resulting increase in demands for participation, many modernization theorists actually took a fairly explicit stance against mass involvement in the politics of developing nations. Based on the rationale that an increasing gap between demands for participation and a political systems capacity to institutionally channel such demands would stall or entirely endanger the modernization process itself, modernization theorists often saw authoritarian methods of rule as necessary stages in a larger process of development (see for example Apter 1965; Rustow 1967; Huntington 1968). The concern for political order clearly trumped the concern for forms of political rule.

Thus, modernization theory generally was based on an elitist model of politics whereby all that was necessary for modernization was the development of a sufficiently educated, westernized elite that adhered to a scientific understanding of reality that supposedly was characteristic of modern thought. This elite would then 2 This project was mainly pursued by Social Science Research Councils Committee on Political Development

under the guidance of Gabriel Almond that commissioned a series of edited volumes on questions of political development that ultimately aimed at understanding such processes against the at least implicit backdrop of the European experience (see Binder et al. 1971; LaPalombara & Weiner 1966; Pye & Verba 1965; Tilly 1975).

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

5

man state institutions and successfully drive modernization from above without the potentially disruptive interference of the masses (Huntington 1968). Hand in hand with modernization theories elitism hence went a strong focus on the state as instrument of modernization in the hands of elites.

Whereas early modernization theory was characterized by a certain optimism with regard to the prospects of a quick recapitulation of the European experience in the new nations in the form of quicker and less painful processes of industrialization and social transformation for such late-comers, this optimism over the years gave way to a more subdued vision where progress was not easily achieved and the transition process from tradition to modernity was fraught with perils. Consequently, hope for fast democratization gave way to wariness about the political stability of these states during the transitional phase: modernity breeds stability but modernization breeds instability (Huntington 1968).

First-wave scholarship in this respect took an increasingly pessimistic stance towards the prospects of democracy in transitional societies and instead described how changes in state-society interaction brought about by the disruptive force of social, economic and political modernization resulted in distinct forms of authoritarian rule. Elites were trying to solve the problems created by modernization and reacted differently to different developmental challenges. The bureaucratic-authoritarian military regimes of Latin America were thus interpreted as emerging from the challenges of capitalist deepening at relatively advanced levels of modernization (see especially ODonnell 1973), while single party regimes in Africa were presented as political elites attempts to overcome problems of national integration and nation-building (Zolberg 1966). In brief, the centralization of political authority in the hands of elites and the suppression of demands for (immediate) participation were seen as the result of and to some extent necessary for achieving specific developmental goals.

In addition to this common perspective, first-wave literature was also characterized by a focus on either a specific region populated predominantly by authoritarian regimes, mainly Latin America and Sub-Saharan Africa, or on a specific subtype of authoritarianism (see e.g. Finer 1988; Jackson & Rosberg 1981; ODonnell 1973; Zolberg 1966). While these authors often constructed a rough overall typology of authoritarian rule as they went along, there are very few systematic attempts to capture the whole range of nondemocratic regimes (see Linz 1975 and 2000; and Perlmutter 1981 as notable exceptions).

In our review on the following pages, we follow the literature in adopting their focus on specific regime types clustered in geographical regions. The discussion of military regimes thus mainly concentrates on Latin America, whereas the debate on single-party rule is primarily concerned with post-independence Sub-Saharan Africa; in the last section, finally, we review some studies on personalism (or patrimonialism) that grew out of debates on the persistence of supposedly traditional modes of political rule.

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

6

Military Authoritarianism

Military interventions apparently are an inseparable part of political modernization whatever the continent and whatever the country (Huntington 1968, 192). This paradigmatic statement by Samuel Huntington aptly illustrates the modernization-theoretic backdrop that informed many of the characteristic works on military regimes specifically and authoritarianism more generally. Authors such as S.E. Finer (1988), for example, held that military rule in its most direct form was always embedded in a context of a low economic and political development, whereas others such as Guillermo ODonnell (1973) argued that relatively high levels of development in some Latin American cases had led to the rise of the special brand of bureaucratic-authoritarianism, combining a highly technocratic coup coalition of military officers, managers and bureaucrats. While there is thus disagreement concerning the exact nature of the socio-economic conditions giving rise to military intervention, most classical authors agree that military rule is intimately tied to a specific stage of the modernization process (Finer 1988; Huntington 1968; Huntington & Moore 1970; Nordlinger 1977; ODonnell 1973; Perlmutter 1981).

The range of different types of military involvement in politics can best be exemplified by Amos Perlmutters study on Modern Authoritarianism (1981). Perlmutter classifies what he calls the non-institutionalized variant of authoritarian regimes3 into corporatist and praetorian subforms, the former of which he further subdivides into exclusionary and inclusionary corporatism, and the latter into personalist, oligarchic, and corporate praetorianism.

Corporatism, according to Perlmutter, is a type of political domination by a coalition of politicians, technocrats, military men and bureaucrats, with the military as the ultimate arbiter and source of elite recruitment, in which different more or less organized and more or less autonomous social groups are linked to the state and its bureaucracy via patrimonial-clientelistic structures of control (Perlmutter 1981, 38 and 117). Praetorianism, on the other hand, essentially refers to a military dictatorship and, depending on the degree of military interference, is subdivided into personalist (direct rule by a military despot), oligarchic (the military ruler is dependent on the military establishment to secure his rule), and corporate regimes (the military is still the most powerful group, but rule is exercised by a coalition of the military and bureaucrats) (Perlmutter 1981, 129). This scale thus reaches from military tutelage over civilian politics to direct rule by military officers.

The exclusionary corporatist pole of Perlmutters authoritarianism scale is further elaborated upon by Guillermo ODonnell (1973) who is mainly concerned with explaining the rise of regimes he branded bureaucratic-authoritarian. At the same time 3 The institutionalized variants consist in essence of the two totalitarianisms of the 20th century Bolshevism

and Nazism as well as Fascism (Italy). Perlmutter thus dissolves the totalitarianism category into the broader authoritarianism category. Besides giving rise to potential theoretical problems, empirically each corresponds to only one case, which renders the utility of this conceptual subdivision somewhat doubtful. Following more conventional usages of the authoritarianism concept, we will thus in the further discussion of his argument exclusively deal with the noninstitutionalized types.

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

7

he explicitly turns against the idea of a uniformly positive relationship between modernization and democracy inherent in modernization theory. Focusing on the development of bureaucratic authoritarianism in Argentina and Brazil, he argues that the process of industrialization in these countries led to an increase in the size of the urbanized labor force, as well as in what he calls technocratic roles.4 As a result of these developments, more social sectors became politically activated, putting increasing demands on the political system. With growth unsustainable over the long run, developmental bottlenecks occurred that reduced the performance of the populist regimes and led to gaps between demands and performance (ODonnell 1973, 74). Efforts to minimize this gap, along with the multiplication of political forces as a result of deepening social differentiation, created new and sharpened existing conflicts over the distribution of economic and political power, while diminishing the problem solving capabilities of the existing regime (ODonnell 1973, 79). The result was mass praetorianism, providing the stage for the take-over of the military. The bureaucratic authoritarian regimes emerging from military intervention attempted to solve these structural problems by excluding the working classes and bringing order to a divisive political environment. The other side of the coin of these basic problems of social structure and modernization is the frustration of elite actors occupying technocratic roles managers, military officers, technocrats who often had acquired their training abroad and were transplanted to the context of the modernizing society. The greater the penetration and linkage of these technocratic roles, the higher the probability that a coup coalition will emerge that aims at reshaping social structures to make them more compatible with their learned role-expectations. This eventually results in an exclusive and highly coercive regime, aimed at the political deactivation of the working classes and the elimination of divisive politics more generally in the service of further economic modernization (ODonnell 1973, 88-91).

