Economic Origins of Dictatorship and Democracy
Page 36
5. Alternative Political Identities
We now return briefly to the model of Chapter 4, in which we considered political conflict along the lines not of socioeconomic class but in terms of group X versus group Z. In the previous chapter, the basic results concerning the mechanisms leading to democracy and the circumstances under which democracy would be created were unchanged in this situation. The main difference was that some of the comparative statics, particularly with respect to inequality, were different.
This model can be extended directly to coups and the study of democratic consolidation. When group X is the majority, democratic redistribution goes from group Z to group X, with the equilibrium amount of redistribution being determined either by the preferences of the poor or rich members of group X depending on whetheris greater or less than ½ . Nondemocracy is rule by group Z and, for simplicity, we assumed that in nondemocracy the tax rate is determined by majority voting in group Z. The equilibrium tax rate is that preferred by the median voter of group Z; in Chapter 6, we considered the case where this median voter was rich. Clearly, members of group Z prefer nondemocracy to democracy, whereas the opposite is true for members of group X.
Imagine now that we are in democracy (rule by group X) but that members of group Z can mount a coup to reinstall nondemocracy - the rule of group Z. The mechanics of the models of this chapter can be applied to this situation. Imagine that the coup decision is made by majority voting within group Z so that the median voter of Z, a rich agent, will make the decision. Facing the threat of a coup, both poor and rich members of group X wish to make a concession by reducing the amount of redistribution from Z to X. Yet, such concessions are not necessarily credible for the same reasons in our analysis; hence, group Z may wish to mount a coup to recover power and induce a credible commitment to pro-Z policies. The basic mechanisms that lead to coups, therefore, are independent of the nature of political identities. Nevertheless, it is easy to construct examples in which the comparative statics with respect to measured inequality are different from those we have so far emphasized in this chapter.
6. Targeted Transfers
We now briefly discuss the implications of targeted transfers for coups and democratic consolidation. In Chapter 6, we showed that allowing for targeted transfers leads to greater political instability because it increases the stakes of the political game. Democracy is better for the citizens and worse for the elites. Simultaneously, nondemocracy is better for the elites and worse for the citizens.
In the context of coups, this implies that the presence of targeted transfers increases the desire of the elites to mount a coup and tends to make democracy less consolidated. For instance, in the context of the static model of this chapter, the introduction of targeted transfers increases the critical thresholdsand ϕ*, implying that the elites will be willing to undertake coups even when they are more costly. This follows because without a coup, the citizens tax the elites more and once the elites take power via a coup, they can tax the citizens - something they could not do before - which increases the benefit from undertaking a coup.
It is interesting that, different from the discussion of democracy in the previous chapter, now targeted transfers unambiguously increase the likelihood of coups against democracy. This is because inter-group inequality makes democracy less attractive for the elites. In contrast, the implications of inter-group inequality on democratization were ambiguous because it affected both the revolution constraint and the willingness of the elites to use repression.
Finally, the effect of targeted transfers on coups suggests that a more flexible fiscal system may be “counterproductive” because of its impact on the political equilibrium.
7. Power in Democracy and Coups
Our basic analysis implies that the origins of coups against democracy lie in the redistributive policies of democracy. An interesting question, therefore, is how alternative arrangements in democracy affect the likelihood of coups. To answer this question, we return to the static model of Section 3 and introduce our richer model of democracy, which can bestow some power to the elites. In the context of our two-class model, this gives a solution for the tax rate τ (χ), where χ is the weight of the elites. When χ = 0, we have our basic model of democracy, in which the poor agent is the median voter and chooses his or her most preferred tax rate, so τ (χ = 0) = τp. In Chapter 4, τ (χ) was implicitly defined by the first-order condition (4.16) and this implied that dτ (χ) /dχ < 0. That is, as the power of the citizens in democracy declines, so does the equilibrium tax rate and the degree to which a democracy redistributes income away from the elites.
The important implication of this model and the analysis of Chapter 6 was that as χ increases, the power of the elites in democratic politics increases, and the value they obtain in democracy is greater. So, we have d Vp (D, χ) /dχ < 0 and dVr (D, χ) /dχ > 0. Consequently, it is easy to see that the addition of variable power has important effects on the coup constraint in our basic extensive-form game of coups. Recall that the coup constraint is Vr (C, ϕ) > Vr (D, χ), or (7.4). The higher χ, the better is democracy for the elites and the less likely is it that (7.4) will bind. Hence, an increase in χ above 0 can lead an unconsolidated democracy to become semiconsolidated. Moreover, a further increase in χ can lead the society to become a fully consolidated democracy. We can also see how (7.6) depends on χ and we can derive a new critical threshold ϕ* (χ):
Because ϕ* (χ < 1) < ϕ* (χ = 1), as the power of the elites increases, it becomes less attractive to mount coups and it becomes more likely that democracy is consolidated.
Proposition 7.3: In the model with variable political power, an increase in χ makes it less likely that the coup constraint will bind and more likely that the society will have a consolidated democracy.
