This implies that whether transfers can be targeted, more generally the form of fiscal redistribution in society, will have important effects on equilibrium political institutions. Nevertheless, our framework does not make unambiguous predictions on whether targeted transfers make democracy more or less likely. Because they make the revolution threat stronger, they may force democratization, when temporary redistribution would have been sufficient without targeted transfers. However, because they make nondemocracy more attractive to the elites, they may also lead to repression, thereby preventing peaceful transitions to democracy.
11. Power of the Elites in Democracy
Let us now return to the class of models where we can discuss various types of democracies, giving different amounts of power to the citizens. Recall that in a fairly generic model of democratic politics, political competition in democracy between parties maximizes a weighted sum of different groups’ utilities. In the context of the two-group model, this gives an equilibrium tax rate in democracy as a function of the parameter χ, which captures the weight on the utility of the elites. We used the notation τ (χ) for this in Chapter 4 with τ (χ = 0) = τP and 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 democracy redistributes income to the citizens. From this, it follows that:
The values of revolution and repression to the elites and the citizens are not affected by this modification in the modeling of democratic politics.
To study some of the implications of this model, we return to the simple static model of Section 6. Note first that the trade-off for the elites between repression and the promise of redistribution when µ ≥ µ* is not altered by this new model of democracy. Therefore, we can concentrate on investigating the implications of χ for, the critical level of the cost of repression at which the elites are indifferent between repression and democracy. Recalling that the critical threshold for the cost of repression,(χ), which we now index by χ, is defined such that:
we have that:
(6.27)
which is similar to (6.10), except that the equilibrium tax rate resulting from political competition with variable political power, τ (χ), replaces the most preferred tax rate of the citizens, τ (χ). Notice that
by the fact that the elites have higher incomes than the citizens, and that dτ (χ) /dχ < 0. Thus, increases in χ, by making democracy less majoritarian, make repression less attractive for the elites. This implies that the ability to increase the power of the elites in democracy often enables a peaceful transition to democracy by making repression less attractive for the elites. Nevertheless, increasing χ is a double-edged sword because as the power of the elites in democracy goes up, democracy becomes less pro-citizen. It is only the fact that the welfare of the citizens is increased by democracy that makes democratization a feasible institutional change to avoid a revolution. When X becomes too high, democracy is no longer a credible commitment to pro-citizen policies; to avoid a revolution, the elites have no choice but to use repression. To summarize this discussion:
Proposition 6.4: In the model with variable power, an increase in χ starting from low values makes democracy less redistributive and makes repression less attractive for the elites. This makes democracy more likely. However, as χ increases further, (6.7) becomes less likely to hold and, therefore, it becomes less likely that democratization will stop a revolution, which induces the elites to choose repression again.
Many interesting examples suggest the importance of Proposition 6.4. For example, the inability of the elites to compete successfully in democratic politics often leads to coups. As discussed in Chapter 1, many scholars argue that the inability of the Conservatives to compete with the Radicals in Argentina after implementation of the Sáenz Peña Law appears to be one of the factors behind the coup in 1930. Traditional elites were willing to grant full democracy, partially because they thought they would command a great deal of power under the new institutions. The failure of the Conservatives then shows that χ was smaller than had been thought at the time of democratization. In contrast, traditional political elites in Colombia have been successful in manipulating political institutions to sustain their power, even after the complete enfranchisement of males in 1936. In particular, by structuring the electoral rules in a way that discouraged entry by third parties, particularly Socialists, they were able to keep dissident factions within the parties and limit demands for radical redistributive policies (see Mazzuca and Robinson 2004). As noted previously, other factors facilitated this strategy in Colombia, particularly the fact that the distribution of land was more egalitarian than in other Latin American countries; thus, there was a substantial middle class with much less interest in redistribution (see Bergquist 2002).
This Colombian example suggests that, at least to some extent, manipulation of institutions can make X endogenous. In support of this, Sáenz Peña also tried to manipulate the electoral system by introducing a system called the “incomplete list.” Under this system, congressional candidates were elected in three member constituencies; however, only two members were elected from the party with the most votes, with the third allocated to the party with the second largest number of votes. Smith (1978, p. 11) notes that this “discriminated sharply against small parties, discouraged the formation of new movements, favored the established interests.” This system was constructed as a way of guaranteeing one third of the seats to the Radicals as a concession to avoid further conflict with the anticipation that Conservatives would secure the two-thirds majority.
A fascinating example of an apparently successful manipulation of democracy is Pinochet’s 1989 constitution. Pinochet lost a plebiscite that he had hoped would further extend the military government. He was faced with the decision about whether to actually democratize or instead ignore the results of the vote and stay in power by using force. In the end, he decided that democracy was the better option, but his preferences were clearly influenced by his success at “designing democracy.” In particular, he managed to write into the electoral rules a systemic gerrymander that overrepresented Conservative groups (Londregan 2000); in our model, this increases X and makes repression less attractive.
