The Reason of Rules: Constitutional Political Economy
By Geoffrey Brennan and James M. Buchanan
- Ch. 1, The Constitutional Imperative
- Ch. 2, The Contractarian Vision
- Ch. 3, The Myth of Benevolence
- Ch. 4, Modeling the Individual for Constitutional Analysis
- Ch. 5, Time, Temptation, and the Constrained Future
- Ch. 6, Politics Without Rules, I
- Ch. 7, Rules and Justice
- Ch. 8, Politics Without Rules, II
- Ch. 9, Is Constitutional Revolution Possible in Democracy
Modeling the Individual for Constitutional Analysis
Any analysis of the effects of alternative rules on patterns of outcomes emergent from social interaction must embody some assumptions about the nature of the persons who interact. In this chapter, our objective is to set out our own assumptions and to defend them. This is no small order. The basic model of the individual that we shall advance as the most appropriate for comparative institutional or constitutional analysis—for investigating the implications of alternative sets of rules—is basically the
Homo economicus construction of classical and neoclassical political economy. Earlier experience demonstrated to us that the use of this model, particularly outside conventional market settings, is widely regarded (even within the economics profession) as one of the most objectionable features of what has been called the new “economic imperialism.” We do not propose to, and indeed cannot, restrict our discussion to behavior in alternative market orders. Much of our attention will be directed toward behavior in nonmarket institutions, and to behavior in political processes in particular. Hence, we shall have to confront the general antipathy to the use of the
Homo economicus model in nonmarket settings and to deal with possible objections.
There is no question that such antipathy exists. In some of our recent work, for example, in which we attempted to develop a theory of the fiscal
*17 and monetary
*18 powers to be assigned to political agents, we modeled those agents precisely as their counterparts in market contexts are modeled by economists. The thrust of that work was to test the adequacy of constraints on the behavior of political agents under prevailing political institutions and to compare such constraints with those that might become operative under alternative structures. In part, we were examining the question as to whether constraints other than those present in the operation of electoral competition under majority rule might be required to secure tolerable outcomes for the citizenry.
Reactions to the work were predictably diverse, but there was no mistaking a widely shared skepticism about the use of the
Homo economicus model for human behavior in political contexts. In one particularly strident commentary, for example, our “cynical” attitude toward human motivation was characterized as having “fascist”
*19 overtones. And in remarks to the American Academy of Arts and Sciences, no less a personage than Paul A. Samuelson advanced similar charges against “the boys from Chicago” and the whole tax limits movement.
We therefore have both a personal and a professional interest in establishing the legitimacy of the
Homo economicus postulate about human motivation, at least in the specific context of evaluating alternative social rules.
*21 In more general terms, however, our experience is simply further evidence for the hypothesis that many social analysts feel most comfortable with some sort of “public interest” or “benevolent despot” model of government (or agents in governmental roles). Preference (often implicit) for the public-interest model may, however, stem not from empirical, intellectual, or ideological foundations but from sheer analytic convenience. The public-interest model enables the analyst to proceed with normative policy evaluation unhindered by any concern with policy implementation. To judge a social outcome “unsatisfactory” by some ethical standard is, almost definitionally, to reckon that the outcome ought to be changed. The manner in which such a change is to be secured is not a matter to be taken too seriously at the purely ethical level, but the implication seems to be that since the problem is a “social” one, “society” must act. And since the notion of society “acting” is strictly meaningless, it seems natural to replace society with its agent, the “government.” In this mind-set, the implicit presumption that government will, in fact, do what is normatively suggested may not appear to be relevant. The implications of “social ethics” as distinct from “private ethics” seem to be that government either has or ought to have the powers to act as the analyst recommends and that the government will so act once the moral force of the normative recommendation is understood. Of course, the analyst might respond by saying that no such implications are intended in his exercise. He might conceive his task to be limited to the specification of what policy options are “good” and what options are “bad”; the implicit model of government and politics becomes a mere methodological convenience for him. He would defend his procedures by arguing that the evaluation of policy options is a useful exercise, quite apart from any attempt to model political behavior explicitly. So it may be; but as we have previously noted, a similar argument could be made in support of exercises involving the imagination of undiscovered realms of resource and energy potential. We suggest that even if advanced solely for convenience in evaluating what the analyst thinks are relevant policy options, the benevolent despot model of politics and government has promoted and sustained monumental confusion in social science, and social philosophy more generally.
We do not propose to introduce the
Homo economicus model for any comparable analytic convenience. Nor do we wish merely to strike some sort of balance between this model, on the one extreme, and the benevolent despot model, on the other. Our use of
Homo economicus stems from our conviction that this model is the most appropriate one for constitutional analysis. The remaining sections of this chapter deal with various lines of defense of the
Homo economicus construction in the constitutional context.
