- 1 Steven Medema (2000) provides a discussion of this history.
- 2 On the evolving definition of economics, see Roger Backhouse and Steven Medema (2009).
1The rational choice or “economic” approach to politics—public choice analysis—brought the self-interest axiom into the analysis of the political sector as part of a move to provide a model of political behavior useful for economic policy analysis and bring a greater degree of theoretical rigor to the analysis of the political process.1 One aspect of this line of inquiry involved the portrayal of voting as a utility-maximizing choice made by rational agents. The problem that immediately arose, though, was that the model implied that rational individuals will not vote, which flies in the face of empirical data on voter turnout. One of the oddities of the application of the rational actor model to politics was that it seemed to offer rather convincing explanations for many forms of legislative and bureaucratic behavior but could not explain this most basic of democratic political decisions. Rather than see this as evidence against the rational choice model, public choice scholars—political scientists as well as economists—attempted to salvage it through various modeling adaptations. This article examines this literature during the formative era of public choice analysis—the late 1960s and early 1970s, with the goal of shedding light on how and why this literature evolved as it did and what this implied for economists’ early moves into areas outside of the subject’s traditional domain.2 It is also hoped that this article will serve as a stimulus for further work by historians of economics and others on the challenges that economists encountered in applying the rational actor model outside of this domain and the responses and adaptations that these challenges engendered.
- 3 See, e.g., Duncan Black (1948), Black and Robert Newing (1952), and Kenneth Arrow (1951).
- 4 Not long after came Tullock’s Toward a Mathematics of Politics (1967), Downs’s Inside Bureaucracy ( (...)
2 The development of modern public choice theory dates to the late 1940s and early 1950s and the work of Duncan Black and Kenneth Arrow.3 Black fleshed out significant aspects of the theory of majority rule and proportional representation—including, most famously, the median voter theorem—in a series of papers that suggested a new way of conceptualizing the political process. It was in the late 1950s and early 1960s, though, that the literature began to mushroom, with the publication of several works that were to become seminal in the field, including Anthony Downs’s Economic Theory of Democracy (1957), Duncan Black’s Theory of Committees and Elections (1958), William Riker’s Theory of Political Coalitions (1962), James Buchanan and Gordon Tullock’s The Calculus of Consent (1962), and Mancur Olson’s Logic of Collective Action (1965).4 Taken together, these books and the publication over the next decade of the many articles and books that were stimulated by these early efforts provided a body of analysis that challenged the orthodox view of the operation of government and the political process.
- 5 See Pierre Force (2003) and Medema (2009) for extensive discussions of various facets of this issue
3The idea that some form of self-interest serves to motivate individual action in the economic realm has been a part of economic discourse since the time of Plato and Aristotle.5 The content given to the notion of self-interest and its interaction with other behavioral variables has varied widely across time and across persons, but the marginal revolution brought with it an increasing refinement of what it meant to behave in self-interested fashion. Utility and profit maximization provided both specificity and a measure congruence with mathematical techniques that allowed for the derivation of determinate results associated with assumed behaviors. The development of game theory brought a further refinement to the idea of rationality and explored how the process of making rational choices might vary across alternative institutional contexts. The increasing acceptance of rational choice theory, in turn, reflected or at least was accompanied by a belief that choosing should be a consistent process: If people make rational choices in one area, such as consumption decisions, they should also be expected to make rational choices in other areas—including occupational choice, family decisions, the legal arena, and the voting booth.
4 This long history of self-interest as an assumed (or even taken as given) motivating force in economic thought suggests some measure of resonance with at least casual empiricism. In making their desired grocery purchases, it seemed likely that people would buy the products that best suit their needs and preferences, given their budget constraints. Profit-seeking could reasonably be supposed to have a good deal to do with how firms went about their business. Other types of activities, such as charitable giving or sending flowers to the bereaved, were thought to lie outside of the realm of self-interest, but were easily explained by factors such as the benevolence and sympathy motives that Adam Smith tag-teamed with self-interest. This is not to say that these activities could not be subjected to economic analysis. If sympathy and benevolence govern charitable giving or the sending of flowers, this does not mean that demand and supply frameworks are not useful here—charitable giving varying with, say, consumer incomes and expenditures on flowers being greater in years of more abundant harvests. But in realms not so obviously market-oriented, the idea that behavior itself is driven by self-interest was not so much countenanced and not so greatly probed.
- 6 See Richard Posner (2001) and Medema (2007).
5 All of this changed dramatically in the second half of the twentieth century with the move to apply economic analysis to a host of areas traditionally considered non-economic: discrimination, law, politics, family life, religion and many others. The idea that self-interest motivates behavior in realms outside the strictly economic is not, though, a uniquely twentieth-century innovation. Jeremy Bentham and Henry Sidgwick described a world in which individuals apply the utilitarian calculus to the legal realm,6 and the self-interest motive can also be found in the history of political theory—for example, in Thomas Hobbes and in classical political theory.
- 7 See, e.g., Medema (2009).
- 8 Buchanan (1960) and Medema (2005) provide surveys of this literature.
- 9 This claim against A.C. Pigou and the Cambridge school has recently been challenged by Roger Backho (...)
