American Political Science Review Vol.87,No.3 September 1993 DICTATORSHIP,DEMOCRACY,AND DEVELOPMENT MANCUR OLSON University of Maryland nder anarchy,uncoordinated competitive theft by "roving bandits"destroys the incentive to invest and produce,leaving little for either the population or the bandits.Both can be better off if a bandit sets himself up as a dictator-a"stationary bandit"who monopolizes and rationalizes theft in the form of taxes.A secure autocrat has an encompassing interest in his domain that leads him to provide a peaceful order and other public goods that increase productivity.Whenever an autocrat expects a brief tenure,it pays him to confiscate those assets whose tax yield over his tenure is less than their total value.This incentive plus the inherent uncertainty of succession in dictatorships imply that autocracies will rarely have good economic performance for more than a generation.The conditions necessary for a lasting democracy are the same necessary for the security of property and contract rights that generates economic growth. n my student days,in reading Edward Banfield's order or to provide other public goods but receives (1958)account of the beliefs of the people in a only a share of the benefits.In a tiny group,such as Lpoor village in Southern Italy,I came upon a a hunter-gatherer band,each person or family will remarkable statement by a village monarchist.He obtain a significant share of the benefits of a peaceful said,"Monarchy is the best kind of government order,and the net advantages of such an order are so because the King is then owner of the country.Like great that even a single family's share of the gains can the owner of a house,when the wiring is wrong,he easily outweigh the sacrifices needed to obtain it. fixes it"(p.26).The villager's argument jarred against Moreover,when there are only a few,the welfare of my democratic convictions.I could not deny that the each noticeably depends on whether each of the owner of a country would have an incentive to make others acts in a group-oriented way.Thus each fam- his property productive.Could the germ of truth in ily,by making clear that cooperation by another will the monarchist's argument be reconciled with the bring forth its cooperation but that noncooperation case for democracy? will not,can increase the likelihood that another will It is only in recent years that I have arrived at an match its behavior,thereby increasing the incentive answer to this question.It turns out that for a each has to act in the group interest.The theoretical satisfactory answer one needs a new theory of dicta- prediction that sufficiently small groups can often torship and democracy and of how each of these organize for collective action is corroborated by types of government affects economic development. countless observations(Olson 1965). Once this new theory is understood,one can begin to This prediction is also in accord with the anthropo- see how autocracies and democracies first emerge.I logical observations of the most primitive societies. shall set out this conception in a brief and informal The simplest food-gathering and hunting societies way and use it to explain some of the most conspic- are normally made up of bands that have,including uous features of historical experience. the children,only about 50 or 100 people.In other The starting point for the theory is that no society words,such a band will normally contain only a few can work satisfactorily if it does not have a peaceful families that need to cooperate.Anthropologists find order and usually other public goods as well.Obvi- that primitive tribes normally maintain peace and ously,anarchic violence cannot be rational for a order by voluntary agreement,and that is to some society:the victims of violence and theft lose not only extent what Tacitus,Caesar,and other classical writ- what is taken from them but also the incentive to ers observed among the less advanced Germanic produce any goods that would be taken by others. tribes.The most primitive tribes tend to make all There is accordingly little or no production in the important collective decisions by consensus,and absence of a peaceful order.Thus there are colossal many of them do not even have chiefs.When a band gains from providing domestic tranquility and other becomes too large or disagreement is intense,the basic public goods.These gains can be shared in ways band may split,but the new bands normally also that leave everyone in a society better off.Can we make decisions by unanimous consent.If a tribe is in conclude that because everyone could gain from it,a the hunting-and-gathering stage,there is also little or peaceful order emerges by voluntary agreement? no incentive for anyone to subjugate another tribe or From the logic of the matter,we should expect that to keep slaves,since captives cannot generate enough in small groups a generally peaceful order will nor- surplus above subsistence to justify the costs of mally emerge by voluntary agreement but that in guarding them.Thus within the most primitive large populations it will not.The key to the matter is tribes of preagricultural history,the logical presump- that each individual bears the full costs or risks of tion that the great gains from a peaceful order can be anything he or she does to help establish a peaceful achieved by voluntary agreement appears to hold true. 567 This content downloaded from 202.120.14.67 on Sun,19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
American Political Science Review Vol. 87, No. 3 September 1993 DICTATORSHIP, DEMOCRACY, AND DEVELOPMENT MANCUR OLSON University of Maryland l Tnder anarchy, uncoordinated competitive theft by "roving bandits" destroys the incentive to fJ Zinvest and produce, leaving little for either the population or the bandits. Both can be better off if a bandit sets himself up as a dictator-a "stationary bandit" who monopolizes and rationalizes theft in the form of taxes. A secure autocrat has an encompassing interest in his domain that leads him to provide a peaceful order and other public goods that increase productivity. Whenever an autocrat expects a brief tenure, it pays him to confiscate those assets whose tax yield over his tenure is less than their total value. This incentive plus the inherent uncertainty of succession in dictatorships imply that autocracies will rarely have good economic performance for more than a generation. The conditions necessary for a lasting democracy are the same necessary for the security of property and contract rights that generates economic growth. In my student days, in reading Edward Banfield's (1958) account of the beliefs of the people in a poor village in Southern Italy, I came upon a remarkable statement by a village monarchist. He said, "Monarchy is the best kind of government because the King is then owner of the country. Like the owner of a house, when the wiring is wrong, he fixes it" (p. 26). The villager's argument jarred against my democratic convictions. I could not deny that the owner of a country would have an incentive to make his property productive. Could the germ of truth in the monarchist's argument be reconciled with the case for democracy? It is only in recent years that I have arrived at an answer to this question. It turns out that for a satisfactory answer one needs a new theory of dicta- torship and democracy and of how each of these types of government affects economic development. Once this new theory is understood, one can begin to see how autocracies