Homo economicus in Politics: The Argument for Symmetry
On the basis of elementary methodological principle it would seem that the
same model of human behavior should be applied across different institutions or different sets of rules. The initial burden of proof must surely rest with anyone who proposes to introduce differing behavioral assumptions in different institutional settings. If, for example, different models of human behavior were used in economic (market) and political contexts, there would be no way of isolating the effects of changing the institutions from the effects of changing the behavioral assumptions. Hence, to insist that the basic behavioral model remain invariant over institutions is to do no more than apply the ceteris paribus device in focusing on the question at issue.
If an individual in a market setting is to be presumed to exercise any power he possesses (within the limits of market rules) so as to maximize his net wealth, then an individual in a corresponding political setting must also be presumed to exercise any power he possesses (within the limits of political rules) in precisely the same way. If political agents do not exercise discretionary power in a manner analogous to market agents, then this result must follow because the rules of the political game constrain the exercise of power in ways the rules of the market do not, which is to say that the constraints are not comparable in the two settings. Otherwise, there must be an error in analysis or observation. No other conclusion is logically possible, given the invariance of the behavioral model across institutions.
This procedure does not, of course, rule out the possibility that actual behavior in differing institutional contexts will be different. What it does exclude is the introduction of behavioral difference as an analytic assumption. If behavioral differences are attributable to differences in rules, it must be possible to link the rules in some way to the behavioral patterns they generate, without resort to separate fundamental models of behavior, which can do nothing but guarantee emptiness in any attempted institutional comparison.
We propose to analyze behavior as economists generally do, that is, in terms of the choices among alternative courses of action faced by individual agents. The conceptual apparatus here involves a radical separation between means and ends—between opportunity sets and preferences. “Explanations” of outcomes are advanced with reference to the relative prices or costs of the options (and of changes in those prices) rather than to the preferences of the agents. In this sense, for the economist, the only differences in institutions that are relevant for explaining behavioral differences are the differences in the prices of the alternatives. Pure differences in preferences (that is, differences that cannot ultimately be traced to differences in relative prices) wrought by institutional change cannot be brought within the explanatory power of the economist.
The analytic approach that relates preference shifts to institutional differences skirts dangerously close to, and may be indistinguishable from, the use of models wholly different from those mentioned earlier. Such an approach might suggest that individuals assume roles that are institution-dependent, that in politics, for example, persons take on character roles as “statesmen,” whereas in the market they take on character roles as “possessive profit seekers.” In ignoring this whole approach and the literature in which it has been developed, we realize that we are limiting our realm of discourse and dialogue. But the analytic presumptions in support of behavioral symmetry at the most basic level seem so strong that the onus of proof must lie with those who would advance the institution-dependent behavioral model.
The symmetry argument may seem elementary and self-evident, but its acceptance surely carries us part of the way toward the use of
Homo economicus in politics, at least among those economists who remain content to adhere to such a behavioral model in their analyses of markets. In a more general sense, of course, the symmetry argument does nothing to establish
Homo economicus as the appropriate model of human behavior. Alternative models may be introduced. The symmetry argument suggests only that whatever model of behavior is used, that model should be applied across all institutions. The argument insists that it is illegitimate to restrict
Homo economicus to the domain of market behavior while employing widely different models of behavior in nonmarket settings, without any coherent
explanation of how such a behavioral shift comes about.
III. Science and the Empiricist Defense
The use of the
Homo economicus model requires more specific support. The most natural, and perhaps most familiar, argument is that people do, in fact, behave as the model suggests, at least on average and in the large. Due allowance must, of course, be made for the degree of abstraction necessary to introduce any generalized model and hence for instances of specific violation of the behavioral postulate. But once such allowance is made, the argument is that
Homo economicus offers a better basic model for explaining human behavior than any comparable alternative. Most modern economists would probably take this position, which receives its most explicit defense from members of the modern Chicago school.
The position is difficult to reject, and for two reasons. First, if one insists on a comparison of
Homo economicus with alternative behavioral models of roughly equal levels of abstraction and generality, many of the grounds for debate are swept away. Models of behavior that are psychologically richer may be rejected because of their failure to meet the implicit austerity test. Second, and more important,
Homo economicus is not well defined, and the would-be critic may find that his quarry has disappeared, only to reemerge in another guise. In specifying
Homo economicus as a net wealth maximizer, for example, one may fail to explain much of what can be observed, but the observations may not be definitive because the defender of the model may resort to changes in the specification of the choosers’ utility functions. In other words, the defender of
Homo economicus deflects the criticism of the content of preferences by the claim that the structure of preferences rather than content is the central element of the model.