On the other end of Perlmutters scale, S.E. Finer (1988), basing his analysis on a distinction between the motive, mood and opportunity for military intervention, finds a correspondence between the degree of intervention influence, blackmail, displacement, or supplantment and the level of political culture of a given society. The level of political-cultural development mainly refers to the attachment to civil institutions within a given population and influences the degree of legitimacy a potential military intervention can claim. Thus, the higher the level of development in political culture, the higher the standards for the legitimation of authority and the less overt the military intervention.5

All these conceptions of military rule whether on the corporatist end of the scale or the praetorian one share the common feature that the military is taken to intervene in societies shaped by overt political conflict and mobilization. In each of these cases, 4 According to ODonnell, technocratic roles are positions in a social structure which require application of

modern technology as on important part of their daily routine. To perform these roles, each incumbent must have prolonged schooling geared to provide the necessary technical expertise. In addition he must keep abreast of developments in the more industrialized societies, where most of these roles originated (1973, 30-31).

5 For a similar conception of military rule see Eric Nordlinger (1977), who argues for a subdivison into moderators, guardians, and rulers based on different degrees of military intervention.

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

8

the military perceives itself (and sometimes is perceived) as an institution standing above the rest of society, either because of their Westernized training and corresponding role expectation (in coalition with similarly minded technocrats), or because of their unitary and more advanced organizational character in under-organized societies an organization superiorly positioned to deal with the modernizing challenges their societies and economies face. It is thus not surprising that, in addition to reshaping their respective systems, the regimes that resulted from military rule more often than not had a highly exclusionary character, aiming at demobilizing their societies rather than allowing them a role in pushing forward the national project.

Single-Party Regimes and National Integration

The conditions that led to the emergence of the second major type of modern authoritarian regimes as seen by the first-wave literature, namely single-party rule, were different from that leading to military intervention.6 As Samuel Hungtington observed, single-party regimes are always the product of nationalist or revolutionary movements from below which had to fight for power (Huntington 1968, 418). The classical literature on single-party rule is thus mainly concerned with the post-independence development of new states emerging from colonialism and focuses on the role of dominant political parties in the processes of nation building and national integration (see Apter 1955 and 1965, 179-222; Coleman & Rosberg 1964; Huntington 1968; Moore 1962; Schachter 1961; Wallerstein 1960; Zolberg 1963 and 1966).7 Exploiting their privileged position as the only organized political force, independence-movements-turned-parties in many cases monopolized political power and established dominant- or single-party regimes.

Juan Linz consequently discusses this form of authoritarianism under the title of post-independence mobilizational authoritarian regimes (Linz 2000, 227-233). In such regimes, empirically mainly located in post-independence Africa (see Brooker 1999, 106), the period of colonialism had destroyed traditional structures of political domination and led to the emergence of a nationalist movement under the leadership of mainly Western-educated elites. Facing economically as well as socially little developed societies with low levels of national integration, in many cases, these movements transformed into dominant or single parties once independence had been achieved. This process is often attributed to the overwhelming economic strain, the difficulties of a nation-building project in poorly integrated societies, or perceived threats from mounting opposition (Linz 2000, 229; also see Schachter 1961; Zolberg 1963 and 1966).

6 In fact, David Apter (1965, 396) described what he called a mobilization system as being very likely to

transform into a military oligarchy once the early stages of mobilizational success are over. 7 We consciously exclude the communist single-party regimes in Eastern Europe from consideration here since

they would have been considered totalitarian in Linzs framework. This leaves us with mainly African single-party cases (see Brooker 1999, 106).

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

9

The process of Creating Political Order by transforming a nationalist movement first into a dominant and then into a single party was aptly described by Aristide R. Zolberg (1966) for five West African cases. Tracing the emergence of unitary ideologies as well as single-party structures in Ghana, Guinea, Ivory Coast, Mali, and Senegal, he analyzes the transformation of nationalist movements into single parties as well as the characteristics of the resulting regimes. According to this analysis, the roots of single-party rule lay in a combination of structural, psychological, cultural, as well as ideological factors. Psychologically, having achieved national independence, African leaders faced a nation building project in economically poor and socially weakly integrated societies, the sheer magnitude of which put enormous psychological strain on them. Ideology in this regard served as a conceptual map created by men facing an unknown political world, pinpointing a way of dealing with the burden of nation building in this case by instituting a one-party ideology that defined opposition as illegitimate (Zolberg 1966, 59 and 91). Whereas governmental bureaucracies were often perceived as alien Western imports and partially remained in the hands of European elites even after independence (Schachter 1961, 294), the party represented a familiar tool of organizing political affairs and often enjoyed an organizational monopoly. In addition, single-party structures were further seen as supported by a cultural understanding of unity as oneness, not as a form of unity achieved via the competition of different interests. The party in this regard was the organizational expression of this unity (Zolberg 1966, 62; also see Linz 2000, 229). The resulting regimes, in Zolbergs words were system[s] of government with a monocephalic and nearly sovereign executive; a national assembly that is consultative rather than legislative and which is based on functional and corporate representation rather than geographical and individual; a centralized political administration []; and a governmental bureaucracy in which the criterion of political loyalty is given overwhelming weight (Zolberg 1966, 108).

It is doubtful, however, to what an extent many of the African cases discussed in this context ever reached a level of organization that justifies speaking of mass parties (Schachter 1961). Thus, as Zolberg acknowledges, the West-African party-states approximate Webers patrimonial type in many important respects. The relationships between the ruling group and their followers are indeed based on personal loyalty (Zolberg 1966, 141; also see Brooker 1999, 125-29; Harik 1973). In the next section, we turn to more explicitly consider this type of political rule that has been classified as pre-modern by much of the literature based on the premises of modernization theory.

Neopatrimonialism and Personalist Rule

Linz (1975 and 2000) explicitly excluded regimes seemingly or de facto based on what Max Weber (1978) called traditional authority from his typology of authoritarian regimes and adopted the Weberian term of sultanism to describe such forms of political rule (see Linz 2000, 151-155; also see Chehabi & Linz 1998). Beginning in the 1970s, however, especially Africanists and scholars working on the Middle East began to reconsider this distinction (see Eisenstadt 1973; Lemarchand & Legg 1972; Roth

FarkMac

FarkMac

FarkMac

FarkMac

10

1968; Springborg 1979). There is a certain tension between the work of these scholars and mainstream modernization theory with the new focus on the role of personal loyalty in authoritarian regimes implying a critical attitude to some of the assumptions of the first studies of modernisation and political development (Eisenstadt 1973, 8). While under the assumptions of modernization theory, clientelist or patrimonial patterns of behavior where relegated to the realm of pre-modern or traditional politics, the term modern neo-patrimonialism first introduced by Samuel N. Eisenstadt (1973) emphasized the fact that some modern political systems seemed to combine legal-rational and patrimonial forms of domination (see also Mdard 1982, 179; Roth 1968).

