2. Growth and Economic Efficiency Under Dictatorship
There has been a lot of research asking the question, which is better for the economy, democracy or dictatorship? The answer is complex, mainly because the economic systems under autocracies vary so much. Those who believe there is some simple formula for distinguishing the economy of dictatorship from that of democracy should compare, for example, the economies of Nazi Germany, Apartheid South Africa, Papa Doc’s Haiti, Pinochet’s Chile and the Former Soviet Union.
2.1. Democratic Inaction
One general proposition which is true of all these systems is that dictators have a greater capacity for action, good or bad. If a dictator wishes to raise taxes, declare war, or take tough measures vs. crime, he may have to deal with some opposition to these policies among his advisers, but by and large, he can do so. Democracies, on the other hand are often mired in inaction.8 The basic reason is that democratic leaders can only act when they can build support for their policies and there may be no consensus as to what to do. Even on problems where there is agreement that something should be done, there may be no agreement on what should be done. In extreme cases, the political system of a democratic country may become paralyzed by conflicts or opposing viewpoints.9 In these circumstances, politicians often prefer to do nothing, to shroud their positions in ambiguity, or to pretend to be on all sides of an issue. The result is that the population can become cynical, and lose trust in the promises of any politician. This can set in motion a downward spiral, since the more this happens and trust is lost, the harder it becomes for politicians to do something by forging a compromise. This is more likely to happen when the pressures for political action on an issue are particularly conflicting, when positions are far apart, when issues are particularly divisive, when the population is divided along racial or ethnic lines, and when there is relatively little trust in politicians by the citizens.
2.2. Economic Growth and Efficiency
2.2.1. Introduction
Some new theorizing and empirical work compares the economic performance of democracies and dictatorships directly. A convenient place to start is Barro’s empirical work (1996a,b). Barro stresses the advantages of dictatorship, which are that it controls rent seeking and other redis-tributory pressures, i.e., the autocrat, unlike the democratic politician, is capable of shutting down or simply ignoring the redistributory demands of interest groups characteristic of democracy (Barro, 1996b, p. 2). His empirical work suggests that more democracy raises growth at low levels of political freedom but depresses growth when a moderate amount of freedom has already been attained. However, the effect of an increase in political freedom on economic growth is not very large and the overall effect "not statistically different from zero" (Barro, 1996b, p. 6). Barro’s results are only obtained once certain variables are held constant, including free markets, the rule of law, and small government consumption. So, really, again, only certain kinds of dictatorship are being discussed. The paper also, finds, perhaps surprisingly, that democracy does not necessarily promote the rule of law.
Przeworski et al. (2000) find that basically there is no difference between the rates of growth in dictatorships vs democracies in their comprehensive examination of the performance of these two kinds of regimes in 141 countries over the 40 years or so after the second world war. But the same study confirms the importance of politics on economic growth. They show that changes in office (political instability) and other forms of unrest such as strikes, demonstrations and riots reduce economic growth substantially under dictatorship, whereas while these are more frequent under democracy they do not cause a reduction in the rate of growth there (Przeworski et al. (2000) pp. 192-193).
Sen (1993) calls the general idea that dictatorship is better suited to economic development than democracy the Lee thesis, after Lee Kwan Yew, the autocratic but economic efficiency-minded ruler of Singapore for many years. Sen raises many questions about Lee’s ideas and suggests instead that democracy is intrinsically important to the process of development. In particular, Sen’s observation that famines only seem to occur under dictatorship is provocative. However, no general theoretical model is presented that compares democracy with dictatorship.
2.2.2. The Predatory State
The most prominent theoretical idea in this literature is undoubtedly Olson’s concept of an autocrat as a "stationary bandit" — at one point he refers to it as "the other invisible hand" — that guides rulers to use their power to at least some extent in the public interest. In his (2000) book, this concept is approached through a criminal metaphor. Each theft reduces the wealth of society and therefore the amount available for the thief to steal. Does this lead the thief to curtail his activity, in order to preserve the wealth of his prey? For the typical criminal, the answer is "no" because his interest is too narrow. The wealth of the society on which he preys is like a public good to the typical small scale criminal, his effort to preserve it would have only a minuscule effect, and so he is better off free riding rather than attempting to conserve it. On the other hand, the Mafia and other criminal organizations which have a monopoly on crime in their area, do have a sufficiently encompassing interest to take the effects of their thefts on the wealth of society as a whole. Thus, Olson asserts, they typically do not steal at all but engage in protection instead, charging the citizens a fee to ensure the safety of their victims both from others and from the protectors themselves.
