(3) If (1) µ < µ*, (8.24) does not hold, and δP< ½ ; or (2) µ < µ*, (8.24) does not hold, K< (τD), and δp≥ ½ ; or (3) µ < µ*, (8.24) holds, and K< (τPD); or (4) µ ≥ µ* and K < , then the rich use repression.
(4) If µ ≥ µ* and K ≥ , the rich prevent democratization by promising to redistribute by setting the tax rate τN= such that VP(N, τN= ) = Vp(R, µ).
To understand the main result in this proposition, note that when the revolution constraint binds there are several possibilities. First, the rich are unable to use concessions to maintain power (µ < µ*), but (8.24) holds and δP ≥ ½ . This implies that a partial democratization is sufficient to avoid revolution essentially because the middle class prefers more redistribution than the rich. Moreover, because δP ≥ ½, full democratization would bring the poor to power, something the rich would like to avoid if possible. In this case partial democratization occurs if K ≥ (τPD) so that repression is relatively costly. Second, full democratization arises because neither concessions nor partial democratization works (i.e., µ < µ* and (8.24) does not hold) and repression is relatively costly (i.e., K ≥(τD)). Because δp ≥ ½ and (8.19) holds, the creation of full democracy leads to a tax rate of τp, which avoids revolution. The third situation is where repression arises. This happens in four types of situations. First, concessions to the poor again do not work, (8.24) does not hold, and δP < ½ . In this case partial democracy is insufficiently redistributive to avoid revolution. Moreover, full democratization leads to the median voter being a member of the middle class, and since this would lead to a tax rate τm the failure of (8.24) to hold implies that this will also lead the poor to revolt. In this case the rich have no option but to repress if they want to avoid a revolution. Second, again neither concessions nor partial democratization can avoid a revolution, but since δP ≥ ½ and (8.19) holds full democratization does. Nevertheless, when K < (τD) repression is preferred to full democracy. Thirdly, concessions do not work but partial democracy does and K <(τPD). Here, though partial democracy would be sufficient to avoid a revolution the rich find it better to repress than enfranchise the middle class. The fourth case is where concessions work but repression is cheaper (µ ≥ µ* and K <). The final case is the familiar one where concessions do work and repression is relatively costly so that the rich maintain power by setting a tax rate sufficiently high to placate the poor.
This proposition is similar to Proposition 6.2. The main difference is that now one of the two key thresholds,(τ), depends on the size and the level of income of the middle class. It is obvious that(τ); is increasing in τ; therefore, a higher level of the tax rate in democracy, τ, makes repression more attractive for the rich. Taxes in democracy are higher when the median voter is a poor agent (i.e., δP ≥ ½ ), which corresponds to the case in which the middle class is small so that the poor are decisive in democracy, or when the median voter is a middle-class agent (i.e., δP < ½ ) but is relatively poor and likes higher taxes.
Therefore, a relatively large and affluent middle class may make democracy less costly for the rich and may act as a buffer between the poor and the rich, making repression less likely. Conversely, when the middle class is small or poor, the rich maybe more inclined to undertake repression. The caveat “may” is necessary because this need not always be the case. For instance, if θm increases, τm falls, and (8.24) becomes less likely to hold. If µ < µ*, (8.24) ceases to hold, and δp< 1/2, then the rich switch to repression because neither partial nor full democracy is redistributive enough to stop revolution.
6. Repression: Softliners versus Hardliners
The previous section discussed a model in which the rich had to choose between repression and democratization to prevent a revolutionary threat from the poor, who were until then excluded from the political system. We also presumed that the middle class, like the poor, was outside the system. Therefore, democratization gave the middle class as well as the poor political power and, in this way, the middle class played an important role in affecting the trade-off between repression and democratization. With a large and relatively rich middle class, the rich anticipated that they would not face high taxes in democracy and were more likely to democratize rather than repress.
In this section, we analyze a similar game; however, both the rich and the middle class are part of the ruling coalition and they have to decide jointly whether to promise redistribution to the poor under the existing regime, democratize, or repress. The key insight of the analysis is that the rich are always more in favor of repression than the middle class. This has a simple reason: the rich have more to lose than the middle class from redistributive taxation.
This difference between the attitudes of the rich and the middle class toward repression provides a way to formalize the often-made distinction between softliners and hardliners in dictatorships. It is argued, especially in the context of Latin American and Southern European transitions to democracy, that there is often a split within the elites controlling dictatorships: hardliners wish to use force to prolong the dictatorship; softliners try to administer a soft landing to democracy.
But who are hardliners and softliners? Elites in nondemocratic regimes are obviously heterogeneous, but what are important sources of heterogeneity? Our three-class model provides a simple answer by mapping the softliners into middle-class agents and the hardliners into rich agents. We show in Chapter 9 that there can be other splits - for example, between landowners and capitalists along the same lines - but for now our focus is with the three-class model in which the only difference is in the levels of income, not from which types of activities these incomes are being generated.