The first author to comprehensively discuss personalism in modern contexts (and on whose work Linz [1975 and 2000] largely bases his own account) was Guenther Roth (1968). In Roths understanding, personalism or patrimonialism refer to a typology of beliefs and organizational practices that can be found at any point of [] a continuum [of political regimes, the authors] (Roth 1968, 197), and thus does not describe a specific type of political rule. Eisenstadt (1973), and following him scholars such as Jean-Franois Mdard (1982), Robert H. Jackson and Carl G. Rosberg (1981), and Peter Pawelka (1985) by contrast, describe what they refer to as modern neo-patrimonial or personalist rule as a specific form of (modern) authoritarian regimes that is highly centralistic in the sense that access to power and resources is concentrated at the center in the hands of political elites that are loyal to the person of the ruler. The basis of regime maintenance in such orders is the distribution of resources, rewards and access to spoils (Eisenstadt 1973; Jackson & Rosberg 1981; Mdard 1982; Pawelka 1985; Roth 1968).

In African studies, the concept of neopatrimonialism (or one of its various forms) became the orthodoxy of the 1970s and early 1980s (Erdmann & Engel 2007, 97), but the notion was also widely employed in the Middle East and North Africa (see Bill & Springborg 1994; Pawelka 1985; Springborg 1979), as well as for a number of regimes outside of these regions (see the contributions in Chehabi & Linz 1998). In most conceptions, the notion not only describes a political regime, but at the same time connotes relatively low state capacities with political control being mostly exercised indirectly via cooptation and clientelism, but with repression remaining an option of last resort (for different forms of personal rule in Africa, see Jackson & Rosberg 1981). As Mdard succinctly put it, under neopatrimonial conditions, the problem is not development, but the maintenance of order and survival. All the energy of the rulers goes into more or less successful efforts to stay in power (Mdard 1982, 163).

While most analyst of neopatrimonialism or personalism agree on the core defining features of these terms (but not on the terms themselves), there is little agreement about whether this phenomenon should be considered an independent regime type or a trait of specific regimes. Taking the first position, Michael Bratton and Nicolas van de Walle argue that while neopatrimonial practices can be found in all polities, it is the core feature of politics in Africa and in a small number of other states, including Haiti, the Philippines, and Indonesia. Thus, personal relationships are a factor at the

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

11

margins of all bureaucratic systems, but in Africa they constitute the foundation and superstructure of political institutions (Bratton & Van de Walle 1994, 459). Other analysts, however, consider personalism a trait that can be found in different forms of (democratic and nondemocratic) political regimes and hesitate to conceptualize it as an independent regime type (Hadenius & Teorell 2007; Lemarchand & Legg 1972; Roth 1968).

The debate on personalist rule is certainly farthest removed from the focus on socio-economic conditions characteristic of first-wave scholarship, although the prevalence of neopatrimonial structures is linked with economic underdevelopment. The informal political processes on which this perspective focuses, however, also play an important role in the second-wave debates on the so called gray zone. Overall, the first-wave literature on authoritarian rule strove to understand the origins of specific forms of authoritarian rule in terms of the socio economic conditions and constellations at the start of the modernization process as in the case of personalist rule, or in the nature and consequences of the process itself, as in the case of military and single party rule. Thus the focus of these studies was much broader, and included economic as well as social factors, with more narrowly political factors such as regime types being mainly thought of as dependent variables. However, while the focus of these studies was on a broader array of factors within the individual countries, there was hardly any comparative effort to delineate differences not just between countries, but rather between different types of authoritarian rule. In terms of the question of how to properly classify authoritarian regimes, efforts largely proceeded inductively, implicitly based on the theoretical question of who holds power, but without attempts to outline the general features of authoritarian regimes and then classify its subtypes along generally identified dimensions.

Post-Democratization Debates and the Second Wave of Authoritarianism Studies

Several developments came together by the end of the 1980s and gave rise to what we call the second wave of authoritarianism studies. Empirically, in the wake of the third wave of democratization (Huntington 1991), many nations seemed to take the road of democratization, but frequently developed into something that might at best be called incomplete democracies, rather than full-fledged liberal democratic regimes. However flawed these new democracies were, the global political changes of the third wave resulted in renewed interest if not in the autocracies from which they resulted, then at least in the conditions under and processes by which they embarked on democratization. What Thomas Carothers (2002) called the transition paradigm became the dominant theoretical lens in the study of authoritarian rule. Under the impression of the successful democratization processes in Southern Europe and Latin America, the conceptual tools developed in this context (see especially the seminal work by ODonnell & Schmitter 1986) were applied on a global scale and produced numerous studies on the progress of and obstacles to democratization or political liberalization in other countries or regions (see for example Bratton & van de Walle 1994 and 1997 on Africa, or Brynen, Korany & Noble 1995 on the Middle East).

FarkMac

FarkMac

12

Despite the initially positive outlook, it soon became clear that democratization was not on the agenda in large parts of the world and that the End of History (Fukuyama 1992) was thus not forthcoming. This realization spawned a series of new debates on the conceptual level. Scholars began to develop new classificatory tools to deal with the allegedly novel (or hybrid) nature of a number of post-third wave regimes, ranging from so called adjective democracies to hybrid regimes and new authoritarianisms (see Collier & Levitsky 1997; Diamond 2002; Levitsky & Way 2002 and 2010; Schedler 2002 and 2006).

Theoretically, the gradual demise of modernization theorys grand theorizing which sought to identify general pathways from tradition to modernity largely independent of specific historical contexts ushered in a phase in which the so called new institutionalisms put an explicit focus on the way in which political institutions shaped actors behavior and thus political outcomes. Thus, while in the immediate tail waters of modernization theory authoritarianism studies had emerged as a more distinct field of study within Comparative Politics, due to the specific logic of the approach, area study approaches dominated the subfield. This was to change with the arrival of the second wave of authoritarianism studies which again sought to identify the mechanisms in which transitions from authoritarian rule proceeded via the mediating factors of mainly formal political institutions within more broadly comparative cross-national research designs often based on quantitative data.

These changes must again be understood against the background of general developments in the discipline. By the late 1970s and more pronouncedly from the 1980s onwards, the behavioral focus on explaining macro level outcomes as aggregates of individual level choices and modernization theorys focus on grand theorizing were challenged on theoretical grounds. Whereas modernization theory sought to find similarities in the transition processes of developing countries on their way to modernity, other approaches started to explicitly focus on differences and tried to trace them back to the institutional design of different polities (March & Olsen 1984; Hall 1996, 936). Hence, a number of approaches subsumed under the headline of the new instiututionalisms developed and would set the tone from there on: [N]ew institutionalists moved away from concepts (like modernity and tradition) that tended to homogenize whole classes of nations, toward concepts that could capture diversity among them. [] These new institutionalists shared the behavioralists concern for building theory. However, by focusing on intermediate institutions, they sought to explain systematic differences across countries that previous theories had obscured (Thelen and Steinmo 1992, 6).

