sign in sign up
<<
Home , Analytical Marxism, Hillel Steiner, John Roemer, Rate of profit

S&S Quarterly, Inc. Guilford Press

Is "Analytical " Marxism? Author(s): Michael A. Lebowitz Reviewed work(s): Source: Science & Society, Vol. 52, No. 2 (Summer, 1988), pp. 191-214 Published by: Guilford Press Stable URL: http://www.jstor.org/stable/40402871 . Accessed: 20/02/2012 19:34

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for- 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].

S&S Quarterly, Inc. and Guilford Press are collaborating with JSTOR to digitize, preserve and extend access to Science & Society.

http://www.jstor.org Science & Society,Vol. 52, No. 2, Summer 1988,pp. 191-214

IS "" MARXISM?

MICHAEL A. LEBOWITZ

A. COHN, , - with- out question,these are prolificwriters with an impres- sive seriesof articlesand books, who have become a significantpresence in commentariesand discussionsof Marxism in recentyears. My firstinkling, though, that more had emerged on thescene came froma 1983article by John Gray (passed on by a skepticalfriend); for, the articlehailed the emergenceof "a powerfulnew schoolof AnalyticalMarxism, by such outstanding figuresas G. A. Cohen,Jon Elster and JohnRoemer, with whose worksthe futureof Marxism,if it has any, musthenceforth be associated"(Gray, 1983, 1461). Is thereindeed such a school?The evidenceof theexistence of some such self-definedgroup is overwhelming.In his Making Sense Marx,Elster indicates that Cohen's 's of ' ' Theory of Historycame as a 'revelation":'Overnight it changedthe standardsof rigourand claritythat were required to writeon Marx and Marxism."Accordingly, he notes,a small group of like-mindedcolleagues formed and begana seriesof annual meet- ings in 1979.Their discussionswere decisive for the shaping of Elster'sbook - and, in particular,the contributions of Roemer (subsequentlystated in his "path-breaking"A GeneralTheory of Exploitationand Class) were "crucial" (Elster,1985, xiv-xv).1 In turn,Roemer begins the latterbook (Roemer,1982) by notinghis particularindebtedness to Cohen and Elster,indicat- ing amongthose who werehelpful several others who also appear

1 Elsterespecially thanks Cohen and Roemer fortheir comments. He does not identifyother group membersbut, included among those thankedfor pre- publication help are ,, Leif Johansen,Serge Kolm, ,Ian Steedman,Robert van der Veen, Phillippe van Parijs and Erik Wright. 191 192 SCIENCE & SOCIETY

on Elster's list.2 Mentioned on both lists, Erik Olin Wrightcor- roboratesthe existenceof the group, its annual meetingsand its ' orientation toward 'Analytical Marxism" in the preface to his recent book, Classes; as well, he testifiesthat its "new ideas and perspectiveshave had a considerableimpact on my thinkingand my work" (Wright,1985, 2).3 Finally, definitivelyembracing the self-designationof "Analytical Marxism" is Roemer's new collec- tion by thatname - a collection which includes threeessays each by Roemer, Elster and Cohen plus individual effortsby several others (Roemer, 1986).4 So, what do the adherentsthemselves see as the constituent elementsin AnalyticalMarxism? For Wright,the central intellec- tual thread is the "systematicinterrogation and clarificationof basic [Marxian] concepts and their reconstructioninto a more coherent theoreticalstructure" (Wright, 1985, 2). Similarly, as noted, Elsteridentified "rigor and clarity"as the underlyingprin- ciple in the formation of the group. The most explicit self- descriptionof Analytical Marxism,however, comes fromRoemer in the Introductionto his collection: "Analyticallysophisticated Marxism" is pursued with "contemporarytools of , mathe- matics,and model building" and committedto "the necessityfor abstraction," to the "search for foundations" of Marxian judg- ments,and to "a non-dogmaticapproach to Marxism" (Roemer, 1986, 1-2). An impressiveset of elements,to be sure. Wheredo we apply for candidate status in this analyticallycorrect fellowship? More than rigor,however, sets Analytical Marxism apart - as John Gray's praise forthis "powerfulnew school" makes clear. For, hailing the early Austrian criticismsof Marx by Böhm- Bawerk,von Mises and Hayek (and thatof right-wingU.S. econo- mist Paul Craig Roberts)and genuflectingbefore "the prodigious virtuosityof " and the marvelsof the ,Gray was far from a sympatheticcommentator on Marxism ("the first world view in human historythat is genuinely self-defeating"); his praise for Analytical Marxism occurs in the context of a lengthy anti-Marxistpolemic ("The System of Ruins"). 2 These include Lief Johansen, Serge Kolm and Erik Wright. 3 Wrightidentifies among the membersof the group: Cohen, Roemer,Elster, van Parijs, van der Veen, Brenner,Przeworski and . 4 Included in this collection are essays by Bardhan, Brenner,Przeworski, Wright and Allen Wood. ANALYTICAL MARXISM 193

