Cost and Choice: An Inquiry in Economic Theory
Private and Social Cost
Equality between marginal private cost and marginal social cost is the allocative criterion of Pigovian welfare economics,*56 and the principle remains acceptable to most modern welfare economists. Corrective taxes and subsidies are deemed to be required in order to satisfy the necessary conditions for optimality when external effects are observed to be present. The subject of discussion here is limited to the cost conception that is implicit in the Pigovian policy criterion; for this reason, there is no need to review recent works in the theory of externality, as such, some of which place major qualifications on the Pigovian norms.*57 The purpose of this chapter is to demonstrate that the Pigovian principle embodies a failure to make the distinction between costs that may influence choice and costs that may be objectively measured.
Consider a standard example where the behavior of one person (or firm) exerts marginal external diseconomies on others than himself. These represent the loss of "goods" to others for which they are not compensated through ordinary market dealings. Application of the Pigovian norm suggests that the costs imposed externally on those who are not party to the decision-making should be brought within the calculus of the decision-maker. These costs should be added to the decision-maker's own internal costs, costs that he is presumed to take into account. The device often suggested is the levy of a tax on the performance of the externality-generating activity, a tax that is equated to the external costs per unit that the activity imposes. Other devices sometimes advanced are institutional arrangements designed to internalize the externality. In all cases, the purpose is to bring the costs that inform or influence the decision-maker into conformity with true "social" costs. The models remain individualistic in the sense that "social" costs are computed by a simple summation over individuals in the relevant community or group.
A Closer Look
According to the Pigovian theory, the change in "costs" which results from an explicitly recommended levy of a tax modifies the behavior of the acting person so that "efficiency" results. But what is meant by "costs" here? This Pigovian framework provides us with perhaps the best single example of confusion between classically derived objective cost concepts and the subjective cost concepts that influence individual choice.
Consider, first, the determination of the amount of the corrective tax that is to be imposed. This amount should equal the external costs that others than the decision-maker suffer as a consequence of decision. These costs are experienced by persons who may evaluate their own resultant utility losses: they may well speculate on what "might have been" in the absence of the external diseconomy that they suffer. In order to estimate the size of the corrective tax, however, some objective measurement must be placed on these external costs. But the analyst has no benchmark from which plausible estimates can be made. Since the persons who bear these "costs"—those who are externally affected—do not participate in the choice that generates the "costs," there is simply no means of determining, even indirectly, the value that they place on the utility loss that might be avoided. In the classic example, how much would the housewife whose laundry is fouled give to have the smoke removed from the air? Until and unless she is actually confronted with this choice, any estimate must remain almost wholly arbitrary. Smoke damage cannot be even remotely approximated by the estimated outlays that would be necessary to produce air "cleanliness." "Clean air" can, of course, be physically defined; the difficulty does not lie in the impossibility of defining units in a physically descriptive sense. Regardless of definition, however, "clean air" cannot be exchanged or traded among separate persons. Each person must simply adjust to the degree of air cleanliness that exists in his environment. There is no possibility of marginal adjustments over quantities of the "good" so as to produce an equilibrium that ensures against interpersonal differences in relative evaluations.
Figure 1 illustrates my argument. There is no way in which the analyst can objectively determine whether the housewife is at position A, B, or C on the diagram, yet it is clear that the utility loss, both at the margin and in total, may be significantly different in the three cases. There is no behavioral basis for observing evaluations here. Figure 1 also suggests that if individual preference functions have the standard properties, the valuations of separate persons probably vary directly with private-goods income. The affluent housewife will value clean air more highly than the poverty-stricken. The reason is obvious. The external diseconomy, "smoke damage," cannot be "retraded" among persons. If it could be, the poverty-stricken housewife might be quite willing to take on an extra share of the damage in exchange for some monetary payment from her affluent neighbor. But since such a trade cannot take place, she must simply adjust to the degree of "bads" in her environment.