6 The modern application of the self-interest axiom to the political realm was in part a reaction against the Pigovian, Keynesian, and standard political science approaches which (implicitly or explicitly) assumed that the primary concern of actors in the political arena, be they voters, legislators, or bureaucrats, is to make policy choices in the best interests of society as a whole—a benevolent public servant approach, if you will. The technical economic formulation of this was that the political process is an exercise in maximizing a particular social welfare function which translates the preferences of the members of society into the optimal outcome for the nation as a whole. Yet throughout the nineteenth and early twentieth centuries, there were hints—some stronger than others—in the economic policy literature that the political process may well not generate the social-welfare-maximizing outcomes contemplated by economic theory, and, indeed, that political actors may not have the larger social interest at heart at all.7 These suggestions of potential inadequacies in government behavior were never formally modeled or probed with any degree of sophistication, and certainly with nothing that resembled what was happening on the market failure side. In essence, there was a dichotomy—some would argue inconsistency—in the theory of market failure, which assumed that people pursue their self-interest in the market place but pursue the public interest in the political arena. This dichotomy was evident in the Pigovian literature, as well as in much of the Italian public finance literature of the late 19th and early 20th centuries, in spite of the latter’s overt consideration of the political process.8 This inconsistency was harshly attacked in the 1960s by scholars including Ronald Coase (1960), James Buchanan (1962) and Harold Demsetz (1969), all of whom argued that the theory of economic policy included a very nice analysis of market failure, but no corresponding category or analysis of government failure.9 Public choice analysis began to change all of that.
- 10 For extensive surveys of this literature, see Dennis Mueller (2003), and William Shughart and Laura (...)
- 11 The Public Choice Society, for example, was formed in the late 1960s. See Medema (2000) for a discu (...)
- 12 Center for Study of Public Choice National Science Foundation proposal, 1970, p. 4. Buchanan House (...)
7 The economic, or rational choice, approach to the analysis of politics brought the model of the self-interested rational actor to bear on the analysis of the political process, including voting, legislative and bureaucratic behavior, interest group activity, and constitutional design.10 Public choice pioneers Buchanan and Tullock, writing in 1970 (and so in the very years of the field’s formalized existence11), characterized this analysis as one which analyzes government “not as an organization that pursues some ‘public interest’ or other higher goals, but rather as a somewhat mundane machine for achieving the concrete desires of individual human beings.”12
8At the level of the individual voter, public choice analysis posited that the choices people make at the polling place are those which they believe will best serve their own interests. As with markets, then, the pursuit of self-interest only serves the interests of society when self- and social interest are aligned. There are certainly instances when a voter’s view of her self-interest and the larger social interest coincide, whether due to preference-related effects or constraints faced by individual voters. However, there are any number of situations in which such coincidence will be lacking. For example, voters make their choices based upon their preferences and the stock of information they have about issues and candidates. Because information is costly to acquire, voters, as rational individuals, will only search for additional information—political or otherwise—so long as the expected benefit is at least as great as the expected cost. The inevitable result is that voters carry a degree of ignorance, perhaps substantial, into the polling place with them. Voters will tend to be most informed about issues that coincide with their particular interests, both because the benefits of having such information are more substantial and because organized groups promoting that interest make it their business to provide information to like-minded voters and thereby lower the cost to those voters of acquiring pertinent information. We thus expect to see members of labor unions particularly well-informed about union-related issues and retired military officers especially knowledgeable about candidate positions on national defense but less well informed about a host of other issues less important to their welfare.
- 13 In Tullock’s essay, there is a typo that causes “voters” to be italicized, rather than “should.”
9 The self-interest model was also employed to suggest that the politician’s interest lies in getting elected and then, as an office-holder, re-elected. This, in turn, implies taking positions on issues and casting votes in the legislature not based on an assessment of the society’s best interest, but upon what the politician believes will appeal to the largest number of voters. As Tullock (1987, 104) put it, politicians take positions “in terms of what they think the voters will reward, not what they think the voters should reward.”13 If politicians do indeed pursue their self-interest in the positions that they take on the issues, then we must expect that the social interest will only be served if self-interest and social interest coincide.
10 It may seem at first glance that attempting to please the voters would be in the best interests of society. Because many citizens do not bother to vote, however, there is already a disconnect between legislative positions that maximize votes and those that serve the larger social interest. And because voters operate with limited information, the legislation that attracts their interest, and thus votes, may have little to do with what is best for society. Thus, we end up with a host of expensive projects that benefit particular constituencies but likely would not pass a social benefit-cost test. Legislative vote-trading ensures sufficient support to get these projects (often referred to as pork-barrel projects) approved, while the cost of these programs—in dollar and opportunity cost terms—is spread across the community at large and thus remains largely hidden from the voters in any given district.
- 14 On rent seeking, see Tullock (1967a; 1993) and Anne Krueger (1974).
11 The vote-maximizing legislator will also be concerned to curry favor with special interest groups that are particularly influential in his district. Should a candidate’s background or voting record suggest an opposition to defense issues, for example, defense-related interest groups will be sure to get that information to pro-defense voters in his district. Given that special interests create voters who are informed and ready to act on issues that concern them, the theory predicts that they will have a disproportionate share of the influence on politicians and on election results. Of course, politicians receive much of their information about what voters will reward from the same special interests that provide information to the voters. This process involves the expenditure of vast sums to influence legislation—that is, rent seeking14—which, being largely an attempt to transfer resources from A to B (rather than to increase national welfare), is largely wasteful from an efficiency perspective.
