Minimalist Conception of Democracy: A Defense
Adam Przeworski
Introduction
I want to defend a "minimalist," Schumpeterian, conception of democracy, by minimalist, Pop-perian, standards. In Schumpeter's (1942) conception, democracy is just a system in which rulers are selected by competitive elections. Popper (1962: 124) defends it as the only system in which citizens can get rid of governments without bloodshed....
Since neither the position I wish to defend nor the claim in its favor are new, what do I defend them from? Perusing innumerable definitions, one discovers that democracy has become an altar on which everyone hangs his or her favorite ex voto. Almost all normatively desirable aspects of political, and sometimes even of social and economic, life are credited as intrinsic to democracy: representation, accountability, equality, participation, justice, dignity, rationality, security, freedom,..., the list goes on. We are repeatedly told that "unless democracy is x or generates x,..." The ellipsis is rarely spelled out, but it insinuates either that a system in which governments are elected is not worthy of being called "democracy" unless x is fulfilled or that democracy in the minimal sense will not endure unless x is satisfied.2 The first claim is normative, even if it often hides as a definition. The second is empirical....
Yet suppose this is all there is to democracy: that rulers are elected. Is it little? It depends on the point of departure.24 If one begins with a vision of a basic harmony of interests, a common good to be discovered and agreed to by a ratio-
Excerpted from: Adam Przeworski, "Minimalist Conception of Democracy: A Defense." In Democracy's Value, edited by Ian Shapiro and Casiano Hacker-Cordon. Cambridge: Cambridge University Press, 1999. © Cambridge University Press, 1999. Reprinted with the permission of Cambridge University Press.
nal deliberation, and to be represented as the view of the informed majority, the fact that rulers are elected is of no particular significance. Voting is just a time-saving expedient (Buchanan and Tullock 1962) and majority rule is just a technically convenient way of identifying what everyone would or should have agreed to. Yet if the point of departure is that in any society there are conflicts, of values and of interests, electing rulers appears nothing short of miraculous.
Let us put the consensualist view of democracy where it belongs—in the Museum of Eighteenth-century Thought—and observe that all societies are ridden with economic, cultural, or moral conflicts. True, as the modernization theory (notably Coser 1959) emphasized, these conflicts can be "cross-cutting": they need not pit class against class or religion against religion. They can be attenuated by an "overlapping consensus": consensus about practicalities compatible with differences of values (Rawls 1993). They may be also moderated by public discussion of both normative and technical reasons, although, as I have argued above, deliberation is a two-edged sword, for it may lead just to solidifying conflicting views. Yet in the end, when all the coalitions have been formed, the practical consensus has been elaborated, and all arguments have been exhausted, conflicts remain.
My defense of the minimalist conception proceeds in two steps. I take it as obvious that
2. Widely cited statements in this vein are Weifort 1992 and Schmitter and Karl 1991, but the phrase is ubiquitous. Here is Shapiro (1996: 108): "If democracy does not function to improve the circumstances of those who appeal to it, its legitimacy as a political system will atrophy." Even Kelsen (1988 [1929]: 38) poses the threat that "Modern democracy will not live unless the Parliament will show itself an instrument appropriate for the solution of the social questions of the hour."
24. Shapiro (1996: 82) also takes this position.
Defining Democracy
13
we want to avoid bloodshed, resolving conflicts through violence.25 Starting with this assumption, I first argue that the mere possibility of being able to change governments can avoid violence. Secondly, I argue that being able to do it by voting has consequences of its own.
Popper's defense of democracy is that it allows us to get rid of governments peacefully. But why should we care about changing governments?26 My answer is that the very prospect that governments may change can result in a peaceful regulation of conflicts. To see this argument in its starkest form, assume that governments are selected by a toss of a, not necessarily fair, coin: "heads" mean that the incumbents should remain in office, "tails" that they should leave. Thus, a reading of the toss designates "winners" and "losers." This designation is an instruction what the winners and the losers should and should not do: the winners should move into a White or Pink House or perhaps even a palacio; while there they can take everything up to the constitutional constraint for themselves and their supporters, and they should toss the same coin again when their term is up. The losers should not move into the House and should accept getting not more than whatever is left.
Note that when the authorization to rule is determined by a lottery, citizens have no electoral sanction, prospective or retrospective, and the incumbents have no electoral incentives to behave well while in office. Since electing governments by a lottery makes their chances of survival independent of their conduct, there are no reasons to expect that governments act in a representative fashion because they want to earn re-election: any link between elections and representation is severed.
25. I am not arguing against Locke that violence is never justified, just that a system that systematically avoids it is preferable to one that does not.
