These are only the initial empirical lessons from the model. In the next three chapters, we discuss many other factors that can be important for determining whether democracy consolidates.
PART FOUR.
PUTTING THE MODELS TO WORK
8
The Role of the Middle Class
1. Introduction
In Part 4, we put the basic models of democratization and coups to work in a number of different settings. To keep things simple, we do this only in the context of our basic extensive-form static game in which the commitment problem is modeled in a reduced-form way. Our aim is to show the utility of the framework we developed for thinking about why some countries are democratic and others are not and illustrate some ways in which the framework can be enriched and extended. We also keep the analysis and statements of results more informal than in the book so far. For instance, we do not present formal specifications of strategies, although it will be straightforward to fill in the details from the analysis already presented.
In this chapter, we extend our framework by allowing for a third group with interests intermediate to those of the elites and the great mass of citizens. For simplicity, we call this group the middle class. Our model has focused on the political conflict between the elites and the citizens. Clearly, real-world societies do not correspond to our simple model with only two groups. Just as the distinctions between the elites and the citizens and the rich and the poor are useful as a tractable device for developing ideas, so is the three-class model (another application of Occam’s razor). Because many individuals see themselves as part of a “middle class” distinct from the rich and the poor, and because political scientists believe that decisive voters in democracy are often from the middle class, we believe this approach is useful.
A long tradition in social science, including Moore’s (1966) Social Origins of Dictatorship and Democracy, views third classes such as the middle class - or, what Moore called, following Marxist terminology, the bourgeoisie - as the key actor in the processes that ultimately lead to democracy. Moore suggested that only societies with a sufficiently strong bourgeoisie would become democratic, whereas societies in which landowners were strong enough that the emerging bourgeoisie had to enter into an alliance with them would turn into dictatorships. Scholars within the modernization tradition have also stressed the importance of the middle class (e.g., Lipset 1959; Dahl 1971). Huntington (1991) suggests the key role of the middle class in reducing the distributional conflicts that make democracy unstable. He argues
... economic development promotes the expansion of the middle class.... Democracy is premised ... on majority rule, and democracy is difficult in a situation of concentrated inequalities in which a large, impoverished majority confronts a small, wealthy oligarchy. (p. 66)
Similarly, the literature on Latin American dictatorships and democratic consolidation emphasizes the important role played by the middle class. We have seen this in our discussion on the role of the Radical party in the history of democracy in Argentina and also discussed the idea that the relatively large middle class in countries such as Colombia and Costa Rica may help to explain why these countries have an unusually democratic political history compared to most of their neighbors.
Motivated by these issues, in this chapter we extend our analysis of political conflict based on the citizens versus the elites by including the middle class. We show how this changes the main results from our basic approach and in what sense the middle class plays an important role in the process of the creation and consolidation of democracy. In this chapter, instead of referring to the elites and the citizens, we refer to these three groups as the rich, the middle class, and the poor. We use this language because we investigate various situations, some in which we can think of the middle class as forming a coalition with the poor against the rich and some in which the middle class forms a coalition with the rich against the poor, thus becoming part of the “elite.” In this case, the composition of the elites could include the middle class so, to avoid confusion, we drop this language from this chapter.
We start with an analysis of the emergence of partial democracy - that is, a situation in which only a limited segment of society participates in voting.21 This segment typically includes the elites and the middle class, while the poor are excluded because of income or literacy restrictions on voting. Although democracy emerged in some Latin American or African countries as a direct move from nondemocracy toward universal mass suffrage (see Chapters 1 and 3), European democracy emerged more gradually. Our three-class model enables an analysis of such incremental democratizations. In particular, when the threat of revolution from the disenfranchised comes as a result of a coalition between the middle class and the poor, it may be beneficial for the rich elites to break the coalition by extending voting rights only to the middle class. Because the middle class is, by definition, richer than the poor, it is easier to dissuade from revolution.
The next step is to analyze the move from a partial to a full democracy. Here, we distinguish between two different approaches. The first, which is often invoked when explaining the 1867 Second Reform Act in Britain, is that competition among elites (e.g., between the rich and the middle class) led to the extension of democratic rights by one of the groups to increase its likelihood of remaining in power. According to this story, Tory Prime Minister Disraeli introduced a Radical reform, enfranchising a large group of the population, to strengthen his party in its competition against the Liberals led by Gladstone. We show how this type of intra-elite competition can be modeled within this framework but then also argue that it is unlikely to provide a satisfactory explanation for either the British or other cases of transition from partial to full democracy. Instead, as in our baseline model of Chapter 6, the revolutionary threat from the disenfranchised poor appears to have been important both in the British and other cases we have studied.