As we shall indicate in Sections IV and V, our defense of the
Homo economicus model is basically methodological rather than empirical. Nonetheless, the methodological defense requires empirical presuppositions, and we can be quite specific in this respect. We simply require that demand curves slope downward, which, in turn, requires that it be possible to identify the “goods” individuals value. That is to say, our empirical presuppositions refer to the signed arguments in individuals’ utility functions and do not involve trade-offs among these arguments. Furthermore, we require that individuals consider their own interests, whatever these may be, to be different from those of others.
For economists whose purpose is quite different from our own, who seek to provide a positive-predictive “science of behavior,” whether in market or political settings, further specification of the model is required. And as research results have indicated,
Homo economicus, as an all-purpose explanatory model, runs into some apparent difficulties. Analysts are hard put to explain such behavior as individual voting in large-number electorates, individual volunteers in defense of the collectivity, and voluntary payment of income taxes. Fortunately, for our purposes, we are not required to discuss possible tests of the usefulness of the
Homo economicus model in this all-inclusive explanatory sense.
The reason is that for our constitutional argument, it is sufficient to defend the
Homo economicus construction as a “useful fiction,” while largely setting to one side the question as to just how “fictional” the individual in the model may be. Indeed, we make the stronger argument that
Homo economicus is uniquely useful for purposes of comparative institutional analysis and for ultimate constitutional design. But as previously noted, the argument here is methodological and analytic rather than empirical.
IV. A Methodological Defense of the Differential Interest Model of Behavior
A central presupposition in our methodological framework, then, is that individuals have interests that conflict. We can show how this presupposition is relevant to our constitutional analysis by introducing a simple analogy.
Suppose that A is thinking of engaging another person, B, to do something that A wishes to be done and that requires a substantial advance payment; for example, B may contract to build a house for A. Any defection on B’s part from his contractual obligation will impose costs on A, costs that may be significant. Hence, A will seek to limit the scope of such possible defection and will do so even if A thinks the likelihood of B’s defection is low. In the example, A will first inquire about the reputation of alternative builders before engaging B and, presumably, would not contract with B at all if the prospects for defection seemed high.
Despite this threshold acceptability of B, once agreement is reached, A may hire a lawyer to draw up a formal contract. And for purposes of the contract itself, A will make the working hypothesis that B is a rogue who is out to defraud him if the opportunity permits. He makes this hypothesis because this is the contingency the formal contract is designed to cover. In other words, the nature of the contract-drawing exercise leads A to make assumptions about B’s motivations that A may well believe to be a poor reflection of the empirical realities.
Nothing here is said about whether drawing up the formal contract is or is not justified. This decision will depend on the relative costs of the exercise. What the argument does suggest is that if the formal contract is to be executed, the ascription of differentially self-interested (
Homo economicus) motivations to the relevant party or parties is a logical part of the operation. The reason for the contract as such is its possible function in modifying B’s private interests in such fashion as to make these interests congruent with A’s. The focus in such a precautionary exercise is necessarily on B’s differential and separable interests.
Note that we do not need to claim that private interest, as it is normally defined, is the only or even the predominant motivation for human action in order to justify such a focus. Once we acknowledge that private and differentially identifiable interest is relevant at all to human behavior, a comparison of alternative institutions must attend primarily to the question of how those institutions operate when individuals act in pursuit of their separately defined interests. If there were no conflict among interacting agents, that is, if interests were not differentially identifiable, then, of course, there would be no concern about how alternative sets of rules might modify and transform such conflicts. But to deny that there are conflicting interests among persons is to engage in an absurd flight of fancy. Individuals’ objectives differ. Conflict exists, and the investigation of alternative institutions is ultimately motivated by some criteria of conflict resolution. In all such meaningful investigation, we simply must assume that agents’ interests conflict in order to focus on the central purpose of the whole analysis. In so doing, we are merely adjusting our analytic microscope to focus it on the subject of our concern.
When we examine the properties of the idealized market order, for example, we can agree with Adam Smith that the consumer does not depend for his supper’s meat on the benevolence of the butcher, without in the least ruling out the existence of such benevolence. The butcher may or may not be benevolent toward his customers. The crucial point is that such benevolence
need not be present, and hence whether or not it is present is essentially irrelevant. Speaking more generally, if we want to examine the extent to which a particular set of rules, such as that of the market order, succeeds in transforming the self-oriented interests of human agents into actions that further the interests of others, it is but natural for us to assume that such agents are entirely self-oriented, even if, empirically, they may not be. If we want to discover how institutional rules can turn conflict into cooperation, we cannot simply assume that persons who operate within those rules are naturally cooperative. Such a procedure would amount to removing the whole problem by assumption.
