this as its mission in life. This power may reflect widely shared values or the intensity with which values are felt. But the price in technical and economic diseconomies are often high. In all one uses as a criterion is the pragmatic test of the firm's "sales" (the disputes it wins, the new programs introduced, the old ones sustained, the share of the budget obtained). One hasn't much. And resources wasted often count as much as resources well used on these criteria.

(c) Even where counteravailing power is present, one cannot assert a high probability that the common interest will benefit. If private firms and organized labor are capable of striking bargains which act against the common interest one shouldn't assume that government agencies are not.

Other examples can be cited: We invest quite a lot to move air passengers from airport to airport but have paid little attention to the increasingly significant links in the journey from portal to and from airports. Our maritime policies which have traditionally been worked out via the bargaining mode include an operating subsidy which is structured so as to create a positive incentive to overmanning of ships. Our water resources policies favor expensive means of reducing water pollution over less expensive means. These policies have also produced flood control projects which have generated incentives for people to overbuild in still vulnerable flood plains. In agriculture we pay both to take land out of agricultural production while bringing reclaimed land in. We have a sugar subsidy program which seems to cost three times the net incomes of the sugar producers. We spend ten times on urban roads as on urban mass transit without the balance between these two types of transportation being examined.

It might be held that some of these examples simply illustrate the principle that our political system has decided to transfer income to specific groups, that a politically feasible way has been found to do this, and the fact that apparently contradictory actions are taken by different parts of the government is either evidence of income being transferred to other groups or is compensatory action to correct undesirable overall effects of particular subsidies.

This is undoubtedly true-in some cases. But it is my belief, that, on the average, instances of this type are at least as much due to the reasons cited above: bureaucratic inertia, random differences in bargaining power, absence of market forces, unregulated intra-governmental monopolistic practices.

WHAT CAN BE DONE?

Neither model will do. Lindblom is right about the undesirability and infeasibility of a rigidly hierarchical system. But he is, I think, too hopeful about the virtues of the largely bargaining system we have. We need analysis as well.

What do I mean by analysis? For present purposes suffice it to mean an attempt to define objectives, to describe alternative means to these ends, to invent new objectives and new alternative means, to assess benefits and costs, to take account of uncertainties, to quantify what looks useful to quantify, to isolate decisions that can be deferred

from those that can't, to create options. All this may appear ordinary. It is, but it is often difficult to do and it hasn't been attempted much in a systematic way on major public decisions. But it has begun to be done in a significant way with results in the Defense Department that are impressive; I predict that results throughout other parts of government will, in time, be at least as impressive.

There are several necessary conditions for doing better: one is that there exist a structure of adversary relationships, that over a wide range of governmental behavior there exist mechanisms for one group to challenge and debate issues of common interest with other centers. This doesn't work well if left to chance. It requires action from a higher level. This is a familiar problem in the operation of big corporations. It is more an important problem in areas where market mechanisms are weak or absent. Therefore, one subject for systematic analysis is to improve the bargaining phenomena.

Another necessary condition is that there be a system of analysis involving many groups working from many points of view. For no one group can assemble all of the relevant data on a complex issue; values and facts do get inter-mixed; ends and means often do inter-act; problems must be decomposed for analysis; analysis must be partial; all optimizations are, in some sense, suboptimizations. One can expect, however, through more systematic analysis to narrow the vast areas in which governmental action is uninformed, arbitrary, and based on unenlightened opinion rather than data and analysis. One can create larger conceptual "islands" in which relatively good predictions can be made about the consequences of taking alternative decisions. One can even expect to connect some islands to each other through the development of broader theories. Just as economic theory was extended over time from separate theories on production and consumption and money into a unified macro-theory with major consequences for the conduct of public affairs, so we should expect to develop broader theories of health, of education, of law enforcement. And some of these might even connect. How far can this process continue? Indefinitely. (But I confess my mind boggles at the notion of the unified theory, for example, of postal service, foreign aid, and outer space.) We needn't be concerned about running out of new phenomena. New ones will be identified or become ripe at at least the rate at which old ones are mastered.

Finally, in carrying out analyses what should be done about the absence of conditions for Pareto optimality? Two things. Firstly, try in making analysis, to make corrections that move the results in what seems to be the right direction. Secondly, take some solace from the bargaining viewpoint: our system doesn't mind making interpersonal comparisons and the interactions, over time, of partisan mutual adjusters will see that rough justice gets done.