and democracies first emerge. I shall set out this conception in a brief and informal way and use it to explain some of the most conspic- uous features of historical experience. The starting point for the theory is that no society can work satisfactorily if it does not have a peaceful order and usually other public goods as well. Obvi- ously, anarchic violence cannot be rational for a society: the victims of violence and theft lose not only what is taken from them but also the incentive to produce any goods that would be taken by others. There is accordingly little or no production in the absence of a peaceful order. Thus there are colossal gains from providing domestic tranquility and other basic public goods. These gains can be shared in ways that leave everyone in a society better off. Can we conclude that because everyone could gain from it, a peaceful order emerges by voluntary agreement? From the logic of the matter, we should expect that in small groups a generally peaceful order will nor- mally emerge by voluntary agreement but that in large populations it will not. The key to the matter is that each individual bears the full costs or risks of anything he or she does to help establish a peaceful order or to provide other public goods but receives only a share of the benefits. In a tiny group, such as a hunter-gatherer band, each person or family will obtain a significant share of the benefits of a peaceful order, and the net advantages of such an order are so great that even a single family's share of the gains can easily outweigh the sacrifices needed to obtain it. Moreover, when there are only a few, the welfare of each noticeably depends on whether each of the others acts in a group-oriented way. Thus each fam- ily, by making clear that cooperation by another will bring forth its cooperation but that noncooperation will not, can increase the likelihood that another will match its behavior, thereby increasing the incentive each has to act in the group interest. The theoretical prediction that sufficiently small groups can often organize for collective action is corroborated by countless observations (Olson 1965). This prediction is also in accord with the anthropo- logical observations of the most primitive societies. The simplest food-gathering and hunting societies are normally made up of bands that have, including the children, only about 50 or 100 people. In other words, such a band will normally contain only a few families that need to cooperate. Anthropologists find that primitive tribes normally maintain peace and order by voluntary agreement, and that is to some extent what Tacitus, Caesar, and other classical writ- ers observed among the less advanced Germanic tribes. The most primitive tribes tend to make all important collective decisions by consensus, and many of them do not even have chiefs. When a band becomes too large or disagreement is intense, the band may split, but the new bands normally also make decisions by unanimous consent. If a tribe is in the hunting-and-gathering stage, there is also little or no incentive for anyone to subjugate another tribe or to keep slaves, since captives cannot generate enough surplus above subsistence to justify the costs of guarding them.' Thus within the most primitive tribes of preagricultural history, the logical presump- tion that the great gains from a peaceful order can be achieved by voluntary agreement appears to hold true. 567 This content downloaded from 202.120.14.67 on Sun, 19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
Dictatorship,Democracy,and Development September 1993 Once peoples learned how to raise crops effec- bandit will take only a part of income in taxes, tively,production increased,population grew,and because he will be able to exact a larger total amount large populations needed governments.When there of income from his subjects if he leaves them with an is a large population,the same logic that shows why incentive to generate income that he can tax small groups can act consensually in their common If the stationary bandit successfully monopolizes interest,tells us that voluntary collective action cannot the theft in his domain,then his victims do not need obtain the gains from a peaceful order or other public to worry about theft by others.If he steals only goods,even when the aggregate net gains from the through regular taxation,then his subjects know that provision of basic public goods are large.2 The main they can keep whatever proportion of their output is reason is that the typical individual in a society with, left after they have paid their taxes.Since all of the say,a million people will get only about one-mil- settled bandit's victims are for him a source of tax lionth of the gain from a collective good,but will bear payments,he also has an incentive to prohibit the the whole cost of whatever he or she does to help murder or maiming of his subjects.With the rational provide it,and therefore has little or no incentive to monopolization of theft-in contrast to uncoordi- contribute to the provision of the collective good. nated competitive theft-the victims of the theft can There is by now a huge theoretical and empirical expect to retain whatever capital they accumulate out literature on this point,and the great preponderance of after-tax income and therefore also have an incen- of this literature agrees that,just as small groups can tive to save and to invest,thereby increasing future usually engage in spontaneous collective action,very income and tax receipts.The monopolization of theft large groups are not able to achieve collective goals and the protection of the tax-generating subjects through voluntary collective action. thereby eliminates anarchy.Since the warlord takes a Thus we should not be surprised that while there part of total production in the form of tax theft,it will have been lots of writings about the desirability of also pay him to provide other public goods whenever "social contracts"to obtain the benefits of law and the provision of these goods increases taxable income order,no one has ever found a large society that sufficiently. obtained a peaceful order or other public goods In a world of roving banditry there is little or no through an agreement among the individuals in the incentive for anyone to produce or accumulate any- society. thing that may be stolen and,thus,little for bandits to steal.Bandit rationality,accordingly,induces the bandit leader to seize a given domain,to make himself the ruler of that domain,and to provide a THE FIRST BLESSING OF THE peaceful order and other public goods for its inhab- INVISIBLE HAND itants,thereby obtaining more in tax theft than he could have obtained from migratory plunder.Thus Why,then,have most populous societies throughout we have"the first blessing of the invisible hand":the history normally avoided anarchy?An answer came rational,self-interested leader of a band of roving to me by chance when reading about a Chinese bandits is led,as though by an invisible hand,to warlord (see Sheridan 1966).In the 1920s,China was settle down,wear a crown,and replace anarchy with in large part under the control of various warlords. government.The gigantic increase in output that They were men who led some armed band with normally arises from the provision of a peaceful order which they conquered some territory and who then and other public goods gives the stationary bandit a appointed themselves lords of that territory.They far larger take than he could obtain without