The focus on institutional factors combined with an emphasis on cross-national comparison on the basis of quantitative indicators would thus come to be one of the defining features of second wave studies on authoritarianism. Whereas the first wave literature mostly sidestepped the question of regime classification, the cross-national comparative focus of second wave studies presupposed the construction of typological systems to capture relevant difference. Again, efforts at delineating regime types proceeded inductively by adding new types more or less ad hoc which, instead of being part of a systematic effort to outline the space of political regimes, blurred

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

FarkMac

13

the line between the different types to accommodate empirical cases that were deemed to not fit either category. This process was largely a reaction to the political changes brought about by the third wave of democratization and the end of the Cold War.

From Adjective Democracies to New Authoritarianisms: The Gray Zone Debate

The first reaction to the Eddies in the Third Wave (Eisenstadt 2000) that became apparent in the second half of the 1990s in this regard was the development of so called adjective democracies (see especially Collier & Levitsky 1997; Collier & Mahon 1993). The debate on adjective democracies is part of the transition and consolidation debates and grew out of the empirical observation that some regimes, even though they might have acquired the form of democracies, continued to lack its substance (see Merkel 2004; Merkel & Croissant 2000; ODonnell 1994 and 1996; Zakaria 1997). As David Collier and Steven Levitsky observed in their seminal 1997 article, this empirical phenomenon led to the proliferation of diminished subtypes8 of democracy in the literature. Concepts such as illiberal democracy (Zakaria 1997), defective democracy (Merkel 2004), or delegative democracy (ODonnell 1994), all have one fundamental point in common in that they serve to highlight a specific regimes democratic deficits by adding a negative adjective that signals in which area the respective regime fails to reach democratic standards: The characteristic feature of Guillermo ODonnells delegative democracy (1994), for example, is that the formal institutional system is counteracted by powerful informal particularistic and patrimonial norms that severely weaken horizontal accountability; Wolfgang Merkel and Aurel Croissant (2000), in turn, argue that democratic deficits in defective democracies in general are due to the existence of informal institutions alongside a formal institutional system and differentiate between different types of defects according to the partial regime affected. In general, the characteristic feature of adjective democracies is the existence of formally democratic institutions that are prevented from working properly by informal institutions and processes. In the context of adjective democracies, moreover, these democratic deficits tend to be interpreted as consolidation challenges, rather than permanent features of alternative regime types, thus establishing an implicit expectation that these regimes will eventually develop into complete liberal democracies.

Partly in opposition to this teleological bias, the discussion on hybrid regimes conceptualized regimes in the gray zone between democracy and authoritarianism as mixed regimes that combine elements of both fundamental types, rather than being simply on the way towards consolidated democracy (Karl 1995; Bogaards 2009; Diamond 2002; Levitsky & Way 2002; Ottaway 2003; Rb 2002; Zinecker 2004).

8 We do not discuss the methodological problems of diminished subtypes and radial categories here (for these

notions, see Collier & Mahon 1993; Lakoff 1987). We argued elsewhere (Koehler & Warkotsch 2010, also see the appendix to this paper) that the concept of diminished subtypes and the underlying notion of radial categories suffer from the so called problem of wide-open texture (Andersen 2000). In less abstract terms, since radial categories by definition lack a core of defining features, their extension cannot be defined sharply.

FarkMac

14

Although the insistence of many theorists of hybrid regimes that such forms of political rule are potentially stable rather than yet-to-be-consolidated democracies represents an important step, there are also some interesting similarities between this debate and the debate on adjective democracies. The most important of these similarities in the given context is the role of formal institutions. Steven Levitsky and Lucan Way (2002, 52) for example, initially defined their concept of competitive authoritarianism as a hybrid regime in which formal democratic institutions are widely viewed as the principal means of obtaining and exercising political authority, although [i]ncumbents violate those rules so often and to such an extent [] that the regime fails to meet conventional minimum standards for democracy. Thus, what distinguishes hybrid regimes from adjective democracies (if anything) is the extent of violations of formal democratic rules, rather than any qualitative differences.9 In other words, it is the extent to which the existing rules are observed by the actors (a question of regime consolidation), rather than the rules themselves (a question of regime type) that distinguishes hybrid regimes from adjective democracies.10

The last step in the classification debate has been taken by theorists of so called new authoritarianisms, with the electoral authoritarian (Schedler 2002 and 2006) type being the most widely used variant. Electoral authoritarian regimes, to use Andreas Schedlers (2002) characterization, are regimes in which opposition parties lose elections.11 The difference between electoral authoritarian and hybrid regimes thus again mainly lies in the extent to which the formal political arena is controlled by the authoritarian incumbents and reflects Barbara Geddes (2005, 6) warning that most authoritarian governments that hold elections are not hybrids but simply successful, well institutionalized authoritarian regimes. In contrast to both, adjective democracies and hybrid regimes, new authoritarianisms are clearly located within the classical three fold typology of political regimes (Linz 1975). They constitute subtypes of authoritarian regimes that allow for some degree of political participation, without, however, crossing the threshold to meaningful political contestation. In a way, the emergence of new authoritarianisms as the (as of yet) last step of the classification debate indicates that analysts have almost come full circle in their conceptual understanding of the gray zone phenomenon. Whereas early contributions emphasized the democratic side and the dynamism inherent in gray zone regimes, more recent work suggests that we might be looking at stable, nondemocratic regimes that are not altogether different from classical forms of authoritarianism.

9 On a more abstract level, another major difference is that in contrast to adjective democracies, hybrid regimes

can be defined clearly in the framework of classical categories (Sartori 1970). In other words, whereas it is impossible to state unambiguously where the boundary between adjective democracies and non-democracy lies, it is possible to define hybrid regimes by a combination of necessary and sufficient features (see Rb 2002 for one of the very few clear definitions).

10 If we follow a classical definition of political regimes (Easton 1965; Munck 1996), then this in the last analysis amounts to saying that on the level of regime types, there are no differences between hybrid regimes and adjective democracies because the defining rules are the same. The difference would rather be on the level of regime consolidation (ODonnell 1996).

11 Of course, Schedlers characterization of EA regimes was inspired by Adam Przeworskis (1991) famous definition of democracy.

15

The three conceptual perspectives outlined above cover different parts of an underlying continuum in terms of the degree to which formal, democratic rules effectively structure political processes. Whereas in adjective democracies, formal institutions provide the main rules of the game but are circumvented by important actors in specific fields (such as the rule of law in illiberal democracies or horizontal accountability in delegative democracies), the same rules are violated systematically in different variants of hybrid regimes without, however, completely eliminating formal political competition; in different types of new authoritarianism, finally, the formal rules are violated to such an extent as to preclude effective contestation for power through formal channels.

There are several ways to critically engage with this second wave literature. One of them is empirical, showing how existing authoritarian regimes deviate from the theoretical expectations expressed in such conceptual systems. This route has been taken by scholars working in the context of authoritarian institutionalism (see Gandhi 2008; Geddes 1999 and 2003; Lust-Okar 2005; Magaloni 2006 and 2008). The main conclusion from these debates is that the effects different authoritarian institutions should be understood in careful empirical analyses, rather than conceptually presupposed. The second way is more fundamental in that it addresses logical problems created by the attempt to conceptualize regime types by relying on continuous scales, rather than discrete criteria.