Objective measurement of externally imposed costs seems more feasible in cases where the removal of the damaging agent results in changes in the production function of firms. If the damaged units should be producing firms, not individuals, there would seem to be no need to get into the complications of evaluating utility losses. A change in the rate of "pollution" can be observed to change the rate of outlay required for producing marketable goods and services. Since these goods and services command prices in markets, objective measurement of their value can be made.*58
If a corrective tax, equal to the costs that are imposed externally upon others (which we shall now assume to be objectively measurable despite the problems noted above), is to generate the behavioral changes predicted by the Pigovian analysis, the internal costs as faced by the decision-maker must also be objectively measurable, at least indirectly. The analysis assumes implicitly that, in the absence of the corrective tax, choices are informed by money outlays made in purchasing inputs in ordinary market transactions. As an earlier discussion has shown, however, there is no logical support for this presumption in the general case. Observed money outlays need not reflect choice-influencing costs, the genuine opportunity costs that the decision-maker considers.
There is an obvious inconsistency. The Pigovian norm aims at bringing marginal private costs, as these influence choice, into line with social costs, as these are objectively measured. Only with objective measurability can the proper corrective devices be introduced. But under what conditions can objectively measurable costs, external and internal, be taken to reflect, with even reasonable accuracy, the costs that the effective decision-maker may take into account. In conditions of ideal competitive equilibrium, the costs that can be measured by the observer provide a reasonable proxy for the subjective evaluations of decision-makers. However, almost by definition, external effects are not imposed in such a setting.
Internal Costs, Equilibrium, and Quasi-Rents
The conditions under which these outlays may be taken to measure, even indirectly, the subjective barrier to choice must be carefully specified. These are as follows: (1) The individual, or firm, must be in full competitive equilibrium with respect to the activity that generates the external diseconomy; (2) at this equilibrium level of activity, and only at this level, losses are avoided and no profits are made; and (3) there are no profits in prospect of being made anywhere else in the economy. Under such conditions, the costs that may be avoided are simply the outlays that must be made. The individual, or firm, has available only one alternative loss-avoiding course of action which is that of not acting. In the latter, he avoids the outlay that the decision to act, considered in total or at the margin, requires. Not acting is clearly the most attractive alternative course of behavior here since all other alternatives must yield net losses.
It is important to note that quasi-rents cannot exist in the competitive equilibrium required in this model. The device of capitalizing differential resource capabilities into quasi-rents so as to equalize costs among separate firms cannot, therefore, be utilized. If it is to exist at all, the bridge between choice-influencing costs and objectively measurable outlays depends critically on the absence of quasi-rents. If such rents exist, either with respect to the personal behavior of an individual or with respect to the productive activity of a firm, there can be no presumption that anticipated outlay measures subjective opportunity costs, those that must influence actual choice behavior. The indirect linkage between subjective opportunity costs and objectively measured outlays which such equilibrium establishes is shattered. The reason is that in the presence of "quasi-rents," the individual or the firm has available more than one loss-avoiding alternative course of action. "Quasi-rents" or their equivalent provide a cushion which allows subjectively relevant elements of the decision calculus to become meaningful. As Frank Knight recognized, even if imperfectly, in his 1935 papers,*59 the allowance for any nonpecuniary aspects in the choice calculus of an individual or a firm plays havoc with the use of measurable outlays as surrogates for the opportunity costs that do, in fact, influence choice behavior. For our purposes at this point, the allowance of "quasi-rents" or their equivalent destroys the underlying logic of the Pigovian policy norms. There is simply no means to make an effective translation between the subjective opportunity costs that influence decision and the objectively measurable outlays that both the decision-taker and others who are externally affected undergo as a result of decision.
An Illustrative Example
Much of the critical analysis may be clarified by a simple illustrative example. Let us suppose that I enjoy foxhunting and that I maintain a kennel of hounds near my residence. I am considering adding one more hound to my already-large pack, and I know with reasonable accuracy the market price for hounds. This price is, let us say, $100.
My neighbor lives within sound range of my kennel, and he (and his family) will suffer some predictable utility loss if I decide to purchase the additional dog. For purposes of analysis here, let us say that this external damage can be reasonably evaluated at $45, presumably by an expert observer and also by both my neighbor and myself. Now let us suppose that I anticipate the incremental benefits of the additional dog at $160. This substantially exceeds the price of $100. Let us also assume that there are no alternative spending outlets where I can secure net marginal benefits. In such circumstances, the opportunity costs arising from the enjoyments that I must avoid by the fact of making the outlay can roughly be measured at $100. However, in addition to these costs, I may well, in my calculus of decision, place some value on the enjoyments that my neighbor must also forego as a consequence of my purchasing the dog. His anticipated suffering, as well as my own, can be an obstacle to my decision.