The practitionersof Analytical Marxism can not, of course, bear the responsibilityfor what others (like Gray) write about them.They bear responsibilityonly fortheir own work.But con- siderthat work. Included by Elsteras "dead" in Marx (in his most recentbook, An Introduction to Karl Marx) are the following: "scientific "; "dialectical "; Marxian eco- nomic theory- in particular,its two "main pillars," the labor theoryof ("intellectuallybankrupt") and the theoryof the falling rate of profits;and, "perhaps the most importantpart of historicalmaterialism," the "theoryof productiveforces and rela- tions of production" (Elster, 1986, 188-94). Similarly,in a long marchthrough in his AnalyticalFoundations of Marxian Economic Theory (Roemer, 1981), Roemer leftintact only the Marxian theoryof exploitation; he then proceeded in Roemer (1982) to findeven thisfinal survivorinadequate. Exploi- tation, Roemer now informsus, is simply inequality. But what, then, is the differencebetween the Analytical Marxist position and that of non-Marxistphilosophers such as Rawls? Roemer answers that "it is not at all clear"; "the lines drawn between contemporaryanalytical Marxism and contemporaryleft-liberal political are fuzzy" (Roemer, 1986, 199-200). One must wonder what really is leftof Marxism in Analyti- cal Marxism. In what follows,we examine some of thiswork (especially that of Elster and Roemer) in order to explore the extentto which it can be considered"Marxist." The conclusion is thatAnalytical Marxism is not Marxism - and that,indeed, it is in essence an¿/-Marxist.

"Neoclassical" or "Rational Choice" Marxism?

There are severalalternative labels which have been attached to Analytical Marxism and its practitioners;they include Neo- , Game-Theoretic Marxism and Rational- Choice Marxism. Consideration of these labels themselves providesa good point of entryinto an examination of Analytical Marxism. "Neoclassical Marxism," as Patrick Clawson describedPhil- lipe van Parijs' article on the falling-rate-of-profitcontroversy, would appear on its face to be an oxymoron(Clawson, 109). How could such a constructexist? After all, neoclassical economic the- 194 SCIENCE & SOCIETY

ory begins from the atomistic individual conceived as ontologi- cally prior to the whole, the particular society. This is the "Cartesian" heritage, so well analyzed by Richard Levins and Richard Lewontin, which it shares with methodological approaches in other spheres:

The partsare ontologicallyprior to thewhole; that is, theparts exist in isolationand come togetherto makewholes. The partshave intrinsic properties,which theypossess in isolationand whichthey lend to the whole. (Levins and Lewontin,269.)

In neoclassical analysis, we have atomisticindividuals who, with exogenously given assets and techniques, enter into relationsof exchange with each other in order to satisfyexogenously given wants; and society is the sum-total of these arrangementsof exchange. Nothing could be furtherfrom Marx's perspective.To begin with the isolated individual forwhom the various formsof social connectednessare a "mere means toward his private purposes" was simply "twaddle" (Marx, 1973, 84). "Private ,"Marx emphasized, "is itself already a socially determined interest, which can be achieved only within the conditions laid down by societyand with the means providedby society."To be sure,it is the interestof individuals, of privatepersons; "but its content,as well as the formand means of its realization,is given by social conditions independent of all" (Marx, 1973, 156). Thus, in the dialectical (in contrastto the Cartesian)perspec- tive, parts have no prior independentexistence as parts. They "acquire propertiesby virtueof partsof a particularwhole, properties they do not have in isolation or as parts of another whole" (Levins and Lewontin, 273, 3). Marx's startingpoint, accordingly,is to develop an understandingof societyas a "con- nected whole," as an organic system;it is to trace the intrinsic connections and to reveal the "obscure structureof the bourgeois ,"the "inner core, which is essential but con- cealed" on the surfaceof society(Marx, 1968,65; Marx, 1981,311). Only then does Marx proceed to explore what is real within this structurefor the individual agents of productionand how things necessarily appear to them. Having developed, for example, "the general and necessary ANALYTICAL MARXISM 195