providing taxed the population heavily and pocketed much of government. the proceeds.The warlord Feng Yu-hsiang was noted Thus government for groups larger than tribes for the exceptional extent to which he used his army normally arises,not because of social contracts or for suppressing bandits and for his defeat of the voluntary transactions of any kind,but rather be- relatively substantial army of the roving bandit,White cause of rational self-interest among those who can Wolf.Apparently most people in Feng's domain organize the greatest capacity for violence.These found him much preferable to the roving bandits. violent entrepreneurs naturally do not call them- At first,this seems puzzling:Why should war- selves bandits but,on the contrary,give themselves lords,who were stationary bandits continuously steal- and their descendants exalted titles.They sometimes ing from a given group of victims,be preferred,by even claim to rule by divine right.Since history is those victims,to roving bandits who soon departed? written by the winners,the origins of ruling dynas- The warlords had no claim to legitimacy and their ties are,of course,conventionally explained in terms thefts were distinguished from those of roving ban- of lofty motives rather than by self-interest.Autocrats dits only because they took the form of continuing of all kinds usually claim that their subjects want taxation rather than occasional plunder. them to rule and thereby nourish the unhistorical In fact,if a roving bandit rationally settles down assumption that government arose out of some kind and takes his theft in the form of regular taxation and of voluntary choice.(These claims have an echo in at the same time maintains a monopoly on theft in his some literature in the"transactions costs"tradition domain,then those from whom he exacts taxes will that attempts to explain the emergence of various have an incentive to produce.The rational stationary kinds of governments partly or wholly through vol- 568 This content downloaded from 202.120.14.67 on Sun,19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
Dictatorship, Democracy, and Development September 1993 Once peoples learned how to raise crops effec- tively, production increased, population grew, and large populations needed governments. When there is a large population, the same logic that shows why small groups can act consensually in their common interest, tells us that voluntary collective action cannot obtain the gains from a peaceful order or other public goods, even when the aggregate net gains from the provision of basic public goods are large.2 The main reason is that the typical individual in a society with, say, a million people will get only about one-mil- lionth of the gain from a collective good, but will bear the whole cost of whatever he or she does to help provide it, and therefore has little or no incentive to contribute to the provision of the collective good. There is by now a huge theoretical and empirical literature on this point, and the great preponderance of this literature agrees that, just as small groups can usually engage in spontaneous collective action, very large groups are not able to achieve collective goals through voluntary collective action.3 Thus we should not be surprised that while there have been lots of writings about the desirability of "social contracts" to obtain the benefits of law and order, no one has ever found a large society that obtained a peaceful order or other public goods through an agreement among the individuals in the society. THE FIRST BLESSING OF THE INVISIBLE HAND Why, then, have most populous societies throughout history normally avoided anarchy? An answer came to me by chance when reading about a Chinese warlord (see Sheridan 1966). In the 1920s, China was in large part under the control of various warlords. They were men who led some armed band with which they conquered some territory and who then appointed themselves lords of that territory. They taxed the population heavily and pocketed much of the proceeds. The warlord Feng Yu-hsiang was noted for the exceptional extent to which he used his army for suppressing bandits and for his defeat of the relatively substantial army of the roving bandit, White Wolf. Apparently most people in Feng's domain found him much preferable to the roving bandits. At first, this seems puzzling: Why should war- lords, who were stationary bandits continuously steal- ing from a given group of victims, be preferred, by those victims, to roving bandits who soon departed? The warlords had no claim to legitimacy and their thefts were distinguished from those of roving ban- dits only because they took the form of continuing taxation rather than occasional plunder. In fact, if a roving bandit rationally settles down and takes his theft in the form of regular taxation and at the same time maintains a monopoly on theft in his domain, then those from whom he exacts taxes will have an incentive to produce. The rational stationary bandit will take only a part of income in taxes, because he will be able to exact a larger total amount of income from his subjects if he leaves them with an incentive to generate income that he can tax. If the stationary bandit successfully monopolizes the theft in his domain, then his-victims do not need to worry about theft by others. If he steals only through regular taxation, then his subjects know that they can keep whatever proportion of their output is left after they have paid their taxes. Since all of the settled bandit's victims are for him a source of tax payments, he also has an incentive to prohibit the murder or maiming of his subjects. With the rational monopolization of theft-in contrast to uncoordi- nated competitive theft-the victims of the theft can expect to retain whatever capital they accumulate out of after-tax income and therefore also have an incen- tive to save and to invest, thereby increasing future income and tax receipts. The monopolization of theft and the protection of the tax-generating subjects thereby eliminates anarchy. Since the warlord takes a part of total production in the form of tax theft, it will also pay him to provide other public goods whenever the provision of these goods increases taxable income sufficiently. In a world of roving banditry there is little or no incentive for anyone to produce or accumulate any- thing that may be stolen and, thus, little for bandits to steal. Bandit rationality, accordingly, induces the bandit leader to seize a given domain, to make himself the ruler of that domain, and to provide a peaceful order and other public goods for its inhab- itants, thereby obtaining more in tax theft than he could have obtained from migratory plunder. Thus we .have "the first blessing of the invisible hand": the rational, self-interested leader of a band of roving bandits is led, as though by an invisible hand, to settle down, wear a crown, and replace anarchy with government. The gigantic increase in output that normally arises from the provision of a peaceful order and other public goods gives the stationary bandit a far larger take than he could obtain without providing government. Thus government for groups larger than tribes normally arises, not because of social contracts or voluntary transactions of any kind, but rather be- cause of rational self-interest among those who can organize the greatest capacity for violence. These violent entrepreneurs naturally do not call them- selves bandits but, on the contrary, give themselves and their descendants exalted titles. They sometimes even claim to rule by divine right. Since history is written by the winners, the origins of ruling dynas- ties are, of course, conventionally explained in terms of lofty motives rather than by self-interest. Autocrats of all kinds usually claim that their subjects want them to rule and thereby nourish the unhistorical assumption that government arose out of some kind of voluntary choice. (These claims have an echo in some literature in the "transactions costs" tradition that attempts to explain the emergence of various kinds of governments partly or wholly through vol- 568 This content downloaded from 202.120.14.67 on Sun, 19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