This criminal metaphor then becomes the foundation for the origins of government. The logic is the same as that just outlined with respect to government by a "roving" vs. that by a "stationary" bandit: the stationary bandit, unlike the roving one, has an encompassing interest in preserving the wealth of the society from which he steals, and therefore limits his "theft" (taxes) and even provides public goods — both to the point where the marginal benefit to him is sufficient to account for his costs in terms of foregone income. The history of the forms of government is then simple to derive: autocracy (the stationary bandit) arises out of anarchy as the bandit(s) with the greatest capacity for violence take over the area and substitutes an encompassing for a narrow interest; democracy arises out of dictatorship when autocracy is overthrown and none of the individuals or leaders involved in the coup has sufficient power to make themselves autocrats.
In the end, just two variables are necessary to compare and analyze governments:
(i) how encompassing (breadth of self interest) is the interest of the ruler,
(ii) how long (time horizon) is his interest.
Thus, in the same way that dictatorship is superior to anarchy because the dictator has an encompassing interest in the society he rules, so democracy is superior to dictatorship because democratic majorities are more encompassing than the interest of the dictator. Secondly, dictators or democracies with long time horizons have more of an interest in preserving or enhancing the wealth of the society they rule than those who rule only for the short term.
Some evidence is presented in Keefer et al. (1996), who argue that any incentive an autocrat has to respect property rights comes from his interest in future tax collections and national income and increases with his planning horizon. They find an empirical relationship between property and contract rights and an autocrat’s time in power.
So, comparing dictatorships, the basic implication is, the more encompassing, the better. Political scientists indeed have a classification that appears to match this: between "mobilizational" regimes which encourage political participation among the ruled and regimes which simply try to stamp out opposition. The problem with Olson’s analysis is that, comparing dictatorships, the worst regimes in human history appear to be precisely those such as Nazi Germany, Soviet Russia, or Cambodia which appear to have been the most encompassing. The reason is simple: it was those regimes that wanted to remold the citizens and the societies under their rule and therefore intervened most dramatically and thoroughly into the lives of their citizens. Whether it is their brutal treatment of minorities or their record on the environment, it is an understatement to suggest that the historical record of these regimes offers little that is to be admired. So the theory appears to be capable, not just of misleading with respect to the understanding of autocratic regimes, but of "getting it wrong" in a spectacular fashion.
The same problem appears with respect to the second variable, the time horizon of the dictator. In Olson’s model, the longer the time horizon, the better, i.e., the more the dictator tends to rule in the social interest. But regimes with a long time horizon have been precisely those in which the leaders had a tighter grip on power, and hence were more capable of molding the society and the individuals within it, i.e., the mobilizational regimes just discussed. Those where the regime is just interested in looting the society typically have a shorter time horizon.
In short, from the point of view of citizens of these regimes, or more specifically from that of the peasants under Stalin, the Jews under Hitler, the blacks in South Africa, and so on, it would no doubt have been better if their bandits had been less stationary!
The alleged superiority of dictatorship over anarchy is also challenged in a major article by Moselle and Polak (2001). In their model, the existence of a state can result in lower levels of both output and welfare than would occur under anarchy. This occurs if the state is "predatory" in the sense that the rulers extract taxes from the population for their own ends. In this framework, even a weak state can be bad for output and welfare and that a "corrupt" state that makes side deals with bandits can be especially bad.