' tendenciesof ' on the basis of theconcept of capital (capi- tal as a whole), it was then possible to demonstratehow "the immanent laws of capitalist production" were manifested through the actions of individual capitalists in (Marx, 1977, 433). As Marx noted repeatedlyin the , "competitionexecutes the inner laws of capital; makes theminto compulsorylaws towards the individual capital, but it does not invent them. It realizes them" (Marx, 1973,414, 552, 651, 751-2). To begin analysis,on the otherhand, with thoseindividual capi- tals (and with the connectionsas theyappear in "the phenomena of competition") produces a distortion of the inner structure because "in competition everythingalways appears in inverted form,always standing on its head" (Marx, 1968, 165). From this perspective,there is absolutely no compatibility between the atomistic approach of and Marxism. "Neoclassical Marxism" is eithernot neoclassical or it is not Marxism. Can we say the same, though, about "Game- Theoretic" or "Rational-Choice" Marxism? In a recent essay, Alan Carling has proposed "Rational-Choice Marxism" as the label most characteristicof the work in question, describingits distinctivepresupposition as the "view that societies are com- posed of human individuals who, being endowed with resources of various kinds, attempt to choose rationally between various courses of action" (Carling, 26-7). But, is this just neoclassical economics by another name? Roemer's of rational choice models (in an essay entitled"Rational Choice Marxism") as "general equilibrium theory,game theoryand the arsenal of modelling techniques developed by neoclassical economics" might seem to suggest as much (Roemer, 1986a, 192). However, it is critical not to confuse particular techniques with theiroriginal emergenceor the use which has been made of those techniques; to do that would be to repeat the unfortunate of Marxian economics with calculus, rejected as "bourgeois" despite Marx's own significantexplorations of this technique (Gerdes; Struik). In short, if it is a question of the appropriation of these techniques within a Marxian framework, then Analytical Marxism may have much to offer. Consider game theoryand game-theoreticapproaches. Char- acteristicof both Elsterand Roemer is a verystrong emphasis on "game-theoretic"modeling; indeed, Roemer's general definition 196 SCIENCE & SOCIETY

