McCloskey 1983 the Rhetoric of Economics

8/12/2019 McCloskey 1983 the Rhetoric of Economics

1/38

American Economic Association

The Rhetoric of EconomicsAuthor(s): Donald N. McCloskeySource: Journal of Economic Literature, Vol. 21, No. 2 (Jun., 1983), pp. 481-517Published by: American Economic AssociationStable URL: http://www.jstor.org/stable/2724987

Accessed: 24/10/2009 16:06

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=aea.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

American Economic Associationis collaborating with JSTOR to digitize, preserve and extend access toJournal

of Economic Literature.

http://www.jstor.org
http://www.jstor.org/stable/2724987?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=aeahttp://www.jstor.org/action/showPublisher?publisherCode=aeahttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2724987?origin=JSTOR-pdf


2/38

Journal of Economic LiteratureVol. XXI (June 1983), pp. 481-517

h e hetoric o conomics

By DONALD N. MCCLOSKEYTheUniversity f Iowa

The length of the acknowledgments here testifies to an unexploredfeature of the rhetoric of economics, the role of the audience: likeoratory, scholarship depends for its virtues on the virtues of itsaudience. I have been fortunate in mine. I must apologize for myamateurish understanding of what is happening in philosophy,mathematics, literary criticism, rhetorical studies, and other placesbeyond my competence, and ask that practitioners in thesefieldsassist in my further education. For their early attempts I thankEvan Fales, Paul Hernadi,John Lyne, Michael McGee,Allan Megill,John Nelson, and Jay Semel of the Colloquiumon Applied Rhetoricat The University of Iowa; Wayne Booth,IraKatznelson, and othersat the University of Chicago in the program in Politics, Rhetoricand Law, before which the earliest version was delivered; RobertBoynton, BernardCohn,John Comaroff;Otis Dudley Duncan, James0. Freedman, Clifford Geertz, William Kruskal, Donald Levine,Laura McCloskey,RichardRorty, RenatoRosaldo; and the Humani-ties Societyat the University of Iowa. Thatthe economists on whomI have inflicted the argument have reacted with such intelligentskepticismand generous encouragementsuggests, as the paperdoes,that we are better scholars than our methodology would allow. Ithank my colleagues in economicsat Iowa, especially the SanctuarySeminar in Economic Argument; Seminarsat the WorldBank andthe National Science Foundation; my colleagues at the Institute ofAdvanced Studies and the Faculty of Economics at the AustralianNational University;seminars at the universities of Adelaide, Auck-land, Melbourne,New South Wales,Tasmaniaand WesternAustra-lia; at Monash, and Iowa State universities; Victoria University ofWellington; and an assemblage of economists elsewhere: WilliamBreit,Ronald Coase,ArthurDiamond, Stanley Engerman,J. M. Fin-ger, Milton Friedman, Allan Gibbard,Robert Goodin, GaryHawke,Robert Higgs, Albert Hirschman, EricJones, Arjo Klamer, HarveyLeibenstein, David Levy, Peter Lindert, Neil de Marchi, MichaelMcPherson,Amartya Sen, Robert Solow, Larry Westphal, GordonWinston, and Gavin Wright. Thomas Mayer's encouragement atan early stage and his detailed comments as referee or this Journalat a later stage were exceptionally heartening and useful.481


3/38

482 Journal of Economic Literature, Vol. XXI (June 1983)ECONOMISTS DO NOT FOLLOW the lawsof enquiry their methodologies laydown. A good thing, too. If they did theywould stand silent on human capital, thelaw of demand, random walks down WallStreet, the elasticity of demand for gaso-line, and most other matters about whichthey commonly speak. In view of the volu-bility of economists the many officialmethodologies are apparently not thegrounds for their scientific conviction.Economists in fact argue on widergrounds, and should. Their genuine,workaday rhetoric, the way they argue in-side their heads or their seminar rooms,diverges from the official rhetoric. Econo-mists should become more self-consciousabout their rhetoric, because they willthen better know why they agree or dis-agree, and will find it less easy to dismisscontrary arguments on merely methodo-logical grounds. Philosophy as a set of nar-rowing rules of evidence should be setaside in scientific argument, as even manyphilosophers have been saying now forfifty years.Economics will not change much in sub-stance, of course, when economists recog-nize that the economic emperor has posi-tively no clothes. He is the same fellowwhether philosophically naked or clothed,in reasonably good health aside from hissartorial delusion. But the temper of argu-ment among economists would improveif they recognized on what grounds theywere arguing. They claim to be arguingon grounds of certain limited matters ofstatistical inference, on grounds of posi-tive economics, operationalism, behavior-ism, and other positivistic enthusiasms ofthe 1930s and 1940s. They believe thatthese are the only grounds for science. Butin their actual scientific work they argueabout the aptness of economic metaphors,the relevance of historical precedents,the persuasiveness of introspections, thepower of authority, the charm of symme-try, the claims of morality. Crude positiv-

ism labels such issues meaningless ornonscientific or just matters of opin-ion. Yet even positivists actually behaveas though the matters are discussable. Infact, most discussion in most sciences, andespecially in economics, arises from them.Nothing is gained from clinging to the Sci-entific Method, or to any methodology ex-cept honesty, clarity, and tolerance. Noth-ing is gained because the methodologydoes not describe the sciences it was oncethought to describe, such as physics ormathematics; and because physics andmathematics are not good models for eco-nomics anyway; and because the method-ology is now seen by many philosophersthemselves to be uncompelling; and be-cause economic science would stop pro-gressing if the methodology were in factused; and, most important, because eco-nomics, like any field, should get its stan-dards of argument from itself, not fromthe legislation of philosopher kings. Thereal arguments would then be joined.I. Rhetoric Is Disciplined Conversation

These points, elaborated below, amountto an appeal to examine the rhetoric ofeconomics. By rhetoric is not meant averbal shell game, as in empty rhetoricor mere rhetoric (although form is nottrivial, either: disdain for the form ofwords is evidence of a mind closed to thevarieties of argument). In Modern Dogmaand the Rhetoric of Assent Wayne Boothgives many useful definitions. Rhetoric isthe art of probing what men believe theyought to believe, rather than provingwhat is true according to abstract meth-ods ; it is the art of discovering good rea-sons, finding what really warrants assent,because any reasonable person ought tobe persuaded ; it is careful weighing ofmore-or-less good reasons to arrive atmore-or-less probable or plausible conclu-sions-none too secure but better thanwould be arrived at by chance or unthink-


4/38

McCloskey: The Rhetoric of Economics 483ing impulse ; it is the art of discoveringwarrantable beliefs and improving thosebeliefs in shared discourse ; its purposemust not be to talk someone else into apreconceived view; rather, it must be toengage in mutual inquiry (Booth, 1974,pp. xiii, xiv, 59, 137).It is what economists,like other dealers in ideas, do anyway: asBooth sayselsewhere, Webelieve in mu-tual persuasion as a way of life; we livefrom conference to conference (Booth,1967, p. 13).Rhetoricis exploring thoughtby conversation.The word rhetoric s doubtless an ob-stacle to understanding the point, so de-based has it become in common parlance.If pragmatism and anarchism had notalready suffered as much, unable to keepclear of irrelevant associations with thebottom line or the bomb, the title mightbetter have been Pragmatism'sConcep-tion of Truth in Economics or Outlineof an Anarchistic Theory of Knowledgein Economics (WilliamJames, 1907; PaulFeyerabend, 1975).But the enemies of so-phisticated pragmatism and gentle anar-chism, asof honest rhetoric,have used theweapons at hand. The results discourageonlookers from satisfying the curiositythey might have had about alternativesto coercion in philosophy, politics, ormethod. A title such as How EconomistsExplain (MarkBlaug, 1980; but see be-low) or Why Methodology Is a Badwould perhaps have been meeker andmore persuasive.1Still, rhetoric ike theothers is a fine and ancient word, whoseproper use ought to be more widelyknown among economists and calculators.The rhetoric here is that of Aristotle,Cicero, and Quintilian among the an-cients, reincarnated in the Renaissance,crucifiedby the Cartesiandogma that onlythe indubitable is true;which in the third

century after Descartes rose from thedead. The faith built on these miracles isknown in literary studies as the New Rhet-oric, new in the 1930s and 1940s fromthe hands of I. A. Richardsin Britain andKenneth Burke in America (Richards,1936; Burke, 1950). In philosophy JohnDewey and Ludwig Wittgenstein had al-ready begun to criticize Descartes' pro-gram of erecting belief on a foundationof skepticism.More recently KarlPopper,Thomas Kuhn, and Imre Lakatos amongothers have undermined the positivistsuppositionthat scientific progressdoes infact follow Descartes' doubting rules ofmethod. The literary, epistemological,and methodological strands have not yetwound into one cord, but they belong to-gether. On the eve of the Cartesian revo-lution the French philosopherand educa-tional reformer, Peter Ramus (fi. 1550),brought to completion a medieval ten-dency to relegate rhetoric to mere elo-quence, leaving logic in charge of reason.In the textbooks that Descartes himselfread as a boy probable argument wasmade thus for the first time wholly subser-vient to indubitable argument. Hostile toclassicalrhetoric, such a reorganizationofthe liberal arts was well suited for theCartesian program extending over thenext three centuries to put knowledge onfoundations built by philosophyand math-ematics. The program failed, and in themeantime probable argument languished.In Richard Rorty's words, followingDewey, the search for the foundations ofknowledge by Descartes, Locke, Hume,Kant, Russell, and Carnap was the tri-umph of the quest for certainty over thequest for wisdom (Rorty, 1979, p. 61; cf.John Dewey, 1929, pp. 33, 227). To rein-state rhetoric properly understood is toreinstate wider and wiser reasoning.The reaction to the narrowing of argu-ment by the Cartesianprogramis by nowbroad. Its leading figures range from pro-fessional philosophers (Stephen Toulmin,

IAfter recognizing the intent, Colin Forster ofAustralianNational University suggested the titleThe Last Paper on Methodology. But ambitionmust have limits.


