Economic Origins of Dictatorship and Democracy

by Daron Acemoglu

  Gapon’s movement gathered momentum in 1904 when rapid inflation caused by the war against Japan (which had started in February) led to a 20 percent decline in real wages. When four members of the Assembly of Russian Workers were dismissed at the Putilov Iron Works, Gapon called for industrial action. Over the next few days, more than 110,000 workers in St. Petersburg went on strike.

  In an attempt to settle the dispute, Gapon made a personal appeal to Nicholas II and in January 1905 he drew up a petition outlining the workers’ sufferings and demands. This petition demanded an eight-hour day; freedom to organize trade unions; improved working conditions; free medical aid; higher wages for women workers; elections to be held for a constituent assembly by universal, equal, and secret suffrage; freedom of speech, press, association, and religion; and an end to the war with Japan.

  On January 22, Gapon led a demonstration to the Winter Palace in St. Petersburg to present the petition to the tsar. When the procession of workers reached the palace, it was attacked by the police and the Cossacks. More than hundred workers were killed and some three hundred were wounded. The incident, known as Bloody Sunday, started a series of events that became known as the 1905 Revolution. Strikes took place all over the country and the universities closed down when the entire student body staged a walkout to complain about the lack of civil liberties. Lawyers, doctor, engineers, and other middle-class workers established the Union of Unions and demanded a constituent assembly.

  In June 1905, sailors on the battleship Potemkin protested against the serving of rotten meat. In response, the captain ordered that the ringleaders be shot. The firing squad refused to carry out the order and joined with the rest of the crew in throwing the officers overboard. The Potemkin mutiny spread to other units in the army and navy.

  Industrial workers all over Russia went on strike and, in October 1905, the railwaymen went on strike, which paralyzed the entire Russian rail network. Later that month, Leon Trotsky and other Mensheviks established the St. Petersburg Soviet. Over the next few weeks, more than fifty soviets were formed throughout Russia.

  Sergei Witte, the new Chief Minister, advised Nicholas II to make concessions. He eventually agreed and published the October Manifesto, which granted freedom of conscience, speech, meeting, and association. He also promised that in the future, people would not be imprisoned without trial. Finally, he announced that no law would become operative without the approval of a new organization called the Duma. Because this was only a consultative body, many Russians felt that the reform did not go far enough. Trotsky and other revolutionaries denounced the plan. In December 1905, Trotsky and the executive committee of the St. Petersburg Soviet were arrested. Nevertheless, the announcement of the concessions made in the October Manifesto had the effect of calming the country and undermining the revolutionary threat.

  The First Duma was elected on the basis of indirect universal male suffrage. The peasants, the townsmen, and the gentry all elected their own representatives. Delegates from all provinces met in the provincial town and chose members of the Duma. However, since publication of the October Manifesto, Nicholas II had already made several changes in the composition of the Duma: he had created a state council, an upper chamber, of which he would nominate half its members. He also retained for himself the right to declare war, to control the Orthodox Church, and to dissolve the Duma. The tsar also had the power to appoint and dismiss ministers. Even before the First Duma met, Nicholas II was backtracking on the promises he had made in October.

  Nevertheless, the First Duma had a left majority consisting of Socialist Revolutionaries, Mensheviks, Bolsheviks, Octobrists, and members of the Constitutional Democrat Party. At their first meeting in May 1906, members of the Duma put forward a series of demands, including the release of political prisoners, trade-union rights, and land reform. Nicholas II rejected all these proposals and dissolved the Duma in July 1906. In April 1906, Nicholas II had forced Witte to resign and replaced him with the more conservative Peter Stolypin. Stolypin attempted to provide a balance between the introduction of much needed social reforms, such as land reform, and the suppression of the radicals.

  Elections for the Second Duma took place in 1907. Stolypin made changes to the electoral law and used his powers to exclude large numbers from voting. The new electoral law also gave better representation to the nobility and greater power to large landowners to the detriment of the peasants. Changes were also made to the voting in towns: those owning their own home elected more than half the urban deputies. This reduced the influence of the left but, when the Second Duma convened in February 1907, it still included many reformers. After three months of heated debate, Nicholas Il dissolved the Duma on June 16, 1907.

  The Third Duma met on November 14, 1907. The former coalition of Socialist Revolutionaries, Mensheviks, Bolsheviks, Octobrists, and the Constitutional Democrat Party were now outnumbered by the reactionaries and the nationalists. Unlike the previous Dumas, this one ran its full term of five years.

  The 1905 Russian Revolution is our final example of how - without fundamental changes in the nature of de jure political power - promises can be reneged on. In response to the uprisings and unrest of 1905, Nicholas II made concessions including, to some extent, the creation of a democratic institution — the Duma. Yet, the Duma was not powerful enough to guarantee that Nicholas II would carryout his concessions; once the revolutionary moment had passed, Nicholas II duly reneged.