of exploitation is explicitlygame-theoretic. Does this approach have a place in Marxisttheoretical work? If we insistthat Marxist analysis must begin from a consideration of the "whole," the establishmentof the inner structureof society,before examining the actions of individuals within that structure,it is not obvious that game theoryas such is inappropriate here. Game theorybegins with the specificationof the "game"; that is, it explicitlysets out the set of relationswithin which the actors perform.(For an introductionto ,see Bradley and Meek.) On its face, thereis nothing inconsistentwith Marx- ism in an approach that begins fromthe specificationof a given set of and then proceeds to explore how the particular actors will behave rationally, giving rise to dynamic properties(the laws of motion) inherentin the particu- lar structure. The key,of course, will be the specificationof the game and the actors. A game, forexample, in which the actorsare identified as the competingsellers of a common ,exploring their rational strategies,inhabits the terrainof thecompetition of capi- tals which, for Marx, executes the inner laws of capitalism but explains nothing about them. By contrast,a game in which the parties are a capitalist and "his" -laborers(and which pro- ceeds to explore the strategiesand actions of each party)would appear to correspondclosely to Marx's own approach. In thislat- ter game, the relations between capitalist and wage-laborerare identical with those of the coalition of capitalists and the coali- tion of wage-laborers;that is, the relation of the capitalist to his own workersis the "essential relation" of capital and wage-labor (Marx, 1973, 420). Similarly, a game which explores the relation between the feudal lord and his peasant tenants(or, between the coalition of lords and that of peasants (the two approaches are seen here as identical so long as neither introduces mattersappropriate to intra-coalitionrelations) would seem to permitan examinationof the essential characterof feudal relations of production. What does the lord want, what are the strategiesavailable to him, what are the potential gains (and risks) fromeach? What does the pea- sant (peasant community)want, what are the potential strategies and returns?What (in a continuing game) is the "appropriate" solution or outcome to the particulargame - and, significantly, ANALYTICAL MARXISM 197 what aspects of the behaviorof the partiesin thisparticular inter- action tend to undermine(rather than preserve)the existing solu- tion/outcomeand, indeed, the particular game itself? Stated in this way, thereappears to be nothing at all incon- sistent between game theoryas such and a Marxist approach; indeed, not only may we speculate that Marx would have been quick to explore its techniquesbut we can go furtherand suggest that Marx's analysis was inherentlya "game-theoretic"perspec- tive. See, for example, Maarek's elaboration of Marx's theoryof using game theory(Maarek, 124-340). However, the above descriptionof the feudal "game" has a certainspecificity; for it is a game which may best be designated as a "collective game." Its actors are classes (or class representa- tives,träger, the bearersof a relation). There is no place here for the autonomous, atomistic individual; nor have we introduced (yet) intra-classinteractions. It is simply assumed that the coali- tion of lords acts in the same way as the AbstractLord; thatour examination of the latter in its specific interaction with the AbstractPeasant yields the essentialinsights into the coalition or class of lords in its interactionswith the class of peasants. In short,in the collectivegame, classes act. The feudal lord and the peasant interact,but individual feudal lords and individ- ual peasants do not interactwith each other. Similarly,in the collective game forcapitalism, capital (the capitalist) and wage- labor (the worker) interact, but factors emanating from the posited existenceof competing capitalists and wage-laborersare seen as secondary to the establishment of the essential capital/wage-laborrelation. As Maarek indicates, in his discus- sion of Marx's theoryof surplus value, "it is just as if therewere a single centerof decision in each class, a 'collective'capitalist and a 'collective' worker,with the two classes confrontingeach other like two autonomous blocks" (Maarek, 132). The collective(or class) game, thus,puts to the side any con- siderationof the abilityof the particularcoalitions to engage suc- cessfullyin collective actions (i.e., the issues posed by Mancur Olson in his The Logic of CollectiveAction) in orderto explore firstin detail the characterof the relation between classes as defined by the relationsof production.All questions of whether individual agents will find it in their individual interest to engage in collective action (to achieve class goals), all matters 198 SCIENCE & SOCIETY relatingto "freerider" problems, etc., are not the principal matter of inquiry of the collective game. Epistemological priorityis assigned to the determinationof the structurewithin which indi- viduals act. Yet, intra-coalitionmatters are not outside the purview of a Marxian analysis (any more than the consideration of how a class-in-itselfbecomes a class-for-itself).The manner in which capital attemptsto divide workersand to encourage competition among them in orderto secure its own goals is an importantpart of Marx's exploration of a rational strategyfor capital in the stra- tegic game of capital and wage-labor (see Lebowitz, 1987a). And, his conclusion that when individual workersact in theirindivid- ual self-interest,the result is the worst strategyfor workersas a whole (Lebowitz, 1987b) is a critical statementabout intra- coalition issues on the side of workers.As importantas Marx's insights on these intra-coalitionmatters are, it is essential to rec- ognize that theycan occur only afterthe prior specification of the collective game. In contrastto the collective game, on the other hand, what we may designate as the "individual game" has a differentstart- ing point. Beginning fromthe position that thereare no supra- individual entities which act in the real world ("capital" does nothing, etc.), it assertsthe necessityto consider the behavior of the individual unit at a pre-coalition level in the war of all against all. Thus, no longer at the core of inquiryis thecharacter of the class relation. Substitutedis a differentproblematic, the neoclassical problematic: the outcomes which emerge from the interactionsof atomistic individuals. At its best, the overriding question in the individual game becomes one of why coalitions emerge,why (and in what sense) thereare classes-for-themselves. Thus, a game-theoreticapproach in itselfcan not be said to be inconsistentwith a Marxist analysis. Rather than rigoras the dividing line between Marxism and "Analytical Marxism," the central issue is the nature of the problematicwithin which such techniques are employed. It is preciselyin thiscontext that "Ana- lytical Marxism" should be considered. ANALYTICAL MARXISM 199