American Political Science Review Vol.87,No.3 untary contracts and the costs of the transactions rent and to levy this monopoly charge on everything, associated with them.See Barzel 1991;Kiser and including human labor. Barzel 1991;North 1981;North and Thomas 1973.) In other words,the autocratic ruler has an incen- Any individual who has autocratic control over a tive to extract the maximum possible surplus from the country will provide public goods to that country whole society and to use it for his own purposes. because he has an"encompassing interest"in it.5 The Exactly the same rational self-interest that makes a extent of the encompassing interest of an office- roving bandit settle down and provide government holder,political party,interest group,monarch,or for his subjects also makes him extract the maximum any other partial or total "owner"'of a society varies possible amount from the society for himself.He will with the size of the stake in the society.The larger or use his monopoly of coercive power to obtain the more encompassing the stake an organization or maximum take in taxes and other exactions. individual has in a society,the greater the incentive The consumption of an autocratic ruler is,more- the organization or individual has,to take action to over,not limited by his personal capacities to use provide public goods for the society.If an autocrat food,shelter,or clothing.Though the pyramids,the received one-third of any increase in the income of palace of Versailles,the Taj Mahal,and even Imelda his domain in increased tax collections,he would Marcos's three thousand pairs of shoes were expen- then get one-third of the benefits of the public goods sive,the social costs of autocratic leaders arise mostly he provided.He would then have an incentive to out of their appetites for military power,international provide public goods up to the point where the prestige,and larger domains.It took a large propor- national income rose by the reciprocal of one-third,or tion of the total output of the Soviet Union,for three,from his last unit of public good expenditure. example,to satisfy the preferences of its dictators. Though the society's income and welfare would Some writers use the metaphor of the"predatory obviously be greater from a larger expenditure on state"but this is misleading,even for autocracies.As public goods,the gain to society from the public we saw earlier,a stationary bandit has an encompass- goods that a rational self-interested autocrat provides ing interest in the territory he controls and accord- are nonetheless often colossal.Consider,for exam- ingly provides domestic order and other public ple,the gains from replacing a violent anarchy with a goods.Thus he is not like the wolf that preys on the minimal degree of public order. elk,but more like the rancher who makes sure that From history,we know that the encompassing his cattle are protected and given water.The meta- interest of the tax-collecting autocrat permits a con- phor of predation obscures the great superiority of siderable development of civilization.From not long stationary banditry over anarchy and the advances of after the first development of settled agriculture until, civilization that have resulted from it.No metaphor say,about the time of the French Revolution,the or model of even the autocratic state can therefore be overwhelming majority of mankind was subject to correct unless it simultaneously takes account of the autocracy and tax theft.History until relatively recent stationary bandit's incentive to provide public goods times has been mostly a story of the gradual progress at the same time that he extracts the largest possible of civilization under stationary bandits interrupted by net surplus for himself. occasional episodes of roving banditry.From about Although the forms that stationary banditry has the time that Sargon's conquests created the empire taken over the course of history are diverse,the of Akkad until,say,the time of Louis XVI and essence of the matter can be seen by assuming that Voltaire,there was an impressive development of the autocrat gets all of his receipts in the form of civilization that occurred in large part under station- explicit taxation.The rational autocrat will devote ary banditry. some of the resources he obtains through taxation to public goods but will impose far higher tax rates than are needed to pay for the public goods since he also uses tax collections to maximize his net surplus.The THE GRASPING HAND higher the level of provision of public goods,given the tax rate,the higher the society's income and the We can now begin to reconcile the village monar- yield from this tax rate.At the same time,the higher chist's insight and the foregoing argument with the the tax rate,given the level of public-good provision, case for democracy.Though the village monarchist the lower the income of society,since taxes distort was right in saying that the absolute ruler has as incentives. much incentive to fix what needs repair as the owner So what tax rate and what level of public good of a house,his analogy is nonetheless profoundly provision will the rational self-interested autocrat misleading.The autocrat is not in a position analo- choose?Assume for the moment that the autocrat's gous to the owner of a single house or even to the level of public-good expenditure is given.As Joseph owner of all housing,but rather to the owner of all Schumpeter (1991)lucidly pointed out,and Ibn wealth,both tangible and human,in a country.The Kalduhn(1967)sensed much earlier,tax receipts will autocrat does indeed have an incentive to maintain (if we start with low taxation)increase as tax rates and increase the productivity of everything and ev- increase,but after the revenue-maximizing rate is eryone in his domain,and his subjects will gain from reached,higher tax rates distort incentives and re- this.But he also has an incentive to charge a monopoly duce income so much that tax collections fall.The 569 This content downloaded from 202.120.14.67 on Sun,19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