This result implies that the citizens in an unconsolidated democracy may wish to limit their own power and bolster that of the elites. Although this reduces their income, other things being equal, it can also remove the threat of a coup. An obvious way for the citizens to do this is to change institutions in such a way as to overrepresent the elites in democracy-give them more power than their numbers alone merit. Nevertheless, even if it is feasible for the citizens, it does not mean that they will choose to do so. In reality, whether a coup will take place or succeed if it is attempted is uncertain. Faced with such uncertainty, the citizens may not want to increase the power of the elites in democracy because it will reduce the payoff of the citizens forever, whereas the coup may fail and the threat vanish in the future. Hence, there is a trade-off in designing institutions that avoids coups. This implies that even when institutions can be designed freely to increase the power of the elites in democracy, it is not always optimal for democrats to undertake such actions; as a result, coups sometimes occur in equilibrium.
As emphasized in Chapter 6, however, many of the relevant institutions are the outcome of long historical processes and highly persistent. By their nature, institutions are difficult to change and it is unrealistic, therefore, to imagine that democrats or even nondemocrats can freely optimize over the structure of political institutions at any date. Indeed, it is interesting that examples of institutional engineering to bolster the power of the elites, such as the Zimbabwean constitution of 1980 or the negotiated settlement that ended apartheid in South Africa, happen only in the context of rather large ruptures in society. Other attempts to redesign institutions, such as the putative shift from a presidential to a parliamentary regime in Brazil after the end of the military dictatorship, typically fail.
The relationship between the institutional structure and the consolidation of democracy has also been emphasized in the political science literature. For example, Rueschemeyer, Stephens, and Stephens (1992) note that:
once democracy was installed, the party system became crucial for protecting the interests of the dominant classes and thus keeping them from pursuing authoritarian alternatives. Democracy could be consolidated only where there were
two or more strong competing political parties at least one of which effectively protected dominant class interests, or where the party system allowed for direct access of the dominant classes to the state apparatus. (p. 9)
They later note (p. 10), “democracy ... could be consolidated only if the interests of the capitalist classes were not directly threatened by it.” We have already discussed two important historical examples of the importance of the party system and the consolidation of democracy: in Argentina before the coup in 1930 and in helping to explain the long democratic history of Colombia. This is obviously an important area for future research.
The idea that has attracted the most attention in this context is that presidential democracies are more prone to coups (Linz 1978, 1994). Przeworski et al. (2000) find that the evidence supports this claim; they conclude:
it is clear that presidential democracies are less durable than parliamentary ones. This difference is not due to the wealth of the countries in which these institutions are observed, nor to their economic performance. Neither is it due to any of the political conditions under which they functioned. Presidential democracies are simply more brittle under all economic and political conditions. (p. 136)
This empirical evidence, therefore, fits well with the idea that presidential democracies are unstable because a president tends to represent the preferences of the median voter. With a parliamentary regime, there are often coalition governments and the preferences of the citizens do not necessarily find full expression in the equilibrium policy. This means that parliamentary regimes may not be so threatening to the elites. In contrast, in a presidential system, more radical policies may come onto the political agenda because they appeal to a presidential candidate trying to gain the support of a majority of the population.
8. Consolidation in a Picture
We are now in a position to rigorously derive Figure 2.2 used in the introduction. This figure shows the relationship between inequality and the cost of a coup. For simplicity in Chapter 2, we did not make a distinction between fully and semiconsolidated democracy, so as with our discussion in Chapter 6, we first build the full picture and then show how it can be simplified to derive the figures in Chapter 2. Consider Figure 7.2: on the horizontal axis is θ, on the vertical axis is ϕ. The first thing to plot is the coup constraint. We can write this now as:
(7.23)
Figure 7.2. Consolidation or Coups?
If θ = δ, so there is complete equality, then (7.23) implies that ϕ = 0. With no inequality, even when the cost of a coup is zero, the elites are indifferent between a coup and democracy. Intuitively, when there is no inequality, there is no income redistribution and thus no incentive for a coup, even if it is costless. Thus, (7.23) starts at the origin and is increasing - as inequality rises, coups become attractive and for the elites to be indifferent, the cost of a coup must be rising. One can see that as inequality rises to θ = 1, we have ϕ = δC(τp(θ = 1)) - τp(θ = 1) (1 - δ) = δ(C(1) + 1) - 1 < 1 because τp (θ = 1) = 1. To the right of this line, inequality is relatively high compared to the cost of coups and, as a consequence, coups will be attractive. To the left is the region of fully consolidated democracy. To distinguish between the situations of semiconsolidated and unconsolidated democracy, consider the function:
which shows pairs of θ and ϕ at which the elites are just indifferent between mounting a coup and accepting the promise of the best possible concession under democracy. It is immediate that this function again goes through the origin, is increasing, and when θ = 1, we have ϕ = (1 - p) (δ (C(1) + 1) - 1) < 1. In Figure 7.2, the implications of this are shown in terms of these two new regions. From Figure 7.2 it is easy to get to Figure 2.2; we just drop the function that determines the boundary line between semiconsolidated and unconsolidated democracies.