Perhaps the most basic problem with Olson’s framework is, I suspect, the lack of emphasis on competition. Once the struggle for power is assumed away, many of the most interesting aspects of the behaviour of dictators become idiosyncratic features of their preferences, and hence largely unpredictable, instead of being derived from the principle of competition. Thus the wars among the monarchies, etc are all aspects of "princely consumption". And how would the model explain Stalin’s war against the peasantry, Hitler’s treatment of the Jews, and the persecution of minorities in other dictatorships? On the bandit model, the only way to understand these forms of behaviour is that dictators have some monopoly power, and that they use this power to implement their preferences which happen to be weird preferences. The reason for this is that the model does not deal with the competitive struggle to acquire and maintain dictatorial powers. So the behavior of the dictator cannot be understood as motivated by competition or survival in office but simply as consumption.
Two other contributions address the problem of why some dictatorships, most notably regimes in East Asia and Chile, appear to be pro-growth while in others the autocrat is "predatory" and simply plunders the economy. Robinson (1997) argues that the likelihood of predatory behaviour may be positively related to the extent to which a regime is encompassing and values the future. He develops a model in which whether or not a state is predatory hinges on the relationship between development and the distribution of political power. Development is typically inconsistent with the preservation of the political status quo and this gives those in power an incentive to oppose it. Predatory behaviour is also more likely the lower the level of income and the more unequal the society. To put it bluntly, from the dictator’s point of view, ruining the economy can sometimes be a good thing! And the regimes of Mobutu and Papa Doc, who both did this, were extremely long lived. A democratic politician cannot hope to profit in the same way.
Michael Spagat’s (2001) paper addresses this problem by suggesting that there is a "bifurcation point" or level of capital below which it does not pay the dictator to try and develop the economy, and above which the dictator pursues rapid growth in order to maximize his personal consumption over time. He develops this idea in a simple formal model. A particularly novel feature of it is that there is an endogenous probability of a political catastrophe which removes the dictator from power, and this in turn depends on the dictator’s capacity to satisfy certain groups which depends on the level of the capital stock. Hence a dictator’s economy sometimes grows faster than a social planner’s might, as capital accumulation wards off the possibility of catastrophe. The authors use simulation analysis to show the existence of bifurcation and to show how it depends on various parameters, and they provide some empirical evidence using Gastil data of the existence of bifurcation, and of their basic prediction that the variance of growth rates in dictatorship is higher than that under democratic regimes.
2.2.3. The Contest for Power
In contrast to economic models which stress the incentives of a ruler, once he is in office, Wintrobe (2002) focuses on the conditions under which the ruler obtains power, and how he can be deprived of it. All political systems contain a mechanism which determines the allocation of political power, and if and how it is reallocated when a transfer would improve the functioning of the system. Among the most obvious and commonly considered types of political system — democracy, dictatorship, anarchy, and hereditary monarchy — only democracy appears to possess a relatively low-cost procedure or mechanism which makes it possible to transfer political power on a regular and systematic basis, where the transfer is accepted by those who lose power as well as those who gain it, and which offers some possibility that these reallocations will tend to shift power into the hands that can use it most effectively.
Thus there is a strict analogy between democracy, based on human rights, and capitalism, based on property rights: democracy makes power transferable just as capitalism makes the ownership of capital assets transferable. This gives democracy an enormous advantage over these other political systems.
To elaborate, the main economic advantage of the election mechanism would seem to be that it allows for the transfer of power at relatively low cost. It solves the contest for power problem. If there are no elections, the only ways to transfer power are by such means as revolutions, insurrections, coups and wars. Compared to these, democratic elections on the basis of inalienable human rights would seem to be, in a word, cheap. Thus the economic attractiveness of the election mechanism is simple: it provides a formal and agreed-upon procedure to decide on the allocation of political power, and one that is explicitly accepted by or consented to in advance by the parties who lose the contest. Among the most commonly discussed systems of government — anarchy, hereditary monarchy, dictatorship and democracy — only democracy possesses this advantage.