26. I want to thank Ignacio Sanchez-Cuenca for posing this question.
Yet the very prospect that governments would alternate may induce the conflicting political forces to comply with the rules rather than engage in violence, for the following reason. Although the losers would be better off in the short run rebelling rather than accepting the outcome of the current round, if they have a sufficient chance to win and a sufficiently large payoff in the future rounds, they are better off continuing to comply with the verdict of the coin toss rather than fighting for power. Similarly, while the winners would be better off in the short run not tossing the coin again, they may be better off in the long run peacefully leaving office rather than provoking violent resistance to their usurpation of power. Regulating conflicts by a coin toss is then a self-enforcing equilibrium (Przeworski 1991: chap. 1). Bloodshed is avoided by the mere fact that, a la Aristotle, the political forces expect to take turns.
Suppose first that the winners of the coin toss get some predetermined part of the pie, 1/2 < x < 1, while losers get the rest.27 Winners decide at each time whether to hold elections at the next time and losers whether to accept defeat or to rebel. If democracy is repeated indefinitely from t = 0 on, the winner at t = 0 expects to get Dw = x + Vw(e, x) and the loser at t = 0 expects to get DL = (1 - x) + VL(1 - e, x), where V stands for the present value of continuing under democracy beyond the current round, e is the probability the current incumbent will win the next toss. Let "democratic equilibrium" stand for a pair of strategies in which the current winners always hold tosses if they expect losers to comply and the current losers always comply if they expect the winners to hold tosses. Then such an equilibrium exists if everyone is better off under democracy than under rebellion: if Dw > Rw and Dl > Rl, where R stands for the expected values of violent conflict for each of the two parties.
27. This analysis is based on joint work with James Fearon, still in progress.
Chapter 1
14
Moreover, the prospect of alternation may induce moderation while in office. Suppose that the current incumbent can either manipulate the probability, e, of being re-elected or can decide what share of the pie, x e [0,1], to take, or both. There are some initial values {e(0), x(0)}; at t = 1 the coin is tossed and it designates winners and losers. Whoever is the winner now chooses {e(l),x(l)}: the rules for this round, etc. Hence, rules are not given ex ante: the incumbent manipulates them at will. Yet there are conditions under which a democratic equilibrium exists in which the incumbents do not grab everything. If the cost of rebellion is sufficiently high for both, each incumbent will prefer to moderate its behavior while in office under democracy rather than provoke a rebellion by the current loser.
As Hardin (1989: 113) puts it, "for the constitutional case, the ultimate source [of stability] is the internal costs of collective action for re-coordination or, in Caesar's word, mutiny." Yet if the threat of mutiny were the only incentive to moderation, why would we ever adopt procedures that subject control over the exercise of rule to a lottery? If the relevant political actors knew what would happen as the result of an open conflict, they could just agree to a distribution that would have resulted from an open confrontation. Instead of a coin toss deciding who gets what, the distribution would be fixed to reflect the strength the conflicting political forces could muster in an open confrontation, x for one, (1 - x) for the other. So why do we have democracy: an agreement to toss a coin with probabilities e and (1 - e)?
The reason, in my view, is that it would be impossible to write a dictatorial contract that would specify every contingent state of nature. In turn, leaving the residual control—control over issues not explicitly regulated by contract— to the dictator would generate increasing returns to power and undermine the contract. Endowed with residual control, the dictator could not commit itself not to use the advantage to under-
mine the strength of the adversaries in an open conflict. Hence, to avoid violence, the conflicting political forces adopt the following device: agree over those issues that can be specified and allow the residual control to alternate according to specified probabilities. In this sense, the constitution specifies x, the limits on incumbents, and e, their chances in electoral competition, but a random device decides who holds residual control.
Yet we do not use random devices; we vote. What difference does that make?
Voting is an imposition of a will over a will. When a decision is reached by voting, some people must submit to an opinion different from theirs or to a decision contrary to their interest.28 Voting authorizes compulsion. It empowers governments, our rulers, to keep people in jail,29 sometimes even to take their life, to seize money from some and give it to others, to regulate private behavior of consenting adults. Voting generates winners and losers, and it authorizes the winners to impose their will, even if within constraints, on the losers. This is what "ruling" is. Bobbio's (1984: 93) parenthetical addition bares a crucial implication of the Schumpeterian definition: "by 'democratic system'," Bobbio says, "I mean one in which supreme power (supreme in so far as it alone is authorized to use force as a last resort) is exerted in the name of and on behalf of the people by virtue of the procedure of elections."
It is voting that authorizes coercion, not reasons behind it. Pace Cohen (1997: 5), who claims that the participants "are prepared to cooperate in accordance with the results of such discussion,
28. This sentence is a paraphrase of Condorcet (1986 [1785]: 22): "il s'agit, dans une loi qui n'a pas ete votee unanimement, de soumettre des hommes a une opinion qui n'est pas la leur, ou a une decision qu'ils croient contraire a leur interet."
29. Indeed, the oldest democracy in the world is also one that keeps more people in jail than any other country in the world.