Suppose that I try as best I can to place a value on this expected loss in utility for my neighbor and that I arrive at a figure of $45, which, as noted above, does roughly represent the value that he himself places on the action. The obstacle to my choice, my choice-influencing cost, will embody two elements. First, there is the evaluation of the alternative uses of the anticipated $100 outlay, which, under the conditions postulated, we measure at $100. Second, there is the evaluation that I place on the anticipated enjoyments that my neighbor must forego, in this case $45. Under such circumstances, I will proceed to carry out the purchase since the anticipated marginal benefits, $160, exceed the evaluation of foregone alternatives, $145.
Note that in the behavior postulated, I am acting in accordance with the Pigovian criterion, treated here as an ethical norm for private behavior. Quite literally, I am treating my neighbor as myself, and my internal decision calculus accurately reflects "marginal social cost" as the obstacle to decision, despite the absence of any corrective tax. Note also, however, that for the discrete choice in question, I shall be observed to impose an external cost on my neighbor for which I do not compensate him. If a Pigovian-trained economist should be called in to advise the government, he would likely recommend that I be subjected to a corrective tax, levied in the amount of the external costs, in this example $45. It is clear that, unless the components of my subjective opportunity costs are directly modified by such a tax, the effect will be to change my decision. Costs that a positive decision embody will now be approximated at $190. Facing these, I shall refrain from purchasing the hound despite the "social" or allocative distortion that my failure to do so generates. In this example, the corrective tax tends to convert a socially desirable choice outcome into a socially undesirable one.
My internal opportunity-cost components may be modified by the imposition of the tax. If I am fully aware that I am being taxed for the express reason that my behavior generates the external economy, I may reduce the valuation that I place on my neighbor's foregone enjoyment of silence. This reaction may be especially likely if the proceeds of the tax are earmarked for direct transfer to my neighbor. Such a direct linkage, and more importantly such a consciousness of the purpose of corrective taxes, has not been emphasized in the Pigovian literature and does not seem remotely descriptive of choice behavior. At best, we may acknowledge some substitution between the tax and the subjective valuation of the "external" component of opportunity cost; surely there is no reason to expect anything like a full offset.
In the simplified example, it is assumed that I value the foregone alternatives of others more or less equally with my own. This extreme altruism need not, of course, be assumed in order to reach the conclusion that the corrective tax produces inefficient outcomes. In the discrete choice discussed in the example, even if I place a valuation of only $16 on the foregone enjoyment of my neighbor, the corrective tax of $45 will cause me to choose the inefficient outcome ($100 + $16 + $45 = $161 > $160). This valuation figure becomes even smaller as the personal "quasi-rent" or "marginal surplus" is reduced. Suppose, for example, that my estimate for marginal benefits is only $146, and that I place only a $2 valuation on the foregone enjoyment of my neighbor. My choice-influencing costs after the tax are then $147 ($100 + $2 + $45), which exceed my anticipated marginal benefits. I shall be led to the inefficient social choice, although the differential inefficiency here will be lower than in those cases where I place a somewhat higher valuation on the prospective utility losses of others.
Pigovian Economics and Christian Ethics
The example above suggests that a defense of the Pigovian policy norm's applicability may lie in the behavioral assumption that each person acts strictly in accordance with his own narrowly defined, materialistic "private" interest. His own behavior may be assumed to be wholly uninfluenced by the effects it exerts on other persons. Under such conditions, it might be argued, the demonstrated conflict between the corrective policy and the achievement of allocative efficiency would not arise. As the following section will show, even this restrictive assumption will not rescue the Pigovian analytics. At this point, however, the legitimacy of the assumption itself must be more carefully examined.
