Democracy as insurance

Public Choice 87: 319-345, 1996. (~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.

319

D e m o c r a c y as i n s u r a n c e *

EINAR OVERBYE Institute of Applied Social Research, Munthes Gate 31, 0260, Oslo, Norway Accepted 6 September 1995 Abstract. Explaining the existence and stability of democracies pose difficulties for the Public Choice-paradigm. The article argues that these difficulties may be resolved by focusing on the insurance-aspect of democratic decision-making procedures. Democratic regimes may be better able than autocratic regimes to provide the inhabitants with insurance against the risk of being subjected to arbitrary suppression from the ruling elite. This propensity of democratic regimes may explain the origin, as well as stability, of such regimes.

1. Introduction

Public Choice has been developed as a theory about democratic decisionmaking procedures. This is evident in Downs' (1957) original work, which already in the title ("An economic theory of democracy") explicitly narrowed the focus to a study of politics confined to democratic regimes. The focus on democratic modes of decision-making has prevailed in the public choiceliterature. Public Choice has only occasionally ventured beyond democracies towards an analysis of the policy-making process in other types of political regimes (Mueller, 1989: 272; Tullock, 1987). This restricted focus on policy processes within democracies has implied a somewhat stepmotherly treatment of at least two problems. First, how to explain why democracies exist in the first place, rather than only autocratic systems, which have a much longer history than democracies, and apparently are not to the same extent vexed with destabilizing propensities such as cyclic preferences (Mueller, 1989: 388). Second, how to explain why this mode of government has replaced autocratic decision-making arrangements in an increasing number of countries at the wane of the 20th century (Diamond, 1992; Fukuyama, 1992: 109ft.). The latter problem is linked to a more general weakness of public choice-theory: how to move from a static towards a dynamic interpretation of politics; how to move from an analysis of a particular (game-theoretic) situation towards an analysis of changes in the situation. * Thanks to Bard-Anders Andreassen, Efling Barth and an anonymous referee for valuable comments to an earlier draft.

320 The purpose of this article is to provide an answer to the classic question of

why "democracies" do exist based on the economic man-assumption of the Public Choice-paradigm, while at the same time suggesting a way to incorporate a transition-to-democracy theory within a Public Choice framework.

2. A closer presentation of the puzzle Public Choice assumes that political behaviour can be analyzed by utilizing the "economic man"-assumption underlying neoclassical economic theory. However, if one maintains this basic assumption, the fact that political actors (parties, interest organizations etc.) usually respect the democratic rules of the game demands an explanation. Downs writes: Our model attempts to forge a positive relationship between individual and social end structures by means of a political device. Because each adult citizen has one vote, his welfare preferences are weighted in the eyes of his government ... we admit openly that we are adopting an ethical principle - equality of franchise. (Downs, 1957: 18; emphasis added) Further, Downs includes in his definition of "democracy" the following assumptions about elite behaviour: The losing parties in an election never try by force or any other illegal means to prevent the winning party (or parties) from taking office ... The party in power never attempts to restrict the political activities of any citizen or other parties as long as they make no attempt to overthrow the government by force. (Downs, ibid. 24) Downs is silent as to why voters or elites usually adhere to the "one-manone-vote" rule, and he is also silent as to why elites often abstain from using force or other illegal means in their attempts to maintain or conquer power. Assuming that both voters and elites are essentially rational and self-regarding individuals, this self-restraint on behalf of elites, as well as why voters do not try to deny the vote to citizens with different preferences, appearpuzzling (in the Kuhnian sense). Why were democratic decision-making procedures introduced in the first place, and why are they often sustained during long spells of time? Why are not the rules of democratic political games inherently unstable (and prone to immediate breakdown), due to pressures from everchanging coalitions of voters and competing elites, bent on breaking the rules to suit their particular interests? One solution is to assume that ethical (normative) preferences underlie the acceptance of the rules of the game; rational, self-regarding preferences only

321 determine how the actors behave within an established legal (democratic) framework. Downs (above quotes) hints in this direction. However, critics of the Public Choice-paradigm may legitimately argue that if it is necessary to introduce ethical (normative) preferences to explain why democracies exist in the first place, then there is no logical reason for assuming that they do not play a role as a guide to behaviour also within a democratic context. The difficulty in explaining the existence of democratic decision-making procedures, based firmly on an economic-man premise, thus emerges as a potential crack in the basement of the Public Choice-paradigm as such. The purpose of this article is to offer an explanation-based on the economic man-premise - as to why democracies do exist, and why they often are fairly stable across time. Very briefly stated, I shall argue that democracies exist because they satisfy a (self-oriented) demand for insurance against certain types of risks; implying that autocratic systems have difficulties in providing such insurance to a similar extent as democratic regimes. By anchoring the existence of democracies in a rational and self-regarding demand for insurance, it is not necessary to seek recourse to more-or-less ad hoc assumptions about "norms" or "ethical behaviour" in order to explain why voters, as well as elites, often respect the rules of the democratic game. Further, I shall argue that the demand for this type of insurance (as well as the technical ability to satisfy it) may well have risen in conjunction with the general industrialization and urbanization process; implying that the insurance-approach may offer an explanation as to the 20th century surge in democratic regimes. I would like to point out from the very outset, however, that I do not want to offer any opinion as to whether or not the existence and spread of democracies is a good or bad thing. I simply want to offer a solution to a potentially detrimental puzzle within the Public Choice-paradigm, i.e. to offer an answer to the question "democracies do exist - how are they possible?".

3. Defining democracy "Democracy" is a vague and value-laden word. For my purposes, it suffices to follow Schumpeter (1942: 271) in defining "democracy" as a political system in which various elites aspire for leadership through a peaceful competition for the people's vote, by way of secret ballot at more-or-less regular intervals. As far as the size of the electorate is concerned, I shall make a distinction between electorial systems which include all grown-up citizens in the electorate and those which, to a varying extent, exclude certain subgroups (e.g. slaves, foreigners, women or the propertyless minors - cf. Elster, 1988: 12). Following Tullock (1987: 4-5), I shall reserve the term "democracy" for governments in which the entire sane and non-criminal adult population can

322 vote, and use the term "electorial system" for regimes with a more restricted electorate. Thus according to this definition democracies are fairly recent phenomena, although the existence of electorial systems can be traced back at least to the Greek city states. Notice the minimalist characteristics of the above definition. By "democracy" I do not connote that the population has become more actively involved in politics (which is at best a dubious claim), nor that a system would be more democratic if they did. Further, most democratic systems contain rules which may limit competition (e.g. by favouring big parties rather than small parties), or enhance the views of some elites rather than others (e.g. the existence of constitutions demanding various types of qualified majorities in order to restrict certain rights, such as the right to own property). I do not label a political system as more-or-less democratic dependent on how strong biases that are built into the rules of the game, although such biases may serve as a fruitful basis for a categorization of different types of democracies. My sole purpose is to explain the existence and spread of various types of democracies during the last century (systems based on competition for votes, but in which this competition may be constrained and/or biased in various ways), thus I need a definition which may serve as a demarcation criterion rather than as a Weberian ideal-type description.

4. Defining the alternative Given the above definition of democracy, we need a definition of the alternative, which I shall label an autocratic system. An "autocracy" is a political system in which one elite monopolizes the decision-making process; implying that the elite recruits itself. The defining feature of an autocracy is that there is no small or large real probability that the prevailing elite may lose power through a secret ballot: decision-making power is monopolized rather than subjected to (peaceful) competition. Unlike a democracy, there is no institutionalized uncertainty as to which elite(s) that are going to be in charge in the future (Przeworski, 1986: 58). We may draw a distinction between autocracies in which one person holds absolute power (dictatorships), and autocracies ruled by a larger group of individuals (e.g. a Junta). If this group is large, the autocratic sysem may border on an electorial system with a limited electorate. 1 Thus dictatorships and full-blown democracies may be presented as two poles along a continuum, the position along the continuum being a function of the size of the group which may have a say in the making of political decisions. Although the grey area is important both in an historical and contemporary perspective, I shall mainly focus on clear-cut cases of autocracies and democracies.