Granted that democracies can transfer power at relatively low cost, does power typically transfer from lower to higher valued uses? Do democracies allocate power properly? In the models of Stigler (1971); Peltzman (1976); Olson (1982), democracy is inefficient because it is dominated by interest groups and the policies pursued by interest groups are inefficient and wasteful. In the rent-seeking framework, it is the contest itself which is inefficient and wasteful. These theories are the foundations of Barro’s (1996) empirical work. However, Becker (1983) showed that under democracy the losses from inefficient policies enter into the workings of the political system and affect its allocation of power. The reason is that the larger the deadweight losses from a policy, the more opposition there will be to it from the groups which bear these losses. Alternatively the more inefficient a subsidy, the less the group which gains from the subsidy will exert pressure to obtain it. Consequently, even in a model such as Becker’s, which focuses solely on interest group competition, the contest is not wasteful, and it tends to select efficient over inefficient policies. To put it simply, if power ends up in the "wrong" hands the democratic political process takes account of this and tends to set it right.10
What about dictatorship? The basic difference between dictatorship and democracy is that dictators have the capacity to repress opposition to their policies (as outlined above). They can silence demonstrations, censor the media, ban opposition parties, put leaders of troublesome groups in jail, and, not uncommonly, torture or execute them. As a consequence, the repressed are, in effect, not allowed to spend resources to exert political pressure: instead they are silenced by the government. It follows that if the costs of public policies can be made to fall on those who are repressed, these costs do not enter into the competition among interest groups.11
Assume for a moment that this is the case, i.e., all of the costs of inefficient policies fall on those whose political demands are effectively repressed by the regime. Since the losses make the repressed worse off, this weakens the capacity of those who are opposed to the regime to resist it.12 This is the strategy of "immiserization" practised most notably, perhaps, by Papa Doc of Haiti.13 Another nice illustration of this is the effect of sanctions against Saddam Hussein, discussed by Kaempfer et al. (2001). The sanctions generate rents, and these are appropriated by those who are close to Saddam. The losses from the sanctions are borne by those who are opposed to the regime, and this in turn weakens their capacity to oppose it, leading to his further entrenchment in power. To put it simply, the sanctions against Saddam Hussein don’t necessarily weaken his hold on power at all.
On the other hand, to the extent that the repressed cannot be made to bear all of the costs of the public sector, some of these costs will fall on other groups — actual supporters, potential supporters and largely passive acceptors of the regime. To the extent that the costs of public expenditures and regulations fall on these groups, they would indeed enter into the competition among groups for subsidies and other rents from the regime under dictatorship, just as they do under democracy. However, even in this case, the mechanism does not work as well as under democracy. The reasons are: (1) The information problem deriving from the Dictator’s Dilemma: In a democracy the different groups competing for redistributory policies or public goods are free to openly debate and criticize existing policies and to expose flaws in each others’ proposals. Under dictatorship, any form of attack on policies which have been or might be favored by the regime can be interpreted as a sign of disloyalty,14 and for this reason people may not be eager to report problems to the autocrat. (2) What incentive is there for the dictator to correct bad policies? After all, among the fruits of dictatorship is "the quiet life" — freedom from competitive pressures so long as he is safely in office. (3) Finally, once decisions are made there may be no mechanism by which to correct them except by the overthrow of the dictator. Since there is no peaceful and regularized way to replace an autocrat, he may tend to oppose any attempt to change the policies, since any change may be threatening to his survival in office.
To sum up, the economic losses from inefficiency may or may not enter into the dictator’s political budget equation, depending on who experiences them. Let us take the two cases in turn: (1) Suppose the losses are experienced by actual or potential supporters. The lack of political competition under dictatorship still implies that the economy may be allowed to deteriorate more (compared to a democracy) before some attempt is made to change the policies or replace the dictator; (2) On the other hand, suppose the economic losses are experienced primarily or wholly by those who are opposed to and repressed by the system. In this case the losses typically weaken rather than strengthen the capacity of those who are opposed to the regime to actually topple it, and this raises the attractiveness of inefficient policies from the point of view of the dictator.
A final issue is the relative influence of producer vs consumer groups under dictatorship vs democracy. Ever since the work of Downs, it has been a standard proposition in the economics of politics that democracy favors producer groups over consumer groups (Downs, 1957; Stigler, 1971; Peltzman, 1976; Olson, 1982; Becker, 1983). The main reasons advanced are that since these groups are small, it is relatively easy for them to overcome the free rider problem, and since their per capita benefits would be large from any subsidy, they have a substantial interest in applying pressure to obtain it. On the other hand, consumer groups are large, and the per capita benefit from any subsidy would be small.
