The Online Library of Liberty

A project of Liberty Fund, Inc.

• Foreword
• Preface
• Acknowlegments
• Cost and Choice
• 1.: Cost In Economic Theory
• Classical Economics
• Marginal-utility Economics
• The Marshallian Synthesis
• Frank Knight and American Neoclassical Paradigms
• 2.: The Origins and Development of a London Tradition
• Wicksteed and the Calculus of Choice
• H. J. Davenport
• Knight On Cost As Valuation
• Robbins, 1934
• Mises, Robbins, and Hayek On Calculation In a Socialist Economy
• Hayek, Mises, and Subjectivist Economics
• The Practical Relevance of Opportunity Cost: Coase, 1938
• G. F. Thirlby and “the Ruler”
• Mises’ Human Action
• The Death of a Tradition?
• Appendix to Chapter 2: Shackle On Decision
• 3.: Cost and Choice
• The Predictive Science of Economics
• Cost In the Predictive Theory
• Cost In a Theory of Choice
• Choice-influencing and Choice-influenced Cost
• Opportunity Cost and Real Cost
• The Subjectivity of Sunk Costs
• Cost and Equilibrium
• 4.: The Cost of Public Goods
• The Theory of Tax Incidence
• Costs and Fiscal Decision-making: the Democratic Model
• Costs and Decision-making: the Authoritarian Model
• Costs and Decision-making: Mixed Models
• The Choice Among Projects
• The Costs of Debt-financed Public Goods
• Ricardo’s Equivalence Theorem
• Tax Capitalization
• 5.: Private and Social Cost
• Summary Analysis
• A Closer Look
• Internal Costs, Equilibrium, and Quasi-rents
• An Illustrative Example
• Pigovian Economics and Christian Ethics
• Narrow Self-interest and Alternative-opportunity Quasi-rents
• Conclusion
• 6.: Cost Without Markets
• Prices, Costs, and Market Equilibrium
• Resource-service Prices As Final-product Costs
• Market Equilibrium, Costs, and Quasi-rents
• The Cost of Military Manpower: an Example
• The Cost of Crime: Another Example
• Artificial Choice-making
• Socialist Calculation and Socialist Choice
• Costs In Bureaucratic Choice

Costs in Bureaucratic Choice

Bureaucratic decision-makers are human beings. This simple fact is only now beginning to be acknowledged in the theories of bureaucracy.3 The individual who is confronted with a choice among alternatives must choose, and the cost that inhibits decision is his own evaluation of the alternative that must be foregone. Rules can be laid down which direct him to adopt criteria that reflect the underlying economic realities. In a world of complete certainty, there is no decision problem. A computer can make all “choices,” if indeed “choices” exist. One of the central confusions leading to the false objectification of costs has been the extension of the perfect knowledge assumption of competitive equilibrium theory to the analysis of nonequilibrium choices, whether made in a market or a nonmarket process. Genuine choice is confronted only in a world of uncertainty, and, of course, all economic choices are made in this context. Any analysis of bureaucratic choice must be based on a recognition of this simple fact.

It will be helpful to construct the simplest possible model. Assume that a civil servant must decide between only two courses of action, a and b. These may be anything, including the production of n or n + 1 units of output. Either of two possible external events may accompany this action, event x or y. Again these may take almost any form, including the state of consumer demand at the margin. Further, let us assume that the total payoff to the

community under each of the four possible outcomes is accurately estimated and that these are indicated by the large numbers in the four cells of Figure 3.4

The choice between a and b will depend, of course, on the subjective probabilities assigned to x and y. Let us assume that the choosing agent assigns each event an equal probability. From the arithmetic, it is then clear that the expected value to the whole community will be higher from a than from b. However, with a change in the probability coefficients, from (.5, .5) to (.4, .6), the expected value to the community becomes higher from b than from a. In genuine uncertainty, the decision-maker must assign such subjective probabilities; there is no objectively determinable set of coefficients. When this is recognized, it is clear that there is simply no means of evaluating the choosing agent’s performance externally and after choice. Each of two separate persons may choose differently when confronted with identical sets of alternatives. There is no “correct” choice independent of the subjective probabilities that are assigned. In our example, one chooser may reject b because its cost exceeds expected returns; the other may reject a for the same reason. There is no way that an external observer or auditor can, ex post, decide which of the two persons followed “the rules” more closely.