323

5. The demand for democracy As a starting point (to be relaxed later), let us assume that the political game is essentially demand-driven, implying that the preferences of the citizens will - at least in the very long run - determine policy outcomes. Given this assumption, the question becomes if there - from a citizen point of view is something to be gained in a democracy which an autocracy is unable to supply. There may indeed be something to be gained in a democracy which an autocracy is hard put providing (or rather: something which an autocracy has difficulty providing to a similar extent). Democracies may be better able to provide the inhabitants with a certain insurance product: namely, insurance against the risk of being subjected to arbitrary suppression from the ruling elite. This includes insurance against the risk of arbitrary detention and/or execution, as well as arbitrary confiscation of property. 2 By providing its citizens with this type of insurance, a democracy may simultaneously be better able to offer its citizens a predictable legal environment in which to live their everyday lives (Dahl, 1971: 28). Provided that the inhabitants in a country want to minimize the above type of risks, they should get hold of rulers who abstain from arbitrary acts of suppression. Even if arbitrary suppression is only directed towards a limited number of individuals, it will - if it is truly arbitrarily applied - lead to an increased level of unpredictable risk across the whole electorate. In a democracy, the ruling elite has a self-interest not to neglect the security interests of large segments of the electorate, as this may weaken their popularity and this reduce their reelection chances. An autocratic ruler has less reason to show restraint in the execution of his power, since he faces less risk of being dethroned if he should attempt to satisfy his personal preferences at the expense of his subjects) For this reason, installing a democracy may be an effective institutional means citizens may employ in order to enhance the probability that their rulers wil not arbitrarily suppress them.

6. The stability of democratic regimes Democracies are riddled with cyclic preferences and other destabilizing propensities. Losers in a redistributional game may offer only marginal improvements to some of the winners to establish a new winning coalition (Tullock, 1981, 1983: 82; Shepsle and Weingast, 1981; Ledyard, 1984: 7-8; Brennan and Lomasky, 1993: 81-83). This holds true not only with regard to political games taking place within a certain institutional setting, but also with regard to the games intended to determine the rules of the game

324 itself (which may be labelled meta-games or constitutional games). If political games (including meta-games) are interpreted only with respect to their redistributional propensities, the apparent stability of democratic institutions appear puzzling. Redistributional games are usually games with empty cores; i.e. they have no equilibrium solutions at all. Provided that no player (or set of players) has the ultimate (dictatorial) authority to fix the rules and/or suppress further coalition-building, there should be endless cycles as regards the choice of rules (Shepsle and Weingast, 1981: 516). Thus why do not democratic decision-making procedures break down almost immediately? Why are not the rules of the political game constantly in flux, reflecting ever-changing coalitions of voters and competing elites bent on breaking existing rules to suit their self-oriented interests? The problem of explaining the apparent stability of political arrangements (including constitutional arrangements) is a general puzzle facing the Public Choice-paradigm, and the literature dealing with this puzzle is enormous (see Mueller, 1989:89 for an overview). I shall not discuss the merits or shortcomings of the various solutions that have been put forward. Rather, I shall argue that if we focus on the insurance aspect of democratic decision-making procedures, rather than on the redistributionalaspects of these rules, the existence of stable democracies may have a straightforward explanation. In the simplest version, the argument may run as follows. Democracies have one advantage to autocracies: they are better able to insure the citizens against the risk of arbitrary suppression. If we may assume that autocracies have no advantages to democracies, this implies that all citizens - even the most risk-willing - should prefer a democracy to an autocracy. For this reason, democratic institutions should usually enjoy the stable support of an overwhelming majority, implying that the insurance-aspect of democratic decision-making procedures may explain why breakdown of democracies happens less often than one should otherwise assume (i.e. they do not break down immediately, and all the time). Admittedly, the argument that autocracies have no advantages to democracies may not be true. At least "enlightened" autocracies may have advantages to democracies, e.g. with regard to furthering economic efficiency. However, even if we (for the sake of argument) accept this possibility, citizens may prefer a democracy to an autocracy; partly because it may be difficult to decide when the autocracy is enlightened and when it is not, and partly because the citizens are not among those who are asked to decide if the autocracy is to be regarded as enlightened or not. To present a formal treatment of the above argument, let U(x~a) denote a citizen's utility from living in an enlightened autocracy, U(xd) the utility of

325 living in a democracy and U(x~a) the utility of living in an unenlightened autocracy. First assume that

U(xea) > U(xd) > U(x o)

(1)

Each citizen ascribes the probability p to the possibility that the autocracy will be enlightened, and the probability (l-p) that it will not be enlightened. Assuming that citizens are risk-neutral, they should prefer a democracy to an autocracy if

pU(xea ) + (1

-

p)U(Xua) < U(Xd)

(2)

Equation (2) obviously presents a very simplified version of the decision problem facing the citizenry. 4 Nonetheless, this simple model suggests that at least two factors are crucial in detecting variation in the stability of democratic regimes across countries, or across time. First, democracies should be more stable if they are not perceived by large segments of the population as less (or much less) efficient than enlightened autocracies; second, democracies should be more stable if the citizens believe that an autocracy will probably not be (or stay) enlightened. 5 There may be situations (e.g. during severe economic depressions) in which an autocracy may seem as a superior system in bringing about order and security. In such situations, a window of opportunity may open for a skilled, non-democratic political entrepreneur (elite) to conquer power. I shall elaborate on this possibility at the end of the article. At this stage, however, let us put a parenthesis around the possibility of a demand-driven democratic breakdown, and first subject the insurance-argument to closer scrutiny.

7. The supply of democracy So far, I have depicted the existence of democracies as resulting from demand characteristics in the population. This is obviously an oversimplification. In practice, supply side characteristics, i.e. the historically derived constellations of various elites, probably have an independent influence on the choice of political system. Rather than to focus exclusively on the demand side, one should focus on the interaction between supply and demand characteristics. More specifically, one should focus on the interaction between subjects/citizens and their aspiring rulers/elites. This relationship may be portrayed as a principal-agenttype of relationship (Banks, 1991: 58). The subjects can be portrayed as the principals (the players with the coarser information positions). The elites are the agents (the players with the finer information positions) (Rasmussen, 1991: 134).

326 The subjects, in their capacity of principals, lack information as to the trustworthiness of an aspiring ruler, i.e. whether or not he will abide to the rule of law and thus secure them a predictable legal environment. They thus face an adverse selection problem in choosing their elites. As principals, the citizens should try to screen the agents (elites) in order to detect which are the most likely to stay committed to a rule of law should they acquire office. The elites (agents) on their side should try to send messages and signals aimed at presenting themselves a's willing to adhere to the rule of law, as well as to the democratic rules of the game, in their competition for office. The elites may thus be engaged in a "verbal arms race", trying to come across as precommitted to the rules of the democratic game, as well as to the rule of law, in order to enhance their election chances.