5/38

484 Journal of Economic Literature, Vol. XXI (June 1983)Paul Feyerabend, Richard Rorty) to amiscellany of practitioners-turned-philos-ophers in chemistry(Michael Polanyi), aw(Chaim Perelman), and literary criticism(Wayne Booth). The reach of the idea now-adays that argument is more than syllo-gism is illustrated well by the lucid treat-ment of it in what would seem an unlikelyplace, by Glenn Webster, Ada Jacox, andBeverly Baldwin in Nursing Theory andthe Ghost of the Received View (1981,pp. 25-35). The reach, however, has notextended to economics. Austrian, institu-tionalist, and Marxist economists, to besure, have for a century been attackingcertain parts of positivism as the basis foreconomic knowledge. But they haveseized on other parts with redoubled fer-vor, and have so expressed their remain-ing doubts as to make them unintelligibleto anyone but themselves. In their ownway they have been as narrowingas thor-oughgoing positivists-the rejection ofeconometrics, for instance, would be rea-sonable only if its more naive claims weretaken seriously. For the rest, economistshave let philosophical scribblers of a fewyears back supply their official thinkingabout what a good argument is.

II. The Official Methodology ofEconomics Is ModernistEconomists have two attitudes towardsdiscourse, the official and unofficial, theexplicit andthe implicit.The officialrheto-ric, to which they subscribein the abstractand in methodological ruminations, de-clares them to be scientists in the modernmode. The credo of Scientific Method,known mockingly among its many criticsas the Received View, is an amalgam oflogical positivism,behaviorism, operation-alism, and the hypothetico-deductivemodel of science. Its leading idea is thatall sureknowledge is modeled on the early20th century's understanding of certainpieces of 19th century physics.To empha-

size its pervasiveness in modern thinkingwell beyond scholarshipit is best labeledsimply modernism, that is, the notion(asBooth puts it) that we know only whatwe cannot doubt and cannot really knowwhat we can merely assent to.Among the precepts of modernism are:

(1) Prediction (and control) is the- goalof science.(2) Only the observable implications (orpredictions) of a theory matter to itstruth.(3) Observability entails objective, re-producible experiments.(4) If (and only if) an experimental impli-cation of a theory proves false is thetheory proved false.(5) Objectivity is to be treasured;subjec-tive observation (introspection) isnot scientific knowledge.(6) Kelvin's Dictum: When you cannotexpress it in numbers, your knowl-edge is of a meagre and unsatisfac-tory kind. 2

(7) Introspection, metaphysical belief,aesthetics, and the like may well fig-ure in the discovery of an hypothesisbut cannot figure in its justification.(8) It is the business of methodology todemarcate scientific reasoning fromnon-scientific, positive from norma-tive.(9) A scientific explanation of an eventbrings the event under a coveringlaw.(10) Scientists, for instance economic sci-entists, have nothing to say as scien-tists about values, whether of moral-ity or art.

2From SirWilliamThomson(LordKelvin),Popu-larAddresses,edition of 1888-1889, quotedin Kuhn,1977, p. 178n. An approximationto this version isinscribed on the frontof the SocialScience ResearchBuilding at the University of Chicago.FrankKnight,the famous University of Iowa economist, is said tohave remarked on it one day: Yes,and when youcan express it in numbers your knowledge is of ameagre and unsatisfactorykind.


6/38

McCloskey:The Rhetoric of Economics 485(11) Hume's Fork: When we run over li-braries, persuaded of these princi-ples, what havoc must we make? If

we take in our hand any volume-of divinity or school metaphysics, forinstance-let us ask, Does it containany abstract reasoning concerningquantity or number? No. Does it con-tain any experimentalreasoningcon-cerning matter offact and existence?No. Commit it then to the flames,forit can contain nothing but sophistryand illusion (italics his [1748], 1955,p. 173).Few in philosophy now believe as manyas half of these propositions. A substantial,respectable, and growing minority be-lieves none of them. But a large majorityin economics believes them all.For instance, the leading methodolo-gists in economics do. It is odd but truethat modernism in economic methodol-ogy is associated with the Chicago School.3The main texts of economic modernism,such as Milton Friedman's The Method-ology of Positive Economics (1953) orGary Becker and George Stigler's DeGustibus Non Est Disputandum (1977),bear a Chicago postmark;and the moreextreme interpretations of the texts flour-ish among economists bearing a Chicagodegree. What is odd aboutit is that a groupso annoying to other economists in mostof its activities should have their assentin the matter of official method: Oddly,a watered down version of Friedman's es-

say is part of the intellectual equipmentof most economists, and its argumentscome readily to their lips.Premeditated writings on method, notexcluding Chicago'sown, are more carefulthan the remark in the course of otherbusiness that reveals modernism in itsrawer form. In precept one can be vagueenough to earn the assent of everyone;in practice one must make enemies. Kal-man Cohen and Richard Cyert, to takeone among many examples of first-chaptermethodology in economics texts, presentin their book an outline of modernism,

which they assert is the method used inall scientific analyses (1975, p. 17). Themethod they then outline, with a bibli-ographyheavily weighted towardslogicalpositivism and its allies, reduces to an ap-peal to be honest and thoughtful. Onlywhen such a phrase as at least in princi-ple testable by experiment and observa-tion (p. 23) is given content by practicedo we know what is at stake. To be sure,vague precepts are not without their uses.When Friedman wrote, for instance, thepractice of economics was split into theorywithout fact and fact without theory. Hismodernist incantations, supported bychorusesof philosophers,were at the timeprobably good for the souls of all con-cerned.Friedman's essay was even then morepost-modernist than one might supposefromslight acquaintancewith its ideas. Hedid, for example, mention with approvalthe aesthetic criteria of simplicity andfruitfulness that an economist might useto select among a multiplicity of theorieswith the same predictions, though in thenext sentence he attempted to reducethem to objective matters of prediction(p. 10). He accepted that questionnaires,forbidden to the modernist in economics,are useful for suggesting hypotheses,though in the next sentence he assertedthat they are almost entirely useless asa means of testing the validity of eco-

3Nothing in this essay is meant to give comfortto the enemies of Chicago.Having long been a vic-tim of their anti-Chicagodogmatism, I am not im-pressed by the assertionthat Chicago economics ispeculiarlydogmatic.Chicago s merely aparticularlyclear and candid versionof a dogmatic impulse com-mon to all economics, expressing itself in methodo-logical imperatives. Economists appear to believethat economics is too important to be left to theopen-minded, and especially must never be left toanyone lacking faith in some approved formula forachieving knowledge. Chicago is no worse than therest. Immo,civis Chicagonussum, subspeciesTP(cfMelvin Reder, 1982).


7/38

486 Journal of Economic Literature, Vol. XXI (June 1983)nomic hypotheses p. 31n). He empha-sized the role of the rhetoricalcommunityto which the scientist speaksin producingconviction-whether made up of sociolo-gists, say, or of economists-though in thenext sentence he returned to an objec-tive theory of testing. Like Karl Popper,Friedman appeared to be struggling to es-cape the grip of positivism and its intellec-tual traditions,though with only sporadicsuccess. Perhaps that the locus classicusof economic modernism contains so muchthat is anti-modernistindicates that mod-ernism cannot survive intelligent discus-sion even by its best advocates.The unpremeditatedremark in the heatof economic argument, however, usuallyhas a crudely modernist content, often inFriedman's very words. An article byRichard Roll and Stephen Rosson finance,for instance, asserts that the theoryshould be tested by its conclusions, notby its assumptions and that similarly,one should not reject the conclusions de-rived from firmprofitmaximization on thebasis of sample surveysin which managersclaim that they trade off profit for socialgood (1980, p. 1093 and footnote). Thesame can be found elsewhere, in nearlyidentical terms, all dating back to Fried-man's essay:WilliamSharpe (1970, p. 77),for instance, writing on the same matteras Roll and Ross,takes it as a rule of politescientificbehaviorthat the realism of theassumptions matters little. If the implica-tions are reasonably consistent with ob-served phenomena, the theory can be saidto 'explain'reality (1970, p. 77). Repeatedoften, and exhibiting modernism as wellin their devotion to objective evidence,quantifiable tests, positive analysis, andother articles of the faith, such phraseshave the ring of incantation. Modernismis influential in economics, but not be-cause its premises have been examinedcarefully and found good. It is a revealed,not a reasoned, religion.

