Economic Origins of Dictatorship and Democracy

by Daron Acemoglu

  This research on state formation inspired an analyses of democratization by Bates and Lien (1985), Bates (1991), Rogowski (1998), and Tilly (2004). These scholars argued that democracy, like the origins of representative institutions more generally, is a concession from authoritarian rulers necessary to raise taxation. The more elastic is the tax base, the more difficult it is for authoritarian rulers to raise taxes without agreement, and the greater the likelihood of concessions - here, democracy. Hence, Bates (1991, p. 25) points out that democracy is less likely in an agrarian society-because land is easier to tax-than it is in a society dominated by physical or human capital. Moreover, he makes the argument that authoritarian rulers will be more willing to abide by democracy if they fear it less. He connects this to their economic power with respect to democracy - democrats cannot hurt previous elites if they have sufficient economic strength, perhaps because taxing the elite leads to a collapse in the economy. Rogowski (1998) similarly emphasizes the impact of the ability of citizens to exit as leading to democracy - a case in which voice prevents exit.

  Finally, our work builds on the literature that emphasizes how political institutions can solve problems of commitment. The seminal paper is by North and Weingast (1989), and this has been a theme of a series of important papers by Weingast (1997, 1998).

  7. Our Contribution

  The ideas presented in this book build on the framework we introduced in Acemoglu and Robinson (2000a,b; 2001, 2002). There, we placed the issue of regime transitions within a framework of redistributive conflict and developed the basic idea of democracy as a credible commitment by the elites to avoid revolution and derived some of the important comparative static results - for instance, the inverted U-shaped relationship between inequality and democratization. Our research provides the first systematic formal analysis of the creation and consolidation of democracy.

  Our analysis of these issues is in the tradition of formal political economy; therefore, we look for simple unified explanations of complex social phenomena. As should be clear from the previous discussion, this is somewhat out of the mainstream of political science literature on regime transitions. Instead, this literature since the 1970s has followed the dictum of Linz and Stepan (1978, p. xi) that “the historicity of macro-political processes precludes the highly abstract generalizing of ahistorical social scientific models ... applicable to all past times and any future cases.” The lack of a theoretical framework with which to analyze regime transitions is even celebrated by some scholars, with O‘Donnell and Schmitter (1986, p. 3) noting “We did not have at the beginning, nor do we have at the end of this lengthy collective endeavor, a ‘theory’ to test or to apply to the case studies and thematic essays in these volumes.” They continue,

  if we ever had the temerity to formulate a theory of such processes, it would have to be a chapter in a much larger inquiry into the problem of “undetermined” social change, of large-scale transformations which occur when there are insufficient structural or behavioral parameters to guide and predict the outcome. Such a theory would have to include elements of accident and unpredictability, of crucial decisions taken in a hurry with inadequate information, of actors facing irresolvable ethical dilemmas and ideological confusions.

  A recent survey of the literature on democratizations in the most prestigious journal in comparative politics, World Politics, noted that

  The literature on the third wave offers a number of general propositions about factors facilitating and obstructing democratization. The following are most notable:

  1. There are few preconditions for the emergence of democracy.

  2. No single factor is sufficient or necessary to the emergence of democracy.

  3. The emergence of democracy in a country is the result of a combination of causes.

  4. The causes responsible for the emergence of democracy are not the same as those promoting its consolidation.

  5. The combination of causes promoting democratic transition and consolidation varies from country to country.

  6. The combination of causes generally responsible for one wave of democratization differs from those responsible for other waves. (Shin 1994, p. 151)

  It seems that the “general propositions” are that there are no general propositions. We do not disagree that democratizations, looked at in microdetail, are tremendously complex social phenomena. Nevertheless, could not the same be said of any issue that social scientists wish to understand? To develop any systematic understanding of the social world, one must proceed by simplifying (Occam’s razor again) and abstracting from much of the details. Perhaps in this book we make the wrong decisions about which factors to emphasize and which to ignore, but whether we do can only be answered by the scientific and empirical usefulness of the theory and not by a priori assessments of how complicated the phenomena of democratization is.

  In our terms, a general proposition about democratization would be an empirical claim, derived from a model with microfoundations, about what forces tend to lead to democratization. In our theory, many factors influence this: inter-group inequality, political institutions, structure of the economy, and nature and extent of globalization, to name a few. Our theory allows us to make comparative static predictions of the form: holding other things equal, a decrease in inequality makes a highly unequal society more likely to democratize. In a particular and highly unequal society, such as South Africa in the 1980s, democratization may be caused by falling inequality. Nevertheless, this does not mean that falling inequality is necessary or sufficient to induce democratization. In another highly unequal nondemocratic society, we might see inequality fall but democratization does not occur because something else changes as well (e.g., the extent of globalization changes) that decreases the appeal of democratization.