8. Will not the rule of law suffice? At this stage, a critic m a y well remark that autocratic rulers have on several occasions also been able to secure a predictable legal environment for their subjects. Historically, the rule or law (Rechtsstaat) preceded democratic decision-making procedures in many countries. By introducing a rule of law, an autocratic ruler should be able to take the "wind out of the sail" of any insurance-based demand for democracy. Nonetheless, autocratic regimes based on a rule of law have later been replaced by full-blown democracies in many countries. This appears to contradict an insurance-based theory of democracy. In answering this objection, it should first be pointed out that it might indeed often be in the "enlightened self-interest" of an autocratic ruler to install a rule of law. This is the essence of Machiavelli's [1514] (1983: 95-97) famous advice that it is better for an autocratic ruler to be feared than loved if he cannot be both, but that above all he must make sure he is not hated; and he avoids being hated precisely by abstaining from arbitrary suppression of his subjects. As long as the subjects are able to predict which actions that may lead to their suppression (say, engaging in a conspiracy), and which actions that will leave them in peace (such as adopting a distinterested attitude towards politics), an autocratic ruler may minimize opposition to his regime, and thus dampen any insurance-based demand for democracy. The point is, however, that an autocratic ruler has a problem assuring his subjects that he will always abide to the rule of law to a similar extent as an elected elite; the reason being precisely that, since the rules are essentially self-imposed, he may change the rules of the political game without anyone being able to - or daring- to object. That is: the regime may slide from being "enlightened" towards the opposite without anyone being able to interrupt the slide. Thus an autocratic ruler will

327 seldom if ever be able to portray himself as self-bound to the same extent as a democratically elected elite. The latter elite may more easily convince its subjects that is will continue to adhere to the rule of law, since the members of the elite know that if they should lose the trust of the electorate in this matter, they may jeopardize their position in future elections. Again, the principalagent propensities of the citizen-ruler relationship is at the core of the matter. An autocratic ruler (agent) is less able to signal commitment to a rule of law, and/or to appear self-bound in the eyes of his subjects (principals), as compared to an elected elite. The citizens should prefer a type of regime (i.e. a democratic regime) where screening and monitoring of the elites (agents) are most easily accomplished. Ceteris paribus, then, a democracy provides a less risky institutional setting for the unfolding of the political game as seen from the point of view of the citizens; a democratic system pose less risk that the legal environment may one day be changed in an unpredictable/arbitrary manner. In the long run, and not least since enlightened autocrats may be succeeded by not-so-enlightened autocrats, the population should go through a learning process leading in the direction of preferring elected elites. 6

9. The decision problem as seen from the point of view of an elite

Up till now, I have analyzed the decision problem mainly from the point of view of the citizenry, assuming that their primary interest is to be ruled by leaders who will secure them a predictable legal environment by abstaining from arbitrary suppression. However, an elite may have a different set of preferences. Ceteris paribus, one should expect an elite to prefer an unconstrained legal setting in which to pursue its own objectives (whichever those may be), and also an interest in limiting or abolishing future competition (elections) once it has acquired power. Provided that the competition for power takes place within a democratic setting, however, an elite which claims that it wants unconstrained power and plans to abandon elections once it is in charge, will usually be defeated; the reason being precisely that this elite pays scant attention to the insurance-interests of the electorate. Provided that an elite prefers some power to no power, it will then try to come across as precommitted to respect the existing rules of the game, in order to enhance its election chances. To present a formal version of the above argument, let us (for the sake of argument) assume that all elites would ideally prefer to win the election without having to respect the existing rules of the game (a). The secondbest option is to commit and then win (b). Not committing but losing is the third-best outcome (c), while the worst possible outcome is to commit but

328 lose anyway (d). If the elite does not commit itself, it faces the probability q of being elected and (1 - q) of losing. If it does commit itself, it faces the probability p of winning and (1 - p) of losing. Since voters prefer committed elites, p > q. An elite will commit to the existing rules of the game if:

p b + ( 1 - p ) d > qa+(1-q)c

(3)

The above equation implies that, as long as the elites believe that the citizens prefer committed elites (i.e. thatp > q), they may regard it in their self-interest to come across as willing to constrain their own power. 7

10. Explaining the existence of constitutions The above line of thought may explain why democratic countries have often set up constitutions which preserve certain "core rules of the game". Constitutions put limits on majority decisions; they constrain the available choices of action a future majority may pursue through parliamentary decisions (Elster, 1988: 3). The problem of justifying why voters and/or politicians at one particular point in time should be allowed to restrict the scope of actions open for future politicians (and/or future majorities of voters) has been a persistent question in normative political theory (Holmes, 1993: 195ff.). The principalagent approach offers the promise of explaining the existence of constitutions without having to assume some normative (as opposed to self-interested) motivational base underlying the willingness of politicians to "tie their own hands". A principal-agent type of explanation may go as follows: from the point of view of the electorate (principals), there is always a potential risk that the elite which succeeds in conquering power may be tempted to throw aside earlier promises and install an autocracy. This risk is larger in countries without constitutions than in countries with constitutions. Competing elites, sensing that the electorate are in demand of a predictable legal environment (in order to minimize the risk that they will be subjected to arbitrary suppression), should then want to appear committed to the existing rules of the game. Those elites which are most convincing in coming across in the eyes of their voters as precommitted not to exploit a future victory, will extract the largest amount of trust in the electorate, and may thus have an advantage in the competition for votes. The point is that a stated willingness on behalf of an elite to support the introduction of a constitution (or to adhere to a constitution already in place), is one of the most convincing ways the members of an elite may communicate an intent not to yield to any temptation to change the rules of the game if they should acquire power. This explanation of the existence of constitutions can be straightforwardly deduced from rational, self-oriented considerations on behalf of voters as well as competing elites.

329 The above explanation may be contrasted to Holmes' Schelling-inspired argument along a somewhat similar line. Holmes argues that constitutions may be regarded as some sort of self-incapacitation devices. He writes: We put the alarm clock across the room so we cannot turn if off without getting out of bed ... An individual can be crafty and strong-willed enough to bind his future selves even if he is too dim and weak-willed to act as he would prefer without precommitments. To achieve his desired ends despite his feeble resolve, an individual must restrict his available options ... It is tempting to think of constitutionalism in these terms, as a cluster of techniques for collective self-management (1993: 236). This explanation, although inspired by rational choice-theory, rests on an assumption that voters have rather schizophrenic selves (while at the same time being aware of, and able to maintain a rational attitude towards, their own schizophrenia). Although one should not rule out the possibility that such voters may exist, it seems far-fetched to assume that a majority of voters should be so distrustful of their own selves as to support constitutions in order to tie their own hands. My solution avoids the "Peter sober/Peter drunk" allegation by framing the issue not as a question of the voters wanting to tie their own hands, but to tie the hands of their politicians.

11. Industrialization and the transition to democracy So far, I have sketched a solution as to why democracies may enjoy stable majority support, and why ruling elites may adhere to constitutions, even though such institutional arrangements constrain their choice of actions while in power. As long as a majority of voters (the median voter) regard their insurance demands as more effectively met in a democracy than in an autocracy, and as long as autocracies do not in any other respect appear to perform better than democracies, the game is in equilibrium. However, since democracies based on mass suffrage are fairly recent phenomena, one is lead to ask why the earlier non-democratic systems broke down in the first place. How can the 20th century trend towards democratic decision-making procedures be explained? 8 In the following, I shall suggest an avenue for incorporating a transitionto-democracy theory within the Public Choice-paradigm. I shall argue that changes in the rules of the game (the meta-game) are brought about by social changes which change the political risk structure: partly by enhancing the repressive potential of central elites vis-h-vis their subjects, and partly by changing the balance of power between various types of elites. More

