Economic Origins of Dictatorship and Democracy

by Daron Acemoglu

  Then, in dramatic fashion, the value of the assets held by the rich recovered thirty years of losses in just five years. These data are consistent with an approach to the motivations of coups that emphasizes distributional conflict.

  Collier (1999) recently argued for the importance of social pressure from the masses as a driving force behind many of the most recent redemocratizations. She argued that

  In ... Peru, Argentina and Spain, massive labor protests destabilized authoritarianism and opened the way for the establishment of a democratically elected government. (p. 114)

  She further argued for an important role for labor activism in inducing democratic transition in Bolivia, Uruguay, and Brazil. Even in Chile, apparently a case where the military withdrew without being pushed too hard, many scholars emphasize the reemergence of civil society during the 1980s. Drake (1998) argued that the reason Pinochet accepted the results of the plebiscite that ended the military regime was that

  ... the foreseeable costs of maintaining the dictatorship probably would have included massive social and political disorder, class conflict, economic disruptions, radicalization of the left, draconian repression, escalating violence. (p. 89)

  Much evidence is, therefore, consistent with the idea that, in Latin America, democracy was forced on political elites by the threat of revolution and by the collective action of the disenfranchised. Moreover, once democracy was created, elites often wanted and were frequently able to mount coups to take back power.

  Figure 3.23. Gini Coefficient (Argentina). Source: Calvo, Torre, and Szwarcberg 2001.

  Finally, following the European discussion, it is interesting to assess what is known about long-run trends in income distribution in Latin America. Unfortunately, much less is known about Latin America than Europe. In Figure 3.23, we plot the Gini coefficient for Argentina since the passing of the Sáenz Peña Law using data from Calvo, Torre, and Szwarcberg (2001). It shows that inequality has changed little in Argentina over the last century. Although the basic trend has been flat, the fluctuations have been interesting. After democratization in 1916, inequality began to fall consistently until the coup of 1930. After this, it was flat but then fell dramatically with the election of Perón’s first government. The coup of 1955 led to a rapid increase in inequality, although this was unwound by the partially democratic regimes that assumed power after 1958. For example, Frondizi tried to court the Perónist vote by adopting pro-union policies. However, real democratization with Perón’s second government after 1973 led to a further fall in inequality, and the coup of 1976 led to a dramatic increase in inequality. Other evidence supports this general pattern. For example, the share of wages in Argentinian national income, which was estimated to be around 28 percent at the time of the introduction of universal male suffrage, increased to 42 percent during the first ten years of democratic politics. Once democracy gave way to a dictatorship, the share of wages started to fall (Diaz-Alejandro 1970; Randall 1978, p. 29). Similarly, during Perón’s first government, the share of wages in national income increased by more than 10 percent in the course of a few years, but all of this gain and more was lost during the military regimes of the 1970s (Di Tella and Dornbusch 1989). These movements are consistent with our framework. Democratization led to the incorporation of poorer groups into the polity and, consequently, resulted in policies designed to favor such groups. Many coups in Argentina were clearly motivated by a desire to reverse such policies. These intentions and political forces show up in the data on income distribution. Democracy tended to promote equality, nondemocracy tended to promote inequality. The exception to this is the rapid rise of inequality since the 1990s under the presidencies of Menem, which abandoned the traditional pro-labor policies of the Perónists.

  For Colombia, Berry and Urrutia (1976) and Londoño (1995) showed that inequality increased between 1938 and the mid-1960s and then fell monotonically thereafter until 1990. Interestingly, the period from 1948 to 1958 was one of nondemocracy. First, under the authoritarian semidemocratic Conservative regimes of Mariano Ospina Pérez and Laureano Gómez from 1948 until 1953, when opposition politicians were harassed and congress was closed, and then under the military until redemocratization in 1958.