I pointed out above that dictators cannot survive in office on the basis of repression alone but need support as well. Which groups can be expected to support dictators? Consumer groups, environmental groups and other groups with a large number of potential supporters, each of which has a small stake in issues like the prices of goods or the state of the environment have difficulty surviving or forming under autocracy. There are typically no laws protecting human rights under dictatorship. Without such laws, it is difficult for large groups — such as consumers — to organize. There is no free press to call attention to pricing or environmental or labour abuses and to aid in the formation of a mass membership and there are no independent courts in which to sue violators. And it is difficult for supporters of human rights, who have been crucial in generating the "rights revolution" (Epp, 1998; Ignatieff, 2000) to mobilize support. In brief, the common weapons of mass organizations — publicity and the courts — are more easily countered by a dictator than a democratic politician.
On the other hand, the weapons of small producer groups such as cash donations actually thrive in the closed environment and tame courts of a dictatorship. In exchange, dictators obviously have much to offer producers for their support including tariffs, subsidies and other rents, fewer problems from labor unions, and the removal of unfavorable regulations. So the possibilities of a trade of rents for support between the dictator and the small, concentrated interest group is actually enhanced under dictatorship, just as trades with representatives of broader public opinion are diminished. This implies that producers typically have more power under dictatorship than democracy.
This analysis also provides an alternative explanation for Barro’s evidence cited above: that the rate of growth is slightly higher under dictatorship than democracy at low levels of dictatorship and lower at high levels of repression. Since producers especially benefit from economic growth, their greater political weight under dictatorship implies that dictators would emphasize this policy. Note, however, that this growth comes as the result of the greater influence of producer groups and is not necessarily a Pareto improvement. Thus the growth could arise to the detriment of the environment, the consumer, etc. Moreover, at high levels of repression, this positive effect on growth is increasingly overwhelmed by the information problems generated by the Dictator’s Dilemma, which increasingly hamper growth and ultimately strangle it.
Finally it is worth pointing out that an extension of the theory of property rights used in this analysis provides a simple economic justification of human rights. Economic efficiency justifies the ownership of private property on the ground that property should be allocated to the party who is most highly motivated to maximize its value. Who is it that can be counted to manage or take care a piece of property best? The owner. Human rights give this privilege of "ownership" of the individual (if you like, of his human capital) to that individual himself or herself. Under dictatorship, it resides with the sovereign. But the dictator, as Sen (1993) suggested , tends to regard the people under his rule as "stock" and cannot be expected to care for their lives the way they would themselves. Perhaps this explains Przeworski et al.’s striking result that the average life span is systematically lower under dictatorship (see Przeworski et al., 2000, chapter 5).
3. Policy towards Dictatorship
3.1. Aid: A Single Standard
What policies should be followed towards dictatorship by democratic regimes interested in promoting freedom?
Suppose, idealistically, that the only goal of Western policy is to reduce repression. The "weapons" in our arsenal are sanctions, trade agreements, imposing human rights constraints, and aid packages. Take a classic example of a tinpot dictator like Ferdinand Marcos. Should we have given aid to his regime? Suppose Marcos’ only goal was to consume as much as possible — in his case, this meant buying shoes for his wife Imelda. What limited his consumption? Why didn’t he spend all of the GNP of the Philippines on shoes for her? The constraint is that he had to stay in office, so he could not allow his power to fall so low that he was in danger of being deposed. As a tinpot, the levels of both repression and loyalty under his regime were just high enough to stay in office.
Suppose first that the tinpot is safely in office, which, at one point, according to accounts of the regime, Marcos felt he was. Then there is no point in giving him aid, because all he will do with the money is to buy more shoes. A trade agreement would have the same effect. On the other hand, suppose he is in danger of being deposed. Then the aid simply props up the regime. So, in neither case does the aid reduce repression. An alternative policy would be to insist on human rights observances as a condition of receiving aid. But if the levels of repression and loyalty were previously just sufficient to stay in office, Marcos will simply be deposed if he lowers repression. So he would have refused this offer, and the policy is ineffectual.
