Economic Origins of Dictatorship and Democracy

by Daron Acemoglu

  Although the discussion of hardliners and softliners in the political science literature has been restricted to discussions of transition to democracy, the underlying logic suggests that such a distinction ought to be important for democratic consolidation as well. There is heterogeneity among those opposed to democracy ; when splits occur among these groups, how their preferences are aggregated is crucial in determining whether democracy survives. Therefore, we follow the previous two sections in focusing on how capital intensity influences democratic consolidation in circumstances where the elites are heterogeneous. This provides some contrast to the Chapter 8 analysis in which elite heterogeneity was discussed only in the context of democratization.

  In this section, we use the same model as in the previous section but with three groups of agents, workers, landowners, and industrialists. We denote the number of industrialists by δk and landowners by δl, such that δk + δl = δ. All capital is held by industrialists and all land is held by landowners. We also continue the analysis of the previous section by assuming that there are no differential costs of a coup for landowners (i.e., ξ = 1), but there are different tax rates imposed on them by the poor workers. The political situation is again described by a similar game. The citizens first decide to set taxes on capital and land,andand they may want to offer promisesK andL, which differ from their ideal tax rates. Then, if the elites decide not to undertake a coup, there is another move by nature, capturing the commitment problem of democracy: with probability 1 — p, the citizens get to reset taxes fromK andL.

  To discuss what the elites want to do, however, we have to propose a way of aggregating the preferences of the capitalists and the landowners. As discussed before there are various ways to do this, but here we follow the model of Chapter 8, Section 6, in which we aggregated the preferences of the rich and the middle class by assuming that decisions were determined using a utilitarian social-welfare function. We make the same assumption here so that the elites are in favor of a coup if this decision maximizes the sum of utilities of the elites - landowners plus industrialists.

  The payoffs are also different now because there are three groups. If the outcome of the game is democracy, the citizens set their most preferred tax rates given by (9.19), and the payoffs to the citizens, the industrialists (capitalists), and the landowners are, respectively:

  with w, r, and ν given by (9.2), and because a democracy chooses Tr = 0, Tp given by (9.21). Here, we are assuming that capital and land are equally owned within each faction of the elites. We have simplified the notation by writing Vk( D) as the value to all industrialists and Vl( D) as the value to all landowners, and we do so with all the value functions in this section.

  If, on the other hand, there is a coup, industrialists and landowners come to power and, in this case, we assume that they jointly choose taxes and transfers. This results in no taxation but, in the process of the coup, a fraction ϕ of the capital stock and land is destroyed. Therefore, the payoffs to a worker, an industrialist, and a landowner are:

  In specifying these payoffs to a coup, we impose that in nondemocracy established after a coup, there is no taxation of the elites. It is possible that industrialists might be in favor of setting> 0 and Tr > 0, taxing landowners to redistribute to themselves. Similarly, landowners may be in favor of taxing industrialists. However, such taxation would be determined here by maximizing the same welfare function that determined whether a coup takes place, and the utilitarian form of the objective function ensures that such taxation never occurs in equilibrium.

  Finally, the expected payoffs when the poor promise redistribution at the tax rates=K and=L - taking into account that they have to adhere to this promise with probability p - are:

  Parallel to the previous analysis, we can again define a coup constraint and threshold values for ϕ such that the elites are indifferent between a coup and living in democracy. If coups are more costly than this critical level, then we are in a fully consolidated democracy. These values now depend on the balance of political power within the elites. We first define the basic coup constraints that imply that Vk(C, ϕ) > Vk(D) and Vl(C, ϕ) > Vl(D). These are, respectively:

  (9.30)

  and

  (9.31)

  If these constraints hold, then democracy must make some type of concession in the high state to avoid a coup and, therefore, is not fully consolidated.

  Therefore, we write the constraints Vk(C, ϕ) > Vk(D,= 0) and Vl(C, ϕ) > Vl(D,which show when a coup will take place even when the poor make the best possible concession they can promise. These conditions determine the conditions under which democracy is semiconsolidated. From this, we have the constraint under which industrialists prefer a coup rather than accept the best possible concession from the citizens:

  (9.32)

  and a constraint that shows the circumstances under which the best possible concession to landowners is worse for them than mounting a coup:

  (9.33)

  recall that these are evaluated at= 0 and= 0.