[Public Administration Review, Vol. XXVI, No. 4, December 1966]

THE POLITICAL ECONOMY

OF

EFFICIENCY:

COSTBENEFIT ANALYSIS, SYSTEMS ANALYSIS, AND PROGRAM BUDGETING

By Aaron Wildavsky*

(Professor, Department of Political Science, University of California, Berkeley) The encroachment of economics upon politics is not difficult to understand. Being political in perspective is viewed as bad; having the perspective of the economist is acclaimed as good. As a discipline, economics has done more with its theory, however inadequate, than has political science. Under some conditions economists can give you some idea of what efficiency requires. It is a rare political scientist who would even concern himself with political rationality. Economists claim to know and work to defend their interests in efficiency: political scientists do not even define their sphere of competence. Thus the market place of ideas is rigged at the start.

There was a day when the meaning of economic efficiency was reasonably clear.

An objective met up with a technician. Efficiency consisted in meeting the objective at the lowest cost or in obtaining the maximum amount of the objective for a specified amount of resources. Let us call this "pure efficiency." The desirability of trying to achieve certain objectives may depend on the cost of achieving them. In this case the analyst (he has graduated from being a mere technician) alters the objective to suit available resources. Let us call this "mixed efficiency." Both pure and mixed efficiency are limited in the sense that they take for granted the existing structure of the political system and work within its boundaries. Yet the economizer, he who values efficiency most dearly, may discover that the most efficient means for accomplishing his ends cannot be secured without altering the machinery for making decisions. He not only alters means and ends (resources and objectives) simultaneously but makes them dependent on changes in political relationships. While he claims no special interest in or expertise concerning the decision apparatus outside of the market place, the economizer pursues efficiency to the heart of the political system. Let us call this "total efficiency." In this vocabulary, then, concepts of efficiency may be pure or mixed, limited or total.

A major purpose of this paper is to take the newest and recently most popular modes of achieving efficiency-cost-benefit analysis, systems analysis, and program budgeting-and show how much more is involved than mere economizing. Even at the most modest level of costbenefit analysis, I will try to show that it becomes difficult to maintain pure notions of efficiency. At a higher level, systems analysis is based on a mixed notion of efficiency. And program budgeting at the highest levels leaves pure efficiency far behind its over-reaching grasp into the

I am more than ordinarily indebted to the people who have improved this paper through their comments. Win Crowther, John Harsanyi, John Krutilla, Arthur Maas, Arnold Meltsner, Nelson Polsby, William Riker, and Dwight Waldo saved me from errors and contributed insights of their own. The responsibility for what is said is entirely my

own.

The paper, written while the author was a Research Political Scientist at the Center for Planning and Development Research, University of California, Berkeley, was originally presented at a conference on public policy sponsored by the Social Science Research Council.

structure of the political system. Program budgeting, it turns out, is a form of systems analysis, that is, political systems analysis.

These modes of analysis are neither good for nothing nor good for everything, and one cannot speak of them as wholly good or bad. It is much more useful to try to specify some conditions under which they would or would not be helpful for various purposes. While such a list could not be exhaustive at this stage, nor permanent at any stage (because of advances in the art), it provides a basis for thinking about what these techniques can and cannot do. Another major purpose of this paper, therefore, is to describe cost-benefit and systems analysis and program budgeting as techniques for decision-making. I shall place particular stress upon what seems to me the most characteristic feature of all three modes of analysis: the aids to calculation designed to get around the vast areas of uncertainty where quantitative analysis leaves off and judgment begins.

COST-BENEFIT ANALYSIS

One can view cost-benefit analysis as anything from an infallible means of reaching the new Utopia to a waste of resources in attempting to measure the unmeasureable.1

The purpose of cost-benefit analysis is to secure an efficient allocation of resources produced by the governmental system in its interaction with the private economy. The nature of efficiency depends on the objectives set up for government. In the field of water resources, where most of the work on cost-benefit analysis has been done, the governmental objective is usually postulated to be an increase in national income. In a crude sense, this means that the costs to whomever may incur them should be less than the benefits to whomever may receive them. The time streams of consumption gained and foregone by a project are its benefits and costs.