In the following pages, we will illustrate both paths. We start by showing that the idea of typological systems mainly differentiated by the degree of democraticness produces logically unsound categories because concept formation presupposes the establishment of clear thresholds. We then discuss empirical results on the functioning of authoritarian institutions that underline the necessity to put institutional factors into perspective and to take variation in institutional strength and effectiveness seriously (Levitsky & Murillo 2009). In the conclusion we take up these arguments and illustrate how a perspective linking the second-wave emphasis on institutional form with the first-wave interest in the social and economic underpinnings of authoritarian rule can help us to understand the dynamics of the Arab Spring and thus provides a fruitful avenue of conceptual development.

Concept Formation, Institutions, and the Continuum Problem

In a nutshell, in the following section we will show how the fact that recent conceptualization strategies relied on the idea of an underlying continuum of political regimes not only produces empirically doubtful results that are difficult to operationalize, but is logically inconsistent with the notion of regime typologies containing qualitatively different regimes. We call this problem the continuum problem. Our main conclusions are that if we want to work with empirically useful and logically sound regime typologies, we should (1) give up the idea of an underlying regime continuum that makes certain nondemocratic regimes more democratic than others (note that already this statement is a contradiction in terms), and that we should

FarkMac

16

(2) go beyond the exclusive focus on formal institutional features for classificatory purposes.

Before we can develop this argument in more detail, we have to briefly take a step back and ask a more fundamental question: Why do we need concepts in the first place? What is their function in the research process? Although it is certainly true that the issue of concept formation in social science research has received relatively little attention especially when compared to the vast literature on indicators and measurement (Goertz 2006, 2; also see Gerring 1999, 358), there nevertheless seems to be a consensus that answering the what-is question (Mair 2008, 179) necessarily has to be the first step in any (social) scientific endeavor (Gerring 1999; Goertz 2006; Mair 2008; Sartori 1970 and 1991). The fundamental epistemological reason for this is that there are a potentially unlimited number of similarities and differences between any two objects (Dupr 2002, 61). Since this is the case, there is at least one perspective under which any two objects can be considered the same (Popper 1973, 376). Concept formation solves this problem. In an effort to provide conceptual containers (Sartori 1970, 1038), the process of concept formation forces us to take a position (Popper 1973, 375)12 and to single out a dimension which we consider essential in the given context, thus establishing a system of similarity and dissimilarity relations among the objects concerned. Only once we have decided under which perspective we compare two objects can we decide if they are different or the same; and only once we answered the what-is question can we approach the how-much question (Mair 2008, 179). Or, to put it in Giovanni Sartoris words, [we] cannot measure unless we know first what it is we are measuring (Sartori 1970, 1038).

This last point is especially important in our context since it is linked with a debate between proponents of dichotomous regime measures and advocates of continuous scales (see Collier & Adcock 1999).13 Without going into too many details with respect to this discussion here, it is important to note two things. First, the view that any concept could be of an inherently continuous nature (Bollen & Jackman 1989, 612) as is sometimes argued for the case of democracy misses the important point that if we conceptualize a concept as continuous, this is a theoretical choice that cannot be justified with reference to the real concept being continuous (also see Mair 2008, 185 for a critique of this idea). To see this, consider that such a claim would force us to accept the philosophically quite strong position that concepts possess some form of objective existence that is independent of the process of concept formation. As long as we consider conceptual systems to be the products of our own attempts to impose order on the world (rather than to reflect an objectively existing order), there is no way

12 Translation from German by the authors. The original quote is Diese Skizzen zeigen, da Dinge in

verschiedener Hinsicht hnlich sein knnen und da beliebige zwei Dinge, die von einem Standpunkt aus hnlich sind, von einem anderen Standpunkt aus unhlich sein knnen. Allgemein gesprochen setzt hnlichkeit und somit auch Wiederholung stets die Einnahme eines Standpunkts voraus []. (Popper 1973, 375; emphases in the original).

13 The debate about continuous vs. dichotomous regime type measures revolves around the question whether regime types, and here especially democracy, should be considered to be qualities that are either present or absent in a given regime (dichotomous), or rather occur in different degrees in any regime (continuous).

17

of maintaining that they possess any inherent qualities. From a theoretical perspective, conceptualizing a concept as continuous or dichotomous is thus a choice that can only be justified by the usefulness of the resulting conceptual containers.

Secondly, it is important to keep in mind the distinction between conceptualization on the one hand, and operationalization and measurement on the other. While the usefulness of operationalizing the features of a certain regime type in a continuous manner depends on the specific research question being asked (Adcock & Collier 1999; Brownlee 2009a; Hadenius & Teorell 2007), on the level of conceptualization the notion of continuous regime types is indeed confused (Alvarez et al. 1996, 21). Either we can order all existing regimes on a single continuous scale, or there are different regime types; to maintain that there are different regime types and that they can at the same time be ordered on such a scale is simply a contradictory statement because a continuous scale by definition does not allow for qualitative differences.14

If we look at recent conceptual discussions in the field of authoritarianism studies from such a perspective, a number of problems on different levels emerge. As we have alluded to above, the different expressions of the gray zone debate, namely adjective democracies, hybrid regimes, and new authoritarianisms all rely on the idea of an underlying regime continuum, with liberal democracy on the one end, and fully closed autocracies on the other. In between these two poles lay a number of different regime types that are conceived of as neither fully democratic, nor completely authoritarian, but exhibit qualities of both regime types to varying degrees. Whether this continuum is expressed in terms of degrees of democracy, competitiveness, civil liberties and political rights, or some other concept does not matter for the given context. The fundamental idea remains that of a continuum on which all political regimes can be projected (see Munck 2006 for an explicit version of this argument).

As Axel Hadenius and Jan Teorell (2007, 144) pointedly observe, however, [i]f the degree of competitiveness were the only dimension along which authoritarian regimes differed, we would need no regime typology. This observation is quite to the point. To see why this is true and how it is linked to the problem of concept formation, let us assume we indeed had such a scale of competitiveness and corresponding values for all regimes around the globe.15 What would we do with these data? Following the logic of the gray zone, we would probably try to specify a number of thresholds that would, for example, result in the well known spectrum of liberal democracy, electoral democracy, competitive authoritarianism, hegemonic authoritarianism, and closed authoritarianism (see Diamond 2002 for a classificatory scheme that comes very close). But how would we decide where to establish the cutoff points? In principle, there are two different possibilities: Either we rely on some kind of arbitrary process

14 The reverse of course is even more obvious: a nominal variable cannot be projected on an interval scale. 15 Some scholars tend to think of the Polity scale, the EIEC (executive index of electoral competition) and

LIEC (legistative index of electoral competition) contained in the World Banks dataset of political institutions (DPI), or even the Freedom House or Polity scores in such terms. While all of these scales certainly have their problems, our point here is more fundamental. Irrespective of which scale we use, establishing thresholds is either arbitrary (and thus empirically useless), or presupposes a conceptual dichotomy that justifies this particular threshold.

18

(e.g. random choice or statistical procedures such as mean or median cutoffs, clustering, etc.), or we have to provide some kind of theoretical justification for each cutoff point. The first method is very unlikely to produce empirically valid regime categories since it entirely relies on the (arbitrary) distribution of our observations on the scale. Or, to again quote Giovanni Sartori, [t]here is a fantastic lack of perspective in the argument that these cutoff points can be obtained via statistical processing, i.e., by letting the data themselves tell us where to draw them. For this argument applies only within the frame of conceptual mappings which have to tell us first of what reality is composed (1970, 1038; emphasis in the original).