  4.3 Modeling Commitment Problems in Nondemocracy

  We now start laying the scene by introducing simple ways of modeling potential commitment problems in politics. Let us first return to the game shown in Figure 5.1, the key feature of which is that the elites decided the tax rate before the citizens made the revolution decision. Now imagine an alternative game shown in Figure 5.2, in which the citizens decide whether to make the revolution decision first; then, if there is no revolution, the elites set the tax rate. The difference between the two games may appear minor, but there is, in fact, a major difference: in the game shown in Figure 5.1, there was no commitment problem. The elites set the tax rate before the revolution decision of the citizens and could use the tax rate to avoid the threat of revolution. Now, the elites no longer have that option because they set the tax rate after the revolution decision.

  Figure 5.2. The Commitment Problem in Nondemo-cracy.

  Let us analyze the subgame perfect equilibrium of this game. As usual, we do this by backward induction, starting in the last subgame, which is the one after the citizens decide not to undertake a revolution. In this subgame, the elites have to decide the tax rate, the tax rate gets implemented, and the game ends. Because there are no longer any constraints left, they simply choose their most preferred tax rate, τr = 0, giving payoffs:

  (5.11)

  to the citizens and the elites. We use the notation Vi(N) as the value to i = p, r in nondemocracy when the elites set their ideal policy. Moving to the previous stage of the game, the citizens have to decide between revolution, which will yield them the payoff VP( R, µ) as given by (5.1), or no revolution, which will give them the payoff Vp(N) = V (yp|τN = τr). The former is greater whenever (5.4) holds, so the citizens undertake a revolution whenever (5.4) holds.

  In specifying the equilibrium, we again use the notation σp = {ρ} and σr= {τN}. The citizens play first and choose p ∈(0, 1} (i.e., whether to revolt), while the elites play second and choose the tax rate τN. Because the elites only get to play if ρ = 0, we specify this as a choice (not a function) τN ∈ [0, 1]. Then, a subgame perfect equilibrium is a strategy combination, {r,p} such thatp andr are best responses to each other in all proper subgames.

  We can see that the following strategy profiles are the unique equilibria. When θ≤ µ, we have ρ = 0 and τN = 0. In this equilibrium, the revolution constraint does not bind so the citizens do not revolt, and the elites set their preferred tax rate of zero. When θ > µ, then the following strategy profile is the unique equilibrium: p = 1. In this
case, revolution is the optimal action and the poor undertake it. We now have the following proposition:

  Proposition 5.2: There is a unique subgame perfect equilibrium {r,p} in the game described in Figure 5.2, and it is such that

  • If (5.4) does not bind, then p = 0 and τN = 0.

  • If (5.4) binds, then p = 1.

  The results of this proposition are different from those of Proposition 5.1, and an equilibrium revolution happens for a much larger set of parameter values. This reflects the commitment problem of the elites. In the game described in the previous subsection, there was no commitment problem because the elites moved before the citizens had to decide whether to undertake a revolution. Now there is a serious commitment problem. To highlight the essence of this problem, think of the elites as “promising” redistribution to avoid revolution. However, this is not credible because, according to the game in Figure 5.2, they move after the revolution decision of the citizens, and whatever promise they make will not be credible.

  This game illustrates the more general commitment problem outlined previously : those with political power - here, the elites - cannot promise to make transfers in the future as long as they hold onto their political power. In the game shown in Figure 5.2, the taxation decision of the elites was made after the revolution decision of the citizens; this implies that the elites have to promise to make transfers in the future. It is this promise about the future that is not credible. This is in some sense quite a reduced-form situation, however, because there is no real sense of present or future, and we can talk of promises only in a loose sense because the game does not really involve promises. We gradually enrich this game and use it as a building block for our analysis of democratization in Chapter 6. In the next section, we introduce a version of the simple game used throughout this book, which is in turn a simplification of a full dynamic game, introduced in the subsequent section.

  5. A Simple Game of Promises

  We have so far discussed the revolution constraint and how the elites can try to prevent revolution by making promises of redistribution, and we indicated why these promises may not be credible because the elites hold onto political power and, given their political power, they can renege on their promises. Two important elements are missing from this picture: ( 1 ) an effective threat of revolution is a rare event and occurs only when the citizens manage to solve the collective-action problem inherent in revolution; and (2) we have so far analyzed games in which either the elites move before the revolution decision and there is no commitment problem, or they move after the revolution decision and there is no possibility of promises. Instead, we would like a game that has some possibility of promises by the elites, but these promises are only partially credible.

  Figure 5.3. A Game of Promises.

  Figure 5.3 shows the simplest game incorporating these features. Nature moves first and selects between two threat states, low and high; S = L or H. The motivation for introducing these two states is to emphasize that only in some situations is there an effective threat of revolution. In general, this could be because some circumstances are uniquely propitious for solving the collective-action problem - such as a harvest failure, a business-cycle depression, the end of a war, or some other economic, social, or political crisis. We assume that the effectiveness of the revolution threat differs between these two states. In particular, we assume that the payoff to the citizens from revolution in the state S is:

  (5.12)

  where we think that the low-threat state corresponds to the case in which it is relatively costly for the citizens to solve the collective-action problem or face other problems in organizing revolution, so µL is high. To simplify the discussion, we take the extreme case in which µL = 1. In contrast, in the high-threat state, the citizens are able to solve the collective-action problem relatively costlessly and/or the elites are not well organized in their defense, so there may be an effective threat of revolution, which we capture by assuming that 1 > µH > 0. Because µL does not play any real role in our analysis - indeed, we suppress this state later in the book to simplify the game trees - from now on, we use the notation µH = µ.