The economic model is the same as before with three groups of agents. In nondemocracy, when the rich and the middle class have different preferences, we have to propose a way to aggregate their diverging preferences, specifically with respect to the decision about whether to repress the poor. In this chapter so far when we modeled partial democracy, we considered it a situation in which the preferences of the middle class determined the policy outcome, at least if unconstrained by the threat of revolution. Here, we adopt a different approach that allows the preferences of both the rich and the middle class to matter. We assume policy decisions in nondemocracy are made according to a utilitarian social-welfare function, meaning that repression takes place if the repression decision maximizes the weighted sum of utilities of the rich and the middle class.
There are various ways in which such an approach can be justified, but it is a natural extension of our model in Chapter 4 in which the parameter χ represented the power of the elites in democracy. We argued and substantiated this argument in the appendix to the chapter that many models of democratic politics boiled down to different microfoundations for x. In general, therefore, we can think of the democratic tax rate as maximizing a weighted sum of utilities, with the median-voter model being a special case with X = 0. Analogous reasoning suggests that we can treat the intra-elite preference-aggregation problem in the same way and imagine that the repression decision was simply that which maximizes a weighted sum of utilities of the rich and the middle class. For instance, we can think of elite control as a limited type of democracy (e.g., most European and Latin American countries before the creation of universal suffrage) in which political parties compete only for the votes of members and factions of the elites. For simplicity and without affecting our main results, we proceed by assuming that the weights on the preferences of different subsets of the elites are the same so that the repression decision simply maximizes the sum of utilities of the elites.
There is again the democratization option and the feature that the promise to redistribute by the rich is imperfectly credible because the elites can reset the tax rate after the threat of a revolution has subsided with probability 1 — p.
We assume that returns from a revolution are similar to before, with the poor sharing the returns only among themselves. The return to the poor from undertaking a revolution is VP(R
, µ) = (1 - µ)/δp, with Vm(R, µ) = Vr(R, µ) = 0.
As usual, the revolution constraint is binding if the poor prefer a revolution to the existing system or if (1 — µ)The relevant revolution constraint can be written as:
(8.30)
In this section, we assume that this condition holds.
The values to the three different groups when the existing system is maintained and redistribution at the tax rate τPD is promised are given again by Vi (P D, τPD) in (8.17). Because both the rich and the middle class are part of the ruling coalition, we refer to this regime as partial democracy.
The values in full democracy depend on whether the median voter is a poor or a middle-class agent. Recall that this depends on whether δp is less than or greater than ½ . Here, we assume that δp≥ ½ , so:
As in our previous analysis and in Chapter 6, the promise of redistribution is only imperfectly credible and it prevents a revolution only if Vp (PD, τPD) ≥ Vp(R, µ). Again, we can determine a critical value, µ*, such that at µ*, we have:
To simplify the discussion, we are going to focus on the case in which µ < µ* so that the promise of redistribution is not sufficient to prevent a revolution. The choice is, therefore, between democratization and repression. The payoffs from repression are given by (8.25).
We again determine two threshold values, but now one refers to the rich and the other to the middle class, making the respective group indifferent between democratization and repression. Let these two critical values ber andm for the rich and the middle class, respectively. They are defined by:
or, more explicitly:
(8.31)
As before, the rich prefer repression to democratization when K
It is important that because θr/δr > θm/δm by the fact that the rich are richer than the middle class, we have thatm
or if:
Now, substituting for the definitions of yr and ym and dividing through by y, we find that repression is chosen if:
Using the definitions ofr andm, we also have:
wheree∈ (m,r). If κ
As before, we also need to ensure that democratization prevents revolution; the condition for this is (8.19). This analysis leads to the following proposition:
Proposition 8.4: Assume that δp≥ ½ , µ < µ*, 1 - µ > θp, and (8.19) holds so that democratization prevents revolution. Then, in the unique subgame perfect equilibrium:
• If k ≥ r, then both the rich and the middle class prefer democratization to repression, and democratization occurs as a credible commitment to future redistribution.
• IfK< m, then both the rich and the middle class prefer repression to democratization, and they use repression to prevent a revolution.
• IfK ∈ [m, r), the rich prefer repression to democratization, whereas the middle class prefers democratization to repression. If K< e, the elites use repression to avoid democratizing, whereas ifK> e, they democratize.
This proposition, especially the case in which K ∈ [m, r), captures the different attitudes of the softliners (here, the middle class) and the hardliners (here, the rich). The hardliners have more to lose from democratization and prefer to use repression even when softliners prefer a transition to democracy.