In any evaluation of alternative institutions, therefore,
Homo economicus is a uniquely appropriate caricature of human behavior, not because it is empirically valid but because it is analytically germane. Where the question of empirical validity arises is in evaluating the importance of the whole contractual or constitutional exercise. An institutional setting that operates so as to transform private self-interest into behavior that is profitable to individuals other than the actors, and that does so more effectively than other institutions, is relatively more valuable to the extent that private self-interest does motivate human behavior. If individuals tend naturally to be other-regarding in all but a few minor aspects of their behavior, then a preference for institutions that channel self-interest toward furtherance of the general interest is less pronounced, and other possible criteria for institutional evaluation become relatively more important. At base, therefore, empirical issues determine the significance of the whole constitutional exercise. But the analytic method for constitutional analysis is a separate issue, and in our dealing with the question of analytic method, empirical considerations do not enter, save in the threshold manner already indicated.
V. Social Evaluation and Quasi-Risk Aversion
We shall proceed with a somewhat varied elaboration of the theme introduced in Section IV. We have perhaps not yet fully established the case for the use of
Homo economicus in the generation of normative conclusions about the choice among alternative sets of rules. In particular, we have not discussed the implications of the elementary fact that more restrictive rules will not only help to prevent the occurrence of disaster but also often preclude actions that may be intended to promote desirable outcomes.
Consider an issue that was widely discussed in the early 1980s: the proposed “balanced-budget amendment” to the United States Constitution. We can analyze the role of constraints within existing political institutions
*24 by means of the self-interest behavioral model for actors. We can make some predictions as to the effects of the existing constraints of electoral competition on the proclivity of political agents, or coalitions, to create budget deficits. We can then ask how the behavior of these agents might be modified by the introduction of an enforceable balanced-budget constraint. But any complete analysis would also have to reckon with the possibility that such a balanced-budget rule, if operative, might sometimes restrict well-intentioned and far-seeing politicians from securing macroeconomic stability. (We do not propose to enter into the debate concerning whether systematic governmental intrusion in the macroeconomy can, even in theory, exert a stabilizing influence.) In one perspective on politics at least, any implied reduction in the governmental flexibility of response to unforeseen circumstances will embody potential costs that must be taken into account.
It is obvious that the degree of disinterested and far-seeing behavior on the part of political agents will be relevant to the comparison of expected costs and benefits of alternative fiscal rules. The model of self-interest, or
Homo economicus, will tip the balance of argument in favor of assigning less discretionary power to political agents than would be the case under the benevolence model. In this sense, the
Homo economicus model is not innocent, and its claim to empirical relevance must be addressed.
Even at this level of inquiry, however, the mere empirical record can be very misleading, particularly if this record is interpreted in a strictly predictive manner. That is to say, if we seek a model of human behavior that corresponds to some “best” prediction, either in the technical sense that the variance from observed values is minimized or in the heuristic sense that some fair average of observed behavior seems to fit, and then, at the second stage, try to compare institutions on the basis of such “best-fit” models, the results will be systematically biased in the direction of inadequate constraints. For these reasons, the
Homo economicus model may be justified despite the fact that it embodies more cynicism about persons’ behavior patterns than the simple evidence warrants. To put the same point differently, if we array models of behavior along a conceptual spectrum from “worst-case” to “best-case” poles, the model that is appropriate for making a comparison among social arrangements is somewhat closer to the worst-case pole than that corresponding to the simple “average” description of behavior.
The line of reasoning here is that there is, in the evaluation of institutional alternatives, an intrinsic feature that imposes a sort of risk aversion on the evaluator. We should emphasize that we do not assume that individual citizens, either behind some veil of ignorance or as located in society, are inherently risk-averse in the normal meaning of this term. For our argument here, we may take individuals to be strictly risk-neutral. It is the peculiar setting of choice that causes the individual to behave
as if he were risk-averse—hence, our use of the phrase “quasi-risk aversion” in the title of this section.
Our claim is that because of the nature of what is to be evaluated, the gains attached to an “improvement” secured by departures of behavior from the modeled are less than the losses imposed by corresponding departures of behavior in the opposing direction, that is, toward behavior worse than that represented in the model itself. To express the argument in terms of the worst-case, best-case spectrum, as we move from the best-case pole to the worst-case pole, predicted social losses (costs) increase at an accelerating rate. The harm inflicted on his fellows by a person who behaves “worse” than the average person in the community is greater than the benefits provided by another person who behaves “better” than the average person. Accordingly, the average-person model understates the average harm done. In imagining scenarios that might emerge under various sets of rules (a process that is essential before a choice is made), citizens will act as if they were risk-averse. There will be a rational “bias” toward avoidance of the worst-case prospects.