III. Modernism Is a Poor MethodModernism Is Obsolete in Philosophy

There are a great many things wrongwith modernism as a methodology for sci-ence or for economic science.4Even whenphilosophically 'inclined, economists ap-pear to read about as much in professionalphilosophy as philosophers do in profes-sional economics. It is unsurprising,then,that the news of the decline of modernismhas not reached all ears. From a philoso-pher's point of view the worst flaw in thehostility to the metaphysics that mod-ernism sees everywhere is that the hostil-ity is itself metaphysical. If metaphysicsis to be castinto the flames,then the meth-odological declarations of the modernistfamily from Descartes through Hume andComte to Russelland Hempel and Popperwill be the first to go. For this and othergood reasonsphilosophersagree that strictlogical positivismis dead, raisingthe ques-tion whether economists are wise to carryon with their necrophilia.5In the economic case the metaphysicalposition akin to logical positivism is notwell argued, probablybecause its roots liemore in the philosophizing of physicistsfrom Mach to Bridgeman than in the par-allel thinkingof professionalphilosophers.It is at least obscure what might be theappeal of operationallymeaningful state-ments (Paul Samuelson, 1947, p. 3 and

4 The overdiscussedquestionof whether there canbe a value-free social science will not be much dis-cussed here, but it must be accounted one of thechief failingsof modernismthat it placesmoralargu-ment outside the pale of rationaldiscussion.In thisconnection it should be more widely known thatMorrisSchlick, the founder of the Vienna Circle oflogical positivism and a vigorous lecturer on thetheme that moral knowledge is no knowledge at all,was murdered in 1936 by one of his students.6See John Passmore, 1967. Karl Popper quotesPassmore with approvalfor the motto of a chapterof his own entitled WhoKilledLogicalPositivism?(Popper, 1976, pp. 87-90), in which he confesses tothe murder.


8/38

McCloskey: The Rhetoric of Economics 487throughout)or valid and meaningful pre-dictions about phenomena not yet ob-served (Friedman, p. 7) as standardsagainst which all but mathematical asser-tions are to be judged. Samuelson, Fried-man, or their followers do not present rea-sons for adopting such metaphysicalpositions, except for confident assertions,at the time correct, that they were thereceived views of philosophers on themethod of science. The trustin philosophywas a tactical error, for the philosophy it-self has since changed. Some philosophersnow doubt the entire enterprise of epis-temology and its claim to provide founda-tions for knowledge (Richard Rorty,1982b).A great many doubt the prescrip-tions of modernist methodology.Falsification Is Not Cogent

A prescriptionthateconomic methodol-ogists have in common, for instance, is anemphasis on the crucialfalsifyingtest, sup-posedly the hallmark of scientific reason-ing. But philosophershave recognized formany decades that falsificationruns afoulof a criticism made by the physicist andphilosopher Pierre Duhem in 1906, evi-dent at once without philosophical read-ing to an economist who has tried to usefalsification for science. Suppose that thehypothesis H ( Britishbusinessmen per-formed very poorly relative to Americansand Germans in the late 19th century )implies a testing observation 0 ( Mea-sures of total factor productivity in ironand steel show a large difference betweenBritish and foreign steelmaking ); it im-plies it, that is, not by itself, but only withthe addition of ancillary hypotheses H1,H2, and so forth that make the measure-ment possible ( Marginal productivitytheory applies to Britain 1870-1913 ;British steel had no hidden inputs offset-ting poor business leaderships ; and soforth).Then of course not-0 implies not-H-or not-H1 or not-H2 or any number

of failures of premises irrelevant to themain hypothesis in question.The hypothe-sis in question is insulatedfrom crucialtestby the ancillary hypotheses necessary tobring it to a test. This is no mere possibilitybut the substance of most scientific dis-agreement: Your experiment was notproperly controlled ; You have notsolved the identification problem ; Youhave used an equilibrium (competitive,single-equation) model when a disequi-librium (monopolistic, 500-equation)model is relevant. And even if the onehypothesis in question could be isolated,the probabilistic nature of hypotheses,most especially in economics, makes cru-cialexperiments non-crucial:chance is theever present alternative, the Hn thatspoils falsificationism.Prediction Is Impossible in Economics

The common claim that prediction isthe defining feature of a real science, andthat economics possesses the feature, isequally open to doubt. It is a cliche amongphilosophersand historians of science, forinstance, that one of the most successfulof all scientific theories, the theory of evo-lution, has no predictions in the normalsense, and is therefore unfalsifiableby pre-diction. It is at least suggestive of some-thing odd in prediction as a criterion foruseful economics that Darwin's theorywas inspired by classicaleconomics, a sys-tem as it happens erroneous in most ofthe predictionsit made. With no apparentawareness of the incongruity, Friedmanquoted Alchian'srevival of the connection(ArmenAlchian, 1950) in the midst of hismost famous piece of predictionist meta-physics ( the leaves are positioned as ifeach leaf deliberately sought to maximizethe amount of sunlight it receives ).In any event, predicting the economicfuture is, as Ludwig von Misesput it, be-yond the power of any mortal man (1949,p. 867). What puts it beyond his power


9/38

488 Journal of Economic Literature, Vol. XXI (June 1983)is the very economics he uses to make theprediction. When the economist for a bigbank predicts lower interest rates afterChristmas, and has not before the predic-tion placed his net worth in margin loanson bonds, properly hedged and insuredagainst variance, he is behaving either ir-rationally or self-deceivingly. If he knowsthe expected value of the future, he forsome reason chooses not to take the unlim-ited wealth that such Faustian knowledgecan surely bring, and is willing for somereason instead to dissipate the opportunityby the act of telling others about it. If hedoes not really know, then he faces nosuch unexploited opportunity. But thenhe has perhaps no business talking asthough he does. Predictionism cannot berescued by remarking that the big bankeconomist makes only conditional predic-tions. Conditionalpredictions sell for theirvalue in a soft market: if the sea were todisappear,a rock would accelerate in fall-ing from sea level to the sea floor at about32.17 feet per second per second. But aserious prediction has serious boundaryconditions. If it does it must answer againthe American Question:If you're so smartwhy aren't you rich? At the margin (be-cause that is where economics works)andon average (because some people arelucky) the industry of making economicpredictions, which includes universities,earns only normal returns.ModernismItself Is Impossible,and Is NotFollowedThe most damaging, however, of theselesser criticisms of the modernist method-ology is that if takenat its wordit is narrowto absurdity.Consider again the steps tomodernistknowledge, frompredictionismthrough Kelvin's Dictum to Hume's Fork.If economists (or physicists) confinedthemselves to economic (or physical)propositions that literally conformed tosuch steps they would have nothing to say.Cartesian or Humean skepticism is too

corrosive a standard of belief for a realhuman. As the chemist and philosopherMichael Polanyi put it, the methodologyof modernism sets up quixotic standardsof valid meaning which, if rigorouslyprac-ticed, would reduce us all to voluntaryimbecility (1962, p. 88). Modernismpromises knowledge free from doubt,metaphysics,morals, and personal convic-tion; what it delivers merely renames asScientific Method the scientist's and espe-cially the economic scientist's metaphys-ics, morals, and personal convictions. Itcannot, and should not, deliver what itpromises. Scientific knowledge is no dif-ferent from other personal knowledge(Polanyi, 1962). Trying to make it differ-ent, instead of simply better, is the deathof science.In other words, the literal applicationof modernist methodology cannot give auseful economics. The best proofs are his-torical.In his Against Method (1975) PaulFeyerabend uses an interpretationof Gal-ileo's career to attack the claims of pre-scriptive methodology in physics; thesame point can be made abouteconomics.Had the modernist criterionof persuasionbeen adopted by Galileo's contemporar-ies, he argues, the Galilean case wouldhave failed. A grant proposal to use thestrange premise that terrestrial optics ap-plied also to the celestial sphere, to assertthat the tides were the sloshing of wateron a mobile earth, and to suppose thatthe fuzzy views of Jupiter'salleged moonswould prove, by a wild analogy, that theplanets, too, went around the sun as didthe moons aroundJupiterwould not havesurvived the first round of peer reviewin a National Science Foundationof 1632,at any rate if that one (unlike ours)werewedded to modernist ideology. The argu-ment applieswidely to the historyof phys-ics:observationalanomalies in the experi-ments testing Einstein's theories wereignored for many years, to be revealedas errors of measurement long after the


10/38

McCloskey: The Rhetoric of Economics 489theories had been embraced, embracedon groundsof the reason of the matter,as Einstein was fond of saying (Feyera-bend, 1975, pp. 56-57).Historians of biology have uncoveredone after another case of cooking the sta-tistical results to fit modernist precepts ofwhat counts asevidence, fromPasteur andMendel down to the present. The mea-surement of IQ has been a scandalof self-deception and bold fraud in the name ofscientific method from its beginning (Ste-phen Jay Gould, 1981). Perhaps modern-ism fits poorly the complexities of biologyand psychology: straining after evidenceof a sorttypicallyavailableonly in the sim-plest experiments in physics may not suittheir frontiers. It suits the frontiersof eco-nomics poorly enough. For better orworse the Keynesian revolution in eco-nomics would not have happened underthe modernist legislation recommendedfor the method of science. The Keynesianinsights were not formulated as statisticalpropositionsuntil the early 1950s, well af-ter the bulk of younger economists hadbecome persuaded they were true. By theearly 1960s liquidity trapsand acceleratormodels of investment, despite failures intheir statistical implementations, weretaught to first-yearstudents of economicsas matters of scientific routine. Modernistmethodology would have stopped all thisin 1936: where was the evidence of anobjective, statistical,controlled kind?Nor was the monetarist counterrevolu-tion a successfor modernist methodology,though so powerful had the methodologybecome by the 1960sin the minds of econ-omists and especiallyof monetarist econo-mists that most of the explicit debate tookplace in its terms. Yet in truth crude ex-periments and big books won the day, bytheir very crudeness and bigness. TheKennedy tax cut boosted the Keynesiansto their peak of prestige; the inflation ofthe 1970s broughtthem down again, leav-ing the monetaristsas temporary kings of

the castle. An important blow for mone-tarism was Friedman and Schwartz' bigbook, A Monetary History of the UnitedStates,1867-1960. It established a correla-tion, which Keynesians would not deny,between money and money income. Thesignificance of the correlation, however,depended on the assumptionthat moneycaused prices and that money was deter-minable by the monetary authority (in1929-1933, for example)despite the open-ness of the American economy to tradein both goods and money itself. Nonethe-less, what was telling in the debate wasthe sheer bulk of the book-the richnessand intelligence of its arguments, how-ever irrelevant most of the argumentswere to the main point.A modernist method thoroughly ap-plied, in other words, would probably stopadvances in economics. What empiricalanomaly in the traditional tale inspiredthe new labor economics or the new eco-nomic history? None: they were merelyrealizations that the logic of economicshad not exhausted its applicabilityat con-ventional borders. Whatobservableimpli-cationsjustify the investment of intellectsince 1950 in general equilibriumtheory?For all the modernist talkcommon amongits theorists, none; but so what? Could ap-plications of economics to legal questionsrely entirely on objective evidence? No;but why would one wish to limit the playof understanding? And so forth. There isnothing to be gained and a great deal tobe lost by adopting modernism in eco-nomic methodology.The very point is economic. In orderfor an economic theory to be tested, Ron-ald Coase points out, some economistsmust care enough about it to bother. Theycare only when it is believed by some in-vestigators-they and their allies or somesignificantgroupof opponents. Only whenmany believe is there a demand for tests.Fortunately, economists, or at any rateenough of them, do not wait to discover