330 specifically, I shall argue that those changes which may conveniently be summed up as "increased industrialization and urbanization" have changed the political risk various subgroups of the population are exposed to, which in its turn has led them (and the elites which try to appeal to their interests in order to elevate themselves to political office) to push for changes in the rules of the political game, in an attempt to counter their increased exposure to such risk. As a starting point consider the argument that, at least in Europe, democracies replaced feudal power structures. Feudal power structures are characterized by a weak central authority, implying that decision-making power within a certain territory is in reality to a large extent shared between various elites (Moore, 1966: 415ff.; Anderson and Hall, 1986: 25ff.). Feudal power structures can thus be perceived as a third type of regime, alongside both democracies and autocracies (Tullock, 1987: 2). Although feudal societies were usually led by a central elite (e.g. a king), there were in reality a number of formal as well as informal checks and balances built into these power structures, which in effect constrained the actions of various elites. The kings did not have an efficient monopoly on the use of force, nor in the collection of tax revenues. In practice, they had to engage in a never-ending balancing act vis-a-vis local elites (landed nobility), as well as with the Church and moreor-less autonomous cities (Moore, ibid.). Through these checks and balances, the central elites were constrained in any attempt to satisfy their personal whims at the expense of the rest of the population. Gradually, however, central elites in various parts of Europe were able to suppress local elites: the landed nobility was replaced by a court nobility, much more dependent on being in favour by the king (Rokkan, 1975; Elias, 1982). Central elites were thus able to monopolize the right to coerce, as well as the right to collect taxes, within larger geographical areas; although the success of central elites to suppress local elites varied between regions (Rokkan, ibid.). At the same time, the gradual emergence of a monetary economy vastly improved the power to tax, and thus enhanced the ability of the central elites to interfere in the lives of their subjects (partly by enhancing their ability to extract revenue, and partly since this increased revenue made them able to finance a larger standing army, as well as a more efficient tax collection system). The rapid spread of mass transport and mass communication devices (e.g. railways, autobahns and telegraph facilities) made possible by the industrial revolution also increased the ability of the central elite to push its will through in the various corners of the territory it controlled. Taken together, these changes implied that the individuals living in the territory became gradually more exposed to the risk of being subjected to suppression by an ever-stronger central elite. In this situation, they were furnished with an incentive to push

331 for the introduction of new types of checks and balances, in order to constrain the decision-making ability of the central elites, and thus reduce risk to an acceptable level. Tullock (ibid.) argues that the balancing of various local and central elites which characterized medieval Europe was a fairly uncommon phenomenon in other parts of the world. Most ancient regimes were autocratic rather than feudal (i.e. they were to a much larger degree led by a central elite which usually had a fairly solid grip on power). However, even if Tullock should be right that the feudal experience is a peculiarity of European history, this does not imply that the mechanisms which have lead to the introduction of democracies in this part of the world (as well as in most of the immigrant nations which were colonized mainly by europeans) is limited by the idiosyncratic European experience. Even if an autocratic rather than a feudal power structure may have been the dominant ancient regime in most other parts of the world, these autocracies were not very efficient from a modem point of view. Lack of efficient means of transportation and communication limited the ability of any central elite to make its presence felt in the territory it controlled, implying that local vassals had a larger room for manoeuvre than the regional authorities in a modem state. 9 The absence of a monetary economy also reduced the power to tax, further limiting the coercive potential of any autocratic ruler. Thus in all ancient regimes - be they feudal or autocratic - the ruling elites simply lacked the technical means for achieving the almost total control an autocratic regime in one of today's most advanced industrialized states might potentially be able to accomplish. Consequently, the demand for insurance among ordinary citizens against the risk of being at the mercy of the more-or-less arbitrary will of a self-appointed absolute ruler should be much greater in the "modem" age than in any previous epoch. 1° Speer's last speech at the Ntimberg trial succinctly illustrates the point:

Hitler's dictatorship was the first dictatorship of an industrial state in this stage of modem technology, a dictatorship which employed to perfection the instruments of technology to dominate its own people... By means of such instruments of technology as the radio and public-address systems, eighty million persons could be made subject to the will of one individual. Telephone, teletype and radio made it possible to transmit the commands of the highest level directly to the lowest organs where because of their high authority they were executed uncritically... The instruments of technology made it possible to maintain a close watch over all citizens ... Dictatorships of the past needed assistants of high quality in the lower ranks of the leadership also - men who could think and act independently. The authoritarian system in the age of technology can do without

332 such men.., the more technological the world becomes, the greater is the danger. (Speer, 1971: 693) Since other parts of the world now experience a similar industrialization and urbanization process as has already taken place in most European and AngloAmerican countries (and Japan), central elites everywhere will gradually be provided with a similar vasty enhanced ability to control and interfere in the daily lives of their inhabitants. This increased exposure to risk, as seen from the point of view of the citizenry, should then prompt them to push for procedures constraining the scope of actions available to central elites. In practice, this demand may take the form of offering support to aspiring elites (political entrepreneurs) bent on using this security-demand as a device to elevate themselves to political offices, through more-or-less peaceful transitions to democracy. In this context, it is important to notice that the technological innovations which have enhanced the ability of central elites to make their presence felt in all comers of their territories, simultaneously offer the citizenry - for the first time in history - the technical capability not only of electing, but also of monitoring the behaviour of their rulers almost on a day-to-day basis: cf. e.g. the importance of the printing press and (later) television and radio broadcasting technology in achieving such monitoring. Thus the new technology provided through the industrial revolution may work "both ways". The last argument may also explain why democracies are fairly recent phenomena, despite a presumption that the inhabitants of all countries, at all times, have probably preferred to be insured against the risk of arbitrary suppression. Before the monitoring devices provided through the industrial revolution were well in place, and before effective communication and transportation facilities were available, autocratic elites could (fairly) easily control their disjointed subjects by hitting one trouble spot at a time. Thus before the industrial revolution, with its various appendages, such as mass education, various types of autocracies - or limited electorial systems - were the only stable options. Constant monitoring implies that the rulers have difficulties in getting away with arbitrary acts of suppression. A legal system granting media an autonomous position vis-~t-vis the state enhances effective monitoring. Utilizing principal-agent imagery, mass media can be portrayed as complementary agents, specializing in providing the electorate with information that makes screening and monitoring easier to accomplish. Notice that although a ruling elite may ideally prefer to censor the media in order to escape monitoring, an act of censorship may easily be interpreted by the electorate (principals) as a sign that the elite (agent) has something to hide (is untrustworthy). Besides, any attempt by a ruling elite to censor the media is likely to be utilised by

333 aspiring elites in upcoming elections. For these reasons, elites may regard it in their self-interest to abstain from censorship. The assumption that industrialization, urbanization and the spread of mass communication devices may in the long run induce a change in political regimes towards systems in which the inhabitants are able to control and monitor the behaviour of their rulers may be regarded as a (conditioned) prediction derived from the insurance-approach to democratic transitions; a prediction which is in principle falsifiable. To the extent that democratic decision-making procedures spread through the world, they should be hard-wired to an ever-increasing process of industrialization, urbanization and spread of mass communication. If these processes should for some reason break down in the future, the tendency towards increased emphasis on democratic decision-making procedures across countries should be expected to break down as well. It should further be noticed that there is fairly solid empirical evidence for assuming a linkage between industrialization and urbanization ("modernization") and the introduction of democracy (see e.g. Dahl, 1971: 62-80; Lipset, 1981: 31-45, 469-476; Diamond et al., 1987: 10-11; Diamond, 1992; Lipset et al., 1993: 156-164). However, unlike these scholars, the insurance-approach is able to account for this correlation without having to assume any role of norms (as opposed to self-interest) in providing the motivational foundations behind this transition; i.e. it is able to provide the "modernization"-argument with a less vague microfoundation.