Defining Democracy
15
treating those results as authoritative," it is the result of voting, not of discussion, that authorizes governments to govern, to compel. Deliberation may lead to a decision that is reasoned: it may illuminate the reasons a decision is or should not be taken. Further, these reasons may guide the implementation of the decision, the actions of the government. But if all the reasons have been exhausted and yet there is no unanimity, some people must act against their reasons. They are coerced to do so, and the authorization to coerce them is derived from counting heads, the sheer force of numbers, not from the validity of reasons.
What difference, then, does it make that we vote? One answer to this question is that the right to vote imposes an obligation to respect the results of voting. In this view, democracy persists because people see it as their duty to obey outcomes resulting from a decision process in which they voluntarily participated. Democracy is legitimate in the sense that people are ready to accept decisions of as yet undetermined content, as long as they can participate in the making of these decisions. I do not find this view persuasive, however, either normatively or positively. Clearly, this is not the place to enter into a discussion of a central topic of political theory (Dunn 1996a: chap. 4) but I stand with Kelsen (1998 [1929]: 21) when he observes that "The purely negative assumption that no individual counts more than any other does not permit to deduce the positive principle that the will of the majority should prevail," and I know no evidence to the effect that participation induces compliance.
Yet I think that voting does induce compliance, through a different mechanism. Voting constitutes "flexing muscles": a reading of chances in the eventual war. If all men are equally strong (or armed) then the distribution of vote is a proxy for the outcome of war. Referring to Herodotus, Bryce (1921: 25-6) announces that he uses the concept of democracy "in its old and strict sense, as denoting a government in
which the will of the majority of qualified citizens rules, taking qualified citizens to constitute the great bulk of the inhabitants, say, roughly three-fourths, so that physical force of the citizens coincides (broadly speaking) with their voting power" (italics supplied). Condorcet claims that this was the reason for adopting majority rule: for the good of peace and general welfare, it was necessary to place authority where lies the force.30 Clearly, once physical force diverges from sheer numbers, when the ability to wage war becomes professionalized and technical, voting no longer provides a reading of chances in a violent conflict. But voting does reveal information about passions, values, and interests. If elections are a peaceful substitute for rebellion (Hampton 1994), it is because they inform everyone who would mutiny and against what. They inform the losers—"Here is the distribution of force: if you disobey the instructions conveyed by the results of the election, I will be more likely to beat you than you will be able to beat me in a violent confrontation"—and the winners—"If you do not hold elections again or if you grab too much, I will be able to put up a forbidding resistance." Dictatorships do not generate this information; they need secret police to find out. In democracies, even if voting does not reveal a unique collective will, it does indicate limits to rule. Why else would we interpret participation as an indication of legitimacy, why would we be concerned about support for extremist parties?
In the end, the miracle of democracy is that conflicting political forces obey the results of
30. "Lorsque l'usage de soumettre tous les individus ä la volonte du plus grand nombre, s'introduisit dans les societes, et que les hommes convinrent de regarder la decision de la pluralite comme la volonte de tous, ils n'adopterent pas cette methode comme un moyen d'eviter l'erreur et de se conduire d'apres des decisions fondees sur la verite: mais ils trouverent que, pour le bien de la paix et l'utilite generale, il falloit placer Yau-torite oil etoit la force" (Condorcet 1986 [1785]: 11; italics supplied).
Chapter 1
16
voting. People who have guns obey those without them. Incumbents risk their control of governmental offices by holding elections. Losers wait for their chance to win office. Conflicts are regulated, processed according to rules, and thus limited. This is not consensus, yet not mayhem either. Just limited conflict; conflict without killing. Ballots are "paper stones," as Engels once observed.
Yet this miracle does not work under all conditions.31 The expected life of democracy in a country with per capita income under $1,000 is about eight years.32 Between $1,001 and $2,000, an average democracy can expect to endure eighteen years. But above $6,000, democracies last forever. Indeed, no democracy ever fell, regardless of everything else, in a country with a per capita income higher than that of Argentina in 1976: $6,055. Thus Lipset (1959: 46) was undoubtedly correct when he argued that "The more well-to-do a country, the greater the chance that it will sustain democracy."
Several other factors affect the survival of democracies but they all pale in comparison to per capita income. Two are particularly relevant. First, it turns out that democracies are more likely to fall when one party controls a large share (more than two-thirds) of seats in the legislature. Secondly, democracies are most stable when the heads of governments change not too infrequently, more often than once every five years (although not as often as less than every two years). Thus, democracy is more likely to survive when no single force dominates politics completely and permanently.
Finally, the stability of democracies does depend on their particular institutional arrangements: parliamentary democracies are much
31. The forthcoming paragraphs are based on Prze-worski, Alvarez, Cheibub, and Limongi 1996, and Przeworski and Limongi 1997b.