  To see under which circumstances a coup will take place we have to study whether a coup maximizes a utilitarian welfare function of the elites. It will do so if:

  Here, δkVk(C, ϕ) + δlVl(C, ϕ) is the sum of the utilities of industrialists and landowners when the elites mount a coup against democracy. The notation δkVk(D,= 0) + δlVl(D, = 0) is the sum of utilities when industrialists and landowners accept the best possible concession and do not mount a coup. It can be the case that Vl(C, ϕ) > Vl(D, = 0) so that landowners are in favor of a coup, whereas Vk(C, ϕ) < Vk(D,0) so that industrialists are against it. Now:

  Similarly:

  Thus, a coup occurs when democracy makes the best possible promise it can (i.e.,= 0,= 0) if:

  (9.34)

  A coup occurs when democracy makes no concessions when:

  (9.35)

  Note that (9.35) is the same equation as (9.26). Moreover, (9.34) implies exactly the same critical value for the cost of a coup ϕ* given in (9.27). Thus, the analysis of Proposition 9.5 applies in this case. For example, we can define critical levels of capital intensity, k* and, such that if k < k*, a coup will occur. For k ∈ [k*, ), democracy can survive by making concessions and is therefore semiconsolidated, whereas if k ≥, democracy is fully consolidated.

  It is interesting that situations now emerge where the interests and preferences of the elites diverge. Becauselandowners are naturally more inclined to have a coup than industrialists. However, as capital intensity increases, industrialists gain in power relative to landowners; therefore, the coup decision increasingly reflects their interests (i.e., because they lose less from democracy, they are less inclined to mount coups). Thus, we can have an interesting situation in which there is a split in the elites. Landowners want a coup but industrialists do not; the preferences of the industrialists dominate when the capital intensity becomes high enough.

  What matters here is not simply that the elites as a whole are becoming more pro-democratic as the economy develops and capital and industry become more important. In contrast and somewhat more realistically, there are divisions within the elites; the old aristocratic landowners are always more opposed to democracy because they pay a greater price and fear an even greater price in the future from democratic politics. Their attitudes are not changing very fast but with industrialization, the structure of the economy is changing, new factions of the elite are becoming more powerful, and industrialists have more to lose from coups and less to fear from democracy. As these new segments gain more power, democracy has less to fear from the elites. This result follows from the fact that as capital intensity increases, the intensity of preference of the different elite factions changes with industrialists becoming increasingly opposed to a coup and landlords less in favor. Relative intensity of preference maps into relative political power.

  8. Industrialists, Landowners, and Democracy in Practice

  How do the perspectives developed in this chapter help us understand cross-country diffe
rences in the creation and consolidation of democracy? The comparison between Latin America and Western Europe is particularly telling. When European countries such as Britain and France moved toward full democracy in the 1870s, they were primarily urban societies; when Brazil, Guatemala, and Venezuela democratized in the 1940s, they were primarily rural. In the European cases, although democracy created redistribution of income and economic and social policies that favored the poor, no radical program of asset redistribution emerged. Although European socialists certainly talked about the “socialization of the capital stock,” it was never actually proposed as a serious electoral strategy, except perhaps in the context of nationalization of industry. Yet, nationalization, at least in the British case, was often of industries that had heavy losses and whose owners were always compensated. Serious redistribution of capital took place only after communist revolutions. In Latin America, however, the newly enfranchised rural poor demanded agrarian reform-the wholesale redistribution of land (Lapp 2004). This happened consistently in Latin American democratizations except in the more urbanized countries such as Argentina and Uruguay, where politics had evolved around a rural-versus-urban cleavage.22 The response to demands for radical land redistribution in Brazil in 1964, Guatemala in 1954, Venezuela in 1948, and Chile in 1973 was a coup. Thus, the notion that industrialists, because they have less to fear from redistribution, are less anti-democratic than landowners seems consistent with the cross-country historical experience.

  The idea that industrialists and landowners may have different preferences toward democracy can also help explain the dynamics of democratization in Central America in the 1990s. In El Salvador, for instance, economic diversification took place after the 1940s with new import substitution industries in the towns and a move from coffee to cotton (Williams 1986; Paige 1997). Because cotton was more mechanized, there was significant shedding of rural labor and workers moved to towns and urban areas. The concentration of people in urban areas seems to have added considerably to the political instability of the country. Moreover, a new breed of industrialists invested in cotton and industry emerged. This new elite suffered much heavier losses from the fighting and was central to the push for compromise that began in the 1980s. Thus, the idea that repression is more costly for industrialists fits well with the Central American evidence, as does the idea that increasing political power of industrialists can lead to a split in the regime and democratization.

  Wood (2000) presents an interesting extension of these ideas to South Africa, arguing that a similar transformation took place with white landowners becoming less important relative to industrialists who benefited less from the apartheid regime (because they were hurt by the restriction stopping Africans accumulating human capital) and also lost more through repression and the international sanctions placed on South Africa.