Another potentially important example is from Rokkan (1970), who argued that proportional representation was introduced in many Western European countries at the time of mass democratization by Conservatives trying to protect their power. In our framework, if Rokkan is right, then this switch in electoral rules may have played an important role in preserving democracy in such countries as Sweden, Belgium, and Norway (although Rokkan did not explain why the rule changes were permanent once the Socialists took power, as they did in Sweden and Norway; see Mazzuca and Robinson 2004).
The results in this section also throw some interesting light on the claims made in the comparative politics literature about how political elites try to “manage” transitions (e.g., Linz and Stepan 1996). For example, it is often argued that because the dictatorship in Argentina collapsed after the Falklands War in 1983, it had little ability to influence the design of democratic institutions. On the other hand, because the Brazilian dictatorship managed to organize a relatively orderly transition to democracy in 1985, it was able to significantly influence the form of political institutions and the outcomes in the nascent democracy. Our model shows that the ability to manipulate democracy may lead to a peaceful transition to democracy whereas otherwise there would have been repression. Thus, the fact that the Brazilian military was able to control the process of democratization in the 1980s may have actually facilitated it.
What explains why in some places the elites were able to install a limited democracy whereas in others they were not? What explains why in some circumstances the majority are willing to design institutions to limit their own power?
First, in many circumstances the relevant institutions may essentially be historically determined and difficult to change. By their nature, institutions te
nd to persist over time (Acemoglu, Johnson, and Robinson 2001, 2002) and, for the purpose of understanding regime dynamics, must be taken as given. An interesting example of this would be the fact that all Latin American countries have presidents. The consensus on the origins of presidentialism in Latin America is that when these countries became independent, they took the form of political institutions in the United States as a blueprint for how to organize a republic. Hence, they adopted presidential forms of democracy that have persisted over time.
Second, designing institutions involves both costs and benefits, both of which are uncertain. Take the decision of the ANC to build guarantees for whites into the South African constitution. This limited its power and, other things being equal, was undesirable from its point of view. One part of these concessions was the introduction of proportional representation. Reynolds (1999) notes:
One of the least contentious issues throughout the entire negotiation process was the agreement of almost all the key players on the use of a proportional representation system (PR) to elect the Constitutional Assembly in 1994. The whites-only parliament had inherited the British single member district (SMD) plurality system ... and it was long thought that the ANC would seek to maintain the system ... because they perceived electoral advantage in doing so. (p. 183)
However, it was also clear to the ANC that an electoral system that underrepresented whites could be dangerously destabilizing. Reynolds (1999, p. 184) records that “the 1980 census showed whites to be in a majority in only five ... districts ... the ANC appreciated the way in which PR could facilitate an inclusive polity which would convert potentially anti-system minority parties into pro-system parties with incentives to play their democratic roles” and “the NP quickly realized that the existing SMD plurality system had the potential to devastate their seat winning abilities” (Reynolds 1999, p. 185). Not only did the ANC worry that the whites would be underrepresented, it also worried they would be overrepresented. For example, one problem with the SMD system was that “it would have given the ANC enough of a ‘seat bonus to push them over the two-thirds threshold, giving them enough seats to write the permanent constitution alone” (Reynolds 1999, p. 185). As noted in Chapter 5, the ANC recognized that it was not advantageous to be able to independently rewrite the constitution. As a result, the ANC quickly agreed to switch to PR.
Whether the ANC would want to make such concessions would depend on its perception of the possible actions that the white minority could take. For example, if it expected the whites to sponsor a coup against democracy or flee the country with their wealth, building guarantees into the constitution would be more attractive. In reality, it is also uncertain whether any particular institutional guarantees work. For example, Robert Mugabe’s regime in Zimbabwe has been able to override most of the checks and balances placed on it by the 1980 constitution, including the clauses designed to bolster the political power of the whites.
12. Ideological Preferences over Regimes
In our analysis so far, the only reason that agents care about political institutions is because of their different economic consequences. An alternative and complementary perspective is to recognize that individuals may also have ideological preferences over regimes. For example, after the Enlightenment in Europe, it may have been the case that the elites preferred democracy to nondemocracy for purely ideological reasons.
How does incorporating such ideological concerns change our analysis? At some level, a lot - at some other level, not that much. Of course, if ideological preferences are primary much of our analysis is not relevant. However, if ideological preferences are present, but not large enough to totally swamp the relevant economic concerns of individuals as well, much of our analysis and many of the insights developed so far continue to apply.
Let us introduce ideological concerns in our baseline model of democratization in Section 5 without repression. In particular, imagine that people’s utility functions are additive in consumption and a term that captures an intrinsic preference for democracy. In democracy, the utilities of a poor citizen and rich elite agent who consume incomes yp and yr are yp + Bpand yr + Br, where we normalize by average income. Here, Bp > 0 and B, > 0 capture the positive utility from living under democratic institutions. In contrast, if society is a nondemocracy, then agents do not receive these extra utility “benefits.” All agents aim to maximize their expected utility.