12. The transition game: A further elaboration

So far, I have focused on citizens' demands, as well as citizen-elite interaction, in explaining the existence and spread of democratic regimes. Demand pressures may be an underlying tide creating ever-more favourable opportunities for political entrepreneurs to use this demand to elevate themselves to political positions, by more-or-less peaceful transitions to democracy. Moreover, citizen-elite interaction (perceived as a principal-agent relationship) may explain why competing elites willingly constrain themselves to respect the rules of the game in campaigning for office. However, two remaining problems must be solved in order to ground the existence and spread of democracies in the economic man-premise of the Public Choice-paradigm. First, to explain the origin of new, democratically minded political entrepreneurs from an autocratic or feudal setting. Second, to explain why an elite - once it has won an election - does not simply break its earlier promises, tear up (or "re-interpret") the constitution, and use its monopoly on power to suppress its competitors and install a new autocracy.11 In trying to answer these questions, it might be fruitful to introduce a third topic in the analysis, alongside citi-

334 zen demands and citizen-elite interaction. This third topic is the relationship between various elites (elite-elite interaction). In the following, I shall argue that the genesis of democratically oriented elites, as well as the reason why elected elites often stay committed to earlier promises once in office, may be understood by focusing on the relationship between the elites themselves. 12.1. Power-sharing as mutual insurance First, it should be noticed that competition between elites is in no way limited to democratic regimes. A similar competition for power takes place also in autocratic regimes, only in these systems it is usually hidden from the public eye. Tullock (1987: 17) is probably right when arguing that the internal history of an autocracy is largely a jockeying for power. 12 Indeed, competition may be more intense in an autocracy, because the stakes are higher: the losers of the game may eventually end up killed rather than merely out of office. Besides, even if they should survive they are seldom (as in a democracy) allowed a second try. If the members of various autocratic elites are risk-averse, they have an interest in being insured against the risk that a defeat have too disastrous personal consequences, such as being jailed, executed or robbed of all possessions. Thus they may prefer the political game to unfold within an institutional setting granting that the losers are not too vigorously persecuted by the winners (Przeworski, 1986: 56). Such institutional rules may serve as mutual insurance against the risk of imprisonment, death or loss of property. One strategy for providing this type of mutual insurance is to institutionalize some type of power sharing, implying that no elite is provided with the legal and/or physical ability to suppress all the others. 13 Thus also at the elite level, we may identify insurance-interests on behalf of various elites providing them with incentives to reign each other into a web of mutual checks and balances. In this context, it is illuminating to notice that the first rudimentary democracies within Europe emerged in countries which were recovering from intense civil wars (Kriele, 1981: 47-65). The experience of civil wars furthered the war-fatiguedelites with a strong incentive to reach some type of mutual agreement as to how their competition could be regulated, in order to bring civil war to an end and reduce the risk that future political games should take a similar violent/repressive turn. 14 The separation of legislative, judicial and executive power advocated by Montesquieu [1748] (1989) can be interpreted as a means to provide various elites with such mutual insurance, by creating an institutional power structure in which no elite is allowed to monopolize all the three main types of decisionmaking authority within the state. It is worth noticing that Montesquieu, when emphasizing how a separation of powers within the state would enhance the

335 liberty of the individual citizen, defined "liberty" as synonymous to security from arbitrary power, i.e. as insurance against a particular type of risk or "fear" (cf. also Locke [1680] 1993: 163). Montesquieu writes: Political liberty in a citizen is that tranquillity of spirit which comes from the opinion each one has of his security, and in order for him to have this liberty the government must be such that one citizen cannot fear another citizen. When legislative power is united with executive power in a single person or a single body of the magistracy, there is no liberty, because one can fear that the same monarch or senate that makes tyrannical laws will execute them tyrannically. Nor is there liberty if the power of judging is not separate from legislative power and from executive power. If they were joined to legislative power, the power over the life and liberty of the citizens would be arbitrary, for the judge would be the legislator. [1748] (1989: 157; emphasis added) The need for some explicit power-sharing devices is enhanced by the fact that the elites may not trust each other to stay committed to an electoral rule of the game should they win the first election. In a situation of mutual distrust, constitutionally derived power-sharing principles (such as those advocated by Montesquieu) may be a vital necessity in order to prevent a relapse into autocratic rule. By explicitly dividing the power of the state so that no elite is able to wholly dominate the others, the various elites may feel reasonably assured to take the risk of entering into a peaceful competition for (some of the) power. Framed yet another way: by maintaining some modicum of separate power-bases not up for the vote, the elites may be willing to agree on some limited competition for a limited amount of the state's power. A division of parliament into an upper and lower House, elected (or appointed) from different constituencies, further seeks to diversify power and limit the ability of one elite to suppress all the others) 5 The existence of consociational democracies further illustrates how elites may tie each other down in a web of mutual checks and balances. According to Lijphart (1977: 25), consociational democracies are likely to be found in deeply divided (plural) societies. Lijphart characterizes consociational democracies as follows: First, they are governed by some type of grand coalition of the political leaders of all significant segments of society. This can take several different forms, such as a grand coalition cabinet in a parliamentary system, a "grand" council or committee with important advisory functions, or a grand coalition of a president and other top officeholders in a presidential system. Second, they are characterized by mutual veto or "concurrent majority" rule, which serves as an additional protection of vital minority interests. Third, proportionality is the principal standard of political represen-

336 tation and allocation of public funds. Lastly, each segment has a large degree of autonomy to run its own internal affairs. Lijphart (op.cit.: 15) presents the Netherlands, Belgium and Austria as examples of consociational democracies, reflecting deep national cleavage lines (see also Dahl, 1971:116-119). More generally, various types of power-sharing institutions can be thought of as risk-reducing devices. Thus the insurance-approach may also offer an "economic man"-explanation as to the various institutional propensities of democratic regimes. 12.2. The gradual stabilization of a new electorial system As stated above, the transition-towards-democracy game might easily relapse into autocratic rule, since each elite faces uncertainty as to whether or not any other elite - should it acquire power - will stay committed to the rules of the game. While each elite lacks information as to the trustworthiness of other elites, they are aware that the other elites face a similar adverse selectionproblem when facing them. This implies that the relationship between the elites may also be regarded as a principal-agent type of relationship, the important point being that each elite can be perceived simultaneously as a principal and an agent: Each elite is a principal when observing the other elite(s), while at the same time being an agent in the eyes of the other(s) (cf. Overbye, 1995: 373). By portraying elite-elite interaction as a two-way principal-agent relationship, we may detect a mechanism which may enhance the stabilization of an electorial regime, once it has been in operation for some time. As a starting point for this line of thought, consider the argument that one of the major strategies players (in their capacity of agents) may utilize in order to overcome an adverse selection problem, is through signalling (Rasmusen, 1991: 205ff.). Signalling is a way for an agent to communicate his "type" under adverse selection. The distinguishing feature of signalling is that the signal, although not directly related to the agent's future course of action (in this case: will he stay committed to the rules of the game or not) is related to his commitment to perform a future action (Rasmusen, 1991: 210). In order for the elites to assure each other that they will stay committed to electorial rule, the players (agents) must be willing to trust the other players not to cheat. Now the question is: how can the elites signal trustworthiness to each other? Obviously this is a difficult task, since no observable, "objective" indicator of trustworthiness exists. Simply telling the other elites that one will not suppress all the competitors once in office will often not suffice, since any power-hungry liar may offer such reassurances. The distinguishing feature of signals, as opposed to mere "messages", is that signals are not cost-free (Rasmusen, 1991: 134). The problem with all types of verbal commitments,

337 including stated commitments to respect a constitution, is that they do not involve costs and thus cannot be regarded as signals proper. However, in the absence of "objective" indicators, the elites may try to create reputations for trustworthiness. This can be done in much the same way as a private entrepreneur accumulates favourable references in order to convince future clients that he is able to do a reliable job. If the members of an elite can refer to earlier occasions in which they carried the costs of staying committed to the rules of the game (by not giving in to any temptation to exploit the rules once in office), they have invested in a reputation which may be a valuable assetin future political games. Their reputation serves as a signal to the other elites (as well as to voters) that they can probably be trusted (Overbye, 1995: 374). Now the point is that once an electorial system has been formed and has been in existence for some years, ruling elites will have had the opportunity to build this type of reputations. Trustworthy reputations are developed by abstaining from breaking the rules once in office (i.e. by incurring the costs of lost opportunities to hurt or eliminate one's competitors once in office) (cf. Wilson, 1985). Through this type of signalling, the elites will - across time be able to mutually assure each other that they are committed to respecting the rules of the electronic game, implying that the electorial regime is gradually stabilized. Framed differently, but with the same meaning: By staying committed to the rules of the game while in office a ruling elite allwos itself to be screened by competing elites as well as by voters, who may then be assured that the ruling elite is committed to the rules of the game. 16 In the initial stages of the game, signalling through building and maintaining reputations is not possible, for the simple reason that the system has not been in operation for a sufficiently long period for reputations to evolve (Dahl, 1971: 39, 151-152, 217). In the initial stages, the best the elites can do is usually to send messages to each other that they will stay committed to the rules of the game, should they win the first election. Since messages alone will often not suffice (it being too risky for the other elites to rely on messages alone), the end result of the game may well be democratic breakdown. This is precisely the reason why, in the initial stages of the game, explicit power-sharing devices may be necessary in order to prevent a relapse into autocratic rule. There is indeed a common assumption in the empirical literature that the probability of a democratic breakdown is highest in the first years after the new regime has been installed (Dahl, 1971: 39; Diamond et al., 1987: 6-7). However, the longer the system has prevailed, the more stable it should become. 17 Thus, over the long pull, explicit power-sharing mechanisms may be of less importance to secure the stability of electorial rule, as the reputations of various elites may then serve as sufficient mutual assurance that an elected elite will not exploit its monopoly on power. 18 For -