On the other hand, suppose the aid is tied to human rights observances in a particular way. In order to keep receiving the aid, repression must be steadily relaxed over time. Then the dictator has an incentive to use the aid to improve the welfare of his people. The reason is that if their welfare improves, and he can claim credit for this, loyalty or support for him will tend to increase. As a result, he can afford to relax repression, and still buy the same number of shoes for Imelda as before.
Now look at totalitarian regimes or tyrannies, defined as regimes whose rulers are uninterested in consumption, but in power. Should we aid them? Again, suppose that, as the result of either policy, economic growth improves. This gives the rulers an opportunity to accumulate more power, and since power is the only thing they care about, they take this opportunity, in the same way that a businessman who is already rich will grab an opportunity to make more money. So, for these regimes, aid which is untied to human rights observances is not merely wasted, but counterproductive — repression increases when the economy improves. This is what happened under Hitler and Stalin: the more popular they were, the more they took these opportunities to put the screws to all those elements of the population whose absolute loyalty was uncertain. In the same way, the enormous economic growth in China has resulted in not the slightest degree of relaxation in the level of repression there.
It might seem obvious that we would not aid these regimes, since the aid money would be spent on accumulating more power over the population, including repressing them. But, again, if the aid is tied to a human rights constraint, which becomes progressively more stringent over time, the policy will work in the right direction. If the economy improves as a result, support increases, and the rulers can afford to relax repression and still have the same level of power as before. The human rights observances constraint is absolutely necessary if this is to lead to a fall in repression and not an increase.
So we have a very simple guide — a single standard — to the policies which should be pursued by foreign governments interested in reducing repression. This is to make human rights observance the cornerstone of Western policy. Aid to any type of regime can be expected to produce beneficial effects provided it is accompanied by a long term human rights constraint, one which becomes progressively more stringent over time. Without the human rights standard, the effects of aid will be ineffectual or perverse.
3.2. Trade
Another policy dilemma is whether to trade with dictatorships. Trade policy is a bit more complicated than aid. We can distinguish the following effects:
1. Trade may be expected to increase national income of the target regime, as productivity there will rise due to the availability of imported inputs at a lower price, and the demand for the target’s exports increase. To the extent that the regime can successfully claim the credit for this improvement in welfare, loyalty to the regime may be expected to increase.
2. The rise in income will also increase tax revenues, giving the dictator more resources at his disposal. These may be used either for his own consumption, or to further his hold on power through increased expenditures on either repression or loyalty.
3. Since the richer people are, the more they tend to demand liberty, the increase in income tends to reduce loyalty to the dictatorship as people increasingly demand their rights (Bilson, 1982; Londregan and Poole, 1996). However, note that the estimated size of this effect is very small. Thus, as Londregan and Poole conclude their analysis of this effect in non-European countries, "Those expecting income growth to promote the development of democratic institutions must be very patient indeed" (pp. 22-23).
4. The increase in trade creates further links to foreign businesses and among domestic producers, possibly resulting in the development of independent power bases within the target regime. This is particularly likely when the trade is not organized through the central government (as it is in Cuba, for example). Thus, in China, regional governments in particular have built up substantial connections with outsiders and with the private sector, and are much more independent of the central government for revenue than they were before Deng launched his "social market" revolution. To the extent that this happens, loyalty to the regime may fall. On the other hand, it has been argued that trade between different types of civilizations actually increases mistrust, as the increased intensity of contacts simply breeds hostility. For example, World War I occurred at precisely the last peak of the "openness" of the international system. In that case, while there may be a short-run fall in loyalty due to the initial increase in contacts, in the longer run, further contacts simply breed nationalism and possibly increased support for the dictatorship in the target regime.