- 15 See, for example, the treatments of bureaucracy in Tullock (1965), Downs (1967), and Niskanen (1971 (...)
12 Public choice analysis also extended the self-interest assumption to the bureaucracy, which is charged with implementing legislation. Political theory had traditionally characterized bureaucrats as public servants who operate in the public interest, following the directions set out by the legislators or their superiors within the bureaucracy. If, instead, they are rational maximizers, the immediate question is what the pursuit of self-interest equates to within the bureaucratic realm. The answer given to this was that the self-interested bureaucrat seeks to expand his or her sphere of bureaucratic influence: the size of the budget, the number of people supervised, the range of activities overseen, perquisites of office, promotions—all of these being indicative of power, influence, and importance.15 If every bureaucrat has an incentive to increase his domain, we would expect to see significant growth over time in the size of the government bureaucracy, particularly given that information problems limit the ability of legislators to effectively oversee bureaucratic activity. To the extent that the bureaucrat’s self-interest and the social interest fail to coincide, then, we expect to see inefficiencies at the bureaucratic level.
- 16 This issue is taken up in Medema (2009).
- 17 The most well-known of the critiques is Donald Green and Ian Shapiro (1994). On the present subject (...)
13 This combination of voters voting their self-interest, politician-legislators taking positions on issues and legislation that they believe will maximize their likelihood of (re-)election, and budget-maximizing bureaucrats called into question the ability and even willingness of government to enact and implement the types of efficiency-enhancing policies that were contemplated by neoclassical welfare theory. By pointing to a wide range of disparities between self-interest and social interest, the public choice models of political behavior pointed back to the types of concerns about state action evidenced in the literature from Adam Smith to John Stuart Mill to Henry Sidgwick and A.C. Pigou.16 At the same time, these models introduced to economic analysis a symmetry in the modeling process that cast governmental agents and their decisions in the same frame as those of individual producing and consuming agents. The operative issue, of course, was the extent to which this move was legitimate—theoretically, descriptively, and empirically—about which there was no small amount of controversy,17 and with the defenders of traditional approaches (whether in economics or political science) looking for chinks in the public choice armor. The paradox of the rational voter seemed to provide just such a chink.
- 18 This, of course, is reflected in the assumption of stable preferences, on which see George Stigler (...)
14From its inception, the axiom that has driven public choice analysis is the assumption that voters, politicians, and bureaucrats are self-interested rational actors. The extension of this axiom to the political realm assumed an underlying consistency in human behavior, that the same forces motivating behavior in one context also do so in other contexts.18 And if, as the economist assumes, it is self-interest that drives behavior in the marketplace for goods and services, this consistency demands that the self-interest must also govern behavior in other decision-making contexts. This suggested approach to thinking about behavior in the political realm placed public choice scholars in the vanguard of a movement that brought the application of the model of the self-interested rational actor to a wide range of traditionally non-economic subjects.
- 19 See, for example, the discussions in Bernard Berelson, Paul Lazarsfeld, and William McPhee (1954), (...)
- 20 See, for example, Sonja Amadae (2003).
15 Traditional political science approaches to voting behavior were grounded in sociology and social psychology. Here, the individual was assumed to vote out of feelings of civic duty, political interest, and social pressure, with the choices actually made at the polls reflecting party affiliation, the attributes of candidates, and the nature of the issues.19 The hypothesis of voter rationality and the probing of the implications of this assumption began to gain momentum with the publication by Anthony Downs of An Economic Theory of Democracy (1957). Downs was a student of Kenneth Arrow and, like Arrow, was affiliated with RAND in the 1950s. Given the role played by economists and others at RAND in pushing the boundaries of rational choice theory, it is not surprising (at least in retrospect) that the origins of the application of the rational choice model to the larger democratic process are to be found in this environment.20
16 For Downs, the assumption of voter rationality was a natural extension of the work of an economist, since “Economic theorists have nearly always looked at decisions as though they were made by rational minds” (1957, 4). While admitting that this constituted a simplification of reality, Downs considered it justified on the grounds that an assumed ordering of preferences is necessary if the analyst is to be able to generate predictions, and rationality is the ordering assumption made by economists. For Downs, then, the rational actor model was not meant to be a description of voter behavior, but rather a means of generating testable predictions, along the methodological lines laid out by Milton Friedman (1953).
17 It is important to understand what Downs meant when he used the term, “rational.” Rationality, here, meant nothing other than that an individual “moves toward his goals in a way which, to the best of his knowledge, uses the least possible input of scarce resources per unit of valued output” (1957, 5). Downs allowed that this leaves out certain aspects of the individual’s personality, “the rich diversity of ends served by each of his acts, the complexity of his motives, the way in which every part of his life is intimately related to his emotional needs” (1957, 7). His political man, like the rational consumer, remained “an abstraction from the real fullness of human personality,” approaching “every situation with one eye on the gains to be had, the other eye on costs, a delicate ability to balance them, and a strong desire to follow wherever rationality leads him” (1957, 7-8). These men are “artificial,” he said, and so are of limited comparability to individuals in the real world.