338 the same reason, consociational democracies may gradually be transformed towards "ordinary" democracies, as observed by Lijphart (1977: 2-3). A gradual extension of suffrage - i.e. a move from an electoral system towards a full-blown democracy - may also be easier to accomplish as mutual trust is enhanced, at least as long as existing elites do not believe that an extension of suffrage will strengthen the position of elites which are considered "unreliable" (Dahl, 1971: 33-40). In this fashion, we may account for the gradual emergence of more democratically oriented elites (elites willing to extend suffrage to ever-larger groups), as well as the gradual stabilization of new electorial regimes. Who will be the most successful in extracting the trust of others? Perhaps those who are able to convince themselves that they do not want to hurt or eliminate their competitors, even if given the technical capability of doing so (Overbye, 1995: 381-383). Competing elites may become involved in a "signals arms race", in which less trustworthy elites/members of elites are gradually weeded out of office. Through this weeding out-mechanism, democratic decision-making procedures may possibly have a long-run impact on "the stuff rulers are made of".

13. Explaining occasional breakdowns of democratic rule Democracies do on some occasions break down, to be replaced with various forms of autocratic regimes. Further, some countries tend to shift between democratic and autocratic regimes, without settling permanently for one or the other regime type. The latter type of regime has been especially prominent in Latin America (Kaufmann, 1987; Lipset et al., 1993: 164). If it is true that industrialization and urbanization induces a long-run tendency towards democratic decision-making procedures, we are then faced with the puzzle of explaining why democracies are not always stable, i.e. to account for the existence of democratic breakdowns, and the existence of unstable (on-off) democracies. A number of idiosyncratic, historical factors may speed up or delay the introduction of democratic rule, or cause the (at least temporary) downfall of democratic regimes. I shall not attempt to present an account of the variety of factors which may be at work in this respect (see Linz and Stephan, 1978 and O'Connor and Schmitter, 1986 for empirical studies and extensive discussions of such factors). However, if I may hazard a guess at a destabilizing mechanism which may be at work across several cases, I would argue that a new democracy is in grave danger if its introduction severely increases the perceived level of risk faced by large or powerful segments of the population (Przeworski, 1986: 60). If this is the case, then those segments which feel

339 that the introduction of democracy has jeopardized their safety, may push for a return to autocratic rule. Let me provide three illustrations as to how a democracy may in some instances enhance rather than reduce risk. First, suppose that a country is divided into several ethnic groups (tribes), and that one ethnic group is much larger than the rest. If democracy is introduced, and if voting takes place along ethnic cleavage lines, the smaller ethnic groups may fear that the new regime places them in a more vulnerable position than in an autocracy, at least provided that an autocracy can be furnished with some power-sharing institutions not found in the present democratic regime. 19 In this situation, the smaller groups may unite in order to abolish majority rule, in effect limiting the dominant position of a majority tribe. 2° Second, suppose that an election elevates a new elite to power which attempts to curb the societal position of older/more established elites (e.g. the military leadership, large property owners or top-level state bureaucrats). If so, these elites may feel that an autocratic system provides them with a less risky political environment in which to secure the continuation of their present social position (Dahl, 1971: 51). 21 Being powerful elites, they may succeed in abolishing democratic rule, at least temporarily.22 Third, suppose that a democratic regime is hit by a severe economic depression, implying that both employment and savings become much less secure. In a situation of increased economic uncertainty, and provided that the elected elites do not appear able to reduce uncertainty to acceptable levels, ordinary citizens may support an autocratic leader promising order and stability. The fate of the Weimar republic may illustrate the point. 23

14. Concluding remark According to Thomas Kuhn (1970: 23-24, 35-36), puzzle-solving is the primary purpose of so-called "normal science". By trying to account for empirical facts which initially appear puzzling within the boundaries of a particular research paradigm, adherents of the paradigm try to fend off fundamental criticism. To the extent that they succeed, the merits of the paradigm is strengthened; if they show persistent failure, the paradigm may enter into a period of "crisis". The purpose of this article has been to suggest an explanation, based firmly on the economic man-premise of the Public Choice-paradigm, to two related sets of problems ("puzzles"). The first concerns how to explain the genesis and stability of democratic regimes. The second concerns how to explain the 20th century increase in the number of democracies, as well as why democratic breakdowns do occasionally take place. At this stage, I have avoided formal refinements and rather presented a broad, exploratory outline

340 of how an insurance-approach may provide solutions to these twin sets of problems. What is key is self-oriented insurance-interests on behalf of citizens as well as competing elites, plus the adaption of principal-agent models in understanding the relationship between citizens and political elites, as well as the competitive relationship between elites. Granted, there are also other mechanisms at work which may help explain the existence and spread of democracies; the most striking being foreign conquest. Democracies are sometimes installed by the victor, as happened in Germany, Japan and Italy after World War II (Dahl, 1971: 37-38; Stephan, 1986: 71-72). 24 However, in these cases the introduction of democracy represents no serious puzzle for the Public Choice-paradigm: it being rather obvious that defeated elites will accept the system imposed upon them, as they have no other options (Tullock, 1987: 180). In such cases, the remaining puzzle is to explain why democratic rule is often maintained even after the victor has withdrawn his military presence; a puzzle which can be solved precisely by pointing out how the mutual building of trustworthy reputations may gradually stabilize an imposed regime. 25 Given the thrust of the article (to offer a solution to some interrelated puzzles within the Public Choice-paradigm) I have abstained from discussing in detail to which extent empirical evidence confirms the theory. On some occasions, I have indicated how existing evidence in the form of case studies as well as broad-based statistical studies may offer support to the arguments advanced here. It should be emphasized, however, that real tests of the theory can only be conducted in the future, implying that no one (including this author!) can then be suspected of having fabricated the theory after first having had a quick glance in the answer book. Thus to prevent the theory from degenerating to post hoc-generalizations, I have launched some predictions as to likely political developments in industrializing autocracies and newly established democracies. First, industrializing autocracies should become increasingly prone to internal democratization pressures. Second, new democracies should be expected to set up explicit power-sharing devices, and/or maintain a consociational structure, in particular in deeply divided societies. Third, provided that a backlash is avoided in the initial stages of the process, various checks and balances may gradually decrease in importance, as the elites - through signalling and screening- are gradually able to mutually trust each other to maintain democratic rule. Future research should aim towards more precise elaborations of these - and related - hypotheses, in order to further strengthen (or weaken) the insurance-theory of democracy.27