32. Expected life is the inverse of the probability of dying. The income numbers are in purchasing power parity international dollars of 1985.
more durable than pure presidential ones. The expected life of democracy under presidentialism is twenty-one years, while under parliamentarism it is seventy-two years. Presidential systems are less stable under any distribution of seats; indeed, they are less stable whatever variable is controlled for. The most likely reason presidential democracies are more fragile than parliamentary ones is that presidents rarely change because they are defeated in elections. Most of them leave office because they are obligated to do so by constitutionally imposed term limits. In turn, whenever incumbent presidents can run and do, two out of three win reelection (Cheibub and Przeworski 1996). Presidentialism thus appears to give an excessive advantage to incumbents when they are legally permitted to run for re-election and, in turn, to prevent the incumbents from exploiting this advantage, it obligates them to leave office whether or not voters want them to stay.
Here then are three facts: (1) democracies are more likely to survive in wealthy countries; (2) they are more likely to last when no single political force dominates; and (3) they are more likely to endure when voters can choose rulers through elections. And these facts add up: democracy lasts when it offers an opportunity to the conflicting forces to advance their interests within the institutional framework.
In the end then, the Popperian posture is not sufficient, because democracy endures only under some conditions. Elections alone are not sufficient for conflicts to be resolved through elections. And while some of these conditions are economic, others are political and institutional. Thus, a minimalist conception of democracy does not alleviate the need for thinking about institutional design. In the end, the "quality of democracy," to use the currently fashionable phrase, does matter for its very survival. But my point is not that democracy can be, needs to be, improved, but that it would be worth defending even if it could not be.
Defining Democracy
17
References
Bobbio, Norberto. 1984. The Future of Democracy. Minneapolis: University of Minnesota Press. Bryce, James. 1921. Modern Democracies. New York: Macmillan.
Buchanan, James and Gordon Tullock. 1962. The Calculus of Consent: Logical Foundations of Constitutional Democracy. Ann Arbor: University of Michigan Press.
Cheibub, Jose Antonio and Adam Przeworski. 1996. "Democracy, elections, and accountability for economic outcomes." Revised paper presented at the Conference on Democracy and Accountability, New York University, 27-9 April.
Cohen, Joshua. 1997. "Procedure and substance in deliberative democracy." In Jon Elster (ed.), Democratic Deliberation. New York: Cambridge University Press. Condorcet. 1986 (1785). "Essai sur l'application de 1'analyse a la probability des decisions rendues a la pluralite des voix." In Sur les elections et autres textes. Textes choisis et revus par Olivier de Bernon. Paris: Fayard.
Coser, Lewis. 1959. The Functions of Social Conflict. Glencoe: Free Press.
Dunn, John. 1996a. The History of Political Theory and other Essays. Cambridge: Cambridge University Press.
Hampton, Jean. 1994. "Democracy and the rule of law." In Ian Shapiro (ed.), NOMOS XXXVI: The Rule of Law, pp. 13-45.
Hardin, Russell. 1989. "Why a constitution?" In Bernard Grofman and Donald Wittman (eds.), The Federalist Papers and the New Institutionalism, pp. 100-20. New York: Agathon Press.
Kelsen,  Hans.   1988   (1929).  La  Democratie. Sa
Nature—Sa Valeur. Paris: Economica.
Lipset, Seymour Martin. 1959. "Some social requisites
of democracy: economic development and political
legitimacy." American Political Science Review 53:
69-105.
Popper, Karl. 1962. The Open Society and Its Enemies. London: Routledge and Kegan Paul. Przeworski, Adam. 1986. Capitalism and Social Democracy. Cambridge: Cambridge University Press.
-. 1991. Democracy and the Market. New York:
Cambridge University Press.
Przeworski, Adam, Mike Alvarez, Jose Antonio Cheibub, and Fernando Limongi. 1996. "What makes democracies endure?" Journal of Democracy 7: 39-55.
Przeworski, Adam and Fernando Limongi. 1997a. "Modernization: theories and facts." World Politics 49: 155-183.
-. 1997b. "Development and democracy." In
Alex Hadenius (ed.), Democracy's Victory and Crisis. Cambridge: Cambridge University Press. Rawls, John. 1993. "The domain of the political and overlapping consensus." In David Copp, Jean Hampton, and John H. Roemer (eds.), The Idea of Democracy. Cambridge: Cambridge University Press. Schmitter, Philippe and Terry Lynn Karl. 1991. "What democracy is ... and what it is not". Journal of Democracy 2: 75-88.
Schumpeter, Joseph A. 1942. Capitalism, Socialism, and Democracy. New York: Harper & Brothers. Shapiro, Ian. 1996. Democracy's Place. Ithaca: Cornell University Press.
Weffort, Francisco. 1992. Qual Democracia? Sao Paulo: Companhia das Letras.