18 Downs went a step further, though, by combining the rationality axiom with a self-interest axiom, so that rational behavior, in his model, indicated “rational behavior directed primarily toward selfish ends” (1957, 27). Two points are worth noting here. First, Downs equated self-interest with selfishness—only one of several meanings that can be (and have been) ascribed to this term. But second, note that he used the qualifying language, “primarily.” Again, Downs allows the unrealism of a hard-nosed assumption:
In reality, men are not always selfish, even in politics. They frequently do what appears to be individually irrational because they believe it is socially rational—i.e., it benefits others even though it harms them personally. … In every field, no account of human behavior is complete without mention of … altruism; its possessors are among the heroes men rightly admire.
Nevertheless, general theories of social action always rely heavily on the self-interest axiom. (1957, 27-28)
- 21 “But man has almost constant occasion for the help of his brethren, and it is in vain for him to ex (...)
19Downs went on to cite Adam Smith’s allusion to the self-interest of the butcher and the baker, suggesting that Smith’s reasoning “applies equally well to politics” (1957, 28).21
20 Downs’s application of this conception of individual rationality to the political realm was a particularly narrow one and, as we shall see presently, would become the source of some controversy. Downs explicitly assumed that the only “political” benefits and costs that matter to the voter are those directly related to the outcome of the voting process. He did acknowledge that people may cast their votes rationally with broader concerns in mind—he offered the example of the man who votes in a particular way so as to avoid aggravating his wife (where the costs of aggravating one’s wife outweigh any potential benefits of voting differently)—but he nonetheless chose to restrict his political man to one who considers only the benefits and costs of political outcomes in deciding how to cast his vote. Thus, as Downs described it, “the axiom that citizens act rationally in politics … implies that each citizen casts his vote for the party he believes will provide him with more benefits than any other” (1957, 36). This altered the focus from standard political discourse, in that individuals here are not concerned about policies per se but about their own utility levels (1957, 42).
21Given that political agents are rationally self-interested and that their utility is a function of the outcome of the election, Downs’s argument posited that individuals vote in order to contribute to the achievement of a political outcome that reflects their preferences. Of course, voting involves the absorption of costs as well as the receipt of benefits. Downs thus modeled the calculus of voting as follows:
22R = (BP) - C,
- 22 The costs associated with voting received very little attention during the period considered in thi (...)
where R is the individual’s reward from voting, B is the differential benefit from his more preferred outcome triumphing over his less preferred outcome, P is the probability that the individual’s vote will make the difference between the more preferred outcome winning and losing, and C is the cost of voting. These costs, for Downs, were primarily the costs of time—to register to vote, to acquire information regarding candidates and parties, to evaluate the alternatives, to travel to and from the polling place, and to fill out one’s ballot.22
23 As Downs recognized, and as a litany of subsequent commentators later emphasized, it is unlikely that any one individual’s vote will influence the outcome of an election. This probability is so low, in fact, that BP is likely to be approximately zero. Even so, Downs pointed out that if there is any benefit at all from voting, even the smallest possibility that his vote could influence the outcome of the election, the rational voter will vote—if the cost of voting is zero (1957, 262). However, the cost of voting is never zero, as Downs acknowledged. This suggested two possible outcomes: if the benefits of voting are approximately zero and the costs are non-negligible, no one will vote. On the other hand, a rational voter may reason backward to the conclusion that he should vote, since the odds are that no one else will. Downs’s model thus predicted that either everyone votes or that no one does. Neither of these predictions is accurate, of course, meaning that his model appeared to violate Friedman’s dictum, with corresponding implications for the theory.
- 23 See, e.g., Downs (1957, 266, 269).
- 24 See, e.g., Olson (1965) and Morris Fiorina (1976).
24 Downs was thus faced with the need to address the question of why, in the face of these positive (albeit low, in his mind23) costs of voting, some people choose to vote and some choose not to vote. His answer involved assuming that rational individuals living in a democracy “are motivated to some extent by a sense of social responsibility relatively independent of their own short-run gains and losses” (1957, 267). Underlying this take on the issue was the assumption that people value living in a democracy and recognize that voting is necessary for democracy to function. Because people value democracy differently, assign different subjective values to the closeness of the election, care differently about which party wins, and have different views about how many other citizens will vote, it was apparent to Downs that different people, all of them rational, could perceive very different net benefits from voting (1957, 267-68). The problem with this result, as several commentators were to point out, is that the preservation of democracy has a public good aspect that generates its own collective decision problems.24
- 25 The statement by Hegel is as follows: “As for popular suffrage, it may be further remarked that esp (...)
