JOURNALOF The ideology of Efficiency: Searching for a Theoryof Policy Analysis*DANIEL W. BROMLEYDepartmentof AgriculturalEconomics,Universivof Wmconsin,Madison,Wisconsin53706Received June 1, 1989; revised August 1, 1989The potential Pareto improvement criterion and other measures of economic efficiencydo not pass the test of consistency and coherence within economic theory, nor do suchmeasures accord with what public decision makers seek in policy advice from economists.Such efficiency measures are, nonetheless, durable components of the ideology of economicsin general, and benefit-cost analysis in‘ particular. The objectivity of the policy scientist hasbeen confused with the objectivity of the science. While economic efficiency has no claim toobjectivity, the policy scientist can be an objective analyst of policy choices. o 1990 AcademicPress. Inc.I. ON IDEOLOGYIdeology can be thought of in two quite distinct ways. One connotation is as anemotional or propagandistic position held by someone. The ideologue is one whoengages in a variety of means-somesubtle, some not-inorder that others mightbe swayed. Indeed, one synonym of ideology is creed, which immediately leads oneto such terms as religion, faith, cult, and persuasion. In economics one is thought,on this definition, to be a “market ideologue” or a “collective-action ideologue.”This particular dimension of ideology is perhaps foremost in our minds when westrive to avoid letting subjectivity and the taint of our personal values have aninfluence on economic analysis. We are taught from an early age that objective(positive) analysis is both the goal of economics, and the relentless burden of thegood economist. It is said that we must avoid, at all cost, allowing ideology-bywhich is meant value judgments-tocolor our analysis.The second facet of ideology receives much less attention from the socialscientist-curiously,since this other version accords rather closely with what thesocial sciences are supposedly about. Here I have in mind ideology as an overallview of, or attitude toward, something. On this interpretation, ideology is a sharedsystem of meaning and comprehension. It is a structure within which information issupplied and processed, directions are given and justification for certain behavioris provided.’ Of course this dimension can be comprehended under the firstmeaning of ideology. We can easily understand how a religion represents a shared*I am grateful to Ron Cummings and several anonymous reviewers for helpful suggestions on anearlier draft. An earlier version of this paper was presented as the Kenneth Parson Lecture, Universityof Wisconsin-Madison, April, 1989.‘For a treatment of ideology in this vein see Appleby’s “Economic Thought and Ideology inSeventeenth-Century England” Ill.8600950696/90 3.00Copyright Q 1990 by Academic Press, Inc.All rights of reproductionin any form resewed.

THEIDEOLOGYOFEFFICIENCY87system of meaning and comprehension, how it supplies and aids its adherents toprocess that information, how it gives sanction regarding certain behaviors, andhow it offers justification for other behaviors.I wish to focus on the second-andmore subtle-notionof ideology here,suggesting that it represents a useful metaphor within which to discuss behaviorsand thought processes within a scientific discipline. Recall that ideology is a sharedsystem of meaning and comprehension, and that it is a set of norms for certainbehaviors. To be Kuhnian for a moment, an ideology is a paradigm. Normalscience is an ideology in that the recognized body of practitioners hold similarbeliefs about phenomena and processes that define the accepted domain ofenquiry. Indeed the very act of acquiring training in a particular scientific discipline is to understand and accept its ideology in this latter sense. To be “trained”is to be socialized into the paradigm. To talk of the ideology of a scientificdiscipline is not to imply fervor, propaganda, or preying (and praying) cults. It is,instead, to recognize that the very essence of a discipline is shared beliefs aboutthe meaning of events, about how to process information about those events, andabout how to add to the body of systematic concepts that ultimately differentiateone discipline from another.One abiding truth about a shared belief system is that it appears differentdepending on whether one regards it from within, or instead checks it against theexternal world. A view from within asks just two things of an ideology-isitconsistent, and is it coherent? That is, does it meet the test of logical validity, anddoes it comprehend all of the phenomena to which it claims relevance? Anideology-ora paradigm-isthus rather like a syllogism in logic, in that its validityis determined by a set of rules that have nothing at all to do with its truthfulness.An argument can be valid by the rules of logic and still have no connection withthe real world; validity says nothing about truth content. When one moves frominternal concerns to external matters attention shifts to the problem of concordance-howclosely a model or theory corresponds to the world it purports