  9. Economic Institutions

  The analysis in this chapter (and, for that matter, this entire book) took the structure of economic institutions as given. Nevertheless, it is clear that if those with political power can alter such institutions, it may have important implications for democracy. Imagine, for instance, that instead of markets being competitive, those in power could intervene and distort markets. In nondemocracy run by capital owners and landowners, the elites could intervene to reduce wages, perhaps by creating monopsonies in the labor market. In nondemocracy, this would increase the share of national income going to capital and land, reducing what accrued to labor to below 1 — θ. In such a society, democratization would not only lead to taxation policies that the elites would not like, it would also undermine their preferred economic institutions. For example, once the citizens - who get their income from supplying their labor - dominate democratic politics, they have an incentive to pass laws undermining the market power of industrialists and landowners. Indeed, they have an incentive to increase their own market power, perhaps by facilitating the formation of trade unions, introducing unemployment insurance, minimum wages, and firing costs. This would have the effect of reducing θ in democracy. Democratization in Britain in the nineteenth century led to important changes in labor-market legislation, switching bargaining power away from employers and toward workers (see Chapter 3).

  The effect of allowing labor-market and other economic institutions to be endogenized in this way is to make the elites more antidemocratic and the citizens more prodemocratic. Thus, revolution becomes more attractive because, as in our models with targeted transfers, the nondemocratic status quo becomes worse for the citizens. Simultaneously, democracy becomes worse for the elites and they will therefore be more inclined to use repression to avoid it. Clearly, once democracy has been created, the ability to manipulate economic institutions also increases the incentive of the elites to mount coups. In essence, allowing economic institutions to be endogenous generates results similar to the model with targeted transfers. It increases the stakes from any particular set of political institutions and tends to make society more conflictual and more unstable.

  Although we do not analyze models of endogenous economic institutions in this book, in reality it is an important issue. For example, in the work of Moore (1966) and his many followers, great emphasis is placed on the organization of agriculture. Moore argued that one of the forces that facilitated democracy in Britain was the fact that agriculture was highly commercialized with relatively free factor markets. As we discussed previously, predemocratic labor-market institutions in Britain certainly tried to reduce the bargaining power of workers - for example, by banning trade unions, but they were a long cry from the situation in Eastern Europe. Britain was one of the first countries in Europe to witness a collapse of feudalism, whereas in Eastern Europe it lived on until the middle of the nineteenth century. Moore contrasted the situation in Britain with the “labor-repressive” agriculture in Eastern Europe. This distinction makes sense in our framework when economic institutions are endogenous. In Britain, political elites in the nineteenth century, although they certainly anticipated changes in economic institutions, had much less to lose from democratization than the elites of Russia or Austria-Hungary.

  Moore’s discussion also suggests another connection between land-intensive societies and democratization. It is possible that labor-repressive economic institutions - and, in the extreme, slavery - are less inefficient and/or feasible in conjunction with agricultural technology. For example, this is the standard argument about why slaves were used primarily in the Southern United States before the Civil War (Fogel and Engerman 1974; Eltis 2000). Although we do not know of microfoundations for this claim, it certainly seems consistent with much evidence and would provide another link, this time via economic institutions, between capital-intensive societies and democracy - labor repression is simply less possible or attractive for industrialists.

  Although in Moore’s study, one might take nineteenth-century Britain to be relatively capital-intensive and Russia to be land-intensive, there is much variation that comes from differences in economic institutions even in land-intensive societies. These ideas can also help explain the intra-Latin American variation. Take Central America, for example. Despite being highly specialized in the same economic activities, particularly coffee, there are large differences in the paths of political development experienced by different Central American countries (Williams 1994; Paige 1997). For example, Nicaragua had one of the most pernicious personalistic dictatorships, that of the Somoza family, throughout most of the twentieth century, until it fell to the Sandinista Revolution of 1979. In Guatemala and El Salvador, such a kleptocratic regime did not emerge; instead, landed elites kept a close grip on power with the support of the military. This grip on power loosened only briefly in Guatemala in the 1940s and in El Salvador in the late 1920s. In both countries, elites took the path of repression rather than democracy; as a result, they had to fight sustained guerilla wars. These wars ended through negotiation in the 1990s, but certainly in Guatemala, the same elites still maintain considerable political po
wer. On the other hand, neighboring Costa Rica is perhaps the most democratic nation in Latin America and has been a democracy since 1948; even before then, it experienced relatively democratic and nonrepressive regimes.

  What can explain these differing outcomes? One clear factor is the absence of large landed estates in Costa Rica (Williams 1994; Gudmundson 1995; Paige 1997; Yashar 1997; Lehoucq 1998; Nugent and Robinson 2000; Wood 2000; Mahoney 2001). There, coffee was grown by smallholders and in the early and mid-nineteenth century, the state passed a series of “homestead acts” that basically gave away coffee-growing land to anyone who wanted to farm it. Conversely, in the other Central American countries, the expansion of the world economy in the late nineteenth century led not to homestead acts but to a series of large expropriations of lands by political elites and those with political connections. This led not to a smallholder society like Costa Rica but rather to the creation of large estates and higher land inequality. Most scholars see the different forms of agricultural organization, the existence of a “landed elite,” in most of Central America but its absence in Costa Rica as a key to explaining the different paths of political development in those countries.