25 This paradox of voting, as it came to be called, received virtually no attention in the decade following the publication of Downs’s book. Interestingly enough, even the reviews made no mention of it, though, as we shall see later, some questioned the application of the rationality postulate to politics generally. One reason for this neglect may be that the paradox of voting was not new to political scientists, who regularly cited its appearance in G.W.F. Hegel (1952) and B.F. Skinner (1948, 265), the latter of whom said that the odds of a voter affecting the outcome of a U.S. Presidential election is less than that of being killed while driving to the polls.25
26In spite of the problems posed by this paradox, the next moment in this discussion did not come until a full decade later, when Tullock generalized Downs’s model in his Toward a Mathematics of Politics (1967). Tullock’s innovative move was to allow for the possibility of altruistic behavior within the voting process, which led him to reformulate Downs’s expression of the reward from voting as,
27R = [(BP+BC)P]-Cv-Ci,
- 26 Taken together, Cv and Ci correspond to costs as set out by Downs (1957). Tullock argued that Cv ma (...)
where BP is the direct benefit the voter receives, BC is the benefit he receives from helping others receive benefits, Cv is the cost “in money and convenience” of voting (1967, 110), and Ci is the cost of “obtaining information” related to the voting process (1967, 113).26 As Tullock noted, however, the magnitude of P remained problematic: the probability of an individual influencing the results of the election is so small that the charitable benefit felt by the voter would have to be enormous in order for it to make voting rational. The same logic led him to question whether there was any value of Ci that would result in a positive return on information acquisition. Tullock’s model also suggested that people might vote against their direct self-interest simply to make others better off (1967, 111-112), a notion which seemed to be inconsistent with the larger public choice framework based on the validity of applying the axiom of self-interested behavior to the political process. Tullock was thus left to conclude that voting is likely to be irrational for most people (1967, 114), which meant that the rational voter model essentially remained an empirical failure.
- 27 The ratio of empirical work to theory in public choice remained extremely low for many years, in pa (...)
- 28 It may be of some interest that Riker and Ordeshook were equating “rational” and “reasonable” here. (...)
28Given the increased attention being given to public choice analysis in both economics and political science, the paradox of voting posed a significant problem for this emerging body of analysis. After all, if the theory could not explain a bedrock component of the democratic political process—the decision to cast a vote—how much reason was there for confidence in other aspects of the theory, the empirical evidence for which at that point was largely absent and, in fact, much less easy to establish?27 William Riker and his student Peter Ordeshook (1968), though, took issue with the pessimistic conclusions reached by Downs and Tullock, expressing a particular concern with their conclusion that voting was essentially an irrational act. Riker and Ordeshook found it “bizarre” that two scholars purporting to lay out a rational choice theory of politics would reach such a conclusion. “The function of theory,” they argued, “is to explain behavior and it is certainly no explanation to assign a sizeable part of politics to the mysterious and inexplicable world of the irrational.” They thus set out to, as they put it, “reinterpret the voting calculus” in order to make it “fit comfortably into a rationalistic theory of political behavior,” one that makes it “reasonable” for voters to vote and non-voters to not vote (1968, 25).28
29 In particular, Riker and Ordeshook sought to expand Downs’s analysis, arguing that his definition of rationality and his conception of benefits from voting were both overly narrow. The definition of rationality employed by Riker and Ordeshook invoked the theory of revealed preference in suggesting that, properly conceived, rationality implies “the ability to order preferences and to choose the more preferred action over the less preferred” (1968, 27n). Downs was sufficiently troubled by the tautological nature of this reasoning that he had explicitly rejected it. Yet, said Riker and Ordeshook, the definition that they employed is one “customarily used by economists” and had much to recommend it against the alternative utilized by Downs. Downs’s method, they argued, puts the theorist in the position of having to postulate what are the appropriate goals for individual action within a particular realm. Doing so is “unwise,” they said, in that it leads the theorist into “the trap … of saying … that one goal is rational and another is not.” Riker and Ordeshook, for their part, rejected the notion that it is “possible to judge the rationality of goals” and so were content to adopt the revealed-preference approach in developing their theory, reasoning that it “may more adequately describe behavior” both inside and outside of the political realm (1968, 27n.).
30 The definition of rationality posited by Riker and Ordeshook differed from that employed by Downs in allowing for the inclusion of non-political benefits and even various classes of benefits related to the political process but not strictly associated with political outcomes. Among the other benefits from voting pointed to by Riker and Ordeshook were: (i) “the satisfaction from compliance with the ethic of voting” (part of being socialized in democratic tradition); (ii) “the satisfaction from affirming allegiance to the political system,” (iii) “the satisfaction from affirming a partisan preference” (stand up and be counted, win or lose, which helps explain why people vote for even hopeless candidates), (iv) “the satisfaction of deciding to go to the polls, etc.”—satisfaction out of just participating in the voting process, becoming informed, etc. for its own sake; and (v) the satisfaction of affirming one’s efficacy in the political system” (1968, 28). That is, their model allowed for the incorporation of a wide range of citizenship- and political system-related preferences which could influence individual voting decisions.
31 The incorporation of these factors into the model transformed the calculus of voting into an expression not unlike that of Downs, but with rather different potential implications for voting. Riker and Ordeshook hypothesized that the voting decision takes the form:
32R = (BP) – C + D
- 29 Riker and Ordeshook also noted that the expected cost of voting could be influenced by factors such (...)
where D accounts for the aforementioned benefits. Riker and Ordeshook more or less followed previous treatments on the cost front, though they adopted a somewhat more subjective approach to costs and allowed that their magnitude may be influenced by the factors that make up D—e.g., that someone who believes that it is extremely important to vote “is likely to minimize [in their mind] the costs of voting,” and vice-versa (1968, 37).29 Thus, the benefits captured in D could outweigh the cost of voting for some individuals but not for others and, so modeled, generated a calculus that would account for some large fraction of the population voting, but not necessarily everyone. The case for the rationality of going to the polls was further buttressed by Riker and Ordeshook’s suggestion that agents may tend to overestimate the probability that their vote will influence the outcome of the election (P), the effect of which would be to further boost voter turnout.