toexplain.My purpose here is to explore the ideology of efficiency, both with respect to itsconsistency and coherence within economics, and with respect to its correspondence to the reality with which it must connect. More particularly, I want todiscuss an aspect of the ideology of economics with respect to the emergence ofeficciency as an objective truth rule. By an “objective truth rule” I have in mind anaccepted behavioral norm that allows the economist to offer up an efficientoutcome as both evidence of a “good” thing, and-moreimportantly for theideology-asproof of the scientific objectivity of that particular finding of goodness. I hope to convince the reader that “economic efficiency” has no logical claimto “objectivity.” And, if efficiency has no secure claim to objectivity, then itsrecommendatory value for determining “goodness” is immediately undermined; itsurvives as a mere value judgment of the economist who recommends it.If this is the case, what then is the resource economist to do in the face of anapparent demand for insights regarding appropriate policy responses to problematic situations? Does the loss of an alleged objective truth rule render theeconomist irrelevant to the policy process? I will argue that, quite to the contrary,the abandonment of the usual efficiency norm liberates the economist to focusevaluation and analysis on those aspects of policy choices that matter most to thosein a position to decide. Finally, I will propose that we recognize the important

88DANIELW. BROMLEYdistinction between the objectivity of the science, and the objectivity of thescientist, a step that increases the scope for economic input into policy analysis.II. THE EMERGENCEOF EFFICIENCYTRUTH RULEAS AN OBJECTIVEThe Early PositivistsThe idea of the scientific objectivity-theethical neutrality-ofeconomics hasits grounding in the methodological writings, dating back to the latter part of the19th century, of Nassau Senior [45], John Stuart Mill [32], John Cairnes [14], andWalter Bagehot [51. These writers were united in the belief that economics was, toquote John Neville Keynes, “positive as distinguished from ethical or practical,and in its method abstract and deductive” [31, p. 751. Keynes, building on Comte’spositivism, seems to have popularized the now-familiar distinction between positiveand normative, the former being synonymous with scientific objectivity, the latterconnoting value-laden arguments we know as methaphysics. To state what everyeconomist holds dear, positive economics speaks to what is or what might be,normative economics speaks to what ought to be.2The elder Keynes defined economics as the study of “ . . . those human activitiesthat direct themselves towards the creation, appropriation, and accumulation ofwealth; and by economic customs and institutions. . . of human society in regard towealth . . . Political economy or economics is a body of doctrine relating toeconomic phenomena in the above sense. . ” [31, p. 701. At the time Keynes waswriting, political economy and economics were synonymous, and Keynes saw themas providing information as to the probable consequences of given lines of action,but not passing moral judgments or pronouncements about what ought to be done.At the same time, however, he argued that “ . . . the greatest value is attached tothe practical applications of economic science; and. . the economist ought. . . toturn his attention to them-not,however, in his character as a pure economist, butrather as a social philosopher, who, because he is an economist, is in possession ofthe necessary theoretical knowledge. . . [I]f this distinction is drawn, the socialand ethical aspects of practical problems-whichmay be of vital importance-areless likely to be overlooked or subordinated” [31, p. 761.A little over a decade after Keynes’s writing, Lionel Robbins published his mostinfluential book entitled “An Essay on the Nature and Significance of EconomicScience” [37]. Robbins had been much influenced by the logical positivists of theVienna Circle and he drew upon their ideas to stress several methodological points*George Stigler, no doubt speaking for a number of economists, has declared that “Economics as apositive science is ethically neutral” [46, p. 5221. From this one can reach two quite distinct conclusions.If Stigler meant by “economics” the entire body of economics embodied in, say, the accepted textbooksof the day then he should have said, “economics is a positive science and therefore it is, by definition,ethically neutral.” Alternatively, if Stigler meant that there is a part of economics that is “positive”then he should have said “that part of economics which is a positive science is, by definition, ethicallyneutral.” Under the first interpretation all of economics is declared to be ethically neutral, while underthe second definition only a subset of economics-the“positive” part-is ethically neutral. Either wayStigler is offering us either a tautology-adefinition-orhis personal views; neither is compelling forthere are no external criteria to which Stigler can turn for proof of his assertion.