11/38

490 Journal of Economic Literature, Vol. XXI (June 1983)whether a theory's predictions are accu-rate before making up their minds ; towait in proper modernist style would re-sult in the paralysis of scientific activity(Coase, 1982, p. 14) because no one wouldhave an incentive to choose one out ofthe infinite number of hypotheses for test.Even quantitative studies, he argues, relyheavily on pre-quantitative argumentsfounding belief, and he quotes with ap-proval T. S. Kuhn's remark that the roadfrom scientific law to scientific measure-ment can rarely be traveled in the reversedirection (Coase, p. 18, quoting Kuhn,1977, p. 219). The laws come from a rheto-ric of tradition or introspection, and inphysics as in economics quantitativestudies . . . are explorations with the aidof a theory (Coase, p. 17), searches fornumbers with which to make specific atheory already believed on other grounds(see Edward Leamer, 1978, and thediscussion below). Modernism is imprac-tical.Any Method Is Arrogant and PretentiousThe objections to modernist method sofar, however, are lesser ones. The greaterobjection is simply that modernism is amethod. It sets up laws of argument drawnfrom an ideal science or the underlyinghistory of science or the essence of knowl-edge. The claim is that the philosopherof science can tell what makes for good,useful, fruitful, progressive science. Heknows this so confidently that he can limitarguments that worthy scientists makespontaneously, casting out some as un-scientific, or at best placing them firmlyin the context of discovery. The philoso-pher undertakes to second-guess the sci-entific community. In economics theclaim of methodological legislation is thatthe legislator is not merely expert in allbranches of economic knowledge withinsound of his proclamations but expert inall possible future economics, limiting thegrowth of economics now in order to

make it fit a philosopher'sidea of the ulti-mate good.It is hard to take such claims seriously.Einstein remarked that Whoeverunder-takes to set himself up as a judge in thefield of Truth and Knowledge is ship-wrecked by the laughter of the gods (Ein-stein, 1953, p. 38). Modernism sets up acourt of the Red Queen ( Normativeargu-ment, she says, offwith his head ), andthe gods laugh merrily. Any methodologythat is law-making and limiting will do so.It will do so with the noblest intentions,but economists are fond of pointing outin like cases that noble intentions can havebad consequences. The methodologistfancies himself the judge of the practi-tioner. His proper business, though, is ananarchistic one, resisting the rigidity andpretension of rules. I. A. Richardsappliedthe point to the theory of metaphor: Itsbusiness is not to replace practice, or totell us how to do what we cannot do al-ready;but to protect our naturalskill fromthe interference of unnecessarily crudeviews about it (1936, p. 116).The crudeness of modernist methodol-ogy, or of any methodology reducible torigid precept, is bad;but that it is allowedto interfere with practice is worse. Thecustom of methodological papers in eco-nomics is to scold economistsfor not allow-ing it to interfere more. MarkBlaug'suse-ful book summarizing the state of play ofeconomic methodology in 1980, TheMethodology of Economics:OrHowEcon-omists Explain, is a recent case in point.It would be better subtitled How theYoung Karl Popper Explained, since itrepeatedly attacks extant arguments ineconomics for failing to comply with therules Popper laid down in Logik der For-schung in 1934. Blaug'sexordium is typi-cal of the best of the methodologists ineconomics: Economists have long beenaware of the need to defend 'correct'prin-ciples of reasoning in their subject; al-though actual practice may bear little


12/38

McCloskey:The Rhetoric of Economics 491relationship to what is preached, thepreaching is worth considering on its ownground (Blaug, p. xii). Words like theseflow easily from a modernist's pen. Butwhy would preaching unrelated to actualpractice be worth considering at all? Whydo economists have to defend in the ab-stract their principles of reasoning, andbefore what tribunal?A case for havinga methodology-whether logical positivistor Popperian or Austrian or Marxist-would be expected to give answers to thequestionsof why, but commonly does not.Recent philosophy of science and ordinarygood sense suggest that it cannot. Blaug'speroration is frankly prescriptive, takingeconomic rhetoric directly from philoso-phy:

Whatmethodologycan do is to provide criteriafor the acceptance and rejection of researchprograms, setting standards that will help usto discriminate between wheat and chaff. Theultimatequestionwe can andindeed mustposeabout any research program is the one madefamiliarby Popper:whatevents, if they materi-alize, would lead us to reject that program?A programthat cannot meet that question hasfallen shortof the highest standards hat scien-tific knowledge can attain [1980, p. 264].It sounds grand, but Einstein's gods arerolling in the aisles.Why should a dubiousepistemological principle be any test ofpractice, much less the ultimate test? Anddoesn't science takeplace most of the timewell short of the ultimate?Anyone would commend the vision ofscience that Popper and his followershave-of science as a self-correcting ex-ploration verging on the dialectic other-wise so foreign to the analytic traditionin philosophy. For an economic scientistto adopt an obdurate refusal to considerobjections and to resist offering hostagesto evidence, though as common in mod-ernist as in nonmodernist circles, is notmerely unscientific; it is cowardly. Somuch one can take from the idea of falsifi-cation by evidence. The problem comes,and the modernist preaching begins, with

the word evidence. Shouldit all be ob-jective, experimental, positive, ob-servable ?Can it be? In The Open Societyand Its Enemies (1945) Popper closes theborders of his society to psychoanalystsand Marxistson the grounds that they donot conform to the modernist notion ofevidence prevalent there. He would alsohave to close it to physicists from GalileoGalilei to particle charmers.An economistbracero,surely,would be deported on thenext truck from such an open intellectualsociety.Other Sciences Do Not Follow ModernistMethods

For all its claims to the scientific priest-hood, then, economics is different fromthe man-in-the-street's image of Science.Economists should be glad that their sub-ject fits poorly with this image and wellwith the New Rhetoric,as do studies longforeign to economics such as the study ofliterature or law or politics. Economics,in other words, is not a Science in the waywe came to understand that word in highschool.But neither, really, are other sciences.Other sciences, even the other mathemat-ical sciences, even the Queen herself, arerhetorical. Mathematicsappears to an in-cognoscento to be the limiting exampleof objectivity, explicitness, and demon-strability.Surely here is bedrockfor belief.Yet standards of mathematical demon-stration change. The last fifty years havebeen a disappointment to followersof Da-vid Hilbert and his programto put mathe-matics on indubitable foundations. Thehistorian of mathematics, Morris Kline,wrote recently that it is now apparentthat the concept of a universallyaccepted,infallible body of reasoning-the majesticmathematics of 1800 and the pride ofman-is a grand illusion. Or again:

There is no rigorousdefinitionof rigor.A proofis accepted if it obtains the endorsement ofthe leading specialistsof the time and employs


13/38

492 Journal of Economic Literature, Vol. XXI (June 1983)the principles that are fashionable at the mo-ment. But no standard s universallyacceptabletoday [1980, pp. 6, 315].

The recent flap over a computerized proofof the four-colorproposition is one exam-ple. The more fundamental example issaid to be Kurt G6del's proof fifty yearsago that some true and statable proposi-tions in mathematics are unprovable. Thepoint is controversial.John van Heijenoortwrites that the bearing of G6del's resultson epistemological problems remains un-certain. . . . [T]hey should not be rashlycalled upon to establish the primacy ofsome act of intuition that would dispensewith formalization (1967, p. 357). To besure. But one need not dispense with for-malization and flee to an unexamined actof intuition to think that formalization haslimits.Kline's opinions are somewhat looselyexpressed, and unpopular among mathe-maticians. Apparently less so are those ofPhilip J. Davis and Reuben Hersh, whoserecent book The Mathematical Experi-ence(1981)wasdescribed in the AmericanMathematical Monthly as one of themasterpiecesof ourage. Davis and Hershspeakof the crisis of confidence in modernmathematical philosophy, however, interms nearly identical with Kline's. In thework of The Ideal Mathematician theline between complete and incompleteproof is alwayssomewhat fuzzy, and oftencontroversial (p. 34;cf. p. 40). They quotea living Ideal Mathematician, SolomonFeferman, who writes it is also clear thatthe searchfor ultimate foundations via for-mal systemshas failed to arrive at any con-vincing conclusion (p. 357). Without us-ing the word, Davis and Hersh argue thatwhat is requiredis a rhetoric of mathemat-ics:

The dominantstyleof Anglo-Americanphiloso-phy ... tends to perpetuate identification'ofthe philosophyof mathematics with logic andthe study of formal systems. From this stand-point, a problem of principal concern to the

mathematician becomes totally invisible. Thisis the problemof giving a philosophicalaccount... of preformal mathematics.. . , includingan examination of how [it] relates to and is af-fected by formalization[1981, p. 344].