  Despite the fact that our approach does not easily mesh with much mainstream research on regime transitions in political science, a close reading of the literature confirms that the distributional conflicts on which we focus are considered by all authors to be at the heart of understanding democratization and coups. For instance, although O‘Donnell and Schmitter (1986) emphasized that transitions coincide with splits in the authoritarian regime, they recognized that authoritarian regimes only liberalize when they are forced to (pp. 16-17). In their conclusion, O’Donnell and Schmitter noted that

  ... it is possible to offer a few generalizations.... First, all previously known transitions to political democracy have observed one fundamental restriction ... the property rights of the bourgeoisie are inviolable. (1986, pp. 68-9)

  This conclusion is not surprising given that (p. 52) “Most of the authoritarian regimes in our sample of countries have deliberately favored bourgeois interests.” At other places in their work, they also made statements highly consonant with our approach. For instance, they argued that in democratization (p. 11), “the threat of violence and even frequent protests, strikes, and demonstrations are virtually always present,” which is one of the building blocks of our approach. Interestingly, the analysis of democratic breakdown that emerges in Linz (1978) is also consistent with this basic idea (pp. 14-15, 20).

  Our framework does, however, build on various themes in the literature. At the heart of our theory are trade-offs close to those discussed by Dahl (1971), although we place the trade-off between repression and democratization into a richer setting where political institutions have a real role because of how they influence the ability of actors to commit. Moore’s (1966) work is also a clear inspiration for our research, both for the title of our book and our taxonomy of “paths of democratization” outlined in Chapter 1. We focus only on a subset of the issues that he did, however, because we do not extensively study revolutions and we completely ignore the issue of how fascism originates. Our economic focus also is different from his sociological focus, and our emphasis on methodological individualism means that we provide much more explicit microfoundations than he did. An obvious case is the connection between the strength of the
middle class and democracy. We develop various explicit mechanisms via which this strength can influence the costs and benefits of democracy for different agents and, thus, the likelihood that it will be created.

  Our work also has a similar relationship to that of scholars such as Therborn (1977) and Rueschemeyer, Stephens, and Stephens (1992). Although there are many common themes, our work is substantially different because we develop much more explicitly the mechanisms linking various factors to the rise and consolidation of democracy. Therborn and Rueschemeyer, Stephens, and Stephens see democratization as the result of capitalist development. They emphasize the effects of such development on the balance of class power more than Lipset did, but their account is based on the same empirical fact. However, they also fail to provide a mechanism for how capitalist development causes democracy. In our analysis, although capitalist development may increase the power of the poor to challenge a nondemocracy, it does not necessarily lead to democracy. For example, if such development allowed the citizens to make a permanent threat to the elite, then the elite would be able to avoid democratizing by credibly redistributing. Or, if capitalist development increased inequality sufficiently and did not make repression costly, it would encourage the elite to use repression rather than concede democratization. In our approach, it is the effect of capitalist development both on the strength of the citizens and the trade-off of the elite between repression and concession that determines the fate of democracy. Moreover, our analysis suggests that capital accumulation in itself may not be sufficient to induce democracy. Rather, it is changes in the structure of society’s assets that may be crucial to changing the costs and benefits of democracy to the elite that lead to democratization.

  These ideas point to an aspect of our work that is original and we believe important. No other authors have placed the issue of democratization in a context where the trade-offs among it, other types of concessions, and repression can be evaluated. As suggested in Chapter 2, we find a model that just says the poor want democracy and if they have more power they get it, to be too simple. Moreover, such a theory provides no real role for political institutions — surely, a critical step in explaining why and when democracy originates.

  The closest thing to this is the distinction, initially made by O‘Donnell and Schmitter (1986), between liberalization and democratization. In their schema, a process of liberalization always precedes democratization. By liberalization, they mean (p. 7) “the process of making effective certain rights that protect both individuals and social groups from arbitrary or illegal acts committed by the state or third parties. On the level of individuals these guarantees include the classical elements of the liberal: habeas corpus; sanctity of private home and correspondence; the right to be defended in a fair trial according to pre-established laws; freedom of movement, speech ... and so forth.” To the extent that such liberalization measures are valued by citizens, they constitute a concession of the type we have studied here (although obviously not a monetary one). Significantly, O’Donnell and Schmitter note that

  ... liberalization and democratization are not synonymous, although their historical relationship has been close ... without the accountability to mass publics and constituent majorities institutionalized under the latter, liberalization may prove to be easily manipulated and retracted at the convenience of those in government. (1986, p. 9)

  The research of Moore (1966), Therborn (1977), and Rueschemeyer, Stephens, and Stephens (1992) is also problematical because they assume that political conflict is always along the lines of class. There is much evidence, however, that a richer conceptual framework is needed to provide a satisfactory general approach to democracy. The framework we develop applies to a much wider set of cases.