THEIDEOLOGYOFEFFICIENCY89that survive today. Of foremost pertinence here, Robbins took from logicalpositivism the idea that there were only two kinds of propositions that could becountenanced in a science-thosethat were true by definition (tautologies), calledanalytical statements, and those that were empirical propositions (called syntheticstatements). Propositions that did not fit these two classes were said to be lackingtruth content and hence were value-laden. It is usually held that since the“scientific part” of economics consists exclusively of descriptive statements-eithertautologies or empirical propositions that can be tested-economicscannot haveany ethical entailments, and is therefore value-free.Ends and MeansUnlike Neville Keynes before him, Robbins insisted that economics was thestudy of the allocation of scarce means among competing ends, such ends beingbeyond question to the economist. “Being neutral, the argument proceeds, economics does not choose between or pronounce value-judgments on different ends,and it is implied that no value-judgments are involved in recommending ‘means’ togiven ‘ends’ ” [27, pp. llO-1111. To remain objective, economists should not choosebetween different ends, but must restrict themselves to recommending “means” soas to accomplish given “ends.” A close reading of Robbins reveals that he used theword “means” to refer to factors of production or financial resources that could beallocated to alternative employments. That is, Robbins envisioned an economicsthat was very much like the theory of the firm. Robbins’s definition of economics-that it is the study of choice involving scarcity in which conflicting means areconsidered to reach given ends-isstill the most common definition of ourdiscipline. The principal burden of Robbins’s work was an attempt to demarcatethe scientific part of economics from the value-laden part. He relied upon thedistinction between means and ends to effect this demarcation. The acceptance ofRobbins’s definition of economics, in which economists are said to study choiceamong scarce means to accomplish given ends, places a central burden on ourability precisely to differentiate ends from means. To make a clear differentiationbetween ends and means, however, it is necessary to invoke some external criterionso that the distinction-andthe linkage-betweenthe two is placed in context.That is, one cannot distinguish between ends and means without first having atheoretical basis upon which to ground that distinction. At the most abstract level,we might follow Robbins by suggesting that an “end” is something that enters intoan individual’s utility function, while a “means” would not be found there. This issimply a definition; it fits the positivist idea of an analytical statement-thatis, atautology. But having thus differentiated means from ends what has been accomplished? In one sense, a very helpful analytical start has been made; ends are thosethings that individuals care about, means are mere instruments, of no specialnotice, for accomplishing desired ends.On closer inspection, however, it is seen that a criterion external to theinvestigator is required to determine whether or not something is “in” the utilityfunction of an individual, or a group of individuals. It cannot be our determinationfor that is to impose the value system of the investigator into the analysis. While itmight be possible to ask all those affected by a particular policy whether there iscomplete agreement on our analytical distinction between ends and means, this israrely-ifever-done.Indeed, in most instances the respondents would be hard

90DANIELW. BROMLEYpressed to make a clear-cut distinction. Hence, the dichotomy upon which Robbinsbased his edifice of scientific objectivity against the insidious effects of metaphysicsis nothing more than a convenient assumption. In a simple world, where thedistinction between means and ends may be thought clear, it is necessary to regardthe means as simply factors of production or commodities in which there is nointrinsic merit attached to the components of either. This distinction is meaningless, however, in the real world of policy analysis in which there are few-perhapsno-policies(institutional arrangements) that can be assumed to be neutral meanswithout intrinsic value of their own [27].Utility and OphelimityRobbins regarded the means-ends distinction as central to the discussion ofinterpersonal utility comparisons in that to discuss ends one must make suchcomparisons, while to discuss means is to be ethically neutral. The old welfareeconomics made use of the idea of social utility as a summation of individualutilities so as to discuss the general well-being of the community via somethingcalled “material welfare.” To these economists, utility was an individual concept,while