Inspired by this, we use our three-class model to analyze the response of a partial democracy to a threat of a revolution from the poor. More specifically, we model the situation as one in which the rich and the middle class have voting rights and the poor challenge the system. Once again, the promises of the existing regime to redistribute in the future are not fully credible because with de jure political power in the hands of the rich and the middle class, they will revert to policies they prefer once the threat of a revolution subsides. Therefore, full democracy emerges as a way to change the future distribution of political power, thus creating a credible commitment to future pro-poor policy. The new results in this instance involve the role of the middle class: if the middle class is sufficiently poor, even partial democracy will generate policies (e.g., rates of redistributive taxation) that are close to those the poor prefer. In this case, the poor expect relatively pro-poor policies in the future even if the exact promises made to them while they have an effective threat of revolution are not kept. Here, greater inequality in the form of a relatively poor middle class may make full democracy less likely by enabling the existing regime to commit to policies relatively attractive to the poor without having to actually give the poor democratic rights.
The most interesting new results from our analysis of the role of the middle class come when we introduce the option to use repression. We learned in Chapter 6 that a highly unequal society may not democratize because with high levels of inequality, democracy would adopt policies that are radically different to those preferred by the rich. Anticipating this, the rich are willing to use repression to prevent democratization. In a model of redistributive taxation, the poor, who constitute a majority in democracy, are in favor of high rates of income redistribution. However, if the middle class is sufficiently large, the median voter in a full democracy could be a middle-class agent; moreover, if the middle class is relatively affluent, this median voter would choose only limited redistribution. Therefore, a relatively large and affluent middle class acts as a buffer between the rich and the poor and limits redistribution. By e
nsuring that policies are not too far from those preferred by the rich, it discourages the rich from using repression and makes democracy more likely.
The discussion of the middle class as a buffer focuses on a model in which, initially, the rich are in power and are considering extending voting rights to the rest of the population, which includes the middle class and the poor. An alternative scenario, relevant in many Latin American examples, is that a nondemocracy represents the interests of the rich and the middle class and has to decide whether to move to a democracy with the poor also included in the system. In this case, which we can also think of as a movement from partial to full democracy, the model provides a way to formalize the often-discussed distinction between softliners and hardliners in the political science literature. According to many qualitative accounts (e.g., O’Donnell and Schmitter 1986), when nondemocratic regimes are challenged, there are often divides between hardliners who want to use force to keep the system going and softliners who want to administer a smooth transition to democracy. Transitions to democracy take place when softliners become more influential within the existing regime. In our framework, a natural divide exists between the rich and the middle class. Because the rich have more to lose from the policies adopted by a democracy, they are more pro-repression than the middle class. For an interesting set of parameter values, there will be repression when the rich, who correspond to hardliners, are more influential within nondemocracy but peaceful transition to democracy when the middle class, the softliners, is more influential.
Finally, in Section 7 we analyze coups. We show that the role of the middle class can mirror its role in reducing the inclination of the rich to repress to avoid democratizing. If the median voter in democracy is a middle-class agent and if the middle class is relatively affluent, this mitigates the antirich impact of democracy and makes coups less attractive for the rich. The fact that the middle class can be a buffer between the rich and the poor, therefore, may help consolidate democracy as well as create it in the first place.
Overall, therefore, the analysis in this chapter reveals that the middle class plays an important role in the emergence of democracy in a number of ways: (1) it can be the driving force for democracy, especially for the emergence of partial democracy; (2) it can be in favor of the poor being included in the political arena, facilitating a move from partial to full democracy; (3) perhaps most interesting, it can act as a buffer between the rich and the poor by ensuring that democracy will not be very antirich and, therefore, dissuading the rich from using repression or mounting coups; and (4) when it is in power together with the rich, it can play the role of softliners arguing against repression and in favor of a transition to democracy, which is less costly for the middle class than for the rich.
2. The Three-Class Model
We first generalize our basic two-group model of Chapter 4 to allow for a third income group. There are three groups of agents: the rich of size δr, the middle class of size δm, and the poor of size δp. We normalize total population to 1 as before, thus Σi δi = 1, and assume that δp > δm > δr; that is, the poor are the most populous followed by the middle class, and the rich constitute the smallest group in the population. Also, we denote average income byas before, and introduce the notation that:
(8.1)
This implies that group i has a share θi of the economy’s total income and, naturally, Σi θi = 1. Moreover, we assume that:
(8.2)
so that the rich are richer than the middle class, which is in turn richer than the poor.