This difficulty in evaluating the efficiency of nonmarket decision-making suggests that the institutional pattern of rewards and punishments may be modified to ensure that, regardless of the choices that are made, the chooser will have some personal incentive to perform in accordance with “social” maximization criteria. This will substitute ex ante motivation for individual behavior in the “public interest” for the misguided and hopeless efforts at judging or auditing the results ex post. The necessity for some coordination between the cost-benefit structure as confronted by the decision-maker and the “true” cost-benefit structure of the whole community has, by this time, come to be widely recognized both in theory and in practice.

This institutional device is necessarily limited, however, and for several reasons it cannot fully resolve the dilemma of nonmarket economic choice. Nonmarket choice cannot, by its very nature, be made to duplicate market choice until and unless the ownership-responsibility pattern in the former fully matches that in the latter, an achievement that would, of course, eliminate all institutional differences between the two.

Suppose, as an initial example, that an individual cost-benefit structure is introduced as shown by the single-bracketed terms in Figure 3. Ordinally, at least, the relative payoffs to the decision-maker coincide with those for the community. However, if he assigns equal subjective probabilities to x and to y, his own cost-benefit calculation will lead him to select b, not a. The numerical array is, of course, deliberately designed to indicate this result, but it should be evident that ordinal equivalence between the payoff structure of the decision-maker and that of the whole community is not sufficient to ensure consistency in choices.

Proportionality is suggested. If the personal payoffs to the decision-maker, negative or positive, are made strictly proportional to those of the whole community, then choices made in accordance with expected-value criteria will produce the required coordination. At this point, the relevance of expected-value maximization as a rule for individual choice behavior must be called into question. It is well established that an individual will maximize present value only if he derives no utility or disutility from risk-taking and if the marginal utility of income to him is constant over the relevant outcome range. If the marginal utility of income declines over this range and if the chooser is neither a risk averter nor a risk preferrer, he will tend to have some preference for the safer of the two alternatives, some “nonpecuniary” differential in favor of alternative b in the numerical illustration of Figure 3. The question that then emerges is whether or not this nonpecuniary differential faced by the decision-maker whose payoffs are proportional to those for the whole community need be the same as that which “should” inform the decision made from the community’s point of view. As Domar and Musgrave pointed out in another connection,5 the individual whose payoff structure is only some proportionate share of that which he might confront under full ownership will tend to take more risks. The reason is obvious. Since the nonpecuniary differential arises only because of the declining marginal utility of income, the fact that the outcome range is lower under proportionate share payoffs than under full responsibility and ownership ensures some lessening of this differential.

An additional and important element tends to work in the opposing direction. Given a structure of individual payoffs that are only proportional to total community payoffs, the absolute differences between the expected value of alternatives are lower for the decision-maker than for the community; and the differences in the opportunity costs of two separate alternatives are lower. Considering this, it seems evident that behavior will tend to be less responsive to changes in the underlying conditions under bureaucratic choice than under market choice. The decision-maker in the latter situation cannot perceive changes in signals with the same sensitivity as he could in the former for the simple reason that the signals are stronger in the first case. If we also recognize and allow for threshold-sensitive response in behavior generally, this differential in behavior becomes even more pronounced.6

These separate elements emphasize the fact that proportionality between the decision-maker’s cost-benefit matrix and that of the community will not ensure an approximation to market-choice results in a regime of bureaucratic choice. Costs as confronted by the choosing agents must remain inherently different in the two decision structures, and it is these differences that constitute the basic problem of securing efficiency in nonmarket choice-making.

[3. ]See Gordon Tullock, The Politics of Bureaucracy (Washington, D.C.: Public Affairs Press, 1965).

[4. ]These estimates are necessarily ex ante: only one outcome can be actually observed after choice.

[5. ]E. D. Domar and R. A. Musgrave, “Proportional Income Taxation and Risk-Taking,” Quarterly Journal of Economics, LVIII (May 1944), 388-422, reprinted in American Economic Association, Readings in the Economics of Taxation (Homewood, Ill.: Richard D. Irwin, 1959), pp. 493-524.

[6. ]Devletoglou has argued persuasively that all human behavior must be analyzed in terms of a threshold-sensitive model. See Nicos Devletoglou and P. A. Demetriou, “Choice and Threshold: A Further Experiment in Spatial Duopoly,” Economica, XXXIV (November 1967), 351-71.