Initially, the behavioral assumption seems nothing more than an extension of the "economic man" who roams throughout predictive economic theory. Closer examination reveals, however, that the requirement here is much more restrictive than this. In the traditional neoclassical theory of markets, the implicit behavioral assumption is that of "nontuism," first clarified by Wicksteed. This is merely the assumption that, by and large and on the average, individuals or firms engaged in market-like behavior leave out of account the direct interests of those who are on the opposing side of the trading contract. The "economic man" of Wicksteed can adhere to a Christian ethic without neurosis, since he can, if he so chooses, incorporate in his behavior pattern some recognition of the interests of all his fellows except those with whom he is directly trading. He may continue to "love his neighbor," as long as his neighbor is not trading with him. In the externality relationship, by definition, trade does not take place. It seems reasonable to think that it is precisely in this kind of relationship that genuinely benevolent behavior patterns might be witnessed. Indeed, it might plausibly be argued that in almost all of our nonmarket behavior, there is potential externality and that the ordinary functioning of civil society depends critically on a certain mutuality of respect. When property rights are not well defined and, hence, market-like arrangements are difficult to establish, the very forms of behavior seem to pay at least lip service to something other than narrowly defined self-interest. "May I smoke?" provides a classic illustration.
The departures from behavior patterns based on narrowly materialistic utility functions seem to be almost universal only when personal externality relationships exist. That is to say, the argument against the narrow self-interest assumption applies fully only when the potential externality relationship is limited to a critically small number of persons. In large-number groups, by comparison, there may be little or no incorporation of the interests of "others" in the utility calculus of individuals. Here the individual really has no "neighbors," or may have none in any effective behavioral sense, despite the presence of "neighborhood effects." Under the latter conditions, the Pigovian logic and its policy implications are at least partially restored. The person who litters the nonresidential street in the large city probably does not worry much about the effects of his action on others. This suggests that, for such cases, the corrective devices implied by the Pigovian analysis should not generate conflicts with standard allocative norms provided, of course, that all of the other conditions required for their applicability are met.*60
Narrow Self-Interest and Alternative-Opportunity Quasi-Rents
The preceding section indicated that one means of rescuing the Pigovian policy logic lies in making the explicit assumption that no factor involving "regard for others" influences the choices of the person who exerts external costs. Even with this constraint on individual utility functions, however, conflicts between applications of the policy norms and efficiency criteria will arise if prospective "quasi-rents" exist for alternative courses of action. This can also be shown in terms of the simple illustrative example already discussed.
In the earlier use of the example, we assumed that no "profit" prospects exist for any other spending opportunities. In this case and only in this case will the expected money outlay on resource inputs, $100, reflect at all accurately the internal component of genuine opportunity costs, and the expected marginal tax, $45, the comparable externally imposed component. In such a model, the added assumption that the choosing-acting person places no evaluation on either the utility levels attained by others or the changes in these levels that are the results of his own behavior will restore the consistency between the Pigovian policy logic and overall efficiency norms. What we now must show is that, even if we retain the narrowly defined self-interest assumption about individual behavior, any relaxation of the assumption about "profits" or "quasi-rents" in alternative courses of action will undermine the whole policy apparatus.
Consider the situation where there are anticipated "profit" prospects in alternative spending opportunities. Suppose that in considering the purchase of the additional foxhound, from which I estimate a marginal benefit of $160, I expect the outlay on resource inputs measured at $100, but that I also anticipate that I could invest $100 in some other line of activity yielding an expected marginal benefit which I subjectively value at $115. In this case, $115, and not $100, is the figure that best represents my choice-influencing opportunity cost, the barrier to choice, before the imposition of the tax. Suppose now that the corrective tax of $45 is levied on the marginal purchase, and, as before, let us accept that this accurately reflects my neighbor's own evaluation of the external damage that he will suffer from my action. It follows that "social costs"—those costs that must be borne by all members of the group and which are the result of the marginal choice—are best measured at $160. This figure reflects my own marginal opportunity costs, now measured at $115, plus the external costs borne by my neighbor, measured at $45. Because both the social costs and the social benefits of my acquiring another foxhound are measured at $160, the standard allocative norms suggest that I should be indifferent in the decision. Note, however, that this indifference will not be realized in my own choice calculus once the corrective tax is imposed on my marginal purchase. As I now confront the alternatives, my choice-influencing costs will be $166.75, not $160. Not only must I value the expected outlay on inputs in terms of the foregone alternatives, i.e., $115, but also, I must value the expected marginal tax outlay in terms of foregone alternatives which payment will make impossible to achieve. If the expected "profit" on the $100 outlay in an alternative course of action is $115, we should expect the choice-influencing costs of the expected $45 tax to be roughly $51.75. The choice is no longer marginal in my own decision calculus; the corrective tax has caused choice-influencing opportunity costs—private costs—to exceed marginal social costs. I shall overadjust my behavior, even considering the most restrictive self-interest arguments in my utility function.