MethodologicalIndividualism and

At thecore of AnalyticalMarxism is thecategorical impera- tive:there shalt be no explanationat a levelabove thatof the individualunit. Thus, Elsteropens his MakingSense of Marx by announcingthat he will begin "by statingand justifyingthe principleof methodologicalindividualism." The doctrineis quite uncompromising:"all social phenomena- theirstructure and theirchange - are in principleexplicable in waysthat only involveindividuals - theirproperties, their goals, theirbeliefs and theiractions." To explain, Elsterproposes, it is necessary"to providea mechansim,to open up the black box and show the nuts and bolts,the cogs and wheels,the desires and beliefsthat generate the aggregateoutcomes" (Elster, 1985, 5). Accordingly,methodologi- cal individualismleaves the macro level for the micro, and rejects an explanationwhich does not proceedfrom individuals; it stands in opposition to methodologicalcollectivism, which "assumesthat there are supra-individualentities that are priorto individualsin the explanatoryorder" (6). As Elsteris well aware,however, Marx's discussions about "humanity,""capital," and especially"capital in general"as col- lectivesubjects are inconsistentwith this doctrine of methodolog- ical . Citing one of Marx's statementson competitionin the Grundrisse,Elster indeed comments: "One could notwish for a moreexplicit denial of methodological indi- vidualism"(7). He immediately,however, invokes an alternative authority- John Roemer. In this respect,it is importantto recognizethat Elster has readMarx closely, and thatnot thelack of familiaritywith rele- vantpassages (although his interpretationsare ratherquestiona- ble at times)but, rather, the rejection of theseas grievouserrors and as "near-nonsense"underlies his argument.What is to be rescuedis the Marx who makes"sense," the Marx who sounds like a methodologicalindividualist. Elster's project, simply, is to getrid of thebad Marxand preservethe good - theseparation of the "misguidedframework" from what he sees as valuable in Marx. The verysame themescan be foundin Roemer'sessay on methodin AnalyticalMarxism. Roemer asserts: "Marxian analy- 200 SCIENCE & SOCIETY sis requiresmicro-foundations" (Roemer, 1986a, 192). How, he asks, can we say that the entity,capital, does anything(e.g., dividesand conquersworkers) "when in a competitiveeconomy thereis no agent who looks afterthe needsof capital"?When Marxistsargue in such a manner,he proposes,they are guiltyof "a lazykind of teleologicalreasoning" (191). Again, the identified projectis thenecessity to findmicro-mechanisms: "What Marx- ists mustprovide are explanationsof mechanisms,at themicro level,for the phenomena they claim come about forteleological reasons" (192). The logic behind this AnalyticalMarxist position can be seen mostclearly in PhillipeVan Parijs' responseto thedescrip- tionof his positionas one of "NeoclassicalMarxism." Noting the contrastbetween "rational man (or individualistic)and structural (or systematic)explanations," Van Parijs indicatesthat structural explanations which referto a structuralimperative (e.g., a "fromthe itself")are "unambigu- requirementflowing system" ouslyrejected by 'neoclassical Marxism' (Van Parijs,119). Why? Because"no explanationof B byA is acceptableunless one speci- fies the mechanismthrough which A generatesB." Yet,"mechanism" has a ratherspecific meaning for Van Par- ijs here.For example,the propositions that can be derivedfrom the structuredcollective game of capital and wage-laborwould fail his testfor acceptability. This is clear fromhis subsequent proposition:"Or, equivalently,no explanatorytheory is accepta- ble unlessit is providedwith microfoundations.'1 (How Proposi- tionII is equivalentto PropositionI is somethingthat Van Parijs considersso self-evidentthat it need not be mentioned!)Clearly missingfrom his discussionis a criticalproposition - one which statesthat "the only mechanism by whichone can explainis one withmicrofoundations." This, of course,is theonly mechanism by whichone can getfrom I to II, and it is thecore of thematter. For,if we do acceptthat missing proposition, it ofcourse follows that "Marxismneeds microfoundations"(120). But, whyshould we acceptthe proposition that microfoun- dationsare the only mechanismby whichone can explain?All we have are assertions.But whereis theproof? Where is thedem- onstrationthat "methodological " cannot provide a valid (and, indeed,better) explanation? Where is thebasis forde- scribingit as misguided,near-nonsense, disastrous scientific prac- ANALYTICAL MARXISM 201