To disentangle the implications for policy, suppose first that the net effect of these changes, is, as seems likely, that support for the regime increases as the result of the trade agreement. Suppose also that the ruler is a tinpot. Then it can be argued that, with increased support, the tinpot will be himself motivated to relax repression (so that he can buy more shoes for his wife), and there is no need for a human rights constraint. But note that, even in this case, the human rights constraint does no harm; it simply asks the dictator to do what he would do in any case, and therefore it should be acceptable to him. On the other hand, if, on balance, loyalty to the regime were to decrease, the tinpot would want to raise repression in order to stay in office, and the human rights constraint is absolutely necessary for the trade agreement to lower, not raise, repression.
Suppose now that we are dealing with a totalitarian dictator. Again, if loyalty were to increase, on balance, as the result of the trade agreement, the dictator would tend to raise repression, and the binding human rights constraint is necessary to prevent a loss of freedom. The only case for a trade agreement with a totalitarian regime is where the opposite happens, and loyalty to the regime decreases from the trade agreement. In that case, repression falls as well. This is the only case where trade with a totalitarian regime makes sense. But note that, the totalitarian leader, in pursuing this trade agreement, cannot fail to be aware of the likely consequences of the trade agreement for his hold on power; namely, that his capacity for repression, the loyalty to him of the citizenry, and his power are all going to diminish as a result of his signing up. So, if this analysis were correct, it requires us to believe that the totalitarian is either unaware of, or deliberately acting contrary to, his own long run interest.15 It is noteworthy also that all the totalitarian regimes which have collapsed historically did so as the result of falling, not rising real income, and that the increase in income in China has resulted in not the slightest relaxation of repression there after almost two decades of reform and spectacular economic growth. The case for trade with totalitarian regimes, therefore is particularly weak.16
Finally, suppose that the human rights constraint cannot be implemented, either because the target regime is too powerful, or because no agreement can be reached among the countries involved in implementing the policy. Or alternatively suppose the dictator promises to abide by the human rights constraint and then reneges. Then there is a difficult choice between a policy of sanctions, on the one hand, and trade agreements with no effective human rights constraint, on the other. Of course, the actual choices are never this stark, and the actual policies followed will be a mixture of trade and sanctions, but the basic principle involved in the choice remains one of engagement or isolation. In that case, the analysis here implies that the least harm is likely to come from a trade agreement with a tin-pot regime, the most harm from trade with a totalitarian, with tyranny an intermediate case.
3.3. Sanctions
Historically, the most important alternative to a policy of aid to motivate dictatorships to behave is to use sanctions to punish those that do not. However, it is vital to realize that sanctions are not just the reverse of aid, and that policies like those pursued by the United States and the United Nations vis-a-vis regimes like Castro’s Cuba, Saddam Hussein’s Iraq or Milosevic’s Serbia may superficially resemble those described here, but in fact they work very differently. In all these cases, the U.S. or the UN imposed sanctions, and offered to lift them as a reward for better behavior. Such policies are not necessarily wrongheaded, but they do not work in the manner of those advocated here. The reason is that the sequence is reversed: the regime has to liberalize first, i.e., before the sanctions are lifted, trade allowed to resume and aid to flow. This means that the regime has no chance to use the benefit of aid or trade to build loyalty prior to liberalization, as with the policies advocated here. So the dictator who agrees to liberalize puts himself in immediate danger of being deposed, and it is no surprise that dictators like Castro, Hussein and Milosevic were all reluctant to do so.
Kaempfer et al. (2001) extend Wintrobe’s 1998 model of dictatorship and combine it with the public choice analysis of sanctions. They note that 85% of the time that sanctions are imposed they are imposed on a non-democratic regime. They point out that damaging economic sanctions can be counterproductive, undermining the political influence of the opposition. In the public choice approach, sanctions work through their impact on the relative power of interest groups in the target country. An important implication of this approach is that sanctions only work if there is a relatively well organized interest group whose political effectiveness can be enhanced as a consequence of the sanctions. For example, as the authors note, sanctions vs Iraq have had a devastating on the country but have been ineffective in destabilizing the Hussein regime. The reason, they argue is the fragmentation of the Iraqi opposition. At the other extreme, sanctions against South Africa were highly effective, because, in that case, there was a well organized opposition. The authors suggest that the effectiveness of the opposition is key to the effectiveness of sanctions and they try to show why this is true and to derive implications of this insight.