Thus, we would do better if we chose the second method and found some theoretical argument to justify our thresholds. This second method, however, presupposes exactly what proponents of the continuum-view try to avoid: mutually exclusive concepts that can justify why certain values should fall to the left and others to the right of any cutoff point. These categories can obviously not be generated by the same empirical data but must be built and justified independently. Conceptually, the gray zone debate thus boils down to a very simple alternative: Either we want to work with regime types (and consequently have to give up the continuum-view), or we want to establish some kind of regime continuum (and consequently cannot classify regimes into meaningful categories).16

On a logical level, attempts to use the degree to which nondemocratic regimes are structured by quasi-democratic institutions that dominated the gray zone debate thus runs into considerable difficulties. At the same time, moreover, such interpretations of the effects of formal institutions under authoritarian rule are empirically implausible in the light of a current of authoritarian institutionalist (Malesky & Schuler 2010) arguments that revolved mainly about the problem of explaining regime stability.

How Do Institutions Matter? The Stability Debate

Starting in the early 2000s, scholars increasingly turned away from the idea of formal institutions as liberalizing features and started to examine the extent to which such institutions could perform distinctly authoritarian functions and thus contribute to the stability of authoritarian regimes. Two interrelated trends combined to refocus the academic debate. On the one hand, starting in the second half of the 1990s, the rational choice variant of neo-institutionalism began to be employed more explicitly as the theoretical backdrop of work on authoritarian politics (see Wintrobe 2007). Building on the work of Gordon Tullock (1987), scholars such as Robert Wintrobe (1990, 1998, and 2007), Stephen Haber (2006), or Daron Acemoglu and James Robinson (2006), developed general formal theories of authoritarian rule, mostly focusing on the various ways in which dictators cope with threats emanating from either political elites or from society. Whereas most of these models were too general in nature to be immediately applicable to empirical research, they were arguably

16 Note that this immediately follows from Sartoris famous admonition that concept formation precedes

measurement (1970).

19

influential in helping to give rise to what can now be considered a formal current within the literature on authoritarian rule that recognizes formal institutions as an important part of authoritarian regimes (see Carter 2010; Cox 2009; Gandhi 2008; Gandhi & Przeworski 2006 and 2007; Lust-Okar 2005; Magaloni 2006 and 2008; Miller 2009; Przeworski 2009).

The second factor that led to a re-appreciation of the role of formal institutions as distinctly authoritarian institutions came from a more empirically oriented perspective informed by detailed case studies. Building on earlier work in the Comparative Politics and area studies literature (see e.g. Harik 1973; Hermet, Rose and Rouqui 1978; Perlmutter 1981; Dawisha & Zartman 1988), scholars re-examined the role of imitative institutions (Albrecht & Schlumberger 2004) in the field of incumbent-opposition relations, arguing that the existence of opposition actors can have functional aspects for dictators (see Albrecht 2005, and 2010) and that dictators use formal institutions to structure their political systems through inclusion and exclusion (Lust-Okar 2004, 2005, and 2007). Others focused on formal institutions such as elections and legislatures that were interpreted as mechanisms of co-optation and the distribution of spoils (Blaydes 2010; Gandhi 2008; Koehler 2008; Lust-Okar 2006), or analyzed the role of ruling parties in stabilizing elite coalitions in authoritarian contexts (Brownlee 2007; Langston 2006; Magaloni 2006). These mainly empirically oriented studies produced important evidence for the fact that formal institutions can serve important functions under autocracy without necessarily inducing any kind of regime change.

The main conclusion emerging from this focus on authoritarian institution is succinctly summarized by Ellen Lust-Okar (2005, 1) who maintains that formal institutions matter in authoritarian regimes although [t]hey do so independently of the larger rules of the game that characterize regime types. In the meantime, the debate on authoritarian institutions and regime stability has crystallized around two main sub-debates. The first of these debates focuses on the dynamics of authoritarian elections (see Blaydes 2010; Gandhi & Lust-Okar 2009; Koehler 2008; Lindberg 2009; Lust 2009; Lust-Okar 2006; Magaloni 2006; Schedler 2006), whereas the second mainly analyzes the role of dominant or single parties in the context of authoritarian rule (Brownlee 2007; Kricheli & Magaloni 2010; Langston 2006; Magaloni 2008).

Taken together, the current conceptual state of the art in authoritarianism studies is characterized by two different research agendas that, a common focus on formal institutions notwithstanding, produce rather incompatible results. We will return to this problem in the next section. Before we discuss this issue, however, we will briefly summarize what we think are the most important components of the current state of the art.

Authoritarianism Studies: The State of the Art

One of the most striking features of the overall debate is the gap between first- and second-wave scholarship on authoritarian rule. While classical conceptions of

20

authoritarian rule were primarily interested in the socio-economic conditions leading to different forms of authoritarian regimes, more recent work tends to adopt an institutionalist focus in accordance with the general neo-institutionalist turn in Comparative Politics. At the same time, first-wave scholarship mainly analyzed the emergence of authoritarian rule, whereas the second-wave of authoritarianism studies inherited a focus on regime breakdown and stability from the democratization debate.

It is thus surprising to see that despite these striking differences, there has been little change in terms of developing our general conceptual understanding of authoritarianism as a regime type. Rather, Juan Linzs classical definition remains the only broadly accepted characterization. According to this definition, authoritarian regimes are political systems with limited, not responsible political pluralism, without an elaborate and guiding ideology, but with distinctive mentalities, without extensive nor intensive political mobilization, except at some points in their development, and in which a leader or occasionally a small group exercise power within formally ill-defined limits but actually quite predictable ones (Linz 1964, 255, cited in Linz 2000: 159). Conceptually, the state of the art in authoritarianism studies has barely moved beyond this definition.

On an empirical level, however, there is a growing sense of unease among many analysts concerning the applicability of this concept to a number gray zone regimes. Rather than explicitly rejecting the notion, however, analysts of hybrid regimes and new authoritarianisms presuppose (often implicitly) a form of full-scale authoritarianism (Levitsky & Way 2002), full authoritarian control (Schedler 2002), closed regimes (Howard & Roessler 2006), or closed autocracies (Schedler 2006) against which the more liberal variants in the gray zone can and should be contrasted.

The starting point for this contrast between fully closed variants of authoritarianism and more liberal (and thus somehow less authoritarian) regimes is the presence of ostensibly democratic institutions such as elections, parties, or legislatures. These institutions are seen as possessing an inherent democratic quality and thus as inevitably moving such regimes closer towards democracy on an imagined continuum between the two poles of fully closed authoritarian regimes (lacking any of these formal institutions) and liberal democracy. Thus, as Staffan Lindberg pointedly puts it, the underlying assumption is the more elections, the more democratic the regime and society in general (Lindberg 2009, 9).

As we argued above, this idea of inherently liberal institutions provoked yet another turn in the scholarly literature, namely the debate revolving around authoritarian institutions and regime stability. Empirically, this debate has produced a number of hypotheses about the effects of formal institutions under autocracy that are at least partly at odds with the perspective advanced under the assumptions of the gray zone debate. Thus, more institutionalized authoritarian regimes have been found to be more long-lived than less institutionalized variants (Gandhi & Przeworski 2007; Geddes 2003; Magaloni 2008), to be more successful economically (Gandhi 2008), and to experience violent leadership turnover less frequently (Cox 2009). These

21

hypotheses cannot easily be reconciled with the gray zone perspective that more institutions mean more democracy, but rather suggest that the relationship between institutionalization and authoritarian rule is not as uniform as maintained by these scholars.