  Other facts on long-run income distribution can be deduced from work on the relationship between real wages and real rental rates of land (O’Rourke, Taylor and Williamson 1996; Williamson 1999; Bértola 2000; Bértola and Williamson 2003). These data suggest that in most Latin American countries, inequality rose from around the 1880s until The Great Depression. These authors argue that this was due to the incorporation of these primarily land-abundant countries into the world economy as exporters of agricultural goods. Interestingly, this rise in inequality was especially pronounced in countries such as Argentina and Uruguay, which were the most involved in international trade, and these were the countries that democratized first.

  In the context of the Latin American experience, there are also many examples in which democracies have started important redistributive programs. Even in Costa Rica, a country with a relatively egalitarian history, Chalker (1995, p. 104) argued that “the most remarkable egalitarian measure in Costa Rica occurred in the 1960s and 1970s when the concentration in income distribution was reduced. Interestingly this was an outcome, rather than a cause of, democratic politics.” Engerman, Mariscal, and Sokoloff (1998) establish more generally for Latin America that, as for Europe, there is a strong historical relationship between democratization and educational expansion.

  6. The Literature

  Our analysis of the emergence, consolidation, or collapse of democracy builds on a large and somewhat heterogeneous literature in political science and sociology and a small more recent literature using formal mathematical models by economists. There is a joke in economics that any statement you make is already in Marshall’s Principles of Economics. Similarly, it is difficult to imagine making a claim about either democratization or consolidation that has not appeared somewhere in some form in the literature. For example, Huntington (1991, pp. 37-8 ) lists twenty-seven different factors that he claims have been said to promote democracy. Nevertheless, why anybody believes any particular causal claim is usually unclear, as are the causal mechanisms linking particular putative causes to outcomes. The great strength of the analytical approach we adopt is that these issues are crystal clear.

  Theorizing about the issues we tackle in this book dates back at least to Aristotle and Plato and has become the center of much academic work since the studies of Lipset (1959) and Moore (1966). In this section, we describe how our work fits into the mainstream of the existing literature and outline what we think are our major contributions. As the book proceeds, we discuss extensively how our findings relate to existing work and particular theoretical and empirical claims made in the academic literature.

  Lipset’s (1959) work, inspired by “modernization” theory, was founded on the strong empirical correlation between per capita income and democracy. He argued that democracy emerged in society as it modernized, a process associated with rising urbanization, an increased importance of industry, higher educational attainment, and the increasing “complexity” of society. The work of Moore (1966), to some extent, challenged this focus on the unambiguous implications of modernization by emphasizing three “paths to the modern world,” of which democracy was only one, the other two being fascism and communist revolution. Both scholars emphasized how underlying socioeconomic factors determined when democracy would emerge. Moore’s work and the more recent contribution of Luebbert (1991) linked subsequent political regimes to initial social conditions, such as the class structure and the organization of agriculture, and to the strength of the bourgeoisie. For example, democracy emerged in Moore’s theory when agriculture had commercialized and was no longer characterized by feudal or semifeudal labor relations, and where the bourgeoisie was strong.

  These “structural” approaches came under attack from many political scientists in the 1970s,
particularly Rustow (1970), Linz and Stepan (1978), and Linz (1978), as being too deterministic and apolitical. This criticism came with a change of focus from democratization to the collapse of democracy. The comparative project on the collapse of democracy overseen by Linz and Stepan was particularly important in reorienting the literature. They advocated (1978; p. ix) directing “systematic attention to the dynamics of the political process of breakdown.” In their view, whether democracy collapsed was not determined by socioeconomic structures or conditions but was instead a result of specific choices by the relevant actors, both pro- and anti-democratic (Linz 1978, p. 4). More specifically, Linz (1978, p. 50) proposed that democracy collapses because it loses “legitimacy,” and he argued that democracy collapses because of a failure of democratic politicians to solve political problems. Although the discussion of modernization by Lipset did not focus on choices by individuals or even groups, Moore’s (1966) analysis does incorporate choices - for example, whether the bourgeoisie enters into a coalition with the aristocracy. Nevertheless, it is not clear in his analysis what determines whether such a choice is made.