- 30 Subsequent literature (lying beyond the scope of the present analysis of the early stages of public (...)
33 Given the smallness of BP, the explanation for voting behavior in the Riker and Ordeshook model basically boiled down to an assertion that some people vote because the psychic benefits of voting outweigh the costs, while others do not vote because the costs of voting outweigh the psychic benefits. This meant that explaining the voting decision ultimately came down to explaining why some people perceive big psychic benefits from voting and others do not (John Ferejohn and Morris Fiorina, 1974, 526).30 Brian Barry (1970, 16) argued this only shifted the problem, in that Riker and Ordeshook’s D term, while incorporating factors that would seem to be relevant to the voting decision, included many of the same factors that political science scholars referred to when they talked about voting out of a “sense of duty.” As such, said Barry, the model still did not offer an explanation for why some people perceive those benefits to be greater, and others less. In sum, while Riker and Ordeshook had offered a model intended to rescue the theory of the rational voter by introducing an alternative (in their minds, more proper) interpretation of rational actor theory, the jury was still very much out on the question of whether the model had sufficient explanatory power or offered a truly new paradigm for thinking about the voting decision.
- 31 See Ferejohn and Fiorina (1975, 921), citing unpublished experimental research by Charles Plott and (...)
34The modeling of the rational voter took on a somewhat different cast in the first half of the 1970s, with John Ferejohn and Morris Fiorina’s (1974) argument that the basic flaw in the Downs, Tullock, and Riker and Ordeshook models was that they assumed that rational decision making means expected utility maximization. Against this, Ferejohn and Fiorina pointed out that rationality—which they defined as “purposeful behavior” (1974, 525.n1)—takes other forms as well, as recent advances in the theory of games had demonstrated, and that individuals may well use different decision rules in different contexts.31 Ferejohn and Fiorina (1974, 535) believed that “the concept of rational behavior is more ambiguous than many of us take it to be,” and that this “simple but often overlooked point,” as they put it, was at the root of the problems with earlier rational choice-based voter models. They also suggested that the failure of previous models to explain observed voting behavior derived from a basic flaw in the perspective that framed the analysis: These models had posited voting as an investment decision, where individuals devote resources to voting because they expect to receive a return in excess of the cost. The fact that these models failed to accurately predict voting behavior implied that the notion of voting as an investment decision was wrongheaded, and Ferejohn and Fiorina proposed instead to model voting as an act of consumption.
35 The model that they offered to describe voting behavior employed minimax regret criterion, which holds that an agent will choose the outcome that minimizes his maximum regret associated with the various possible electoral outcomes. The distinction between this approach and the earlier models was made clear in an illustration that they offered:
If asked why he voted, a minimax regret decision maker might reply “My God, what if I didn’t vote and my preferred candidate lost by one vote? I’d feel like killing myself.” Notice that for the expected utility maximizer the probability of such an event is very important, whereas for the minimax regret decision maker, the mere logical possibility of such an event is enough. (Ferejohn and Fiorina, 1974, 535)
- 32 Ferejohn and Fiorina did not discuss the costs associated with voting, noting only that the minimax (...)
36Thus, while the expected utility maximization criterion had predicted that almost no one would vote, the minimax regret criterion provided a straightforward rationale for why people would go to the polls.32
- 33 For example, maximin decision makers would never vote, which might explain why “the poor and cultur (...)
37 Ferejohn and Fiorina acknowledged that minimax regret did not explain the behavior of all voters,33 let alone behavior in a variety of non-voting contexts, but their assumption that people use different decision rules in different contexts freed them from the constraints imposed by the consistency criterion. What remained unexplained by the model, though, was why people use, say, expected utility in some contexts and minimax regret in others—and in the voting context in particular. That is, while minimax regret offered an explanation for why many people—far more than predicted by the expected utility model—turn out to the polls, Ferejohn and Fiorina were not able to offer a convincing rational for why people might use the minimax regret rule to guide their decisions regarding voting.
38 The minimax regret analysis provoked a great deal of reaction in defense of the expected utility approach—far more, in fact, than did Downs’s initial demonstration that suggested voting was an irrational act. Commenting on these reactions a year later, Ferejohn and Fiorina claimed that something other than scientific concerns lay behind these attempts to prop up the expected utility model. They were convinced that the Downs model had “lived on not because it has any descriptive value, but rather because theorists who accept the prescriptive model underlying it refuse to let it die” (1975, 925, emphasis added). That is, they believed that it was the normative notion that the voting process should reflect rational choices that kept scholars wedded to the expected utility maximization model in spite of all of the evidence against it. Against this, Ferejohn and Fiorina argued not so much that their model was the correct way in which to characterize the voting process, but that they were “attempting to fan the fires of agnosticism” against an expected utility model that appeared to them to be fundamentally flawed.