As before, we assume that the political system determines a nonnegative income tax rate τ ≥ 0, the proceeds of which are redistributed lump sum, and there is an aggregate cost of taxation C(τ). Nevertheless, as we showed in Chapter 4, one can easily introduce group-specific transfers without altering the thrust of the results; hence, the setup used here is only for simplicity.
Given this setup, we can define the most preferred tax rates of rich, middle-class, and poor agents. For any group, the most preferred tax rate is that which maximizesi; therefore, the most preferred tax rate of group i satisfies the following condition that we write in the Kuhn-Tucker form to allow for the possibility of a corner solution: - yi + (1 - C‘(τi))= 0 and τi > 0, or - yi+ (1 - C’(τi))≤ 0 and τi = 0. Substituting for the definitions of incomes, we write these two conditions as:
(8.3)
Because yr >by definition, we have that for the rich, (8.3) holds as an inequality and τr = 0 as before. Moreover, because> yp, the most preferred tax rate of the poor is positive (i.e., τp > 0), given by:
(8.4)
The most preferred tax rate of the middle class could be zero or positive depending on whether ym is greater or less than mean income. In most real-world income distributions, the rich are sufficiently rich that the median is less than the mean, so we assume that θm/δm < 1 or> ym. Therefore, we have that r’” is given by:
(8.5)
and τm > 0. This assumption allows us to focus on the most interesting case. However, by virtue of the fact that the middle class is richer than the poor (i.e., θm / δm> θp / δp), we also have that:
so that the middle class always prefers lower taxes than the poor.
We can apply the analysis of Chapter 4 to determine the democratic equilibrium of this model. Consider the game where two “Downsian” political parties noncooperatively offer tax policies in an attempt to win an election. There is majority voting. Because all individuals have single-peaked preferences, the MVT (Median Voter Theorem) applies and the unique equilibrium involves both parties offering the policy preferred by the median voter. The nature of the democratic political equilibrium then depends crucially on the relative sizes of the three groups. In particular, the previous assumption that δp > δm > δr immediately implies δr < ½ , so the rich are not the majority. This leaves us with two interesting cases:
1. δp < ½ , so the poor are not the majority either, and the median voter is a middle-class agent. In that case, majority voting leads to the most preferred policy of the middle class, τm.
2. δp ≥ ½ , so the poor are the majority and majority voting generates their most preferred policy, τp.
We now separately analyze these two cases.
First, suppose that δp ≥ ½ . Then, the poor are the majority and democratic politics will lead to their most preferred tax rate, τp, as given by (8.4). The comparative statics of this equilibrium are similar to those of the two-class model, but what matters now is θp, which is a measure of the gap between the poor and average income, not necessarily the gap between the poor and the rich. For example, when θp declines so that the poor become relatively poorer, their most preferred tax rate, τp, increases; in other words, dτp/dθp < 0. However, this can happen while the gap between the rich and the poor remains constant. For example, we could have a simultaneous decline in θp and θr, compensated by an increase in θm. In this case, the poor would still vote for and obtain higher taxes, but they are not poorer relative to the rich; they are simply poorer relative to average income.
This observation already shows that the relationship between inequality and the equilibrium policy now depends on exactly what measure of inequality we use. For example, a common measure in the literature is the Gini coefficient or the standard deviation of the logarithm of individual income. Now consider a change in income distribution such that the middle class becomes poor (i.e., θm falls) and the rich become richer (i.e., θr increases) without any change in θp or. In this model with δp ≥ 1/2, this has no effect on the equilibrium rate of taxation, whereas according to both measures, income inequality has increased. In fact, if ym <, according to the more rigorous and demanding definition of a mean-preserving spread (Rothschild and Stiglitz 1970, 1971), we have a more unequal distribution - one that is a mean-preserving spread of the original one, meaning that the distribution now has more weight in the tails than the original distribution. Similarly, if we used the measure of inequality that is the gap between the
rich and the poor (e.g., the often-used measure of the ratio of 90th and 10th percentiles of the income distribution), again inequality has increased, but there is no effect on the equilibrium policy chosen. Instead, this model makes a specific prediction: the equilibrium policy should depend on the gap between the poor and average incomes. However, this prediction does not necessarily map into a relationship between policy and a standard measure of inequality.
Next, consider the case in which δp < ½ so that the poor are not the absolute majority and the median voter is from the middle class. In this case, the political equilibrium is given by the tax rate that maximizes the indirect utility of a middle-class agent. The political equilibrium tax rate is given by (8.5). The comparative statics of this equilibrium tax rate are similar to those of the most preferred tax rate of the poor. In particular, we have:
Economic Origins of Dictatorship and Democracy Page 37