  Modernization theory was also attacked by O’Donnell (1973), who argued that the collapse of democracy in Latin America in the 1960s and 1970s undermined confidence in the income-democracy relationship and the idea that modernization promoted democracy. He pointed out that the military coups had happened in the richest Latin American countries - for example, Argentina, Uruguay, and Brazil.

  In the 1980s, following contemporary events, research again refocused, this time back on democratizations. The most influential work was the “transitions” project overseen by O‘Donnell, Schmitter, and Whitehead; their conclusions were presented in a highly influential book by O’Donnell and Schmitter (1986). They followed many of the methodological dicta of Linz and Stepan, arguing that structural explanations on democratizations were inadequate (1986, p. 4). O’Donnell and Schmitter’s book presents a framework for clarifying the relevant processes that might lead to democratization and the various types of actors involved; for example, they made an influential distinction between the “hardliners” and the “softliners” in an authoritarian regime. The book then discusses various interactions between the relevant groups and the types of situations and dilemmas that might emerge between the end of an authoritarian regime and the initiation of democracy. All research in this tradition tends to emphasize that democracy is created by the will and decisions of individuals who are barely constrained by environmental factors (di Palma 1990 is perhaps the most extreme version of such a thesis). As such, the book does not really present an explanation of when democratization occurs, although it does, offer a few generalizations, the most famous of which is as follows:

  We assert that there is no transition whose beginning is not the consequence - direct or indirect-of important divisions within the authoritarian regime itself, principally along the fluctuating cleavage between hard-liners and soft-liners. (O’Donnell and Schmitter 1986, p. 19)

  The most recent incarnation of this transitions literature and the focus of most political science research in the 1990s has been on democratic consolidation. Linz and Stepan’s (1996) work is the most central. This literature emphasizes differences in the nature of democracy and the existence of different paths from authoritarian to democratic regimes. In an early paper, Stepan (1986) proposed the existence of ten alternative paths from nondemocratic regimes to democracy. Central is the idea that the form that democracy takes, once constructed, depends on the nature of the prior regime. For example, Linz and Stepan distinguish between four types of nondemocratic regime: authoritarian, totalitarian, post-totalitarian, and sultanistic. The type of democracy that emerges typically depends, in their view, on the type of nondemocratic regime initially in place. For example, the issues facing those wishing to create consolidated democracy in North Korea (totalitarian) are very different from those faced in the Congo (sultanistic) (Linz and Stepan 1996, p. 55).

  The literature on democratic consolidation has also seen a resurgence in ideas about political culture and how this can be an important factor in determining consolidation (Almond and Verba 1963; Diamond 1999).

  Other works have attempted to integrate both structural- and actor-based approaches to democracy and its consolidation. Huntington (1991) proposed a complex web of factors that influence democratization, and he argued that these vary according to which “wave” of democracy one considered. For instance, with respect to the First wave before the First World War, he emphasized modernization, urbanization, creation of a middle class, and decreasing inequality (p. 39). In the second wave his emphasis shifted to the impact of the Second World War and the collapse of empires (p. 40). With respect to the third wave, Huntington lists five factors as being important (pp. 45-6): (1) a crisis of authoritarian legitimacy created by economic recession induced by the oil shocks of the 1970s and the international debt crisis of the 1980s; (2) the income growth and increase in education experienced in the 1960s; (3) the change in the attitude of the Catholic church; (4) the changes in the attitudes of international institutions, the United States, and the Soviet Union; and (5) the “snowballing” or demonstration effects that led to contagion and the international dissemination of democracy. Huntington’s discussion of the effects of income level on democracy differs little from Lipset’s. He argues (p. 106) that democracy in the third wave was facilitated by “higher levels of economic well-being, which led to more widespread literacy, education and urbanization, a larger middle class, and the development of values and attitudes supportive of democracy.” However, “the emergence of social, economic and external conditions favorable to democracy is never enough to produce democracy. Whatever their motives, some political leaders have to want it to happen” (p. 108).