39 This view of voting as a consumption activity gave rise to a more broad-based conception of rationality in the voting process, via what have come to be called “expressive voting” models. In spite of the fact that the genesis of these models was a reaction against what were seen as previously flawed attempts at the construction of a rational voter model, these models actually took as a point of departure Riker and Ordeshook’s hypothesis about intrinsic benefits that voters receive from going to the polls, which are captured in their “D” term. Fiorina (1976) labeled this the “expressive” aspect of voting, which allows one to “express solidarity with one’s class or peer group, to affirm a psychic allegiance to a party, or simply to enjoy the satisfaction of having performed one’s civic duty” (1976, 393). As Geoffrey Brennan and Alan Hamlin later defined it, expressive voting is “where citizens vote to identify themselves with particular positions and to register support for those positions, rather than to bring certain policies about” (1998, 172). This expressive aspect, then, reflected the depiction of voting found in traditional political science accounts and also offered a response to the problem that the instrumental model’s focus on electoral outcomes did not offer a convincing account of rational voter behavior. So conceived, voting was loosed from its connection to electoral outcomes and, in the words of Brennan and Hamlin, became an activity “much more like cheering at a football match than it is like purchasing an asset portfolio.” Though the rational choice framework was retained, at least one significant aspect of an ‘economics of politics’ was lost in this move, since the implication of this approach was that “any analogy [between voting and] market choice is inappropriate” (1998, 150).
40 Expressive voting models solidified the characterization of voting as an act of consumption, the outcome of a rational decision to express one’s preferences. These models hypothesized that an individual will go to the polls and vote for a particular candidate if that candidate has a bundle of attributes for which the individual wishes to register a preference. Such attributes could include party affiliation, religious or moral stance, ideology, race, gender, and even things like rhetorical skill. A racial difference between candidates may offer no instrumental benefit—that is, it may not make one iota of difference in what policies are enacted—but the individual may derive satisfaction simply from having more members of his race in the legislature and so vote on that basis. The expressive model had the added benefit of offering an explanation for why individuals vote even when a candidate is running unopposed: people wish to express support for the things that they value. From an instrumental perspective, in contrast, there is no reason whatsoever to go to the polls if a candidate is running unopposed, given that the probability of influencing the outcome is zero.
41 When expressive voting ideas were combined with the instrumental aspect (the benefit from influencing the outcome of the election in one’s preferred direction), the result was a more broad-based approach to voting than had been offered by either the Downs or traditional political science models. And, as Fiorina (1976) showed, this combined approach could be modeled within the expected utility framework favored by economists, with the decision to cast a vote expressed by the equation,
42R = (BP) – C + D + B
- 34 Fiorina did not take up the costs associated with the voting process.
43The expressive aspect enters the benefit equation directly though B, which describes to what extent the voter benefits simply from expressing his or her preferences in the polling place.34 One of the interesting aspects of this model is that allowed for direct personal interest and party position to be in conflict with each other—that is, for situations in which the other party’s platform is closer to one’s preferences than is one’s own party. This would be the case when, for example, an individual’s instrumental considerations would point to voting in favor of a candidate who favors free trade or looser immigration policies, but where that individual identifies expressively with the party opposing such policies. And, of course, the expressive model also overcame the central flaw of the earlier models of rational voting by predicting that there will be a positive voter turnout: because individuals derive widely varying benefits from expressing their preferences, some individuals will participate in any given election while others will abstain. It also predicted that those who do vote will hold positions fairly similar to those of the candidates themselves, a prediction which tends to be borne out by the empirical evidence. That said, this approach was certainly not without its shortcomings. For example, while embedding expressive voting within an expected utility framework allowed for the derivation of the general conditions under which an individual will go to the polls, it is difficult to specify expressive issues with precision—particularly with the same functional precision and degree of determinacy that were found in the purely instrumental approach.
44Though public choice analysis had a number of theoretical successes to its credit in the early years—if success is measured by congruence of the theory with observed political or governmental phenomena—the earliest attempts to model voting as a rational investment in achieving a desired political outcome that provides net benefits to the voter failed to achieve the same level of success. Given that the expected benefit from voting in this framework (PB) is effectively zero and the costs of voting are not, the model suggested, as Downs and Tullock pointed out, that it is irrational for an individual to go to the polls. The debate then moved to the question of whether civic duty and related factors could appropriately be classified as aspects of a process of rational choosing. The expansion of the scope given to rationality in the late 1960s and early 1970s allowed for a somewhat less ad hoc approach to the voting decision, as the idea of expressive voting was developed to reflect the idea that people could rationally chose to engage in an activity because it makes them feel good. Just as the decision to attend college can reflect for some an investment in one’s future and in others a preference to be able to discuss Plato intelligently at cocktail parties, so voting could be for some an investment in achieving a particular political outcome and in others a means of preference satisfaction—or reflect the interaction of the two. The key to formalizing all of this was provided by the recognition, by Fiorina, that the expected utility model was broad enough to encompass both instrumental and preference-based concerns.
45 What are we to make of all of this to-and-fro over the rational voter in the early stages of the development of public choice analysis?
- 35 The exception here is the discussion in Barry’s 1970 book, which was really the only opposition to (...)
- 36 See, for example, Green and Shapiro (1994). Medema (2000) offers some speculation as to the reasons (...)