341 No~s 1. For example, Tullock (1987: 5) points out that greek city states were often governed by a council of about 30 heads of noble families: hardly democracies, but not clear-cut cases of autocratic regimes either. 2. Liberty, life and property were the three basic rights and privileges which Locke [1690] (1993: 157) assumed that man had a "natural" right to defend against the "injuries and attempts of other men". 3. For the sake of simplicity I shall use "his" rather than "his/hers" in this article. 4. Equation (2) simplifies the decision problem for a number of reasons. First, autocracies are seldom seen as either enlightened or unenlightened; in reality, this variable represents a continuum rather than a dichotomy. There may also be different degrees, and types, of democracies, which are viewed differently with regard to the utility they provide to various segments of the population. As pointed out by Erling Barth (private conversations), democracies which secure high economic efficiency, e.g. by skewing the rules of the political game in a direction that secures strong governments, may stand the best chances of maintaining stability in the long run (see also Dahl, 1971: 123, 220-221). Second, citizens should take into account that a once-enlightened autocracy may slide across time towards a less enlightened regime, implying that their discount rates should enter into the equation. Third, risk-neutral, risk-willing and risk-averse citizens might chose differently. Risk aversion should sway more citizens in the direction of preferring a democracy rather than an autocracy, since for risk-averse individuals U[pXe,~ + (1 - p)Xua] > pU(Xc,~) + (1 - p)U(X~a). If Bishop et al. (1991: 65-66) are right that a majority of voters are likely to be risk-averse, then part of the popular support democracies enjoy stems not least from their ability to insure the citizens against the risk of waking up one day (or in the middle of the night) to discover that they no longer live in an enlightened regime. Cf. the saying that "the real advantage of living in a democracy is that, if someone should knock at your door in the middle of the night, you know it is only the milkman." 5. The point that a democracy is in jeopardy if it is perceived as inefficient has been emphasized by a number of scholars; see e.g. Dalai, 1971: 147; Lipset, 1981: 64; Diamond et al., 1989: 7. 6. Stephens' (1989: 1030, 1036) observation that autocratic elites often face popular (democratic) revolts following military defeats (implying that they have exposed the population to war-induced risks of injuries, loss of property and death without even being able to win) may fit in with this hypothesis. 7. The decision problem facing an elite is made more complex by the fact that its decision may hinge on which decisions it expects other (competing) elites to arrive at. However, I shall postpone an analysis of elite-elite interaction till the last section of the article. 8. It should be noticed that not every scholar will agree that the 20th century has witnessed a gradual increase in the number of democracies (cf. e.g. Tullock, 1987: 1-2, 190). Admittedly, the process has occurred in leaps and bounds, and with lots of temporary setbacks. Besides, the last surge in such regimes (appearing after 1987) is so recent that setbacks are bound to occur in at least some of the new democracies. Nonetheless, the recent surge is so broad-based that by all probabilities the end of the 20th century will know a much larger number of democracies than at the beginning: especially if we keep in mind that few of the many European colonies prevailing at the beginning of the century had a leadership elected from within (that is: by the native population); and that none of the late 19th century electorial systems (with the exception of New Zealand and Finland) extended suffrage to the whole adult population. 9. Although the transportation problem was easier to manage in sea-states than land-states. Rulers of sea-states such as Britain and Denmark could transport large armies from one trouble spot to another across open waters, while the rulers in the European mainland were constrained by poor roads and high mountains, as well as a string of semi-autonomous

342

10.

11.

12. 13.

14.

15.

cities (the "city belt"), stretching from northern Italy through central Europe to Flanders and the Baltic. As a consequence, sea-states situated at the Northern and Western periphery of the city belt displayed a more centralized power structure in medieval Europe than the central-European land-states (Rokkan, op.cit.). At this point, it might also be worth while to call attention to Montesquieu's distinction between monarchy and despotism. According to Montesquieu (1989: 18, 27-28), monarchies are characterized by a complicated hierarchy of institutions based on various codes of honour; the reason being that the king is to some extent dependent on an aristocracy to maintain and secure his power. The existence of an aristocracy (even a court aristocracy) constrains the actions the ruler may pursue. In despotic regimes (autocracies without any intermediate elites) there is only the ruler and his subjects, and the ruler is thus bound by fewer informal rules of conduct. Notice also that the industrial revolution has not only changed the technical ability of a ruler to suppress and exploit his subjects. By enhancing geographical and social mobility, industrialization and urbanization have also weakened old customs and traditional behavioural patterns; thus reducing the sway of old behavioural codes (including codes of honour) in constraining the choice of actions pursued by various individuals - including rulers. Or perhaps the question should be why new elites do not always behave in this fashion. Undoubtedly, many countries experience that changes in governments are accompanied by new constitutions, or severe amendments of the existing ones. Speer's (1971) lucid account of the never-ending intrigues between warring factions within Nazi Germany aptly illustrates the point. As an appendage to this argument, it may be that democratic procedures are more stable in societies where the state does not play an omnipotent economic role, i.e. where control of the state does not automatically imply a direct control of most of the economic resources in a society. Economic resources held independently by citizens can to some extent function similar to "power bases" not up for the vote, and may thus serve as an additional impediment limiting the temptation - as well as ability - of an elite to reign supreme once it has acquired power (Diamond, 1992: 481--483). As pointed out by Kriele (op.cit.), the experience of civil wars was also used by those who wanted to strengthen the position of a central elite (the king). This is evident in Hobbes' line of argument, which explicitly stress security-interests as the primary motivation for the inhabitants in a country to succumb under the authority of a central elite. Hobbes [1651] (1991:117-120) puts the argument this way: "The final Cause ... of men ... in the introduction of that restraint upon themselves (in which we see them live in Commonwealths) is the foresight of their own preservation ... that is to say, of getting themselves out from the miserable condition of Warre ... For ... Covenants, without the Sword, are but Words, and of no strength to secure a man at all ... Nor is it the joining together of a final number of men, that gives them this security; because in small numbers, small additions on the one side or the other, make the advantages of strength to great, as is sufficient to carry Victory; and therefore gives encouragement to an Invasion ... Nor is it enough for the security, which men desire should last all the time of their life, that they be governed, and directed by one judgement, for a limited time... For though they obtain a Victory by their unanimous endeavour against a forraign enemy; yet afterwards ... they ... fall again into a Warre amongst themselves ... The only way to erect such a Common Power, as may be able to defend them from the invasion of Foreigners, and the injuries of one another, and thereby to secure them in such a sort ... is, to conferre all their power and strength upon one Man, or upon one Assembly of men" (emphasis added). Holmes (1993: 228) presents a somewhat different explanation as to the emergence of power-sharing institutions which may supplement the explanation presented here. Quoting Adam Smith, he regards such institutions primarily as an outgrowth of the increased division of labour characteristic of an industrializing society. The increased complexity of society, stemming from increased business and interdependence, led to a division also of political offices. Holmes further points out that the earliest justification of the separation

343

16. 17. 18.

19.

20.

21.

22. 23.

24.

25.

26.