  After nature reveals the threat state, the elites set the tax rate τN. Observing this tax rate, the citizens decide whether to undertake a revolution. So far, the game is not very different from the game in Figure 5.1. In fact, if it ended here, it would be almost identical, enriched only by having two states instead of one. However, after the revolution decision of the citizens, there is a continuation game capturing in reduced form the problems that those with political power will have in promising to undertake future actions that are not in their immediate interest. In particular, nature moves and determines whether the elites get to reset the tax from τN to a new rate different from that which they promised. More specifically, with probability p, the promise that the elites made to redistribute at the tax rate τN stands. But, with probability 1 — p, the promise is void, and the elites get to reset the tax. We useN to denote this tax rate. At this point, because the opportunity to mount revolution has passed, the elites are unconstrained and set their most preferred tax, τN= τr. We use the notation ν ∈ {0, 1} for nature’s choice, with ν = 1 indicating that the elites can reset the tax rate.

  This continuation game after the revolution decision of the citizens is a reduced form way of modeling the inability of those with political power to commit to future redistribution and taxation decisions. When p = 1, there is no commitment problem and we have the situation depicted in Figure 5.1; whereas when p = 0, there is a complete inability to commit and we have the game shown in Figure 5.2. We can, therefore, use p as a way of parameterizing the ability of the nondemocratic regime to commit. In this game, there is no “future” in the proper sense because there is only one period of redistribution rather than an explicit difference between today and in the future. Nevertheless, the continuation game incorporates, in a relatively simple way, the possibility that after the threat of revolution is gone, the elites can backtrack from their promises. The next section shows that when we have a fully dynamic model in which the revolution threat recurs in the future, the model has a reduced form similar to the simpler game shown in Figure 5.3 that we are analyzing.

  The relevant payoffs are as follows. If the citizens undertake a revolution, the payoffs are VP (R, µs) given by (5.12) and Vr (R, µs) = 0. If the elites get to reset the tax, they will choose their most preferred tax rate, τr, so the payoffs are Vp(N) and Vr (N) given by (5.11). If they are unable to reset the tax and the promised tax rate of τ N stands, then the values of the two groups are V (yp|τN) and V (yr| τN) as given by (5.8). This implies that the expected payoffs at the time the elites make a promise to redistribute at τN are (Vp (N, τN), Vr(N, τN)), such that:

  (5.13)

  which take into account the fact that redistribution at the tax rate τN happens only with probability p, whereas with probability 1 — p, the elites reset the tax to τr. Notice also that we are using the notation Vi(N, τN), which refers to the case in which the elites make a promise of redistribution at the tax rate τN. This is distinct from Vi (N), which refers to the values when the elites are unconstrained. We use this type of notation throughout the book.

  Therefore, after observing the promise of redistribution at the tax rate τN, the citizens have to make a comparison between Vp(N,τN) as given by (5.13) and the payoff from revolution Vp(R, µS) as given by (5.12). Clearly, VP(N, τN) > Vp (R, µL) for any τN by virtue of the fact that µL = 1. Therefore, in the low state, µS= µL, the elites do not suffer a revolution; anticipating this, they make no concessions and simply set their most preferred tax rate, τN = τr = 0 (or, using our notation, τN(µL) = τr).

  In contrast, in the high-threat state S = H, the revolution constraint could be binding. As before, we say that the revolution constraint binds if Vp(R, µH) > Vp(N); that is, if the citizens receive more from revolution than they would when the elites set their most preferred tax rate in nondemocracy. Usin
g (4.7) and (5.12), this revolution constraint is again equal to (5.4). If this revolution constraint does not bind, then even in the high state, the elites are unconstrained and, again, they set their most preferred tax rate. Suppose, on the other hand, that the revolution constraint binds (i.e., θ > µ). What happens then?

  The elites would like to prevent revolution if at all possible. Whether they can do so depends on the value they can promise to the citizens. Clearly, the most favorable tax rate they can offer to the citizens is τN = τ p, as given by (4.11). However, this is not as good as offering τp for certain because of the commitment problem. Whether the elites can prevent revolution depends on whether Vp (N, τN= τp) is greater than Vp(R, µH). Written more explicitly, the key condition is whether:

  recalling that µH takes the specific value µ, or whether:

  (5.14)

  If inequality is limited (i.e., θ is relatively low) or if there is a high probability that the promise made by the elites will be upheld (i.e., p is relatively high), then living under nondemocracy is not too bad for the citizens, and the condition (5.14) will hold and revolution can be avoided.

  To analyze the model, let us determine a critical value of the revolution cost µ* such that (5.14) holds as an equality:

  (5.15)

  Then, when µ > µ*, we have Vp (N, τN=τp) >