Nothing more is required to sustain this claim than the elementary apparatus of economic analysis. We need only assume, first, that sets of rules are instrumentally valued in the sense that they are expected to facilitate the provision of what can be conceptualized as valued goods and services. From this assumption it follows that these goods and services can be quantified, in terms of more or less. Second, we assume that goods and services are not costless; the polity-economy operates within an overall scarcity constraint. Finally, we assume that the demand curves for these goods and services slope downward.
1. The monopoly example
We shall elaborate the argument by means of an extended example, the results of which can then be generalized to institutional comparisons of all types. The example involves a comparison of competition and monopoly. Elementary theorems in welfare economics demonstrate that efficient resource usage occurs when there is complete freedom of entry given specified technical conditions that must be satisfied. In the equilibrium of such an industry, there are no unexploited gains from trade. Implicitly, this model of competition assumes that all persons are net wealth maximizers. By extending the same behavioral assumption to (nondiscriminating) monopoly organization of the same industry, the analysis demonstrates that efficient resource usage does not characterize the equilibrium results. There will be welfare losses, in comparison with the ideal outcome, that stem from restriction on entry: There are unexploited gains from trade.
The magnitude of these welfare losses will depend, however, on the assumptions made about the monopolist’s behavior. In the textbook models, monopolists are modeled as net wealth maximizers, but other possibilities are no less plausible a priori. For any number of reasons, the monopolist may refrain from charging the profit-maximizing price—to discourage entry by other firms, out of sheer inertia, or to further the perceived “public interest.”
If the purpose were to develop a theory of monopoly behavior that would be of assistance in predicting the pricing and the output strategies of firms that are in monopoly positions, appeal to the empirical record would seem entirely appropriate. From such a record, we might be able to derive a maximand for the simple monopolist that would yield the best prediction of price-output behavior, a maximand that would embody both private-interest and public-interest arguments, as empirically relevant. Our point is that such a predictive model is
not, in general, appropriate for the estimation of welfare losses from monopoly organization.
For the sake of simplicity, assume that all monopolists fall in two groups. One-half of all monopolists are strict profit maximizers in the pure textbook sense; the other half are pure “public-interest” firms, and these set prices and outputs at the level attained in full competition. In this setting (and assuming linear demand curves), the best single model is one in which price is set halfway between the profit-maximizing price, say,
p*, and the competitive price, say,
pc. If the good is provided under monopoly conditions, in many separate locations, and the demand curve in each location is that labeled
D in Figure 4.1, then the best single prediction of price in any location will be
pe = ½(
pc), with the corresponding output
If this best-fit model of monopoly behavior is now used to estimate the welfare losses of monopoly organization in general, we should proceed to estimate welfare loss for each location in the amount
ABC in Figure 4.1, with total welfare loss being this quantity summed over the
n locations. This total would then purport to measure the excess value that all persons (consumers and monopolists) would be prepared to pay (in terms of the value of other goods forgone) to secure a guarantee of the efficient outcome.
Such analysis would, however, simply be wrong. The area
ABC does not provide the best measure of the per-location welfare loss of the monopoly form of organization. The expected per-location loss attributable to monopoly is one-half of the loss under the profit-maximizing model plus one-half of the loss under the public-interest model of behavior (the latter is zero by assumption here). In other words, the expected welfare loss per location is one-half of the area
CEF (Figure 4.1), which is larger than the area
If we measure welfare loss as a function of output over the relevant range, we obtain Figure 4.2. The
W curve shows the relation between welfare loss and output and is based on the measurement of the triangles subtended in Figure 4.1. We use the more or less conventional formula
*25W = (½
2. The central point is that
W is a “convex function” of the difference between actual and ideal output. This means that the welfare loss at the best-fit output level
qe, which is
qe) in Figure 4.2, is characteristically less than the best-fit welfare loss, which is ½[
qc)]. That is, the expected cost of monopoly is greater than the cost associated with expected monopoly behavior, because the profit maximizers do proportionately more harm.
There is, of course, a single model of monopoly behavior that will yield an appropriate estimate of the costs of monopoly—namely, the model that generates
q‘ in Figure 4.2 as the average monopoly output choice. Our basic point here is that
this model of monopoly behavior is
systematically more cynical (that is, closer to the worst-case end of the spectrum) than simple empirical inspection of monopoly behavior would suggest. Consequently, something reasonably close to the simple textbook model of the profit-maximizing monopoly may be justified, even though at least some monopolists behave in a way more congruent with “public interest” provided that the objective of the model is to lay a basis for the normative evaluation of alternative institutional forms.