They assert that informal mathematicsis mathematics. Formalization is only anabstract possibility which no one wouldwant or be able actually to carry out (p.349). Real proofs are establishedby 'con-sensus of the qualified ' and are notcheckable . . . by any mathematician notprivy to the gestalt, the mode of thoughtin the particular field. . . . It may takegenerations to detect an error (p. 354).They conclude:

The actual experience of all schools-and theactual daily experience of mathematicians-shows thatmathematicaltruth,like other kindsof truth, is fallible and corrigible.... It is rea-sonable to propose a different task for mathe-matical philosophy, not to seek indubitabletruth, but to give an account of mathematicalknowledge as it really is-fallible, corrigible,tentative, and evolving, as is every other kindof human knowledge [p. 406].

Not much in this line has been done,though one astounding piece has shownwhat can be: Imre Lakatos' Proofs andRefutations: The Logic of MathematicalDiscovery gives an account for a theoremin topology of the rhetoric of mathemat-ics.It appears, then, that some deep prob-lems facing mathematics are problems ofrhetoric, problems in the art of probingwhat men believe they ought to believe.Similar points can be made about othersciences, such as paleontology. The sud-den proliferation of species at the begin-ning of the Cambrian period, one of thegreat puzzles in evolution, was explainedby Steven Stanley in 1973 by supposingthe sudden arrivalof forms of life that fedon other forms of life, single-celled her-bivores, as it were, in a grassy sea. Theirgrazing on the dominant forms allowednew forms to survive the competitionfrom the previously dominant ones, which


14/38

McCloskey:The Rhetoric of Economics 493in turn resulted in new grazers. StephenJay Gould remarks of the arguments of-fered in support of this brilliant and per-suasive theory that:. . .they do not correspond to the simplisticnotions about scientificprogressthat are taughtin most high schools and advancedby most me-dia.Stanley does not invoke proof by new infor-mation obtained fromrigorousexperiment.Hissecond criterion is a methodological presump-tion, the third a philosophicalpreference, the

fourth an application of prior theory. . . . Sci-ence, at its best, interposes human judgmentand ingenuity upon all its proceedings. It is,after all (although we sometimes forget it),practiced by human beings [Gould, 1977, p.125].One can even say the same of physics,that favorite of outsiders seeking a pre-scription for real, objective, positive,predictive science. The sequence Carnap-Popper-Lakatos-Kuhn-Feyerabend epre-sents in the history and philosophy ofphysics a descent, accelerating recently,from the frigid peaks of scientific absolut-ism to the sweet valleys of anarchicrheto-ric (see Popper, 1934, 1976; Lakatos, 1970;Kuhn, 1970; Feyerabend, 1975, 1978). Ifeconomics should imitate other sciences,imitate even the majesty of physics andmathematics (there is, to be sure, consid-erable doubt that it should),then it shouldofficially open itself to a wider range ofdiscourse.IV. The UnofficialRhetoric Is HonorableBut UnexaminedEconometricRhetoric Is Too Narrow

But unofficially t does. The second atti-tude towards discourse is that adopted inactual scientific work in economics. It isdifferent from the official,modernist rhet-oric. What is alarmingabout the workadayrhetoric is not its content but that it isunexamined, and that in consequence theofficial rhetoric pops up in mischievousways.Economistsagree or disagree-theirdisagreements are exaggerated-but they

do not know why. Any economist believesmore than his evidence of a suitablymod-ernist and objective sort implies. A recentpoll of economists, for example, found thatonly three percent of those surveyed flatlydisagreed with the assertion that tariffsand import quotas reduce general eco-nomic welfare. Only two percent dis-agreed with the assertion that a ceilingon rents reduces the quantity and qualityof housing available. Only eight percentdisagreed with the assertionthat the tax-ing and spending of government has a sig-nificant impact on the income of a partlyidle economy (J. R. Kearl, Clayne Pope,Gordon Whiting, and Larry Wimmer,1979). You probably fall into the 97, 98,and 92 percent majorities.The evidencefor the assertions, however, is obscure.How do economists know these state-ments are true? Where did they acquiresuch confidence? The usual answer is thattheory tells us. But great social ques-tions are not answered by looking at a dia-gram on a blackboard,because it is trivi-ally easy to draw a diagram that yieldsthe opposite answer. The factual experi-ence of the economy, certainly, has littleto do with their confidence. No study hasshown in ways that would satisfy a consis-tent modernist, for example, that high tar-iffs in America during the 19th century,on balance, hurt Americans. Yet it is be-lieved that tariffs hurt then and now.6 Nostudy has shown that an inadvertent pol-icy of fiscal ease brought unemploymentdown during the War. Yet it is believedon all sides. Economists have not consid-ered their rhetoric.Everywhere in the literature of eco-nomics one is met with premises that areunargued, tricks of style masqueradingasreason ( it is evident that ),forms of evi-dence that ignore the concerns of the au-

6Charles Peirce, the founder of pragmatism,re-lated in 1877 how he had been entreated not toread a certain newspaper lest it might change myopinion upon free-trade (p. 101).


15/38

494 Journal of Economic Literature, Vol. XXI (June 1983)dience, and other symptoms of a lack ofself-consciousnessin rhetoric. The lack ismost evident in quarrels across researchparadigms. Some economists, (I am one)believe that peasants are rational. Themass of modernist proofs, originatinglargely from Chicago, that resistance tothe Green Revolution or persistence inscattering plots of land are rational leavemany other economists cold. Some econo-mists (I am one) believe that competitionis a robustcharacterization of the modernAmerican economy. The mass of mod-ernistproofs,originating largely from Chi-cago, that, for instance, advertising hassmall effects on profits leaves the otherscold. Why? Why do Chicago proofs leaveTexasinstitutionalistsor NYU AustriansorMassachusettsMarxists or even Berkeleyneoclassicists cold? The non-Chicagoeconomists, of course, believe they havemodernist evidence of their own. But partof the problem is that they also believe,without thinking about it much, that theyhave evidence of a non-modernist sort:stories of peasantsand their lumpish char-acter in the flesh; self-awareness of theforce of advertising. A good part of thedisagreement is over evidence that is notbrought openly into the discussion,though it is used.Even in the most narrowly technicalmatters of scientific discussion economistshave a shared set of convictions aboutwhat makes an argument strong, but a setwhich they have not examined, whichthey can communicate to graduate stu-dents only tacitly, and which containsmany elements embarrassingto the offi-cial rhetoric.A good example is the typicalprocedure in econometrics. From eco-nomic theory, politics, and the workingsof the economist'spsyche, all of which arein the rhetorical sense unexamined, comehypotheses about some bit of the econ-omy. The hypotheses are then specifiedas straight lines, linear models being thosemost easily manipulated. The straight

lines are fitted to someone else's collectionof facts. So far the official and workadayrhetoric correspond, and the one mightwithjustice be called a guide to the other.Presently, however, they diverge. If theresults of the fitting to the dataare reason-able, on grounds that are not themselvessubject to examination, the article is sentoff to a journal. If the results are unreason-able, the hypothesis is consigned to a doloop: the economic scientist returns to thehypotheses or the specifications, alteringthem until a publishable article emerges.The product may or may not have value,but it does not acquire its value from itsadherence to the official rhetoric. It vio-lates the official rhetoric blatantly.But why shouldn't it? Even at the levelof tests of statistical significance theworkaday rhetoric violates the philoso-pher's law. But so what? It is a cliche ofcynicismin economics and related statisti-cal fields to point out that a result signifi-cantly different from one that may havebeen gotten by chance does not have thesignificance it claims if the hypothesis hasbeen manipulated to fit the data. Thatonly significant results get published haslong been a scandal among statistical pur-ists: they fear with some reason that atthe five percent level of significancesome-thing like five percent of the computerruns will be successful.The scandal is not,however, the failure to achieve moderniststandardsof scientific purity. The scandalis the failure to articulatereasonswhy onemight want to ignore them.It would be arrogant to suppose thatone knew better than thousandsof intelli-gent and honest economic scholars whatthe proper form of argument was. TheReceived View is arrogant n thisway, lay-ing down legislation for science on the ba-sis of epistemological convictions heldwith vehemence inversely proportionaltothe amount of evidence that they work.Better to look hardat what is in fact done.In an important book that is an exception