As our brief review of the debate on formal institutions under authoritarianism has shown, one of the main conclusions of this current of literature is that formal institutions can be integral parts of authoritarian regimes. Whereas this conclusion should not come as a surprise in the case of such institutions as authoritarian single parties (Kricheli & Magaloni 2010; Magaloni 2006 and 2008; Smith 2005), it has also been shown to be true for less obvious cases such as opposition parties (Albrecht 2005; Dawisha & Zartman 1988), legislatures (Gandhi 2008), and elections (Koehler 2008; Lust-Okar 2006 and 2009; Magaloni 2006). If these arguments are indeed valid, they raise considerable doubts concerning one of the fundamental assumptions involved in the gray zone-debate. If the formal institutions on which analysts of hybrid regimes and new authoritarianisms focus in their efforts to determine degrees of competitiveness do not function as democratic institutions in the first place, why should we then classify nondemocratic regimes along the dimension of the degree to which these institutions produce dynamics that resemble those found in democratic systems (such as electoral results)? In other words, if it cannot be taken for granted that the presence of formal party systems, elections, or parliaments will automatically increase the degree of competitiveness, using these features as the basis for classification will be of little value. Rather, formal institutions seem to work under some circumstances, while they fail to do so under others (see for example Howard & Roessler 2006; Smith 2005). Classifying gray zone regimes along the very dimension of formal institutional competitiveness prevents us from asking under what kind of regime these formal institutional effects arise in the first place.

Authoritarianism Studies and the Arab Spring: Challenging the State of the Art

So far, we arguably dealt with the easier part of the exercise. Criticizing conceptual frameworks on the basis of logical and methodological arguments, although an important part of coming to terms with conceptual confusion, can only be the first step. Which constructive lessons can we learn from the conceptual confusion in authoritarianism studies? Is there a way of getting the concepts right? Obviously, and maybe disappointingly, we cannot offer a full fledged classificatory system that would overcome the problems we identified in the existing conceptual tools. What we will do, however, is to offer some thoughts on the direction in which such an endeavor should lead.

In order to illustrate these ideas, we confront the state of the art in authoritarianism studies reviewed in the preceding pages with empirical evidence of different regime trajectories in the Arab Spring. Our main concern is to show that an understanding of these trajectories presupposes going beyond institutionalist regime categories and hypotheses and integrating some of the issues emphasized by first wave approaches.

22

Given both the ongoing nature of regime change in the Arab world and our mainly conceptual concerns in this paper, we naturally do not aim to give an account of the Arab Spring as such, but rather want to point out how understanding these developments requires us to go beyond the existing state of the art by linking existing institutionalist notions of authoritarian regimes back to their social origins. In other words, we urge scholars to look beyond institutional effects and to take the non-institutional origins of institutions seriously as factors conditioning institutional effects (see Heydeman 1999; Levitsky & Murillo 2009; Smith 2005; Waldner 1999).

The Arab Spring saw mass mobilization in most countries of the Middle Eastern and North African (MENA) region outside the Gulf,17 yet only a handful of regimes either broke down or exhibited strong signs of instability.18 A more complete explanation for the emergence of and differences between protest movements in different cases certainly has to await both more empirical research and at least a relative stabilization of the situation. Why, for example, is it that the Egyptian uprising was concentrated in population centers such as Alexandria, Cairo, and Port Said and left rural areas (especially in Upper Egypt) relatively untouched, while protests in Syria and Tunisia spread from the periphery to the cities (Anderson 2011; Hibou 2011)? What accounts for the very different role played by political parties, trade unions, and other formal political organizations in different cases? What, if anything, can such differences tell us about likely post-Arab Spring developments?

Despite the fact that the situation is far from settled in many countries and that an attempt to answer most of these questions would thus be premature, some interesting differences in regime trajectories are nevertheless already apparent. Most importantly at the time of writing only two Arab dictators, Zine al-Abidine bin Ali of Tunisia and Husni Mubarak of Egypt, had definitely been forced out of office. In other cases, large scale mobilization led to protracted crises in which concessions, negotiations, repression, and continued mobilization interacted. This last trajectory is exemplified by Syria and Yemen, with Libya forming a somewhat special case due to foreign intervention. Lastly, there are cases of regime stability in the presence of mass protests as in Algeria, Bahrain, Jordan, and Morocco, and cases without major protest events such as Saudi Arabia and the smaller Gulf monarchies. Table 1 summarizes these differences. What we want to do here is not so much to empirically explain these differences, but to show how such an empirical explanation requires us to go beyond current institutionalist approaches focusing on the effects of institutions.

17 And even the Gulf did not remain completely untouched with mass demonstrations in Bahrain. 18 One of the best sources on the dynamics of protest in different countries is a series of reports by the

International Crisis Group, covering protests in Egypt (I), Yemen (II), Bahrain (III), Tunisia (IV), Libya (V), and Syria (VI and VII). We largely draw on these sources which we cite as ICG 2011 followed by the Roman numeral referring to the specific report.

23

Table 1: Regime Trajectories in the Arab Spring Trajectory Cases

Egypt Breakdown Tunisia (Libya) Syria Protracted Crisis Yemen Algeria Bahrain Jordan Stability despite protests

Morocco Oman Kuwait Saudi Arabia Stability without protests

United Arab Emirates

To begin with, it is immediately clear that current institutionalist regime typologies do not offer much explanatory leverage for regime trajectories in the Arab Spring beyond the apparent connection between monarchical rule and regime stability. There is no case of regime breakdown in a monarchy despite massive protests in some cases, notably Bahrain, Jordan, and Morocco, and limited mobilization even in such unlikely cases as Oman and Saudi Arabia. Algeria is the only non-monarchical case to manage mass mobilization without experiencing regime breakdown or entering into a major regime crisis. Since monarchies have been treated with relative neglect in the recent literature, there is little in the institutionalist literature we could start from by way of hypotheses on causal mechanisms, however.19 Nevertheless, the connection seems to be strong enough to warrant further research.

Secondly, turning to the different regime trajectories among non-monarchical cases, institutionalist regime typologies offer little leverage. In Gandhis typology, for example, Algeria, Egypt, Libya, Tunisia, Syria, and Yemen are all classified as military regimes based on the fact that all six presidents all held military titles (Gandhi 2008). Even if such a classification was not utterly unconvincing empirically (especially for the Libyan, Tunisian, and Yemeni cases), it does not offer us any way of accounting why supposedly similar regimes reacted differently to similar challenges. Geddes (1999 and 2003) typology fares slightly better by classifying both Egypt and Syria as

19 Geddes (1999 and 2003) excludes monarchies from her regime typology and only later versions of her dataset

(Smith 2005 and Brownlee 2007) contain monarchical cases. Gandhi (2008), by contrast, sees monarchies as a separate regime category but offers little in terms of discussing what exactly makes monarchies different (beyond the chief executives title). Richards and Waterbury (1996) and Lust-Okar and Jamal (2002) both offer interesting discussions of monarchs preference for tactics of divide and rule over strategies of central control that they attribute to the fact that monarchs draw legitimacy from sources outside the political system (religion, tradition), while presidents have to legitimate themselves within the political arena.