They also extend the model by adding two exogenous variables to it, s the impact of sanctions on the terms of trade; and q, the level of opposition; moreover q depends on s, qs > 0, and by making the price of repression PR (constant in Wintrobe’s model) a variable which depends on q and s (in addition to their other effects on the model) as well as on the country’s economic performance.
In their model sanctions have two main and opposing effects on the dictator’s budget: (1) the budget of the dictator rises through the appropriation of sanctions rents; (2) the budget falls due to the increase in opposition. There are two cases. In the first case, the opposition is significant enough that qs > 0. If, in addition, the second effect is large enough, the budget falls. If in addition, loyalty to the dictator falls due to the sanctions then sanctions are effective.
In the second case, there is no significant opposition. Then the net effect on the dictator’s budget of sanctions is that it rises due to the appropriation of sanctions rents. If in addition loyalty rises because those close to the dictator are happy about their increased capacity to appropriate these rents, then the sanctions are entirely counterproductive, and the budget of the regime, its power, and the level of repression all increase.
4. Conclusion
In recent years a small but now growing literature has looked at dictatorship from the point of view of public choice. While there is no consensus in the literature and it would be too soon to look for one, a number of ideas are attracting interest. The literature looks at (1) the objectives of dictators; (2) the constraints on their behavior; (3) their strategies for staying in office; (4) their incentives to provide public goods compared to that under democracy; (4) the economic efficiency of dictatorship compared to democracy; and (5) policy towards dictatorships. On the objectives of dictators, some models simply assume that dictators maximize lifetime consumption, as in standard economic models. However although this is certainly true of small scale dictators, it hardly fits many of the most important dictatorships like those of Hitler or Stalin and other models explicitly posit a taste for power or, in common with many models of democracy, ideological objectives. Since dictators are by definition monopolistic, the case for including such other objectives is particularly strong. Wintrobe looks at the strategies used by dictators to stay in office and emphasizes the Dictator’s Dilemma — the tradeoff between using repression and building support, noting that only the latter provides a firm foundation for autocratic rule. Recent contributions extend this framework to consider dynamic models of repression, the issue of genocide and the efficacy of sanctions against dictatorships.
The constraint on autocratic maximization is sometimes specified as the maximum revenue available. Other models specify the so-called "irony of absolutism" as the chief limit to a dictator’s power. Wintrobe reasons that as long as more power is available there are ways to extract more revenue from the private sector. Similarly, as long as more revenue is available, it is possible to accumulate more power: For these reasons in his model the equilibrium power and budget of the dictator are determined simultaneously.
The model of the dictator as "stationary bandit" originated by Mancur Olson shows that even a dictator has an incentive to provide public goods in order to raise revenue though arguing that this incentive is less under dictatorship than under democracy.
There is, as always, disagreement about the economic efficiency of dictatorship vs democracy but the disagreement appears to be narrowing. On both theoretical and empirical grounds there appears to be a consensus that high levels of repression are inimical to economic efficiency. Empirical work by Barro and others provides some (very slim, as acknowledged) evidence that growth rates are higher under dictatorship at low levels of repression though even this is challenged in a major empirical study by Przeworski and others. Theoretically, such a result can be explained by a reduction of rent seeking or redistributory pressures, as Barro does, or by the greater influence of producer groups under dictatorship, in which case the growth might come about at the expense of consumers or workers and need not signal greater efficiency. The "contest for power" framework, on the other hand, emphasizes that only democracy provides a mechanism for getting incompetent or corrupt rulers out and suggests that democracy might be more economically efficient than dictatorship when this factor is taken into account.
On policy the Wintrobe model provides a simple guide to the policies that should be pursued by foreign governments interested in reducing repression. This is to implent human rights observances, which become more stringent over time, as a condition for receiving aid or trade. One problem with this is that the dictator may promise to abide by human rights and then renege. The literature continues the standard skepticism of economists on the effectiveness of economic sanctions as a tool for getting dictators to lower repression.