  A long tradition from Moore (1966) and Dahl (1971) onward emphasizes that democracy is not feasible in agrarian societies. Rueschemeyer, Stephens, and Stephens (1992, p. 8) explain the reason for this in the following way: “The landed upper-class which were dependent on a large supply of cheap labor were the most consistently anti-democratic force. Democratization for them posed the possibility of losing their labor supply.” Although this mechanism is plausible, the Latin American evidence is also consistent with landed interests opposing democracy because they anticipated losing their land. It is this idea that we develop more intensively, along with related ideas about how the cost of coups is influenced by the structure of assets.

  We emphasize the fact that democracy is conceded in the face of potential conflict which is internal to a society (Therborn 1977; Rueschemeyer, Stephens, and Stephens 1992). Our reading of the historical literature suggests that the type of democratization that Collier (1999) calls an “elite project,” where political elites create democracy for other reasons without external pressure, is such a rare event that it cannot be the basis of any useful generalizations. We also do not believe that the evidence is consistent with the notion that democracy arises as a by-product of state formation and the expansion of the fiscal base, possibly induced by external threats.

  In contrast, the elite project approach, often associated with O‘Donnell and Schmitter (1986), plays down the role of outside social pressure leading to democratization and instead emphasizes conflict within ruling authoritarian regimes. O’Donnell and Schmitter’s elite splits are part of this and undeniably took place in many democratizations. Collier’s use of this term is broader because she wants to use it also to capture some nineteenth-century democratizations that purportedly took place as nascent political parties extended voting rights to increase their support (the classic example being competition between Disraeli and Gladstone over the Second Reform Act of 1867). Our basic view is that elite splits are a manifestation of heterogeneity among the elite, but they are caused in the first place by the challenge of the disenfranchised citizens to the existing system. This challenge, combined with the intra-elite heterogeneity, leads to different attitudes toward democracy. We believe that this view is consistent with a close reading of O’Donnell and Schmitter and the case-study evidence on which their analysis is based. Haggard and Kaufman (1995) also recognize that splits in authoritarian elites may be important but also argue that these splits may be initiated by crises (pp. 31-2). In terms of Disraeli and Gladstone, as Chapter 8 makes clear, we believe this to be an unconvincing interpretation of events.

  In terms of the recent literature on democratic consolidation, our work focuses on a narrower set of questions. Interestingly, in Linz and Stepan (1996), the reason for the path dependence that they argue exists in every democratization really hinges on the way that the nondemocratic legacy influences the difficulty of creating the different components of consolidation, such as the rule of law or a nonpatrimonial bureaucracy, which are outside the domain of our study, given our focus on a Schumpeterian definition of democracy. Although these questions are interesting, they are not within the scope of the questions asked in this book. What is central to our approach is the idea that the distribution of power in a democracy depends on many factors, including the structure of political institutions. If these vary, then so do the outcomes of democracy.

  Because our approach is game theoretic, we obviously build on the literature that has attempted to use such an approach. This work began following the simple games of democratization sketched in Przeworski (1991), who used them to illustrate some of the conclusions reached by O’Donnell and Schmitter (1986). His approach has been developed by a number of scholars, including Gates and Humes (1997), Crescenzi (1999), and Sutter (2000). Other scholars have appealed to simple games, particularly the prisoner’s dilemma, as metaphors for what happens when there is a coup (Cohen 1994) or a democratization (Colomer 2000).

  Our emphasis on the economic motives of actors involved in creating and undermining democracy is shared by several of the recent key works — for instance, Przeworski (1991) and Haggard and Kaufman (1995). Haggard and Kaufman’s approach is that

  we assume that the opportunities for political elites to
mobilize political support or opposition will depend on how economic policy and performance affect the income of different social groups. Both aggregate economic performance and the distributive consequences of policy are crucial to politics everywhere, affecting the chances of both incumbents and oppositions. (pp. 6 — 7)

  Nevertheless, the majority of game-theoretic models developed so far by political scientists are reduced form, generating few if any testable predictions, and failing to illuminate the causal mechanisms at work. By reduced form, we mean that the payoff to different players (e.g., from democracy or dictatorship) are represented as numbers or perhaps variables like x or y. Then, if I receive a payoff of 2 from democracy and 3 from dictatorship, I prefer dictatorship; alternatively, if x is my payoff from democracy and y my payoff from dictatorship, and x > y, then I prefer democracy. Such models do not reveal why any particular individual or group prefers the regime it does, nor do they allow one to derive testable predictions about the circumstances under which different outcomes arise. More problematical, following O‘Donnell and Schmitter (1986), they define the preferences of individuals in terms of the actions they prefer. Thus, an agent is defined to be a hardliner because he prefers dictatorship. The same problem arises in the use of these ideas in Huntington (1991). Like O’Donnell and Schmitter, he does not explain why certain interactions occurred in some countries and not others, and he does not explain why pro-democratic actors were strong in some countries but weak in others. Moreover, it is again not really explained fundamentally why anybody is for or against a particular type of political regime. Ideally, an individual’s preferences over regime outcomes ought to be derived from more fundamental preferences over income or other things, along with the implications of particular regimes for these preferences.