American Political Science Review Vol. 87, No. 3 untary contracts and the costs of the transactions associated with them. See Barzel 1991; Kiser and Barzel 1991; North 1981; North and Thomas 1973.)4 Any individual who has autocratic control over a country will provide public goods to that country because he has an "encompassing interest" in it.5 The extent of the encompassing interest of an office- holder, political party, interest group, monarch, or any other partial or total "owner" of a society varies with the size of the stake in the society. The larger or more encompassing the stake an organization or individual has in a society, the greater the incentive the organization or individual has to take action to provide public goods for the society. If an autocrat received one-third of any increase in the income of his domain in increased tax collections, he would then get one-third of the benefits of the public goods he provided. He would then have an incentive to provide public goods up to the point where the national income rose by the reciprocal of one-third, or three, from his last unit of public good expenditure. Though the society's income and welfare would obviously be greater from a larger expenditure on public goods, the gain to society from the public goods that a rational self-interested autocrat provides are nonetheless often colossal. Consider, for exam- ple, the gains from replacing a violent anarchy with a minimal degree of public order. From history, we know that the encompassing interest of the tax-collecting autocrat permits a con- siderable development of civilization. From not long after the first development of settled agriculture until, say, about the time of the French Revolution, the overwhelming majority of mankind was subject to autocracy and tax theft. History until relatively recent times has been mostly a story of the gradual progress of civilization under stationary bandits interrupted by occasional episodes of roving banditry. From about the time that Sargon's conquests created the empire of Akkad until, say, the time of Louis XVI and Voltaire, there was an impressive development of civilization that occurred in large part under station- ary banditry.6 THE GRASPING HAND We can now begin to reconcile the village monar- chist's insight and the foregoing argument with the case for democracy. Though the village monarchist was right in saying that the absolute ruler has as much incentive to fix what needs repair as the owner of a house, his analogy is nonetheless profoundly misleading. The autocrat is not in a position analo- gous to the owner of a single house or even to the owner of all housing, but rather to the owner of all wealth, both tangible and human, in a country. The autocrat does indeed have an incentive to maintain and increase the productivity of everything and ev- eryone in his domain, and his subjects will gain from this. But he also has an incentive to charge a monopoly rent and to levy this monopoly charge on everything, including human labor. In other words, the autocratic ruler has an incen- tive to extract the maximum possible surplus from the whole society and to use it for his own purposes. Exactly the same rational self-interest that makes a roving bandit settle down and provide government for his subjects also makes him extract the maximum possible amount from the society for himself. He will use his monopoly of coercive power to obtain the maximum take in taxes and other exactions. The consumption of an autocratic ruler is, more- over, not limited by his personal capacities to use food, shelter, or clothing. Though the pyramids, the palace of Versailles, the Taj Mahal, and even Imelda Marcos's three thousand pairs of shoes were expen- sive, the social costs of autocratic leaders arise mostly out of their appetites for military power, international prestige, and larger domains. It took a large propor- tion of the total output of the Soviet Union, for example, to satisfy the preferences of its dictators.7 Some writers use the metaphor of the "predatory state" but this is misleading, even for autocracies. As we saw earlier, a stationary bandit has an encompass- ing interest in the territory he controls and accord- ingly provides domestic order and other public goods. Thus he is not like the wolf that preys on the elk, but more like the rancher who makes sure that his cattle are protected and given water. The meta- phor of predation obscures the great superiority of stationary banditry over anarchy and the advances of civilization that have resulted from it. No metaphor or model of even the autocratic state can therefore be correct unless it simultaneously takes account of the stationary bandit's incentive to provide public goods at the same time that he extracts the largest possible net surplus for himself. Although the forms that stationary banditry has taken over the course of history are diverse, the essence of the matter can be seen by assuming that the autocrat gets all of his receipts in the form of explicit taxation. The rational autocrat will devote some of the resources he obtains through taxation to public goods but will impose far higher tax rates than are needed to pay for the public goods since he also uses tax collections to maximize his net surplus. The higher the level of provision of public goods, given the tax rate, the higher the society's income and the yield from this tax rate. At the same time, the higher the tax rate, given the level of public-good provision, the lower the income of society, since taxes distort incentives. So what tax rate and what level of public good provision will the rational self-interested autocrat choose? Assume for the moment that the autocrat's level of public-good expenditure is given. As Joseph Schumpeter (1991) lucidly pointed out, and Ibn Kalduhn (1967) sensed much earlier,8 tax receipts will (if we start with low taxation) increase as tax rates increase, but after the revenue-maximizing rate is reached, higher tax rates distort incentives and re- duce income so much that tax collections fall. The 569 This content downloaded from 202.120.14.67 on Sun, 19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
Dictatorship,Democracy,and Development September 1993 rational self-interested autocrat chooses the revenue- favorable to democracy if we assume that the incum- maximizing tax rate. bent party or president will maximize his chances of Though the amount collected at any tax rate will reelection simply by making the electorate as a whole vary with the level of public-good provision,the as well-off as possible. revenue-maximizing tax rate for the autocrat should A candidate needs only a majority to win,and he not.This optimal tax rate determines exactly how might be able to "buy"a majority by transferring encompassing the interest of the autocrat in the income from the population at large to a prospective society is;that is,it determines what share of any majority.The taxes needed for this transfer would increase in the national income he receives.He will impair incentives and reduce society's output just as then spend money on public goods up to the point an autocrat's redistribution to himself does.Would where his last dollar of expenditure on public goods this competition to buy votes generate as much generates a dollar's increase in his share of the na- distortion of incentives through taxation as a rational tional income.At this point,the gain to society will, autocracy does?That is,would a vote-buying demo- as we know,be the reciprocal of his share. cratic leader,like the rational autocrat,have an incen- Though the subjects of the autocrat are better off tive to push tax rates to the revenue-maximizing than they would be under anarchy,they must endure level? taxes or other impositions so high that,if they were No.Though both the majority and the autocrat increased further,income would fall by so much that have an encompassing interest in the society because even the autocrat,who absorbs only a portion of the they control tax collections,the majority in addition fall in income in the form of lower tax collections, earns a significant share of the market income of the would be worse off. society,and this gives it a more