With either democratization or redistribution by the elites, the citizens may still prefer a revolution. Thus, given the actions φ and τN of the elites, the value to the citizens in the state (N, µH) is:
Combining (7.13) and (7.20), we calculate the maximum utility that can be given to the citizens without democratizing. This involves the elites setting the tax rate τN = τp when there is a threat of a revolution so that the continuation value for the citizens is Vp (N, µH, τN = τp). This value satisfies:
(7.21)
which is, of course, the same as (6.15) derived in the previous chapter. The citizens compare (7.21) to Vp (R, µH). This defines a critical value of µH:
(7.22)
such that Vp (N, µH, τN = τp) = Vp (R, µH) when µH = µ*. For 0 < µ < µ*, a revolution is so attractive for the citizens in state µt = µH that even the maximum amount of redistribution by the elites cannot stop it. Democratization is, therefore, the only option left to the elites. Also:
such that high inequality increases the revolution threshold because the citizens are worse off in a nondemocratic regime. Citizens are now willing to undertake a revolution when the cost of doing so is higher.
For µ ≥ µ*, democratization can be avoided by redistributing to the citizens in state (µH, N). In this case, the tax rate that the elites have to set to avoid a revolution is τN= , such that Vp (N, µH, τN =) = Vp (R, µH), which is decreasing in µ and increasing in θ (i.e., increasing in the level of inequality).
Having determined the conditions under which a nondemocratic regime can stay in power by making concessions and when a democracy is or is not consolidated, it remains to consider the implications of repression. Our assumptions about repression are identical as before so that the payoffs from repression are given by (6.18). Again, there are two situations to consider. If µ ≥ µ*, then a nondemocratic regime never needs to democratize, in which case repression is used in equilibrium if it is cheaper than making policy concessions. The conditions under which this is so and, indeed, the threshold level κ*at which the elites are indifferent between promising redistribution at the tax rateand repression, are identical to those derived previously. In particular, K* is again given by (6.20). If µ < µ*, then the elites cannot use concessions to stay in power and they compare the cost of repression to the cost of democracy. In the previous analysis, the cost of democracy was uniquely defined because we assumed that democracy was fully consolidated. However, this is not the case now and the cost of democracy to the elites and, therefore, the attractiveness of repression, depends on the nature of democracy.
If ϕ ≥so that democracy is fully consolidated, then the threshold at which the elites are just indifferent between repression and democratization isgiven by (6.21). If ϕ ∈ [ϕ*,), then democracy is partially consolidated and when a coup is threatened, the tax rate is cut. In this case, we can define a threshold level κ (ϕ) such that the elites are just indifferent between repressing and creating a semiconsolidated democracy. To see the formula for this, first recall that the value of repression is Vr (O, µH | κ) given by (6.19) in Chapter 6. The value of being in an unconsolidated democracy is Vr(D, ϕL), which satisfies (7.14) and (7.15). Thus, κ (ϕ) is such that Vr (O, µH | κ (ϕ)) = Vr (D, ϕL). The higher ϕ, the more costly a coup, the higher the tax in this state, and the greater the cost of creating democracy to the elites. Hence, κ (ϕ) is a strictly increasing function of ϕ because as ϕ increases, the burden of democracy increases for the elites and they are more inclined to use repression. Finally, if ϕ < ϕ*, democracy is unconsolidated and we can define a thresholdsuch that elites are just indifferent between repressing and creating an unconsolidated democracy.
We restrict attention to the area of the parameter space where democratization prevents revolution; that is, Vp (D, ϕL) ≥ Vp(R, µH). Because democracy is not necessarily an absorbing state, the value function Vp (D, ϕL) takes into account the future possibility of coups. The value to the citizens of a semiconsolidated democracy is higher than that of a democracy subject to coups, so it suffices to ensure that the value to the citizens of an unconsolidated democracy is greater than Vr (R, µH). To derive a formula for the value of a citizen of an unconsolidated democracy, we use (7.7) and (7.13) with Vp (N, µH) = Vp (D, ϕL) and Vp (ϕH) = Vp (N, µL), giving:
which are the same as (7.14) and (7.15) from the point of view of the citizens. Solving for Vp (D, ϕL), we find:
and the condition Vp (D, ϕL) ≥ Vp (R, µH) is therefore equivalent to:
which is a condition on the parameters that we simply assume holds. As with the corresponding condition in Chapter 6, this holds when democracy is sufficiently redistributive. This leads to an interesting trade-off: a highly redistributive democracy leads to political instability, but if the potential for redistribution is too limited, democratization does not prevent a revolution.
We can now establish the following result:
Proposition 7.2: There is a unique Markov perfect equilibrium {r,p} in the game G∞ (β). Let , ϕ*, κ*, , κ(ϕ), and be as defined previously. Then, in this equilibrium:
• If µ ≥ µ*, the society remains nondemocratic. When µt= µL, τN= τrand there is no redistribution. If κ < κ*, then when µt= µH, the rich use repression. If κ ≥ κ*, then when µt= µH, τN= , such that Vp(N, µH, τN= ) = Vp(R, µH).