This model can be used to formalize the idea that democratizations occur when the elites “split.” To see this, consider the case in which κ < m, so that initially both factions of the elites favor repression. Now consider a situation in which K increases. For instance, the costs of repression may increase because the end of the Cold War moves the international community in a more prodemocratic manner, or democratizations in neighboring countries make repression less feasible. In this case, K can move into the region in which K E [m,r). Initially, the rich still favor repression whereas now the middle class swings in favor of democracy. Here, the elites split in the sense that different segments now prefer different policies. Nevertheless, as long as K ∈ [m, e), the preferences of the rich dominate and repression is used in equilibrium. However, if κ increases abovee, even though the rich still favor repression, the preferences of the middle class dominate and democratization occurs. At this point, the split in the elites leads to a democratization but only when the power of the middle classes is sufficiently large within the elites. In our model of preference aggregation (i.e., the utilitarian social-welfare function), as K increases, both groups become less in favor of repression, which can lead to a switch in the decision of the elites even when the rich still prefer repression.
It is also interesting that the disagreement between the rich and the middle class regarding repression becomes stronger when the middle class is relatively poor. When the middle class is relatively richer (i.e., when θm/δm is higher), they also have more to lose from redistribution in democracy and they become more favorable toward repression.
7. The Role of the Middle Class in Consolidating Democracy
In this section, we switch attention from the creation of democracy and examine how the middle class may play an important role in democratic consolidation. We show how a large and relatively rich middle class might help consolidate democracy. Intuitively, when the median voter is a middle-class agent, democracy is less redistributive and becomes even less so when the middle class becomes richer. As democracy becomes less redistributive, the rich have less to gain by changing the regime and democracy becomes more likely to survive.
Let us now return to the three-class model. The basic setup is identical to before. We assume that the median voter in a full democracy is a member of the middle class and prefers the tax rate τm > 0. This implies that the values Vi(D) satisfy (8. 10) with τ D = τm. The rich have to decide whether to mount a coup; the payoffs after a coup are:
As before, the median voter may meet the threat of a coup by promising redistribution, which is only a partially credible promise because has a chance to reset the tax with probability p once the coup threat has subsided. The values to the three different groups when there is democracy and a promise of redistribution at the tax rate, τD ≤ τm, are:
Whether a coup is attractive for the rich given the status quo depends on whether the coup constraint, Vr(C, cp) > Vr(D), binds. This coup constraint can be expressed as:
(8.32)
When this constraint does not bind, democracy is not redistributive enough or coups are sufficiently costly that the rich never find a coup profitable. In this case, we refer to democracy as fully consolidated: there is never any effective threat against the stability of democracy. It is clear that (8.32) is easier to satisfy than (7.4) because τm < τp. Because the middle class is richer, it prefers less taxation, and this makes coups less attractive to the rich. Moreover, the greater θm, the lower is τm and the cheaper a coup must be for it to be attractive to the rich.
When this constraint binds, democracy is not fully consolidated: if the middle class does not take action, there will be a coup along the equilibrium path. The action that it can take is to reduce the fiscal burden that democracy places on the rich or, in other words, reduce the tax rate. The value to the rich of the middle-class setting a tax rate ofis Vr (D, τD= ). This strategy of promising less distribution prevents the coup only if this value is greater than the return to the rich following a coup (i.e., Vr(D, τD =) ≥ Vr(C, ϕ)). In other words, democracy survives only if:
As in our analysis of the basic static consolidation game in Chapter 7, we now define a threshold value ϕ** such that when ϕ < ϕ**, the promise of limited redistribution by democracy is not suffic
ient to dissuade the rich from a coup. Of course, the most attractive promise that can be made to the rich is to stop redistribution away from them (i.e., τD = 0); therefore, we must have that at ϕ**, Vr(D, τD = 0) = Vr (C, ϕ). Solving this equality gives the threshold value ϕ** as:
(8.33)
Given this discussion, we can summarize the subgame perfect equilibrium of this game as follows:
Proposition 8.5: In the game described above, there is a unique subgame perfect equilibrium such that:
• If the coup constraint (8.32) does not bind, the coup threat is weak, democracy is fully consolidated, and the middle class sets its most preferred tax rate, τm> 0.
• If the coup constraint (8.32) binds and ϕ ≥ ϕ**, then democracy is semiconsolidated. The middle class sets a tax rate, < τm, such that Vr(D, τD =) = Vr(C,ϕ).
• If the coup constraint (8.32) binds and ϕ < ϕ**, then democracy is unconsolidated. There is a coup, the rich come to power, and set their most preferred tax rate, τD= τr.
The main insight that this model adds is that the preferred tax rate of the middle class is now crucial in the coup constraint and the definition of (8.33).
Economic Origins of Dictatorship and Democracy Page 40