of powers emphasized its power-enhancing rather than power-controlling function: for example, it was argued that an independent judiciary would increase the king's capacity to govem. This approach may also be adopted when studying elite-elite interaction across countries, e.g. the relationship between US and USSR in the days of the cold war. Although the risk of a democratic breakdown will never be completely overcome, and may be temporarily heightened by various situational factors. In this context, consider the fact that modem parliamentary democracies are characterised by a strong linkage between legislative and executive power, since pafliamentarism usually implies that whichever elite controls the legislature also controls the executive (unlike the situation in presidential democracies such as the United States, in which Montesquieus power-sharing principle is still more strongly adhered to). Cf. also the gradual historic decline in mature democracies of the importance of an upper House in parliament, at least in systems in which the upper House is elected (or appointed) from a different constituency than the lower House (e.g. in Britain). This argument begs the question of how the existence of ethnic loyalties can be explained within a Public Choice-framework. Unfortunately, space does not allow a discussion as to if and how the existence of such loyalties (and group loyalties in general) can be explained without violating the "economic man"-assumption (but cf. Overbye, 1995: 384-385). Alternatively, the tribes may try to secede from the larger state if ethnic and geographical boundaries coincide, or if higher correlation can be accomplished through ethnic cleansing. Thus the introduction of democracy may in some cases trigger a falling apart of the old autocratic state. Present developments in former Yugoslavia may illustrate the point. Notice the similarities with Przeworski's (1986: 60ff.) argument that, if a transition is to be successful, new elites must try to reach a compromise with the elites of the old regime, and in any case avoid provoking them more than necessary. The Chilean experience after Allende may provide a case in point. A democratic system may thus be in jeopardy if it is perceived by the population as much less economically efficient than an autocracy, as illustrated in equation (2). Democratic regimes may also become somewhat less popular in less extreme circumstances. For example, if it is true that "democratic overload" - due to an ever-expanding number of interest groups - limit economic efficiency in a democracy relative to an autocracy, voters may regard the future benefits they can derive from increased efficiency as sufficiently large to overrule the immediate benefits of being insured against arbitrary oppression. However, I doubt if such "academic" considerations are of vital importance to many voters. First, since a democracy will usually be able to offer a more predictable legal environment than an autocracy, it may actually have a competitive advantage vis-il-vis an autocracy, at least in attracting long-term investments. Second, the existence of rent-seeking interest groups is hardly confined to democratic regimes, as pointed out e.g. by Anderson and Boettke (1993). Third, even if we for the sake of argument accept the assumption that autocracies on average are economically more efficient than democracies, self-oriented voters should not care unless they personally expect to benefit from the increased efficiency an autocracy may offer. Also, the balance of power between internal elites may be influenced by various types of external pressures aimed at enhancing the position of democratically minded political entrepreneurs; see Dahl (1971: 189-201), Whitehead (1986), Lowenthal (1991) and Diamond (1992: 480--481) for examples and a discussion of this type of international pressure. In Germany and Italy it was also possible to rely on previous experiences of democratic rule; implying that internally subdued elites, which already trusted each others adherence to democratic rule, could be hoisted to power (Dahl, 1971: 38). A final note of reconciliation towards those insisting that various normative considerations do also underlie some of the support enjoyed by democratic regimes. I do not want to rule

344 out the possibility that such norms may exist. My argument is of the "Occam's razor"type: Although normative considerations may underlie some of the support enjoyed by democratic regimes, it is not necessary to assume that such normative considerations do exist, in order to explain the incidence, as well as prevalence, of democratic decisionmaking procedures in "modem" societies.

References Anderson, G.M. and Boettke, EJ. (1993). Perestroika and public choice: The economics of autocratic succession in a rent-seeking society. Public Choice 75:101-118. Anderson, J. and Hall, S. (1986). Absolutism and other ancestors. In J. Anderson (Ed.), The rise of the modern state. Brighton: Harvester Press. Banks, J.S. (1991). Signalling games in political science. Chur, Switzerland: Harwood Acaden'tic Publishers. Bishop, J.A., Formby, J.E and Smith, W.J. (1991). Incomplete information, income redistribution and risk averse median voter behavior. Public Choice 68: 41-55. Brennan, G. and Lomasky, L. (1993). Democracy & decision. Cambridge: Cambridge University Press. Dahl, R.A. (1971). Polyarchy. New Haven: Yale University Press. Diamond, L., Lipset, S.M. and Linz, J. (1987). Building and sustaining democratic government in developing countries: Some tentative findings. World Affairs 150(1): 5-19. Diamond, L. (1991). Economic development and democracy reconsidered. American Behavioral Scientist 35(4/5): 450--499. Downs, A. (1957). An economic theory of democracy. New York: Harper and Row. Elias, N. (1982). Power and civility, vol II. New York: Pantheon Books. Elster, J. (1988). Introduction. In J. Elster and R. Slagstad (Eds.), Constitutionalism and democracy. Cambridge: Cambridge University Press. Fukuyama, E (1992). The end of history and the last man. New York: The Free Press. Hobbes, T. (1991). Leviathan. Cambridge: Cambridge University Press. Holmes, S. (1988). Precommitment and the paradox of democracy. In J. Elster and R. Slagstad (Eds.), Constitutionalism and democracy. Cambridge: Cambridge University Press. Kaufmann, R.R. (1987). Liberalization and democratization in South America: Perspectives from the 1970s. In G. O'Donnell, E Schmitter and L. Whitehead (Eds.), Transitions from authoritarian rule. Comparative perspecives. Baltimore: The Johns Hopkins University Press. Kriele, M. (1981). Einfiihrung in die Staatslehre. Die geschichtlichen Legitimitgitsgrundlagen des demokratischen Verfassungsstaates. Wiesbaden: Westdeutscher Verlag. Kuhn, T.S. (1970). The structure of scientific revolutions. Chicago: Chicago University Press. Ledyard, J.O. (1984). The pure theory of large two-candidate elections. Public Choice 44: 7-41. Lijphart, A. (1977). Democracy in plural societies. New Haven and London: Yale University Press. Linz, J.J. and Stephan, A. (1978). The breakdown of democratic regimes. Baltimore: The Johns Hopkins University Press. Lipset, S.M. (1981). Political man. Baltimore: The Johns Hopkins University Press. Lipset, S.M., Seong, K.R. and Torres, J.C. (1993). A comparative analysis of the social requisites of democracy. International Social Science Journal 136:155-176. Locke, J. (1993). Two treasties of government. London: Everyman. Lowenthal, A.F. (Ed.) (1991). Exporting democracy. The United States and Latin America. Baltimore: The Johns Hopkins University Press. Machiavelli, N. (1983). The prince. Harmondsworth: Penguin Books.

345 Montesquieu, C. (1989). The spirit of the laws. Cambridge: Cambridge University Press. Moore, B. (1966). Social origins of dictatorship and democracy: Lord and peasant in the making of the modern world. Boston: Beacon Press. O'Donnell, G. and Schmitter, E (1986). Transitionsfrom authoritarian rule. Tentative conclusions about uncertain democracies. Baltimore: The Johns Hopkins University Press. Overbye, E. (1995). Making a case for the rational, self-regarding, "ethical" voter ... and solving the "Paradox of Not Voting" in the process. European Journal of Political Science 27: 369-396. Przeworski, A. (1988). Some problems in the study of the transition to democracy. In G. O'Donnell, EC. Schmitter and L. Whitehead (Eds.), Transitions from authoritarian rule. Baltimore: The Johns Hopkins University Press. Rasmusen, E. (1991). Games and information. Oxford: Basil Blackwell. Rokkan, S. (1975). Dimensions of state-formation and citizenship. A possible paradigm for research on variations within Europe. In C. Tilly (Ed.), The formation of nation-states in Europe. Princeton: Princeton University Press. Schumpeter, J.A. (1942). Capitalism, socialism and democracy. New York: Harper and Row. Shepsle, K.A. and Weingast, B.R. (1981). Structure-induced equilibrium and legislative choice. Public Choice 37(3): 503-519. Speer, A. (1971). Inside the Third Reich. London: Sphere Books. Stephan, A. (1986). Paths toward redemocratization: Theoretical and comparative considerations. In G. O'Donnell, E Schmitter and L. Whitehead (Eds.), Transitions from authoritarian rule. Comparative perspectives. Baltimore: The Johns Hopkins University Press. Stephens, J.D. (1989). Democratic transition and breakdown in Western Europe 1870-1939: A test of the Moore thesis. American Journal of Sociology 94(5): 1019-1077. Tullock, G. (1981). Why so much stability? Public Choice 37(2): 189-202. Tullock, G. (1983). Economics of income redistribution. Boston: Kluwer Nijhoff Publishing. Tullock, G. (1987). Autocracy. Boston: Kluwer Academic Publishers. Whitehead, L. (1986). International aspects of democratization. In G. O'Donnelt, E Schmitter and L. Whitehead (Eds.), Transitions from authoritarian rule, Comparative perspectives. Baltimore: The Johns Hopkins University Press. Wilson, R. (1985). Reputations in games and markets. In A.E. Roth (Ed.), Game-theoretic models of bargaining. Cambridge: Cambridge University Press.