welfare was an aggregate concept. Utility, on this definition, meant usefulness-ratherlike the current dictionary definition [17]. Jevons [28] transformed theterm “utility” into a synonym for “desires” or “preferences,” a notion that Paretohad referred to as “ophelimity.”Prior to Jevons, utility-unlikeophelimity-wasnot subjective. Once the term “utility” took over both meanings-usefulnessanddesires-its practical content diminished. When the old welfare economists-Pigouand Marshall-thoughtof interpersonal comparisons of utility they thought interms of the general well-being of people, and the usefulness of policies to addresstheir problems. Public programs for the deprived certainly had utility in that theywere useful to the needs of the homeless or the ill-fed. But to ponder and toascertain the desires of people for public housing programs introduced a seriouscomplication. Hence, the old welfare economists could be concerned with thegeneral usefulness of alternative social states for accomplishing certain socialobjectives. On this definition of utility, Pigou could argue that the material welfareof the homeless could be increased more than the loss in material welfare of therich if taxes were raised some small percentage to provide housing for the poor. Itis difficult to say that under the current definition of utility.3Building on Jevons’s work, Robbins further muddied the distinction betweenophelimity and utility [17]. With the assistance of early ordinalists such as Hicksand Allen, he applied the term utility to the notion of desires and preferences,thereby purging from economics any discussion of usefulness. Henceforth31n this regard, Peter Hammond has observed that “Many succeeding welfare economists completely misinterpreted Robbins and took a particularly unfortunate step that proved to be a majorhandicap throughout the ensuing thirty years. In an entirely misguided attempt to be ‘scientific’, inRobbins’ sense, many welfare economists saw fit to exclude even the slightest possibility of makinginterpersonal comparisons” [21, p. 4061. See Ng [36] for a discussion of different types of interpersonalutility comparisons-notall of which are value laden. Hammond has earlier shown that “. . . ifinterpersonal comparisons of a certain kind are introduced, it is possible to construct a generalizedsocial welfare function (GSWF) satisfying appropriately modified forms of the Arrow conditions”122, p. 7991.

THEIDEOLOGYOF91EFFICIENCYeconomists could talk only of desires- or states of desire. It was then easy to claimthat economists could not make interpersonal comparisons of utility since utilitynow referred to unobservable preferences. Recall that the logical positivistsregarded unobservable phenomena to be outside the domain of science. Robbins’sadmonition about interpersonal comparisons of utility carried the day and thusseemed to undermine any hope for a scientific welfare economics.Policy AnalysisIt was about this time that the formal field of policy analysis was born, bringingwith it a renewed interest in the ability to pronounce on what would be “good”public policy. Policy analysis got its start, at least in the United States, with theFlood Control Act of 1933 (amended in 1936) in which it is stated that thegovernment would undertake public works on rivers and harbors if “the benefits towhomsoever they may accrue are in excess of the estimated costs, and if the livesand social security of people are otherwise adversely affected” [18, p. 21. At thetime of passage it was not immediately obvious what constituted a “benefit,” whilecosts were rather better understood as the necessary expenditures to bring aboutthe planned project.While creating a new branch of applied economics, which we know asbenefit-cost analysis, the legislation also compelled all government agencies tomake “explicit estimates of the gains and losses to be expected from theirproposals, and to defend the proposals in the light of these estimates” [18, p. 31.About this same time the field of welfare economics was greatly influenced by thework of John Hicks [26] and Nicholas Kaldor [30]. The combined effect of theirwork was to revive welfare economics in its “new” version via the expedient of aconsumer theory based on preferences and the concept of indifference, rather thanon utility of the old kind. The key to Kaldor’s method was to separate productionfrom distribution, a task that Pigou could not accomplish because of his utilitarianism. To Kaldor, splitting production away from distribution avoided the problem ofinterpersonal comparisons of utility since production dealt only with outputs perunit of input, and every economist “knows” that people prefer more to less. Thisaccomplishment tended to reinforce the point that economics was about production (or about efficiency)