2. The general case
The central point is much more general than the monopoly example. Models of behavior used in social analysis are often evaluated simply by appeal to the “facts.” It seems clear that to many analysts these “facts,” distilled from simple observation, from introspection about themselves in policy roles, or more elaborately from consultation of the historical record, suggest that those who hold discretionary power under a particular institutional regime will often be constrained by internal moral considerations from acting in a self-interested way. Suppose that this is so. Nevertheless, any model of behavior derived from a simple “average” of observed behavioral patterns will not be sufficient for comparative institutional analysis. An appropriate behavioral model will have to reckon with the fact that the harm inflicted by those who behave “worse” than the notional average will be proportionately greater than the “good” done by those who behave “better” than the average. Accordingly, a bias toward the worst-case end of the behavioral spectrum is entirely justified. Specifically,
Homo economicus can be used as a model for comparative institutional analysis even when the empirical record (however described) indicates that its allowance for the relevance of public-interest motivations is inadequate.
We have presupposed that political and market institutions are valued not in themselves but for their potential capacity to allow the generation of desired goods and services. If this analytical framework is accepted, it seems natural to conceptualize individuals at the constitutional level as making determinants over possible institutional arrangements in a manner analogous to the choice between monopolistic and competitive market forms. Using the
Homo economicus behavior model in constitutional analysis, and justifying this use on analytic rather than empirical grounds, is a procedure we have borrowed from the classical political economist-philosophers in their analysis of political institutions. And we can, perhaps, do no better in this connection than appeal to David Hume: “In constraining any system of government and fixing the several checks and controls of the constitution, every man ought to be supposed a knave and to have no other end, in all his actions, than private interest.”
VI. Gresham’s Law in Politics
Our final argument in defense of the
Homo economicus model is at least as old as Thomas Hobbes. It embodies the notion that when many persons are involved in a social interaction, the narrow pursuit of self-interest by a subset will induce all persons to behave similarly, simply in order to protect themselves against members of the subset. As Hobbes stated, “Though the wicked were fewer in number than the righteous, yet because we cannot distinguish them, there is a necessity of suspecting, heeding, anticipating, subjugating, self-defending, ever incident to the most honest and fair-conditioned.”
Hobbes might well be interpreted here as presenting a version of the risk-aversion argument similar to that elaborated in the previous section. But he might also be interpreted as claiming that there is what we would call a sort of Gresham’s law in social interactions such that bad behavior drives out good and that all persons will be led themselves by the presence of even a few self-seekers to adopt self-interested behavior.
One way of constructing the relevant analytics would be to follow the obverse of Gary Becker’s theory of social altruism.
*29 In Becker’s theory, if an agent, A, acts benevolently toward a person, B, and B then has the opportunity to undertake some activity (at small cost) that will in turn benefit A, then B can be induced to take such action out of self-interest. For example, suppose A gives B one-quarter of each dollar that A receives at the margin. Suppose, furthermore, that B can, at the cost of one dollar, undertake some act that secures a benefit to A worth more than four dollars (or equally prevents a harm to A worth more than four dollars). Then B will rationally undertake that act and secure the benefit for A: A’s initial altruism toward B stimulates a reciprocal altruism on the part of B, despite the fact that B does not really care about A at all. For the more inclusive social group, the argument suggests that a critical mass of altruists may tend to create an apparently altruistic society, one in which some share of mutual gains is exploited through reciprocal nonexchange behavior.
Suppose that in such a society, the number of altruists declines or that the directly reciprocal relationships required for Becker’s model to work become blurred and ambiguous. The incentives for nonaltruists to behave altruistically may then dissipate rapidly; the structure of expected reciprocation unravels. The rational person, facing choice at the constitutional level, may seek to select institutions that depend only minimally on altruistic behavior as a protection against any erosion of reciprocation that might be present.
The Hobbes argument could, therefore, be made consistent with the Beckerian vision of social interaction. But a more direct, though related, interpretation may be more relevant. We might think of a situation directly contrary to that analyzed by Becker, one in which some person, say M, enjoys the imposition of harm on another, N. Suppose that it costs M one dollar to impose a dollar’s worth of harm on N and that M spends one-fifth of his income indulging in this sort of malice. Suppose, furthermore, that N does not care about M, one way or the other. Nonetheless, if N can act so as to reduce M’s income by one dollar, he has saved himself twenty cents worth of harm. Hence, N will be led to act to damage M, despite his total disinterest. That is to say, M’s malice is contagious. Moreover, if, as Hobbes suggests, the M’s in the community are ex ante undetectable, the N’s may be led to act in part maliciously toward everyone on the chance that anyone encountered may be an M, at least to the extent that such behavior is not excessively costly.
Consider a more plausible situation in which some persons in the community are mildly altruistic and some are truly selfish. Suppose that individuals are linked together through some interactions that are prisoners’ dilemma settings of the sort depicted in Matrix 4.1. (The construction of a matrix of this type was discussed in Chapter 1.) In this case, the numbers in the cells of the matrix represent money returns (in dollars) to the players. If each player acts so as to maximize his own money payoff, the socially disastrous outcome emerges.