  Thus, the structural conditions are necessary but not sufficient for democratization to occur. Huntington’s analysis of the process of democratization in many respects mirrors that of O’Donnell and Schmitter (1986). He outlines a set of stylized actors in the regime and the opposition and argues that democracy emerges when specific groups are strong or when specific sets of interactions occur (pp. 123-4, 142). Rather than provide a theory, Huntington uses this discussion to produce a taxonomy of different cases, and he focuses on three paths of democratization.

  Closer to our work is that of Dahl (1971), who proposed a simple and appealing framework for understanding democratization. He argued that the basic issue with democratization is that

  From the perspective of the incumbents who currently govern, such a transformation carries with it new possibilities of conflict as a result of which their goals (and they themselves) may be displaced by spokesmen for the newly incorporated individuals, groups or interests.

  The problem of their opponents is the mirror image of the problem of the incumbents....

  Thus the greater the conflict between government and opposition, the more likely that each will seek to deny opportunities to the other to participate effectively in policy making ... the greater the conflict between the government and its opponents, the more costly it is for each to tolerate the other. (pp. 14-15)

  Dahl’s theory of democratization is that incumbents will democratize when either the cost of tolerating the opposition falls, so that they are prepared to enfranchise them, or the costs of suppression become too high (1971, pp. 15- 16). He then makes a series of empirical claims about factors that are likely to influence these costs and, hence, the likelihood of democratization. In terms of mechanisms, Dahl emphasized that democracy arose when power was widely distributed in society, a situation he called a “pluralistic” order. It was when society became pluralistic - something induced, for example, by income growth and industrialization - that the costs of suppression became high and simultaneously, the costs of toleration became low.

  In contrast to Moore’s (1966) emphasis on the bourgeoisie and the middle classes, subsequent important and ambitious work - especially Therborn (1977) and Rueschemeyer,
Stephens, and Stephens (1992) - noticed the important role that the poor and the working class played in the democratization process. In their theory, the working classes are pro-democratic and, when they are powerful enough, they can force democracy. Power relations are determined by three sets of forces (p. 5). As Rueschemeyer, Stephens, and Stephens noted:

  There is first the balance of power among different classes and class coalitions. This ... is complemented by two other power configurations-the structure, strength, and autonomy of the state apparatus and its interrelations with civil society and the impact of transnational power relations on both the balance of class power and on state-society relations.

  The main driving force behind democratization in their theory is capitalist development which increases the power of the working classes (p. 58).

  Another important work is Haggard and Kaufman (1995), who concentrate on demonstrating the importance of economic crises for precipitating democratizations and then focus on the interaction among democratization, economic policy reform, and democratic consolidation. Their work suggests that the prime transmission mechanism between crises and democratic transitions is that crises breed social discontent against nondemocratic regimes. For instance, in their case studies,

  ... mounting economic difficulties encouraged opposition within the private sector and contributed to the mobilization of broader social and electoral movements. (p. 45)

  They also found that

  “direct action campaigns” - anti-regime protests, general strikes, and demonstrations-also figured prominently in the authoritarian withdrawals. (p. 63)

  An alternative theoretical approach to democratization stems from the sociological literature on the origins of state institutions. This argument, associated most with Tilly (1990) and applied recently to Africa by Herbst (2000), sees the origins of democracy in the process of state formation. Kings needed resources, particularly taxes, to fight wars. To induce elites to pay taxes, kings had to make concessions, one form of which was the creation of representative institutions. In this account, democracy emerges as a quid pro quo between kings and elites, in which elites are granted representation in exchange for taxes. In Africa, the lack of democracy is a consequence of the particular process of pre- and post-colonial state formation, which meant that political elites never had to make concessions to citizens in exchange for taxes to fight wars.