encompassing inter- There is no lack of historical examples in which est in the productivity of the society.The majority's autocrats for their own political and military purposes interest in its market earnings induces it to redistrib- collected as much revenue as they possibly could. ute less to itself than an autocrat redistributes to Consider the largest autocratic jurisdictions in West- himself.This is evident from considering an option ern history.The Bourbon kings of France were (es- that a democratic majority would have if it were at the pecially on the eve of the French Revolution)collect- revenue-maximizing tax rate.At the revenue-maxi- ing all they could in taxes.The Hapsburg kings of mizing tax rate,a minuscule change in the tax rates Spain did the same.The Roman Empire ultimately will not alter tax collections.A minuscule increase in pushed its tax rates at least to the revenue-maximiz- the tax rate will reduce the national income by ing level. enough so that even though a larger percentage of income is taken in taxes,the amount collected re- mains unchanged,and a tiny reduction in the tax rate will increase the national income so much that even THE REACH OF DICTATORSHIPS AND though a smaller percentage is taken in taxes,receipts DEMOCRACIES COMPARED are unchanged.This is the optimal tax rate for the autocrat because changes in the national income How would government by a rational self-interested affect his income only by changing tax collections. autocrat compare with a democracy?Democracies But a majority at the revenue-maximizing tax rate is vary so much that no one conclusion can cover all bound to increase its income from a reduction in tax cases.Nonetheless,many practical insights can be rates:when the national income goes up,it not only obtained by thinking first about one of the simplest like the autocrat,collects taxes on a larger national democratic situations.This is a situation in which income but also earns more income in the market.So there are two candidates for a presidency or two the optimal tax rate for it is bound to be lower than well-disciplined parties seeking to form the govern- the autocrat's.The easiest arithmetic example comes ment.This simplifying assumption will be favorable from supposing that the revenue-maximizing tax rate to democratic performance,for it gives the democracy is one-third and that the majority earns one-third of an "encompassing"interest rather like the one that the national income in the marketplace.The rational motivates the stationary bandit to provide some pub- autocrat will then find that the last dollar in taxes that lic goods.I shall make the opposite assumption later. he collects reduces the national income by three But throughout,I shall avoid giving democracy an dollars.One-third of this loss is his loss,so he just unfair advantage by assuming better motivation.I breaks even on this last dollar of tax collection and is shall impartially assume that the democratic political at his revenue-maximizing rate.But if a majority leaders are just as self-interested as the stationary mistakenly chose this same tax rate,it would be bandit and will use any expedient to obtain majority hurting itself,for it would lose two dollars(the same support. dollar lost by the autocrat plus one dollar of market Observation of two-party democracies tells us that income)from the last dollar it collected in taxes.Thus incumbents like to run on a "you-never-had-it-so- a majority would maximize its total income with a good"record.An incumbent obviously would not lower tax rate and a smaller redistribution to itself leave himself with such a record if,like the self- than would be chosen by an autocrat. interested autocrat,he took for himself the largest More generally,it pays a ruling interest(whether possible net surplus from the society.But we are too an autocrat,a majority,or any other)to stop redis- 570 This content downloaded from 202.120.14.67 on Sun,19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
Dictatorship, Democracy, and Development September 1993 rational self-interested autocrat chooses the revenue- maximizing tax rate. Though the amount collected at any tax rate will vary with the level of public-good provision, the revenue-maximizing tax rate for the autocrat should not. This optimal tax rate determines exactly how encompassing the interest of the autocrat in the society is; that is, it determines what share of any increase in the national income he receives. He will then spend money on public goods up to the point where his last dollar of expenditure on public goods generates a dollar's increase in his share of the na- tional income. At this point, the gain to society will, as we know, be the reciprocal of his share. Though the subjects of the autocrat are better off than they would be under anarchy, they must endure taxes or other impositions so high that, if they were increased further, income would fall by so much that even the autocrat, who absorbs only a portion of the fall in income in the form of lower tax collections, would be worse off. There is no lack of historical examples in which autocrats for their own political and military purposes collected as much revenue as they possibly could. Consider the largest autocratic jurisdictions in West- ern history. The Bourbon kings of France were (es- pecially on the eve of the French Revolution) collect- ing all they could in taxes. The Hapsburg kings of Spain did the same. The Roman Empire ultimately pushed its tax rates at least to the revenue-maximiz- ing level. THE REACH OF DICTATORSHIPS AND DEMOCRACIES COMPARED How would government by a rational self-interested autocrat compare with a democracy? Democracies vary so much that no one conclusion can cover all cases. Nonetheless, many practical insights can be obtained by thinking first about one of the simplest democratic situations. This is a situation in which there are two candidates for a presidency or two well-disciplined parties seeking to form the govern- ment. This simplifying assumption will be favorable to democratic performance, for it gives the democracy an "encompassing" interest rather like the one that motivates the stationary bandit to provide some pub- lic goods. I shall make the opposite assumption later. But throughout, I shall avoid giving democracy an unfair advantage by assuming better motivation. I shall impartially assume that the democratic political leaders are just as self-interested as the stationary bandit and will use any expedient to obtain majority support. Observation of two-party democracies tells us that incumbents like to run on a "you-never-had-it-so- good" record. An incumbent obviously would not leave himself with such a record if, like the self- interested autocrat, he took for himself the largest possible net surplus from the society. But we are too favorable to democracy if we assume that the incum- bent party or president will maximize his chances of reelection simply by making the electorate as a whole as well-off as possible. A candidate needs only a majority to win, and he might be able to "buy" a majority by transferring income from the population at large to a prospective majority. The taxes needed for this transfer would impair incentives and reduce society's output just as an autocrat's redistribution to himself does. Would this competition to buy votes generate as much distortion of incentives through taxation as a rational autocracy doer? That is, would a vote-buying demo- cratic leader, like the rational autocrat, have