In this model, the threat of revolution is not the only way democratization may arise. If Br is relative large, then the elites prefer to democratize even though they could avoid doing so by redistributing income themselves. This was a result that we could never have before because democratization was always worse for the elites than making concessions. This corresponds to a “purely ideological” democratization, driven by the social values of the elites, arising when Vr (D) ≥ Vr(N,τN =).
To see how this new feature influences the model, note that because neither the revolution constraint nor the equation determining µ* depends on the value from democracy, they are unchanged by the introduction of ideological preferences. The only difference is that before, democracy arose only if θ > µ and µ < µ*. Now it is possible that even if µ > µ*, so that democratization could be avoided by concessions, the elites democratize. Moreover, even if θ ≤ µ, so that the revolution constraint does not bind, Br can be sufficiently large to ensure Vr(D) ≥ Vr(N), thus creating an ideology-driven democratization.
The crucial issue, naturally, is whether the elites have a strong enough preference for democracy. To study this, we need to define two cutoff levels:is the cutoff level such that when Br ≥, even when θ ≤ µ, so that the threat of revolution does not bind, the elites democratize. This is clearly given by:
in which the right-hand side is the net transfers away from the elites when the tax rate is the one that will be chosen in a democracy, τp. This is what the elites pay in democracy as net transfers away from them but, in return, they obtain the ideological benefit of having established democracy, Br.
However, when θ > µ, the comparison is not between no taxation and democracy but rather between limited taxation and democracy. Therefore, the relevant threshold is:
which takes into account that even without democracy, there will be net redistribution away from the elites equal to(δ - θ) - δC() < 0 with probability p. Clearly, we have that:
Analysis of these equations shows that bothandare increasing in θ: in other words, the higher is inequality, the higher areand. For example:
which follows immediately from noting that - (δ - θ) > 0 and recalling that drp/dθ > 0. This is because with greater inequality, democracy is more costly for the elites (because it redistributes more away from them) and, as a result, their ideological preferences have to be stronger for them to prefer democracy to nondemocracy.
Finally, note that now, compared to (6.7), the condition that democracy prevents revolution is easier to satisfy because there is an extra utility benefit from democracy that does not accrue if there is revolution. Taking this into account, we can restate (6.7) as:
(6.28)
We now have the following result:
Proposition 6.5: There is a unique subgame perfect equilibrium such that:
• If θ < µ and Br< , then the revolution constraint does not bind and the elites stay in power without democratizing or redistributing income. If Br> , and (6.28) holds, then the elites democratize.
• If θ > µ, then the revolution constraint binds. In addition, let µ* be defined by (6.6). Then:
(1) If µ ≥ µ*, and Br≤ , the elites do not democratize and set the tax rate to redistribute enough income to avoid a revolution.
(2) If µ < µ*, or µ ≥ µ* and Br> , and (6.28) holds, then the elites democratize.
(3) If (6.28) does not hold, there is a revolution.
There are a couple of interesting points: (1) if ideological considerations are not important, our previous analysis applies identically because this implies that Br is suffici
ently small, so Br ≤and Br≤ will be the relevant part of the parameter space, where the implications of Proposition 6.5 become identical to Proposition 6.1; and (2) when ideological considerations are sufficiently important, they may induce transitions to democracy that would not have taken place for purely economic reasons. Nevertheless, even in this case, economic incentives are potentially important. For example, both Br > B and Br >are more likely when inequality is low. As inequality increases, the redistribution away from the elites in democracy becomes larger, and for a given ideological benefit of democracy, these two conditions are less likely to hold.
Figure 6.4. Concessions, Repression, or Democracy?
13. Democratization in a Picture
We have now sufficiently developed our framework that we can rigorously construct Figure 2.1 from Chapter 2. We do this in the context of the static model of Figure 6.2 and assume that democracy is always sufficiently redistributive that it stops a revolution. Consider Figure 6.4. On the horizontal axis is inequality and on the vertical axis the cost of repression, is plotted. First note that when inequality is low - in particular, when µ > θ - there is no threat of a revolution; this is the region to the left of the vertical line at µ in Figure 6.4. Next, note that whether a nondemocratic regime can stay in power by making policy concessions and redistributing income is also independent of K and is thus another vertical line. This is derived from (6.6). We used the notation θ* in Section 6.1 to refer to the critical level of inequality at which this equation is satisfied, which is shown on Figure 6.4. Therefore, we have two vertical lines that divide the box into three regions. On the left, is the political status quo with no repression. Next is a region where there are concessions but no need to create democracy. Finally, there is a region where inequality is so high that there will be revolution unless democracy is created or repression is used.
Economic Origins of Dictatorship and Democracy Page 31