16/38

McCloskey:TheRhetoric of Economics 495to the general neglect of rhetorical consid-erations in economics Edward Leamerasks what purpose the workaday proce-dures in econometrics may be serving(Leamer, 1978, esp. p. 17). Instead of com-paring them with a doctrine in the philos-ophy of science he compares them withreasons that ought to persuade a reasona-ble person, with what really warrants as-sent, with, in short,economic rhetoric. AsChristopher Sims points out in a review,there is a myth that there are only twocategories of knowledge about theworld-'the' model, given to us by 'eco-nomic theory,' without uncertainty, andthe parameters, about which we knownothing except what the data, via objec-tively specified econometric methods,tells us. . . . The sooner Leamer's cogentwritingscan lead us to abandon this myth,to recognize that nearly all applied workis shot through with applications of uncer-tain, subjective knowledge, and to makethe role of such knowledge more explicitand more effective, the better (Sims,1979, p. 567). Yes. The very title of Lea-mer's book is an outline of rhetoric ineconometrics: Specification Searches: AdHoc Inference with NonexperimentalData.Examples of the search abound. It iscommon in a seminar in economics forthe speaker to present a statistical result,apparently irrefutableby the rules of posi-tive economics, and to be met by a chorusof I can't believe it or It doesn't makesense. Milton Friedman's own MoneyWorkshopat Chicagoin the late 1960s andthe early 1970s was a case in point. Putin statisticallanguage, the rhetorical con-text that creates such skepticism can becalled a prioribeliefs and can be analyzedin Bayesian terms. It seldom is, but sucha step would not be enough even if it weretaken. That the rhetorical community ineconomics might reject solid results, forinstance that oil prices appear in a regres-sion explaining inflation, and accept

flimsy ones, for instance that moneycauses inflation, shows the strength ofprior beliefs. (The beliefs can be reversedwithout changing the example.) To leavethe discussion at prior beliefs, however,perhaps formalizing them as prior proba-bility distributions, is to perpetuate thefact-value split of modernism, leavingmost of what matters in science to squealsof pleasure or pain. What is required isan examination of the workadayrhetoricthat leads to the prior beliefs. It is notenough, as ThomasF. Cooley andStephenF. LeRoy do in their recent, penetratingpaper on Identification and Estimationof Money Demand (1981) to merelystand appalled at the infection of econo-metric conclusions by prior beliefs. Ifeconometric argument does not persuadeit is because the field of argument is toonarrow, not because the impulse towardsthoughtfulness and explicitness which itembodies is wrong. The arguments needto be broadened, not merely dismissed.The Controversy Over Purchasing PowerParity Is an Example of UnexaminedRhetoric

A good example of how the officialrhet-oric-in the absence of an examination ofthe workadayrhetoric-can lead a litera-ture in economics astray, especially ineconometric matters, is the debate aboutpurchasing power parity.It is worth exam-ining in detail as a case study in unex-amined rhetoric and the need to broadenit (Donald McCloskey and J. RichardZecher, 1982).The questionis: is the inter-national economy more like the economyof the Midwest, in which Iowa City andMadison and Champaign all face givenprices for goods; or is it more like the solarsystem, in which each planet's economyis properly thought of in isolation?If theIowa City view is correct, then the pricesof all goods will move together every-where, allowing for exchange rates. If theMartian view is correct they will move


17/38

496 Journal of Economic Literature, Vol. XXI (June 1983)differently. If the Iowa City view is cor-rect, then all closed models of economies,whether Keynesian or monetarist or ra-tionally expecting, are wrong; if the Mar-tian view is correct, then economists can(as they do) go on testing macroeconomicfaiths against American experience sincethe War.The question of whether prices areclosely connected internationally, then, isimportant. The official rhetoric does notleave much doubt as to what is requiredto answer it: collect facts on prices in, say,the United States and Canada and . . .well . . . test the hypothesis (derived inorthodox fashion from a higher order hy-pothesis, using objective data, lookingonly at observable facts, controlling theexperiment as much as possible, and soforth, according to the received view). Alarge number of economists have donethis. Half of them conclude that purchas-ing power parity works; he other half con-clude that it fails. A misleading but none-theless superb paper by Irving Kravis andRobert Lipsey on the subject concludesthat it fails,in terms that are worth repeat-ing:

We think it unlikely that the high degree ofnational and international commodity arbi-trage thatmanyversions of the monetarist[sic]theory of the balance of paymentscontemplateis typical of the real world. This is not to denythat the price structuresof the advancedindus-trial countries are linked together, but it is tosuggest that the links are loose rather thanrigid [1978, p. 243, italics added].Every italicized word involves a com-parison against some standard of whatconstitutes unlikelihood or highness ortypicality or being linked or looseness orrigidity. Yet here and elsewhere in thetortured literature of purchasing powerparity no standardis proposed.The narrowesttest of purchasingpowerparity,and the one that springsmost read-

ily to a mind trained in the official rheto-ric, is to regress the price in the United

States (of steel or of goods-in-general, inlevels or in differences) against the corre-sponding price abroad, allowing for theexchange rate. If the slope coefficient is1.00 the hypothesis of purchasing powerparity is said to be confirmed; if not, not.Kravisand Lipsey performsucha test. Be-ing good economists they are evidentlymade a little uncomfortableby the rheto-ric involved. They admit that Each ana-lyst will have to decide in the light of hispurposes whether the purchasing powerparity relationships fall close enough to1.00 to satisfy the theories (p. 214). Pre-cisely. In the next sentence, however, theylose sight of the need for an explicit stan-dard if their argument is to be cogent:As a matter of general judgment we ex-press our opinion that the results do notsupport the notion of a tightly integratedinternational price structure. They donot say what a general judgment is orhow one might recognize it. The purposeof an explicit economic rhetoric would beto provide guidance. The guidance Kravisand Lipsey provide for evaluating theirgeneral judgment is a footnote (p. 214)reporting the general judgments of Hou-thakker,Haberler, andJohnsonthat devi-ations from parity of anything under 10to 20 percent are acceptable to the hy-pothesis. It happens, incidentally, that thebulkof the evidence offeredby KravisandLipsey passesrather than fails such a test,belying their conclusions. But acceptingor rejecting one unargued standard bycomparingit with another unarguedstan-dard does not much advance the art ofargument in economics.Kravisand Lipsey, to be fair,are unusu-ally sensitive to the case for having somestandard, more sensitive than are mosteconomists workingthe field.They returnrepeatedly to the question of a standard,though without resolving it. On page 204they reject in one irrelevant sentence theonly standard proposed in the literatureso far, the Genberg-Zecher criterion, de-


18/38

McCloskey: The Rhetoric of Economics 497scribed below. On pages 204-05 and on235 and again on 242 they draw a distinc-tion between the statistical and the eco-nomic significance of their results. So fre-quently do they make the point that itmust be counted one of the major onesin the paper. On page 205 they remark,for example, that even small differencesbetween domestic and export prices canmake a big difference to the incentive toexport: this is a case in which statisticalsignificance [that is, a correlation of thetwo prices near 1.0, which one might mis-takenly suppose to imply that they wereinsignificantly different] does not neces-sarilyconnote economic significance. Yetthey do not turn the sword on themselves.No wonder: without a rhetoric of eco-nomic significance, and in the face of amodernist rhetoric of statistical signifi-cance with the prestige of alleged sciencebehind it, they are unaware they arewielding it.The abuse of the word significant nconnection with statistical arguments ineconomics is universal. Statistical signifi-cance seems to give a standardby whichto judge whether a hypothesis is true orfalse that is independent of any tiresomeconsideration of how true a hypothesismust be to be true enough. The point inthe present case is that the failure ofpurchasing power parity in a regressionof the usual type is not measured againsta standard. How close does the slope haveto be to the ideal of 1.00 to say that pur-chasingpower parity succeeds? The litera-ture is silent. The standard used is the ir-relevant one of statistical significance. Asample size of a million yielding a tightestimate that the slope was .9999, signifi-cantly different from 1.00000, could beproduced as evidence that purchasingpower parity had failed, at least if thelogic of the usual method were to be fol-lowed consistently. Common sense, pre-sumably, would rescue the scholar fromasserting that an estimate of .9999 with

a standard error of .0000001 was signifi-cantly different from unity in a significantmeaning of significance. Such commonsense should be applied to findings ofslopes of .90 or 1.20. It is not.7The irrelevance of the merely statisticalstandard of fit does not undermine onlythat half of the empirical literature thatfinds purchasing power parity to bewrong. Towards the end of a fine articlefavorable to purchasing power parity,Paul Krugman writes:

There are several ways in which we might tryto evaluate purchasing power parity as a the-ory. We can ask how much it explains [thatis, R-square];we can ask how large the devia-tionsfrompurchasingpower parityare in someabsolute sense; and we can ask whether thedeviations from purchasing power parity arein some sense systematic [1978, p. 405].The defensive usage in some absolutesense and in some sense betrays hisunease, which is in the event justified.There is no absolute sense in which adescriptionis good or bad. The sense mustbe comparative to a standard, and thestandard must be argued.Similarly, Jacob Frenkel, an enthusiastforpurchasing power parityas suchthingsgo among economists but momentarilybewitched by the ceremony of regression,says that if the market is efficient and ifthe forward exchange rate is an unbiasedforecast of the future spot exchange rate,the constant [in a regression of the spotrate today on the future rate for todayquoted yesterday] . . . should not differsignificantly from unity. In a footnoteon the next page, speakingof the standard

7An example is J. D. Richardson'spaper, SomeEmpirical Evidence on Commodity Arbitrage andthe Lawof One Price (1978).He regressesCanadianon American prices multiplied by the exchange ratefor a numberof industries and concludes: Itis nota-ble that the 'law of one price' fails uniformly.Thehypothesis of perfect commodity arbitrage is re-jected with 95 percent confidence for every com-modity group (p. 347, italics added).The questionis, why in an imperfect world would it matter thatperfect arbitrageis rejected?