1.What stands out is that the debate over the rational voter model was largely between rational choice scholars—be they economists or political scientists—rather than between proponents and opponents of the rational choice approach to politics. Even the reviews of Downs’s book did not mention the issue, and it was only much later that those critical of the rational choice approach to politics entered the fray in any significant way.35 There was certainly criticism of the larger rational choice approach to politics within the political science community in the early years, though perhaps not as much as one might expect, and it was not until the 1980s that political scientists began to seriously attack rational choice political theory.36
2. What is clear is what the reaction did not involve. There was no sentiment expressed that the modeling of voters as rational choosers should be jettisoned. What ensued, rather, were a series of attempts to prop up the rational voter model by bringing into it a series of motives that were part of the very body of ideas that the rational choice approach was said to replace. “Soft” concepts such as civic duty, habit, and other things not consistent with making adjustments at the margin on the basis of changes in relative benefits and costs were said to be factors in the rational decision-making process.
It is interesting that, as we noted earlier, very little attention was paid to the cost side of the voting decision during the period considered here, particularly after the initial work by Downs. Down’s view of voting costs as consisting of the costs of gathering information and going to the polls, conceived of in terms of the opportunity cost of time, was explicitly or implicitly carried through the subsequent literature. Given the paucity of discussion related to the cost side, it is difficult to know what explains this lack of attention. One possibility is that scholars simply felt that the general opportunity cost conception initially posited by Downs captured virtually all of the relevant costs and so had nothing more to add to the discussion. The sense that one gets from this literature, though, is that the answer lies at least as much in the view that the problems with the model were on the benefit side, and that some scheme needed to be developed to model benefits in a way that would generate a rational explanation for voting. Scholars thus devoted their efforts to this aspect of the model, rather than the cost side, in order to develop a rational choice system within which voting made sense—that is, offered some non-negligible expected benefit that was sufficient to outweigh the costs for reasonable numbers of voters.
3.Why did some rational choice scholars work so hard to rescue the rational voter model and keep it alive? Was it a matter of ideology—a normative motivation—as some have suggested? And if so, in what sense? Was it that some of the proponents of the rational choice approach were simply interested in putting governmental policy on a rational footing and so in line with the principle espoused by some that government should, to the extent possible, function according to the same set of principles which govern activity in the market context? Or was it instead a belief that the application of rational choice principles in the governmental context would lead to a more minimalist state?
Of course, the normative side of rational choice politics is only one possible rationale for all of this. We must also bear in mind that the extension of the rational actor model into politics was only one facet of a larger movement of economic analysis in general and rational choice theory in particular into the social sciences generally. These extensions were not without opposition, either within the disciplines being “invaded” or within economics itself. It may be that some rational choice scholars felt that in order to be viewed as a potentially progressive comprehensive research program the rational actor model had to be able to explain something as fundamental to democratic life as the decision to vote—particularly in light of the ostensible links between democratic politics and the market system—or, more generally, that the view in certain quarters that economics was the ‘science of choice’ required that there be a symmetry between models of agent behavior across the choice spectrum.
- 37 See, e.g., Mueller (1979) and Bernhard Grofman (1993).
4. What does all of this say about the concept of rationality? Rational choice scholars were unwilling to allow that irrational elements explained the decision to vote and responded to Downs’s and Tullock’s suggestions of irrationality in one of three general ways. Some took voting motives previously said to be irrational and wrapped the notion of rationality around them. Others suggested that rationality can take different forms in different contexts. Still others (perhaps the majority of public choice scholars) recognized the problems associated with the rational voter model but were of the mind that these attempts to rescue the model were unsatisfactory and so were content to live with the paradox.37 These responses raise questions about the definition use of the concept of rationality—of what it means to be rational, of the role of the consistency criterion within the rationality axiom, and how economists respond when rationality-based models do not perform as hoped. These were issues that were very much in play in economic analysis generally during this period.
- 38 As Brennan and Hamlin (1998, 167) later put it, “whether [people] vote instrumentally or expressive (...)
46 Downs developed an abstract model of the rational voter that was intentionally targeted toward generating testable predictions rather than toward embodying descriptive accuracy. The model failed the tests. The expressive voter models moved us away from the arenas of abstraction and testable predictions, toward a very descriptive, realistic portrayal of voter behavior. What was being argued, at least implicitly, is that voting decisions are influenced by social norms, and that these norms are themselves the outcomes of a rational choice process. While the models could be evaluated according to their consistency with observed voting behavior, they did not allow for any sort of test of the rational voter hypothesis. In effect, expressive voting models took a “what is, is rational” approach, and so remind us of the concerns about the potentially tautological nature of the rationality assumption first expressed in this context by Downs. The model moved, then, from testable implications to irrefutable assumption.38 This move was not unique in the history of ‘economics imperialism,’ but public choice was at the forefront of this process and, in raising the issue of what it meant to be rational and how economists could model a wide range of behaviors in a rational choice context, was a harbinger of things to come.
The author wishes to thank Philippe Fontaine, Roger Backhouse, participants in the ANR History of Altruism Research Group Workshop, the editors, and three anonymous referees for their insightful comments on previous drafts of this article. The financial support of the Agence Nationale de la Recherche is gratefully acknowledged, as is permission to quote from materials in the Buchanan House Archives at George Mason University.