The distinguishing feature of the game depicted in Matrix 4.1 is that the gain to each player from the selection of strategy 2 is much smaller when the other player selects strategy 1 than when the other player selects strategy 2. Note that the incremental gain to A from playing
a2 is only one dollar if B plays
b1, but the gain to A from playing
a2 is twenty dollars if B plays
b2. Suppose, now, that A is indeed mildly altruistic and derives some value from B’s receipt of payoffs, say ten cents on the dollar. This altruism is not sufficient to induce A to make direct transfers to B, since such transfers would cost A one dollar for each dollar receipt for B. But A may well refrain from inflicting harm on B to secure a gain for himself, provided that the “terms of trade” are within certain limits. If we depict the payoffs to A in Matrix 4.1 in terms of their utility equivalents, rather than as money payoffs, then under the assumption of the mild altruism postulated, we get Matrix 4.2. Note that in this setting, there is no dominant strategy for A. His preferred strategy now depends on what B is predicted to do. If A knows that B is symmetrically altruistic, A will then play
a1 confident that B will respond by playing
b1. If, on the other hand, A knows that B is a narrow maximizer of self-interest, then A also will be induced to play as a narrow maximizer. In this game, A’s mild altruism becomes behaviorally relevant only if A believes B to be mildly altruistic; otherwise, the socially disastrous outcome emerges as before.
We shall return to the Hobbesian citation and postulate that A does not really know whether or not B is a maximizer. In this case, A would need to have an expectation that among the various B’s he encounters in the prisoners’ dilemma setting, 90 percent would behave symmetrically in order to induce A to play
a1. There is a threshold number of “righteous” altruists below which all will behave as if purely selfish, simply to protect themselves against the prospect of playing against a narrow dollar maximizer. Consequently, even if the “virtuous” are more numerous than the “wicked,” all may be induced to behave “nonvirtuously,” with predicted consequences.
This result is reinforced if we model social interactions in many-person terms. If each person finds himself in prisoners’ dilemma settings in which many other persons are involved—in which there are many B’s and not just one, as in the simple matrix illustration—then predictions must be made about the behavior of all other persons in the interaction. For example, suppose that A faces ten B’s in a setting like that shown. Most of the B’s can be predicted to behave reciprocally in response to altruistic action by A, but so long as one person in the B group does not, the gains from this mode of behavior on the part of A may well be dissipated.
An additional element emerges, particularly in the many-person interaction, that may prevent voluntary altruism from exerting much behavioral influence on social outcomes. Consider the setting just discussed, in which a single person, A, faces ten B’s. Suppose that A continues to act altruistically, within limits, despite his prediction that at least one B will try to exploit the situation and to secure differential gains. The remaining B’s, initially, will respond symmetrically to A’s gestures of goodwill. But as these B’s observe the single defector to be securing differentially high gains, their inherent sense of unfairness may induce them to act nonreciprocally toward A. The attribute of fairness summarized in the phrase “getting my share” may be an important motivation that prevents the spread of other-regarding behavior.
In summary, the spirit of Hobbes’s analysis is that although altruistic and public-spirited motivations may be widespread among the population, these are delicate flowers, and crucial to their blooming may be the existence of institutions that do not make social order critically dependent on their effectiveness. To this extent, the implications of the Hobbesian argument are that institutions should be designed with
Homo economicus in mind, and that altruism, like good manners, can be appreciated but not “presumed upon.”
There is, finally, a quite different, non-Hobbesian sense in which something like Gresham’s law may apply in social interactions. Lord Acton’s famous dictum that power tends to corrupt, and absolute power corrupts absolutely was, no doubt, based on some predicted psychological destruction of the moral fiber of the despot. Such considerations are beyond our purview in this book, but there is a related, if quite different, point to be made. If institutions are such as to permit a selected number of persons to exercise discretionary powers over others, what sort of persons should be predicted to occupy these positions?
In yet another market analogy, suppose that a monopoly right is to be auctioned; whom will we predict to be the highest bidder? Surely we can presume that the person who intends to exploit the monopoly power most fully, the one for whom the expected profit is highest, will be among the highest bidders for the franchise. In the same way, positions of political power will tend to attract those persons who place higher values on the possession of such power. These persons will tend to be the highest bidders in the allocation of political offices. Economists have only recently become interested in the welfare properties of the political bidding process, under the rubric of “rent seeking.”