19/38

498 Journal of Economic Literature, Vol. XXI (June 1983)errors of the estimates for such an equa-tiori in the 1920s, he argues that whilethese results indicate that markets wereefficient and that on average forward rateswere unbiased forecasts of future spotrates, the 2-8 percent errors were signifi-cant (1978, pp. 175-76, italics added). Heevidently has forgotten his usage of sig-nificant in another signification. What heappears to mean is that he judges a 2-8percent error to be large in some unspeci-fied economic sense, perhaps as offeringsignificant profits for lucky guessers of thecorrect spot rate. In any event, it is un-clear what his results imply about theirsubject, purchasing power parity, becausesignificance in statistics, however usefulit is as an input into economic significance,is not the same thing as economic signifi-cance.The point is not that levels of signifi-cance are arbitrary. Of course they are.The point is that it is not known whetherthe range picked out by the level of signifi-cance affirms or denies the hypothesis.Nor is the point that econometric tests areto be disdained. Quite the contrary. Thepoint is that the econometric tests havenot followed their own rhetoric of hypoth-esis testing. Nowhere in the literature oftests of purchasing power parity doesthere appear a loss function. We do notknow how much it will cost in policywrecked or analysis misapplied or reputa-tion ruined if purchasing power parity issaid to be true when by the measure ofthe slope coefficient it is only, say, 85 per-cent true. That is, the argument due toNeyman and Pearson that undergirdsmodern econometrics has been set asidehere as elsewhere in favor of a merelystatistical standard, and an irrelevant onerelated to sampling error at that. We aretold how improbable it is that a slope coef-ficient of .90 came from a distribution cen-tered on 1.00 in view of the one kind oferror we claim we know about (unbiasedsampling error with finite variance), but

we are not told whether it matters to thetruth of purchasing power parity wheresuch limits of confidence are placed.Silence on the matter is not confinedto the literature of purchasing power par-ity. Most texts on econometrics do notmention that the goodness or badness ofahypothesis is not ascertainableon merelystatistical grounds. Statisticians them-selves are more self-conscious, althoughthe transition from principle to practiceis sometimes awkward. A practical diffi-culty in the way of using the Neyman andPearson theory in pure form, A. F. Mood

and F. A. Graybill say, is thatthe loss function is not known at all or else itis not known accurately enough to warrant tsuse. If the loss function is not known, it seemsthata decision functionthatin some sense min-imizes the error probabilitieswill be a reason-able procedure [1963, p. 278].

The phrase in some sense appears tobe a marker of unexplored rhetoric in theworks of intellectually honest scholars. Inany event, the procedure they suggestmight be reasonable for a general statisti-cian, who makes no claim to know whatis a good or bad approximation to truthin fields outside statistics. It is not reason-able for a specialist in international tradeor macroeconomics. If the loss functionis not known it should be discovered. Andthat will entail a study of the question'srhetoric.One standard of economic significancein questions of parity, for example, mightbe the degree to which the customaryre-gressions between countries resembledsimilarregressionswithin a single country.We agree, for purposes of argument, thatthe United States is to be treated as a sin-gle point in space, as one economy acrosswhich distances are said not to matter forthe purposes of thinking about inflationor the balance of payments. Having doneso we have a standard: s Canadaeconomi-cally speaking just as closely integratedwith the United Statesas is Californiawith


20/38

McCloskey: The Rhetoric of Economics 499Massachusetts?Is the Atlantic Economyas closely integrated as the AmericanEconomy? The standard s called the Gen-berg-Zecher criterion, after its inventors(Hans Genberg, 1976; McCloskey andZecher, 1976). It is not the only conceiv-able one. The degree of market integra-tion in some golden age (1880-1913 per-haps;or 1950-1970) might be a standard;the profits from arbitrage above normalprofits might be another standard;the de-gree to which an X percentage deviationfrom purchasing power parity does ordoes not disturb some assertionabout thecauses of inflation might be still another.The point is to have standards of argu-ment, to go beyond the inconclusive rhet-oric provided by the pseudo-scientificceremony of hypothesis-regression-test-publish in most of modern economics.

V. The Rhetoric of Economics Is aLiterary MatterEven a Modernist Uses,and Must Use,Literary DevicesThe mere recognition that the officialrhetoric might be dubious, then, frees thereason to examine how economists reallyargue. Obscured by the official rhetoricthe workaday rhetoric has not receivedthe attention it deserves, and the knowl-edge of it is therefore contained only inseminar traditions,advice to assistantpro-fessors, referee reports (a promising pri-mary source for its study), and jokes. Itis significantthat George Stigler, a leaderof modernism in economics, chose to ex-press his observations about the rhetoricof economics in a brilliantly funny TheConference Handbook ( IntroductoryRemarknumber E: 'I can be very sympa-thetic with the author; until 2 years agoI was thinkingalongsimilar ines' )ratherthan asa seriousstudyin one of the severalfields he has mastered, the history of eco-nomic thought (1977).The attitude of theHandbook is that rhetoric is mere rheto-

ric, mere game playing in aid of ego grati-fication; the serious business of Sciencewill come on that happy day when theinformation theory of oligopoly or the vul-gar Marxisttheory of the state is broughtto a critical test under the auspices ofnaive falsificationism.Economists can do better if they willlook soberly at the varieties of their argu-ments. The varieties examined here canonly be crude preliminaries to a fullerstudy, a study that might dissect samplesof economic argument, noting in the man-ner of a literary or philosophical exegesisexactly how the arguments sought to con-vince the reader. It is not obvious a prioriwhat the categories might be; in view ofthe methodological range of modern eco-nomics they doubtless would vary muchfrom author to author. A good place tostart might be the categories of classicalrhetoric, Aristotle's divisions into inven-tion, arrangement, delivery, and style, forinstance, with his paired sub-headings ofartificial(i.e., argumentative) and inartifi-cial (i.e., factual) proofs, syllogismand ex-ample, and the like. A good place to con-tinue would be the procedures of modernliterary critics, bright people who maketheir living thinking about the rhetoric oftexts.The purpose would not be to make theauthor look foolish or to uncover fallaciesfor punishment by ridicule-fallacymon-gering is evidence of a legislative attitudetowards method, and it is no surprise thatJeremy Bentham, confident of his abilityto legislate for others in mattersof methodas in education, prisons, and government,had compiled from his notes The Book ofFallacies (1824). David Hackett Fischer'sbook, Historians'Fallacies (1970), has thisflaw: that it takes as fallacious what maybe merely probable and supporting argu-ment.The purpose of literary scrutiny of eco-nomic argument would be to see beyondthe received view on its content. Two


21/38

500 Journal of Economic Literature, Vol. XXI (June 1983)pages (pp. 122-23) chosen literally atrandom from that premier text of the re-ceived view, and a local maximum in eco-nomic scholarship, Samuelson's Founda-tions, will sufficefor illustration:(1) To begin with he gives a generalmathematical form from which detailedresults in comparative statics can be ob-tained by reading across a line. The impli-cation of the lack of elaboration of themathematics is that the details are trivial(leadingone to wonder why they are men-tioned at all).An interesting special caseis left as an exercise to the interestedreader, drawing on the rhetorical tradi-tions of applied mathematics to direct thereader'smind in the right directions. Themathematics is presented in an offhandway, with an assumption that we all canread off partitioned matrices at a glance,inconsistentwith the level of mathematicsin other passages.The air of easy mathe-matical mastery was important to the in-fluence of the book, by contrast with theembarrassedmodesty with which Britishwriters at the time (Hicks most notably)pushed mathematics off into appendices.Samuelson's skill at mathematics in theeyes of his readers, an impression nur-tured at every turn, is itself an importantand persuasive argument. He presentshimself as an authority, with good reason.That the mathematicsis so often pointless,as here, is beside the point. Being ableto do such a difficult thing (so it wouldhave seemed to the typical economistreading in 1947) is warrant of expertise.The argument is similar in force to thatof a classical education conspicuouslydis-played. To read Latin like one's mothertongue and Greek like one's aunt'stongueis extremely difficult, requiring applica-tion well beyond the ordinary; herefore-or so it seemed to Englishmen around1900-men who have acquiredsuch a skillshould have charge of a great empire.Likewise-or so it seemed to economists

around 1983-those who have acquireda skill at partitioned matrices and eigen-values shouldhave charge of a great econ-omy. The argument is not absurd or a fal-lacy or mere rhetoric. Virtuosity issome evidence of virtue.8(2) There are six instances of appeal toauthority (C. E. V. Leser, Keynes, Hicks,Aristotle, Knight, and Samuelson; appealto authorities is something of a Samuel-sonian specialty). Appeal to authority isoften reckoned as the worst kind ofmere rhetoric. Yet it is a common andoften legitimate argument, as here. No sci-ence would advance without it, becauseno scientist can redo every previous argu-ment. We stand on the shoulders of giants,and it is a perfectly legitimate and persua-sive argument to point this out from timeto time.(3) There are several appeals to relax-ation of assumptions. The demand formoney is really interesting . . . when un-certainty . . . is admitted. Again, the im-plicit assumption in Hicks that moneybearsno interest is relaxed, unhitching theinterest rate from the zero return onmoney. Relaxation of assumptions is theliterature generating function of moderneconomics. In the absence of quantitativeevidence on the importance of the as-sumption relaxed it is no modernist evi-dence at all. Samuelson is careful to stickto the subjunctivemood of theory (moneywould pass out of use ), but no doubtwants his strictures on a theory of the in-terest rate based merely on liquidity pref-erence (that is, on risk) to be taken seri-ously as comments on the actual world.They are, surely,but not on the operation-alist grounds he articulateswhen preach-ing methodology.(4) There are several appeals to hypo-

8The limiting case is spelling. Most college teach-ers will agree that those who do not know how tospell consensus lose some of their authority tospeak on it.