• If µ < µ*, then:
(1) If ϕ ≥and κ ≥, we are in a fully consolidated democracy. The society switches to democracy the first time µt= µHand remains democratic thereafter, and taxes are always given by τD= τp.
(2) If ϕ* ≤ ϕ < and κ ≥ κ (ϕ), we are in a semi-consolidated democracy. The society switches to democracy the first time µt= µHand remains democratic thereafter. When ϕt = ϕL, τD = τp. When ϕt = ϕH, democracy sets the tax rate τD= < τpsuch that Vr (N, µL) - ϕ= Vr (ϕH, D, τD= ).
(3) If ϕ < ϕ* and κ ≥ , we are in an unconsolidated democracy. The society continuously switches regimes. In a nondemocratic regime, when µt= µL, the elites set τN= τr; when µt= µH, they democratize. In a democracy, when ϕt= ϕL, τD= τp; when ϕt= ϕH, there is a coup.
(4) If ϕ ≥ and κ < , or ϕ* ≤ ϕ < and κ < κ (ϕ), or if ϕ < ϕ* and κ < , when µt= µL, τN= τr, and there is no redistribution and when µt= µH, the elites use repression to stay in power.
The main message from Proposition 7.2 is that democracy again arises because the elites cannot commit to future policies while they maintain a monopoly of political power. However, once created, democracy is not necessarily consolidated. Despite the fact that rational individuals anticipate that coups against democracy may occur in the future, the creation of democracy may nevertheless stop a revolution in the same way as described in Chapter 6. This is because to mount a coup, the elites must have de facto power and whether they will have it in the future is uncertain. This being the case, the citizens value the creation of democracy, which moves de jure power in their direction even when they understand that democracy may not be permanent.
We now discuss the conditions in the proposition in more detail. In the first type of equilibrium where µ ≥ µ* a revolution is sufficiently costly that, given the amount of inequality and the value of q, the elites can avoid it by redistributing. Therefore, in state µt = µL, the elites set their preferred tax rate of zero (i.e., τN = τr = 0), whereas in state µH, if repression is sufficiently costly, they redistribute by setting the tax rate τN =, which is just enough to stop a revolution. If repression is relatively cheap, however, the elites respond to the threat of a revolution by repressing the citizens. In this equilibrium, there is never democratization and the amount of redistribution is relatively limited (or zero) if the elites choose repression. If redistribution takes place, inequality nonetheless increases the level of redistribution in this regime because the elites are forced to choose higher taxes to prevent a revolution in the state (N, µH).
Consider what happens when µ < µ*. When the society transits into state µH, the elites can no longer maintai
n their political power via redistribution and must either repress or democratize. There are four types of equilibria depending on the values of ϕ and κ. The first possibility is that ϕ ≥and κ ≥ . Democracy, once created, is fully consolidated and repression is sufficiently costly that democracy will be created even though the elites know that the citizens will always set τD = τp from then on. In this type of society, the amount of redistribution is at its highest level, there is little or no fiscal volatility, and the threat of a coup plays no role once the society becomes democratic. We interpret this case as similar to the situation in most OECD countries. It is more likely to arise when θ is low - that is, when the society is fairly equal as long as θ > µ so that the revolution constraint binds.
The second possibility is that ϕ* ≤ ϕ
The third type of equilibrium involves ϕ < ϕ* and κ ≥ so that democracy is unconsolidated: when the state moves to ϕH, a coup is relatively attractive for the elites and cannot be halted by reducing taxes. As a result, the economy fluctuates randomly between democracy and nondemocracy. More specifically, when repression is not attractive, the economy starts with the elites in power and they set τN = τr. Whenever the state moves to µH, they democratize, after which the citizens set τD = τp. But, as soon as the state goes from (D, ϕL) to (D, ϕH), the elites mount a coup, regain political power, and set τN = 0. The variability of policy is, therefore, highest in this equilibrium, and the amount of redistribution is less than in the second and third case but more than in the first case. Higher inequality increases redistribution in this regime because it increases the tax rate when there is democracy, whereas there is never any redistribution in nondemocracy. In this case, when the citizens are in power, they set the maximum tax rate, fully anticipating that redistribution will eventually come to an end as a result of a coup. This result may help to explain the existence of highly redistributive but relatively short-lived populist regimes in Latin America (e.g., see Kaufman and Stallings 1991).
The final type of equilibrium involves repression by the elites to maintain the nondemocratic regime. This arises in various circumstances if the cost of repression is sufficiently low. Because> ĸ (ϕ) >, repression is most attractive for the elites when they anticipate that they will have to create a fully consolidated democracy. It is interesting, therefore, that our analysis suggests it is more likely that an unconsolidated rather than a fully or semiconsolidated democracy will be created.
As with democratizations, coups happen only in the high state, which can be interpreted as a relatively unlikely or unusual state. In this context, one appealing interpretation is that the high state corresponds to periods of recession or economic crises. During such crises, undertaking a coup may be less costly because society is in disarray and a proportional loss of income or output due to turbulence and political instability may be less severe because output is already low. This interpretation - which suggests that regime changes, particularly coups, are more likely during recessionary periods - is in line with the broad patterns in the data. Many coups happen during recessions or during periods of economic difficulties, such as those in Brazil in 1964, Chile in 1973, and Argentina in 1976 (see the evidence in Chapter 3). The relationship between volatility and coups suggests that a possible reason for the greater success of richer societies in consolidating democracy is their economic stability (Acemoglu and Zilibotti 1997).