*31 In the rent-seeking literature, the focus of attention has been on the net resource wastage involved. Here we voice a different concern. Is there any presumption that political rent seeking will ultimately allocate offices to the “best” persons? Is there not the overwhelming presumption that offices will be secured by those who value power most highly and who seek to use such power of discretion in the furtherance of their personal projects, be these moral or otherwise? Genuine public-interest motivations may exist and may even be widespread, but are these motivations sufficiently passionate to stimulate people to fight for political office, to compete with those whose passions include the desire to wield power over others? If procedures are such that power
is allocated to those who value it most highly, then there is some presumption that those who might want the values of
all to weigh in political decisions will be driven out. Since the demand for discretionary power is highest for those individuals who desire social outcomes different from outcomes that perhaps most others would choose, political institutions will be populated by individuals whose interests will conflict with those of ordinary citizens. Citizens will need to plan for their institutional life accordingly.
Homo economicus, the rational, self-oriented maximizer of contemporary economic theory, is, we believe, the appropriate model of human behavior for use in evaluating the workings of different institutional orders. The central feature of the
Homo economicus model in this connection is its presumption of the ubiquity of conflict among interacting agents; it is this presumption that underlies the skepticism toward the possession of power that characterizes our attitude (and that of classical political economists) toward the design of institutions. Such skepticism means that it cannot be presumed that discretionary power possessed by agents under a particular institutional regime will be exercised in others’ interests, unless there are constraints embedded in the institutional structure which ensure that effect. In this sense, our model lies a great distance from the predominant “benevolent despot” model of politics in which public-interest orientations are assumed simply as a matter of course.
In mounting our defense of the
Homo economicus alternative, we have focused on analytic and methodological arguments rather than purely empirical ones. Our arguments are several:
1. There is a strong analytic case for undertaking the comparison of institutions with the
same basic behavioral model, rather than shifting behavioral horses midstream.
2. To the extent that institutional design is supposed to transform private-interest motivations into public-interest behavior, it makes analytic sense to focus on private-interest motivations and abstract from any purely behavioral altruism.
3. If market and political institutions are valued instrumentally for their capacity to produce goods and services that citizens want, and if the preferences for these goods and services exhibit standard convexity properties, it is rational for citizens to design institutions assuming a model of political agents’ behavior that generates “worse” outcomes than the empirical record would suggest emerge on the average.
4. Unless there is a critical mass of altruistically inclined individuals, which might be a substantial majority of citizens, it may well pay even the altruistically inclined to behave selfishly.
5. Because those who bid most for power in institutional orders will tend to be those whose private projects require major modification in the behavior of others, all citizens can rationally expect discretionary power to be exercised in ways that will be uncongenial to themselves.
There is, of course, empirical content in these arguments. However, in contrast with the purely predictive “science of behavior” models of conventional economics, empirical aspects are not entirely decisive. The significance of this fact is not that we believe that the purely empirical case for the use of
Homo economicus is weak, although we believe that case to be rather weaker than some of our more zealous colleagues do. The significance is rather that the empirical record is singularly difficult to unravel, not least because observed behavioral patterns may be substantially influenced by the prevailing institutional structure, so that when that structure is altered there are entirely predictable but (necessarily) currently unobservable behavioral changes. Consequently, rather than attend to ultimately inadequate observations, we have attempted to develop our argument on the basis of reasoned speculation. Of necessity, such reasoned speculation makes up a large part of constitutional analysis. And as for our political-economist forebears, so for us: The
Homo economicus-derived model of social conflict and cooperation seems uniquely appropriate for our constitutional speculations.
The Power to Tax (Cambridge University Press, 1980).
Monopoly in Money and Inflation (London: Institute of Economic Affairs, 1981).
Monopoly in Money and Inflation in
Economic Journal 91 (December 1981): 1105.
Bulletin of the American Academy of Arts and Sciences 34 (May 1981): 44.
International Review of Law and Economics 1 (December 1981): 155-66, and “Predictive Power and the Choice among Regimes,”
Economic Journal 83 (March 1983): 89-105, represent earlier efforts to justify the use of the
Homo economicus construction in constitutional analysis.
American Economic Review 67 (March 1977): 76-90.
The Economist as Preacher (delivered as the Tanner Lectures, Harvard University, 1980).
Democracy in Deficit (New York: Academic Press, 1977); and Brennan and Buchanan,
Power to Tax.
Journal of Economic Literature 9 (September 1971): 785-97.
Essays, Moral, Political and Literary, Vol. 1 (London: Oxford University Press, 1963).
De Cive (1642) (New York: Appleton-Century-Crofts, 1949), p. 12.
Democratic Political Theory (Princeton University Press, 1979), pp. 521-24.
Journal of Economic Literature 14 (September 1976): 817-26.
Economic Commentaries (London: Staples Press, 1956).
Toward a Theory of the Rent-Seeking Society (College Station: Texas A&M University Press, 1981).