an incen- tive to push tax rates to the revenue-maximizing level? No. Though both the majority and the autocrat have an encompassing interest in the society because they control tax collections, the majority in addition earns a significant share of the market income of the society, and this gives it a more encompassing inter- est in the productivity of the society. The majority's interest in its market earnings induces it to redistrib- ute less to itself than an autocrat redistributes to himself. This is evident from considering an option that a democratic majority would have if it were at the revenue-maximizing tax rate. At the revenue-maxi- mizing tax rate, a minuscule change in the tax rates will not alter tax collections. A minuscule increase in the tax rate will reduce the national income by enough so that even though a larger percentage of income is taken in taxes, the amount collected re- mains unchanged, and a tiny reduction in the tax rate will increase the national income so much that even though a smaller percentage is taken in taxes, receipts are unchanged. This is the optimal tax rate for the autocrat because changes in the national income affect his income only by changing tax collections. But a majority at the revenue-maximizing tax rate is bound to increase its income from a reduction in tax rates: when the national income goes up, it not only, like the autocrat, collects taxes on a larger national income but also earns more income in the market. So the optimal tax rate for it is bound to be lower than the autocrat's. The easiest arithmetic example comes from supposing that the revenue-maximizing tax rate is one-third and that the majority earns one-third of the national income in the marketplace. The rational autocrat will then find that the last dollar in taxes that he collects reduces the national income by three dollars. One-third of this loss is his loss, so he just breaks even on this last dollar of tax collection and is at his revenue-maximizing rate. But if a majority mistakenly chose this same tax rate, it would be hurting itself, for it would lose two dollars (the same dollar lost by the autocrat plus one dollar of market income) from the last dollar it collected in taxes. Thus a majority would maximize its total income with a lower tax rate and a smaller redistribution to itself than would be chosen by an autocrat.9 More generally, it pays a ruling interest (whether an autocrat, a majority, or any other) to stop redis- 570 This content downloaded from 202.120.14.67 on Sun, 19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
American Political Science Review Vol.87,No.3 tributing income to itself when the national income To reach the maximum income attainable at a given falls by the reciprocal of the share of the national tax rate,a society will also need to enforce contracts, income it receives.If the revenue-maximizing tax rate such as contracts for long-term loans,impartially;but were one-half,an autocrat would stop increasing the full gains are again reaped only in the long run. taxes when the national income fell by two dollars To obtain the full advantage from long-run contracts from his last dollar of tax collection.A majority that, a country also needs a stable currency.A stationary say,earned three-fifths of the national income in the bandit will therefore reap the maximum harvest in market and found it optimal to take one-fifth of the taxes-and his subjects will get the largest gain from national income to transfer to itself would necessarily his encompassing interest in the productivity of his be reducing the national income by five-fourths,or domain-only if he is taking an indefinitely long view $1.25,from the last dollar that it redistributed to and only if his subjects have total confidence that itself.Thus the more encompassing an interest-the their"rights"to private property and to impartial larger the share of the national income it receives contract enforcement will be permanently respected taking all sources together-the less the social losses and that the coin or currency will retain its full value from its redistributions to itself.Conversely,the Now suppose that an autocrat is only concerned narrower the interest,the less it will take account of about getting through the next year.He will then the social costs of redistributions to itself. gain by expropriating any convenient capital asset This last consideration makes it clear why the whose tax yield over the year is less than its fotal assumption that the democracy is governed by an value.He will also gain from forgetting about the encompassing interest can lead to much-too-optimis- enforcement of long-term contracts,from repudiating tic predictions about many real-world democracies. his debts,and from coining or printing new money The small parties that often emerge under propor- that he can spend even though this ultimately brings tional representation,for example,may encompass inflation.At the limit,when an autocrat has no only a tiny percentage of a society and therefore may reason to consider the future output of the society at have little or no incentive to consider the social cost of all,his incentives are those of a roving bandit and that is what he becomes. 10 the steps they take on behalf of their narrow constit- uencies.The special interest groups that are the main To be sure,the rational autocrat will have an determinant of what government policies prevail in incentive,because of his interest in increasing the the particular areas of interest to those interest investment and trade of his subjects,to promise that groups have almost no incentive to consider the he will never confiscate wealth or repudiate assets. social costs of the redistributions they obtain.A But the promise of an autocrat is not enforceable by typical lobby in the United States,for example,rep- an independent judiciary or any other independent resents less than 1%of the income-earning capacity source of power,because autocratic power by defini- of the country.It follows from the reciprocal rule that tion implies that there cannot be any judges or other such a group has an incentive to stop arranging sources of power in the society that the autocrat further redistributions to its clients only when the cannot overrule.Because of this and the obvious social costs of the redistribution become at least a possibility that any dictator could,because of an hundred times as great as the amount they win in insecure hold on power or the absence of an heir, redistributional struggle(Olson 1982). take a short-term view,the promises of an autocrat It would therefore be wrong to conclude that are never completely credible.Thus the model of the democracies will necessarily redistribute less than rational self-interested autocrat I have offered is,in dictatorships.Their redistributions will,however,be fact,somewhat too sanguine about economic perfor- shared,often quite unequally,by the citizenry.Dem- mance under such autocrats because it implicitly ocratic political competition,even when it works very assumed that they have (and that their subjects badly,does not give the leader of the government the believe that they have)an indefinitely long planning incentive that an autocrat has to extract the maximum horizon. attainable social surplus from the society to achieve Many autocrats,at least at times,have had short his personal objectives. time horizons:the examples of confiscations,repudi- ated loans,debased coinages,and inflated currencies perpetrated by