24

regimes combining military, single party, and personalist elements, while considering Algeria after 1997 a military regime, Tunisia a single party regime (thus, in contrast to Gandhi [2008] taking into account the marginal role of the Tunisian military), and both Libya and Yemen personalist regimes. While it remains unclear how exactly these differences are measured empirically, regime trajectories still cut across regime type categories. Finally, the third major regime typology developed by Hadenius and Teorell (2007) classifies Egypt, Tunisia, and Yemen as limited multiparty regimes, while Algeria is categorized as a military multiparty and Syria as a military one-party regime. Libya in this system falls into the other-category. Again, it is difficult to see how such classifications are arrived at empirically and the regime trajectories cut across regime types.

While it is not surprising that regime typologies can easily be criticized on the basis of individual cases, what is more striking is the degree of disagreement among the three most wide-spread classificatory systems. This fact which to our knowledge went unnoticed (or at least un-discussed) in the literature raises serious doubts as to whether quantitative findings based on different typologies can be easily be compared or even integrated.20

Beyond the rather crude instrument of regime typologies, we might turn to specific hypotheses about authoritarian institutions to account for the different regime trajectories. Two such hypotheses immediately come to mind: The rather well-corroborated idea that ruling parties contribute to regime stability (Blaydes 2010; Brownlee 2007; Geddes 1999; Levitsky & Way 2010; Magaloni 2006 and 2008; Smith 2005), and the less well-researched hypothesis that different authoritarian regimes are more or less vulnerable to mass protests (Ulfelder 2005).

The basic idea behind the hypothesis of ruling party effects is that as soon as dictators transfer some of their discretionary power to the institutional processes of a party organization, political elites have good reasons to expect that they might profit from continued loyalty to the regime. By lengthening actors time horizons, ruling parties allow elites to trade off current losses against expected benefits and thus increase their incentives to remain loyal (see Brownlee 2007; Magaloni 2006 and 2008). Jay Ulfelder

20 This is not an effect of our cases. I merged the Gandhis and Geddes datasets to compare the two typological

systems on the basis of the set of country-year observations in which they overlap (Gandhis dataset includes all dictatorships between 1946 and 2002, whereas Geddes data [in the version of Smith 2005] run from 1930 through 2000). A bivariate correlation between the country-years classified as military regimes in both datasets yields a coefficient of 0.28, indicating that there is substantial disagreement as to which regimes should be classified as military dictatorships. This is due to the fact that of the 2,475 country-year observations common to both datasets, Gandhis typology classifies 44 percent as military dictatorships, whereas Geddes coding rules produce only 7 percent. The picture changes if we collapse all regimes with military involvement from Geddess typology (military regimes, military/personalist regimes, and military/personalist/single-party regimes) into a single category. The correlation coefficient then is 0.52, and the percentages are 44 (Gandhi) and 19 percent (Geddes), respectively. The differences are not as pronounced for the monarchy type. The correlation coefficient is 0.86 and the percentage is 11 (Gandhi) and 10 percent (Geddes). The remaining regime categories (civilian dictatorship and single-party regime or personalist regime) are not directly comparable. Nevertheless, these differences suggest that the results based on these two classificatory systems cannot be easily compared or integrated because they are using very different categories despite employing partially the same terminology.

25

(2005), on the other hand, has argued that single party regimes (in Geddes sense) should be most vulnerable to mass protests because they draw legitimacy from mass support.

Both hypotheses are certainly plausible and more or less well-corroborated by cross-national quantitative evidence (see Magaloni 2008; Ulfelder 2005; Smith 2005). Neither of them, however, allows us to differentiate between regime trajectories among the non-monarchical cases Arab cases. Both cases of regime breakdown (Egypt and Tunisia) had hegemonic party organizations, as do two cases of protracted crisis (Syria and Yemen), while one case (Libya) does not have a ruling party. The presence or absence of ruling parties thus does not seem to hold much explanatory potential in the Arab Spring. Neither did party regimes proof more stable as expected by the ruling party effects hypothesis, nor did they proof generally prone to regime breakdown in the face of mass protest as expected by Ulfelders (2005) theory. Rather, some party regimes broke down while others entered protracted crises, but none remained stable.

The evidence concerning other institutionalist hypotheses is similarly mixed: The Middle East and North Africa with its history of competitive clientelist electoral politics does not accord well either with Lindbergs (2007 and 2009) theory of democratization by elections, or with Brownlees (2009a and 2009b) or Hadenius and Teorells (2007) ideas about the positive relation between the degree of competitiveness and democratization. At the same time, at least at first sight a linkage/leverage type of explanation following Levitskys and Ways excellent study of competitive authoritarianism (2010) does not seem to hold too much potential either. In short, the state of the art after the second wave of authoritarianism studies does not seem to offer conceptual tools that could help us to understand the events of the Arab Spring.

We hold that this is not coincidental, but rather the effect of two interrelated problems in institutionalist studies of authoritarian rule. The first problem has been discussed in the general institutionalist literature as the problem of functionalism (Pierson 2000) but has barely found its way into the debates of authoritarian institutionalism (but see Gandhi & Lust 2009). This problem basically refers to the fact that many institutionalist scholars of authoritarianism do not sufficiently differentiate between describing the function of institutions and explaining their existence. Rather, they tend to conclude that institutions exist because they perform specific functions, an argument that is both difficult to maintain on a logical level and historically unconvincing (Little 1991).

The second problem is what we call a formalism problem (see also Levitsky & Murillo 2009). With this we refer to the fact that institutional form has received ample attention in recent debates whereas institutional strength or effectiveness has not (see Levitsky & Way 2010; Smith 2005 for exceptions). It should certainly not come as a surprise to students of institutions that institutional strength is an important intervening variable. Despite the rather consensual nature of this proposition, however, the question of institutional strength has rarely been examined by scholars

26

of authoritarianism. In other words, we do not know under which conditions authoritarian institutions are or are not effective.

We suggest that both problems can be solved by complementing existing institutionalist accounts with what David Waldner (1999) has referred to as the non-institutional origins of institutions. This argument is anything but revolutionary. There is a reach tradition of comparative historical research examining the emergence of political regimes and other institutions that stretches back to Max Weber and Barrington Moore and comprises such more recent works as the Colliers Shaping the Political Arena (1991) and Rueschemeyers, Stephens and Stephens Capitalist Development and Democracy (1992). What these works and studies with a more limited scope (see Angrist 2004 and 2006; Heydemann 1999; Smith 2005; Waldner 1999) have in common is to understand institutions and institutional configurations such as political regimes as contingent upon the distribution of power among social actors, whether characterized as classes or in other terms. Although it is a characteristic feature of institutions to not immediately adapt to changes in such underlying conditions (see March & Olsen 1989), recent institutional scholarship in comparative authoritarianism lost sight of the social roots of institutions.

This produced the twin problems of functionalism and formalism: Since institutional configurations were taken as given, the only explanation that could be provided for their existence were the functions they performed. At the same time the form of institutional configurations took precedence over their actual reach. Since the connection between institutions and their social bases went almost completely unresearched, we just simply do