I should emphasize that this chapter is not designed as a general critical analysis of the Pigovian policy norms. Such an analysis would have required the treatment of many interesting issues that have been ignored here. My purpose has been to utilize this familiar branch of applied economic theory to demonstrate the desirability of clarifying the basic notions of opportunity costs. To those who fully accept and understand the London-Austrian contributions, the internal inconsistencies in the Pigovian logic will be apparent. To those who have been trained in the neoclassical paradigms of opportunity cost, recognition of the inconsistencies may require a working out of elementary examples. It is not easy to question long-accepted precepts, and in the several versions of this chapter, I have found it difficult to prevent the analysis from lapsing into the kind of conventional methodology that I have often used in other works. The result may give the appearance of complexity despite the elementary nature of the points being made. In effect, the incorporation of the London conception of opportunity cost amounts to transforming one of the foundation stones of economic theory. Only when this basic modification is completed can real progress toward changing the superstructure be attempted on a large scale. Meanwhile, only the most exposed aspects of this superstructure—the Pigovian welfare analytics, for example—can be related directly to the particular flaw in one of the theory's cornerstones.
Notes for this chapter
The companion criterion, equality between marginal private product and marginal social product, reduces to the cost criterion when the latter is stated in opportunity-cost terms. The failure to take action that exerts external benefits can be treated as analytically equivalent to the taking of action that exerts external costs. In his own formulation, Pigou used the product terminology almost exclusively, although he referred to both types of divergence. See A. C. Pigou, The Economics of Welfare (4th ed.; London: Macmillan, 1932), esp. pp. 131-35.
Notably, R. H. Coase, "The Problem of Social Costs," Journal of Law and Economics, III (October 1960), 1-44; Otto A. Davis and Andrew Whinston, "Externality, Welfare, and the Theory of Games," Journal of Political Economy, LXX (June 1962), 241-62.
It seems likely that this helps to explain the source of the confusion. Marshall and Pigou developed the externality notion within the context of interfirm models, implicitly assuming competitive structures. As we shall see, the relevance of objectively measurable costs is limited even in this model, but the errors are of a different order of magnitude from those that arise when the externalities refer to an interpersonal interaction or to an interfirm interaction where utility functions are employed. The possibility of objectively measuring external costs does not, of course, ensure that the policy of levying a corrective tax is desirable. Under competition, this policy can be plausibly defended within certain limits. In noncompetitive structures, by contrast, the attempt to levy corrective taxes on an externality-generating firm may do more harm than good. On this elementary point, see my "External Diseconomies, Corrective Taxes, and Market Structure," American Economic Review, LIX (March 1969), 174-77.
F. Knight, "Notes on Utility and Cost" (Mimeographed, University of Chicago, 1935). Published as two German articles in Zeitschrift für Nationalökonomie (Vienna), Band VI, Heft 1, 3 (1935).
It is perhaps worth noting here the interesting difference in emphasis between political scientists and economists, both of whom discuss essentially the same behavioral interactions. In politics, primary emphasis has traditionally been placed on political obligation, on the duty of the individual to act in the "public interest." This represents an attempt to improve results through modifying the individual's utility function in the direction of causing him to place a higher valuation on the utilities of others. Relatively little attention has been given until quite recently to the prospects of making institutional changes that will channel private choice in the direction of producing more desirable social results. In economics, by contrast, institutional or policy changes have been the center of attention, and relatively little discussion has been devoted to norms for individual behavior. As our analysis shows, economists have implicitly assumed that individuals act in accordance with quite narrowly defined self-interest, and they have developed policy norms which may prove inapplicable if this underlying behavioral postulate is not descriptive of reality. For an earlier discussion of this difference between the two disciplines, see my "Marginal Notes on Reading Political Philosophy" included as Appendix I in James M. Buchanan and Gordon Tullock, The Calculus of Consent (Ann Arbor: University of Michigan Press, 1962; Paperback Edition, 1965).
End of Notes
Return to top