2

The aim of cost-benefit analysis is to maximize "the present value of all benefits less that of all costs, subject to specified restraints." A long view is taken in that costs are estimated not only for the immediate future but also for the life of the project. A wide view is taken in that indirect consequences for others variously called externalities, sideeffects, spillovers, and repercussion effects are considered. Ideally, all costs and benefits are evaluated. The usual procedure is to estimate the installation costs of the project and spread them over time, thus making them into something like annual costs. To these costs are added an estimate of annual operating costs. The next step involves estimating the average value of the output by considering the likely number of units produced each year and their probable value in the market place

1 A. R. Prest and R. Turvey, "Cost-Benefit Analysis: A Survey." The Economic Journal, Vol. LXXV. December, 1965, pp. 683-735. I am much indebted to this valuable and discerning survey. I have also relied upon :

Otto Eckstein, "A Survey of the Theory of Public Expenditure Criteria," in Public Finances: Needs, Sources, and Utilization, National Bureau of Economic Research (New York, Princeton University Press, 1961), pp. 439–504.

Irving K. Fox and Orris C. Herfindahl, Attainment of Efficiency in Satisfying Demands for Water Resources," American Economic Review, May. 1964. pp. 198–206.

Charles J. Hitch, On the Choice of Objectives in Systems Studies (Santa Monica, The RAND Corporation, 1960).

John V. Krutilla, "Is Public Intervention in Water Resources Development Conducive to Economic Efficiency," Natural Resources Journal, January, 1966, pp. 60–75.

John V. Krutilla and Otto Eckstein, Multiple Purpose River Development (Baltimore, Johns Hopkins Press, 1958).

Roland N. McKean, Efficiency in Government Through Systems Analysis with Emphasis on Water Resources Development, (New York, 1958).

* Prest and Turvey, ibid., p. 686.

of the future. Intangible, "secondary," benefits may them be considered. These time streams of costs and benefits are discounted so as to obtain the present value of costs and benefits. Projects whose benefits are greater than costs may then be approved, or the cost-benefit ratios may, with allowance for relative size, be used to rank projects in order of desirability.

Underlying Economic and Political Assumptions

A straightforward description of cost-benefit analysis cannot do justice to the powerful assumptions that underlie it or to the many conditions limiting its usefulness. The assumptions involve value judgments that are not always recognized and, when recognized, are not easily handled in practice. The limiting conditions arise partly out of the assumptions and partly out of severe computational difficulties in estimating costs, and especially benefits. Here I can only indicate some major problems.

3

Cost-benefit analysis is based on superiority in the market place, under competitive conditions and full employment, as the measure of value in society. Any imperfection in the market works against the validity of the results. Unless the same degree of monopoly were found throughout the economy, for example, a governmental body that enjoys monopolistic control of prices or outputs would not necessarily make the same investment decisions as under free competition. A similar difficulty occurs where the size of a project is large in comparison to the economy, as in some developing nations. The project itself then affects the constellation of relative prices and production against which its efficiency is measured. The assumption based on the classical full employment model is also important because it gives prices special significance. Where manpower is not being utilized, projects may justified in part as putting this unused resource to work.

be

The economic model on which cost-benefit analysis depends for its validity is based on a political theory. The idea is that in a free society the economy is to serve the individual's consistent preferences revealed and rationally pursued in the market place. Governments are not supposed to dictate preferences nor make decisions.

This individualist theory assumes as valid the current distribution of income. Preferences are valued in the market place where votes are based on disposable income. Governmental action to achieve efficiency, therefore, inevitably carries with it consequences for the distribution of income. Projects of different size and location and composition will transfer income in different amounts to different people. While economists might estimate the redistributive consequences of various projects, they cannot, on efficiency grounds, specify one or another as prefcrable. How is this serious problem to be handled?

Benefit-cost analysis is a way of trying to promote economic welfare. But whose welfare? No one knows how to deal with interpersonal comparisons of utility. It cannot be assumed that the desirability of rent supplements versus a highway or dam can be measured on a single utility scale. There is no scientific way to compare losses and gains

* In many important areas of policy such as national defense it is not possible to value the product directly in the market place. Since benefits cannot be valued in the same way as costs, it is necessary to resort to a somewhat different type of analysis. Instead of costbenefit analysis, therefore, the work is usually called cost-effectiveness or cost-utility analysis.