Four other conclusions can be drawn from this analysis. The first links inequality to regime changes. An increase in θ increases µ*, ϕ*, , κ*,, κ (ϕ), and. Thus, higher inequality makes revolutions, coups, and repression all more attractive. As in the model of Chapter 6, which assumed that democracy was always consolidated, there is an inverted-U-shaped relationship between inequality and democratization. Highly equal or highly unequal societies are unlikely to democratize. Rather, it is societies at intermediate levels of inequality in which we observe democratization. The model of this chapter predicts that having democratized, democracy is also more likely to consolidate in more equal societies. Thus, we might expect to see very equal societies, such as Singapore, remain nondemocratic. Societies with higher levels of inequality will democratize and become fully or semiconsolidated democracies, whereas societies with greater inequality may democratize but be unconsolidated. These two cases may fit the historical evolution of Britain and Argentina. Finally, a very inegalitarian society may never democratize in the first place, which fits the South African experience. Of course, these statements apply relative to other things being equal.
The second conclusion pertains to the link between inequality and redistribution. To see this, fix the cost of a coup ϕ and define θH > θL such that ϕ =(θL) and ϕ = ϕ* (θH). Moreover, suppose that µ < µ* (θH). When θ < θL, ϕ ≥ (θ), so inequality is sufficiently low that democracy is fully consolidated. Now consider an increase in inequality (i.e., an increase in θ). This increases redistribution at first as in the standard models of voting over redistribution (e.g., Meltzer and Richard 1981) because dτp/dθ > 0. However, as θ rises above θL, democracy is no longer fully consolidated but rather semiconsolidated (i.e., ϕ ∈ [ϕ*(θ),(θ)). In this case, the citizens are forced to reduce taxes from τp toin the state (D, ϕH), so overall redistribution falls. In fact, in a semiconsolidated democracy, the relationship between inequality and taxation is ambiguous. The average tax rate is τa = (1 - s)τp + s. The tax rate τp is increasing in inequality whileis decreasing. If the cost of taxation C(τ) is highly convex, then the second effect dominates and the average tax rate falls as inequality rises. Intuitively, higher inequality makes a coup more attractive for the elites so, to prevent the coup, the citizens have to reduce the tax rate substantially in the state ϕ, leading to lower redistributive taxation on average. As inequality increases further, we have θ > θH so ϕ < ϕ*, and democracy is now unconsolidated with lower overall redistribution than both in fully and semiconsolidated democracies. Therefore, there is a nonmonotonic relationship between inequality and redistribution, with societies at intermediate levels of inequality redistributing more than both very equal and very unequal societies.
The third implication of our analysis is related to fiscal volatility. The relationship between fiscal volatility and inequality is likely to be increasing. Within each regime, higher inequality leads to more variability. Moreover, higher inequality makes unconsolidated democracy, which has the highest amount of fiscal variability, more likely. This may help explain why fiscal policy has been more volatile in Latin America than in the OECD countries (Gavin and Perotti 1997).
The fourth implication of our analysis is that the costs of redistribution also have an impact on the equilibrium political system. Suppose that the cost of taxation becomes less convex, so that C(τp) is unchanged but C’(τp) decreases. Because deadweight losses from taxation are now lower, the median voter chooses a higher level of taxation. However, as τp increases, so will - (τp (- yr) - C(τp)); therefore, democracy becomes more costly to the elites and less likely to be consolidated. This implies that in societies where taxation creates less economic distortions - for example, where a large fraction of the GDP is generated from natural resources - democracies may be more difficult to consolidate. This result has an obvious parallel to the result discussed later; that is, targeted transfers also make coups more likely. These two results together imply that a more efficient or flexible fiscal system may not always be preferable once its implications for the political equilibrium are considered.
Although we do not consider them in this book, the implications of social mobility for regime transitions were investigated in the model of this chapter by Leventoglu (2003a,b), building on work by Wright (1996) and Benabou and Ok (2001). She shows that when there is social mobility-in the sense that an individual who is poor at t may be rich at date t + 1 and vice versa - and when taxation decisions are “sticky” - in the sense that the tax rate set today influences future tax rates - then the rate of social mobility has important implications for regime transitions. Consider the preferred tax rate in democracy of the poor median voter. The main result here is that a poor person who expects to be rich in the future prefers a lower rate of taxation than a poor person who expects to remain poor. Hence, the greater the extent of social mobility, the less support there is politically for high taxes and the less redistributive is democracy. As a result, democracy is more willingly conceded by the elites and more likely to be consolidated because coups are less attractive in a society with high rates of social mobility. This may help to explain why a country like the United States in the nineteenth century, which had high rates of social mobility, was able to consolidate its (white male) democracy.
Economic Origins of Dictatorship and Democracy Page 35