monarchs and dictators over the LONG LIVE THE KING course of history are almost beyond counting. Perhaps the most interesting evidence about the We know that an economy will generate its maximum importance of a monarch's time horizon comes from income only if there is a high rate of investment and the historical concern about the longevity of mon- that much of the return on long-term investments is archs and from the once-widespread belief in the received long after the investment is made.This social desirability of dynasties.There are many ways means that an autocrat who is taking a long view will to wish a king well;but the king's subjects,as the try to convince his subjects that their assets will be foregoing argument shows,have more reason to be permanently protected not only from theft by others sincere when they say "long live the king."If the but also from expropriation by the autocrat himself.If king anticipates and values dynastic succession,that his subjects fear expropriation,they will invest less, further lengthens the planning horizon and is good and in the long run his tax collections will be reduced. for his subjects. 571 This content downloaded from 202.120.14.67 on Sun,19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms
American Political Science Review Vol. 87, No. 3 tributing income to itself when the national income falls by the reciprocal of the share of the national income it receives. If the revenue-maximizing tax rate were one-half, an autocrat would stop increasing taxes when the national income fell by two dollars from his last dollar of tax collection. A majority that, say, earned three-fifths of the national income in the market and found it optimal to take one-fifth of the national income to transfer to itself would necessarily be reducing the national income by five-fourths, or $1.25, from the last dollar that it redistributed to itself. Thus the more encompassing an interest-the larger the share of the national income it receives taking all sources together-the less the social losses from its redistributions to itself. Conversely, the narrower the interest, the less it will take account of the social costs of redistributions to itself. This last consideration makes it clear why the assumption that the democracy is governed by an encompassing interest can lead to much-too-optimis- tic predictions about many real-world democracies. The small parties that often emerge under propor- tional representation, for example, may encompass only a tiny percentage of a society and therefore may have little or no incentive to consider the social cost of the steps they take on behalf of their narrow constit- uencies. The special interest groups that are the main determinant of what government policies prevail in the particular areas of interest to those interest groups have almost no incentive to consider the social costs of the redistributions they obtain. A typical lobby in the United States, for example, rep- resents less than 1% of the income-earning capacity of the country. It follows from the reciprocal rule that such a group has an incentive to stop arranging further redistributions to its clients only when the social costs of the redistribution become at least a hundred times as great as the amount they win in redistributional struggle (Olson 1982). It would therefore be wrong to conclude that democracies will necessarily redistribute less than dictatorships. Their redistributions will, however, be shared, often quite unequally, by the citizenry. Dem- ocratic political competition, even when it works very badly, does not give the leader of the government the incentive that an autocrat has to extract the maximum attainable social surplus from the society to achieve his personal objectives. LONG LIVE THE KING We know that an economy will generate its maximum income only if there is a high rate of investment and that much of the return on long-term investments is received long after the investment is made. This means that an autocrat who is taking a long view will try to convince his subjects that their assets will be permanently protected not only from theft by others but also from expropriation by the autocrat himself. If his subjects fear expropriation, they will invest less, and in the long run his tax collections will be reduced. To reach the maximum income attainable at a given tax rate, a society will also need to enforce contracts, such as contracts for long-term loans, impartially; but the full gains are again reaped only in the long run. To obtain the full advantage from long-run contracts a country also needs a stable currency. A stationary bandit will therefore reap the maximum harvest in taxes-and his subjects will get the largest gain from his encompassing interest in the productivity of his domain-only if he is taking an indefinitely long view and only if his subjects have total confidence that their "rights" to private property and to impartial contract enforcement will be permanently respected and that the coin or currency will retain its full value. Now suppose that an autocrat is only concerned about getting through the next year. He will then gain by expropriating any convenient capital asset whose tax yield over the year is less than its total value. He will also gain from forgetting about the enforcement of long-term contracts, from repudiating his debts, and from coining or printing new money that he can spend even though this ultimately brings inflation. At the limit, when an autocrat has no reason to consider the future output of the society at all, his incentives are those of a roving bandit and that is what he becomes.10 To be sure, the rational autocrat will have an incentive, because of his interest in increasing the investment and trade of his subjects, to promise that he will never confiscate wealth or repudiate assets. But the promise of an autocrat is not enforceable by an independent judiciary or any other independent source of power, because autocratic power by defini- tion implies that there cannot be any judges or other sources of power in the society that the autocrat cannot overrule. Because of this and the obvious possibility that any dictator could, because of an insecure hold on power or the absence of an heir, take a short-term view, the promises of an autocrat are never completely credible. Thus the model of the rational self-interested autocrat I have offered is, in fact, somewhat too sanguine about economic perfor- mance under such autocrats because it implicitly assumed that they have (and that their subjects believe that they have) an indefinitely long planning horizon. Many autocrats, at least at times, have had short time horizons: the examples of confiscations, repudi- ated loans, debased coinages, and inflated currencies perpetrated by monarchs and dictators over the course of history are almost beyond counting. Perhaps the most interesting evidence about the importance of a monarch's time horizon comes from the historical concern about the longevity of mon- archs and from the once-widespread belief in the social desirability of dynasties. There are many ways to wish a king well; but the king's subjects, as the foregoing argument shows, have more reason to be sincere when they say "long live the king." If the king anticipates and values dynastic succession, that further lengthens the planning horizon and is good for his subjects. 571 This content downloaded from 202.120.14.67 on Sun, 19 Feb 2017 15:11:57 UTC All use subject to http://about.jstor.org/terms