9. Defensive Coups
So far, we have focused on coups in democracy that are aimed at limiting redistribution away from the elites. Another plausible idea is that the elites support coups when they are afraid that democracy will fall to a revolution. We can think of such coups as “defensive” in the sense that those supporting coups view them as a defense against a much worse outcome for themselves: a revolution. Such a scenario may arise when a revolution against democracy is easier than a revolution against nondemocracy. We now discuss a model exhibiting these features.
To model defensive coups, consider a variant of our basic static model. Again, the citizens who have control of politics in democracy move first and decide on the tax rate. After this tax rate, the elites decide whether to undertake a coup. We now assume, however, that after the elites’ decision, the citizens may decide to undertake a revolution, which is different than before. The return from a revolution differs between the two states but also depends on whether there has been a coup. So, we denote this by µ (ζ) when the coup decision is ζ
The crucial assumption, which we view as plausible, is that:
which means that it is easier and more effective to take revolutionary action against democracy than against nondemocracy (recall that the cost of a revolution is µ). Although there could be exceptions - for example, when a nondemocratic regime is unfair and brutal, thus helping the citizens to solve their collective-action problems as a reaction to its injustices - it must typically be the case that overthrowing democracy is easier than a well-organized military regime.
How does this affect our results? We first simplify the analysis by assuming that µ (ζ = 1) → 1, so that following a coup, there is no effective revolution threat. We can now write the relevant value functions. When democracies are unconstrained, we have the values Vp(D) and Vr (D) given by (6.4). After a coup, we have (7.2) as in Section 7.3, which for the current purpose incorporates the fact that the revolution threat disappears after a coup. The values of the promise of less redistribution under democracy are identical to what they were in (7.3). In addition, we have the values from a revolution, similar to those in Chapter 6:
which condition the return from a revolution on whether there is a coup.
Here, we informally outline the results from this model. First, if µ (ζ = 0) → 1, then the basic proposition, Proposition 7.1, applies. There is no effective revolution threat against democracy, and the coup decision is taken simply by trading-off the costs of redistribution against the cost of a coup.
However, if µ (ζ = 0) < 1, then there are new results from this model. Naturally, there will be a coup whenever there was a coup before, but there might also be a coup in some additional cases. To see this, first compare:
to:
withbeing the tax rate that prevents a coup. This reduction in the tax rate is necessary because otherwise the elites will necessarily undertake a coup. However, given this reduction in the tax rate, democracy is less attractive for the citizens, and it can be the case that:
If this is the case, the elites anticipate that the citizens will undertake a revolution rather than live in this democracy, which is not very redistributive toward themselves; because a revolution is the worst outcome for them, the elites prefer to undertake a coup to prevent a revolution.
We can think of this as a “defensive coup” because the elites are not undertaking the coup to reduce redistribution but rather to prevent a revolution. Many military coups against democracy in Latin America claimed that they were there to protect the capitalist system or even democracy from a revolution - a salient case being Chile in 1973. This model shows that there might be some truth to those claims.
Nevertheless, it is interesting that there is still an important interaction between this and redistribution. We have that:
but it might still be the case that:
That is, a revolution would not have been attractive for the citizens if democracy were not trying to defend itself against a coup! Therefore, the reason why a revolution might become a threat in the first place is the fact that the coup constraint is preventing democratic politics from catering to the wishes of the citizens.
10. Conc
lusion
In this chapter, we introduced a model of coups. We showed how to integrate this theory of coups with our theory of democratization in Chapter 6. This extended model allows us to study the conditions under which democracy is not only created but also consolidated - surely a question of equal importance. Many democracies, once created, quickly collapse, so here we built a framework to understand why. We showed that many of the same issues that arose in modeling democratization arise in studying democratic consolidation. In particular, coups arise because democrats cannot credibly promise not to use their power to enact pro-citizen and anti-elite legislation and policies. To avoid this, the only solution is for the elites to take power - to mount a coup.
We showed howwhether democracy was consolidated depended on inter-group inequality, although whether this comparative static maps into a statement about observed measures of inequality depends on the nature of political identities. When political conflict is between the poor and the rich, we expect, for example, higher inequality to lead to more coups. We also showed that the power of the elites mattered for democratic consolidation. If the elites have sufficient power, they do not need to undertake coups. This suggests that there might be institutional solutions to avoiding coups, just as we argued that democracy is an institutional solution to avoid revolutions. Perhaps democrats could alter institutions and by doing so give more power in democracy to the elites. This would limit the power of democracy, but it might help consolidate it; democracy would be consolidated but limited. Nevertheless, there are dangers inherent in such a strategy, even if it is feasible. If democrats, in their desire to consolidate democracy, give the elites too much power, then the democracy that they consolidate may be so limited in its ability to transform society that it is not stable because the mass of citizens may push for a revolution and more radical social and political change.