22/38

McCloskey:The Rhetoric of Economics 501thetical toy economies, constrained to oneor two sectors, from which practical re-sults are derived. This has since Ricardobeen among the commonest forms of eco-nomic argument, the Ricardian vice. It isno vice if done reasonably. It would bequite possible to have an economy inwhich money did not exist, and in whichthere was still a substantial rate of inter-est. Yes, of course.(5) There is, finally,one explicit appealto analogy, which is said to be not ...superficial. Analogy, as will be shown indetail in a moment, pervades economicthinking, even when it is not openlyanalogical: transaction friction, yieldspread, securities circulating, moneywithering away are inexplicit exampleshere from one paragraph of live or onlyhalf-dead metaphors. Yet analogy andmetaphor, like most of the other piecesof Samuelson'srhetoric, have no standingin the officialcanon.Most of the Devices Are Only DimlyRecognizedThe range of persuasive discourse ineconomics is wide, ignored in preceptwhile potent in practice. At the broadestlevel it is worth noting that the practiceof economic debate often takes the formof legal reasoning, for, as Booth put it,the processes developed in the law arecodificationsof reasonable processes thatwe follow in every part of our lives,even the scientific (1974, p. 157). Econ-omists would do well to study jurispru-dence, then, with some other aim thansubordinatingit to economic theory. Forinstance, economists, like jurists,argue byexample, by what Edward Levi calls thecontrolling similarity between the pres-ent and prior case (1967, p. 7).The details of the pleading of cases ateconomic law have little to do with theofficial scientific method. Without self-consciousness about workaday rhetoric

they are easily misclassified. A commonargument in economics, for example, isone fromverbal suggestiveness. The prop-ositionthat the economy is basicallycom-petitive may well be simply an invitationto look at it this way, on the assurancethat to do so will be illuminating. In thesame way a psychologist might say weare all neurotic -it does not mean that95 percent of a randomly selected sampleof us will exhibit compulsive handwash-ing; rather,it is merely a recommendationthat we focus attention on the neuroticingredient in us all (Passmore,1966, p.438). To misunderstand the expression asa properly modernist hypothesis would beto invite much useless testing. The caseis similar to MV=PT understood as anidentity. The equation is the same term-for-term as the equation of state of an idealgas, and has the same statusas an irrefuta-ble but useful notion in chemistry as ithas in economics. The identity can be ar-gued against,but not on grounds of fail-ing a test. The arguments against it willdeny its capacity to illuminate, not itsmodernist truth.Another common argument in econom-ics with no status in the official rhetoricis philosophical consistency: If you as-sume the firm knows its own cost curveyou might as well assumeit knows its pro-duction function, too: it is no more dubi-ous that it knows one than the other. Theargument, usually inexplicit though sig-nalled by such a phrase as it is naturalto assume, is in fact characteristic ofphilosophical discourse (Passmore, 1970).It is analogous to symmetry as a criterionof plausibility, which appears in manyforms and forums.A labor economist tellsa seminar about compensating differen-tials for the risk of unemployment, refer-ring only to the utility functions of theworkers. An auditor remarks that thevalue of unemployment on the demandside (that is, to the firm) is not included.


23/38

502 Journal of Economic Literature, Vol. XXI (June 1983)The remark is felt to be powerful, and along discussion ensues of how the demandside might alter the conclusions. The argu-ment from the-other-side-is-empty is per-suasive in economics, but economists areunaware of how persuasive it is.Likewise (and here we reach the borderof self-consciousness n rhetoric), ad hoc-cery is universally condemned by semi-nar audiences. An economist will cheer-fully accept a poor R2 and terrible andunderstated standard errors if only shehas a theory for the inclusion of such-and-such a variable in the regressions.

Havinga theory is not so open and shutas it might seem, depending for instanceon what reasoning is prestigious at the mo-ment. Anyone who threw accumulatedpast output into an equation explainingproductivity change before 1962 wouldhave been accused of ad hoccery. Butafter Arrow's essay on The Economicsof Learning By Doing (which as it hap-pened had little connection with max-imizing behavior or other higher orderhypotheses in economics), there was sud-denly a warrant for doing it.An example of the rhetoric of econom-ics which falls well within the border ofself-consciousness is simulation. Econo-mists will commonly make an argumentfor the importance of this or that variableby showing its potency in a model withback-of-the-envelope estimates of the pa-rameters. Common though it is, few booksor articles are devoted to its explication(but, see Richard Zeckhauser and EdithStokey, 1978). It would be as though stu-dents learned econometrics entirely bystudying examples of it-no bad way tolearn, but not self-consciousin groundingthe arguments. What is legitimate simula-tion? Between A. C. Harberger's modestlittle triangles of distortion and JeffreyWilliamson's immense multiequationmodels of the American orJapaneseecon-omies since 1870 is a broad range. Econo-mists have no vocabulary for criticizing

any part of the range. They can deliversummary grunts of belief or disbelief butfind it difficult to articulate their reasonsin a disciplined way.VI. Economics Is Heavily Metaphorical

Models Are MetaphorsThe most important example of eco-nomic rhetoric, however, falls well outsidethe border of self-consciousness.It is thelanguage economistsuse, and in particularits metaphors. To say that marketscan be

represented by supply and demandcurves s no less a metaphor than to saythat the west wind is the breath of au-tumn's being. A more obvious exampleis game theory, the very name beinga metaphor. It is obviously useful to havein one'shead the notion that the arms raceis a two-person, negative-sum cooperativegame. Its persuasiveness is instantly ob-vious, as are some of its limitations. Eachstep in economic reasoning,even the rea-soning of the officialrhetoric, is metaphor.The world is said to be like a complexmodel, and its measurements are said tobe like the easily measured proxyvariableto hand. The complex model is said to belike a simpler model for actual thinking,which is in turn like an even simplermodel for calculation.For purposesof per-suading doubters the model is said to belike a toy model that can be manipulatedquickly inside the doubter's head whilelistening to the seminar. John Gardnerwrote:

There is a game-in the 1950s it used to beplayed by the members of the Iowa Writers'Workshop-called 'Smoke.' The player who is'it' [thinks of] some famous person .. . andthen each of the other players in turn asks onequestion. .. such as 'What kind of weatherare you?'. . . Marlon Brando, if weather, wouldbe sultry and uncertain.... To understandthat Marlon Brando is a certain kind of weatheris to discover something (though somethingneither useful nor demonstrable) and in


24/38

McCloskey:The Rhetoric of Economics 503the same instant to communicate something[Gardner, 1978, pp. 118-19].

On the contrary, in economics the com-parable discovery is useful and by re-course to rhetorical standards demonstr-able.Metaphorsin Economics Are NotOrnamental

Metaphor, though, is commonly viewedas mere ornament. From Aristotle untilthe 1930s even literary critics viewed itthis way, as an amusing comparison ableto affect the emotions but inessential forthought. Men are beasts : if we caredto be flat-footed about it, the notion was,we could say in what literal way wethought them beastly, removing the orna-ment to reveal the core of plain meaningunderneath. The notion was in 1958 com-mon in philosophy, too:

With the decline of metaphysics, philosophershave grown less and less concerned about God-liness and more and more obsessed with clean-liness, aspiring o ever higherlevels of linguistichygiene. In consequence, there hasbeen a ten-dency for metaphorsto fall into disfavour,thecommonopinion being thatthey are a frequentsource of infection [H. J. N. Horsburgh, 1958,p. 231].Such suspiciontowardmetaphor is widelyrecognized by now to be unnecessary,even harmful. That the very idea of re-moving an ornament to reveal aplain meaning is itself a metaphor sug-gests why. Perhaps thinking is metaphori-cal. Perhapsto remove metaphor is to re-move thought. The operation on themetaphoric growth would in this case beworse than the disease.The question is whether economicthought is metaphoricalin some nonorna-mental sense. The more obvious meta-phors in economics are those used to con-vey novel thoughts, one sort of noveltybeing to compare economic with noneco-nomic matters. Elasticity was once a

mind-stretching fancy; depression wasdepressing; equilibrium compared aneconomy to an apple in a bowl, a set-tling idea; competition once inducedthoughts of horseraces; money's veloc-ity thoughts of swirling bits of paper.Much of the vocabularyof economics con-sists of dead metaphors taken from non-economic spheres.Comparing noneconomic with eco-nomic matters is another sort of novelty,apparent in the imperialism of the neweconomics of law, history, politics, crime,and the rest, and most apparent in thework of that Kipling of the economic em-pire, Gary Becker. Among the least bi-zarre of his many metaphors, for instance,is that children are durable goods. Thephilosopher Max Black pointed out thata memorable metaphor has the powerto bring two separate domains into cogni-tive and emotional relation by using lan-guage directly appropriate to the one asa lens for seeing the other (1962, p. 236).So here: the subject (a child) is viewedthroughthe lens of the modifier (adurablegood). A beginning at literal translationwould say, A child is costly to acquireinitially, lasts for a long time, gives flowsof pleasure during that time, is expensiveto maintain and repair, has an imperfectsecond-hand market.... Likewise, a du-rable good, such as a refrigerator..That the list of similarities could be ex-tended further and further, graduallyre-vealing the differences aswell- children,like durable goods,are not objectsof affec-tion and concern ; children,like durablegoods, do not have their own opinions -is one reason that, as Black says, meta-phorical thought is a distinctive mode ofachieving insight, not to be construed asan ornamental substitute for plainthought (p. 237). The literal translationof an important metaphor is never fin-ished. In this respect and in others an im-portant metaphor in economics has thequality admired in a successful scientific


25/38

504 Journal of Economic Literature, Vol. XXI (June 1983)theory, a capacity to astonish us with im-plications yet unseen.9But it is not merely the pregnant qualityof economic metaphors that makes themimportant for economic thinking. The lit-erary critic I. A. Richard

Documents

McCloskey 1983 the Rhetoric of Economics