tice (Elster,1985, 4)? Are we to assume thatthe point, drawingits force from neoclassical conventionalism,is self-evident? Even if the Analytical Marxistsare able to find examples of functionalistor teleological argumentsconducted at the supra- individual level, it would not prove that methodologicalcollecti- vism necessarilyleads to functionalistor teleological argument. (While noting thata methodologicalcollectivist explanation "fre- " quently takes the formof functionalexplanation, Elsteradmits "thereis no logical connection" (6).) Indeed, Przeworski,Brenner and Elster themselvesall explore collective games in essays in Analytical Marxism. Further, an acceptable methodological individualist (or micro) explanation would not constitutea sufficientrefutation of an explanation of social phenomena based upon the concept of supra-individualentities. Marx's argumentthat the competition of capitalists executes the inner laws of capital is a rejectionof methodological individualism and microfoundations- but not of the real existenceof individual capitals and micro-phenomena. The conclusion that only microfoundationscan explain aggre- gate outcomes thus requiresfar more demanding proof than Ana- lytical Marxism offers. Ultimately,of course, the proofof the pudding is in the eat- ing. So, ratherthan criticizingthe Cartesian reductionismof the above arguments abstractly,let us consider specificallyElster's answer to Marx's explicitdenial of methodologicalindividualism - Roemer's "pathbreaking" work on exploitation. Elster des- cribes this centerpiece of Analytical Marxism as an approach "generating class relations and the capital relationship from exchanges betweendifferently endowed individuals in a competi- tive setting. . . . The overwhelminglystrong argument for this procedureis that it allows one to demonstrateas theoremswhat would otherwisebe unsubstantiatedpostulates" (Elster, 1985, 7). What, however,is wrong with the so-called "unsubstantiated postulates"?Recall thatMarx's procedurewas to begin his exami- nation of capitalism from the postulate of capitalist and wage- laborer in which the relation is specified as one in which the worker has sold the propertyright over labor-power with the necessaryresult both that the workerworks under the directionof the capitalist and that the workerhas no propertyrights in the productof labor. Marx, in short,begins fromthe specificationof 202 SCIENCE & SOCIETY a particularset of relationsof production. Now, we may ask: wheredid thoseunsubstantiated postu- lates come from?And theanswer is obvious:from history, from real life,the real concrete.The sale of labor-power,work under thedirection of thecapitalist and theabsence of workers' property rightsin the productof labor are the historicalpremises of the discussion;and theyare broughtto the theoreticaldiscussion of capitalismas theexogenous point of entry.So, thereis indeeda theoreticallyunsubstantiated postulate, the capital/wage-labor relation.What is critical,too, is whatMarx proceeds to do on the basis of thispremise. He exploresthe natureof the interaction betweencapitalist and workersin thecollective game and gener- ates the dynamic propertiesinherent in that structured relationship. Now considerwhat Elster has said aboutRoemer's approach: Roemerwill generatethe class relationshipfrom individuals; he will demonstratethe capi tal/ wage-labor relation as a "theorem." An immediateresponse might be: but,this is a differenttheoreti- cal object;what Marx takesas his startingpoint, Roemer sees as his result.Yet, it is importantto rememberthat, in Marx'sdialec- tical analysis,a centralrequirement will be to demonstratethat whatwas a merepremise and presupposition(an unsubstantiated postulate)of the theoryis itselfreproduced within the system - i.e., is also a result.In thisrespect, both Marx and Roemerhave thesame object- to demonstratethe production of theclass rela- tion.But, their starting points are different: Marx beginning from theobservation of theconcrete relationships and Roemerfrom . . . . Roemerfrom where? We will put thatquestion aside forthe moment. Let us ask first:what are we to concludeif both Roemerand Marx,having startedout fromdifferent places, arrive at thesame destination? Are we to concludethat Roemer's successful arrival (the deriva- tion of theclass relationfrom atomistic individuals) proves that you can't getthere from Marx's starting point? Obviously not. To concludethis would be to confuseexplanation and necessitation. At best,Roemer's arrival will have demonstratedMarx's argu- mentthat competition executes the inner laws of capital - i.e., that many capitals, the necessaryform of existenceof capital, manifestthrough competition the inner nature of capital. On the otherhand, if we have Roemer'sderivation, do we needMarx's? ANALYTICALMARXISM 203

But,there is a beggedquestion in all this:does Roemerreally arriveat thatsame point,the point which for Marx is bothpre- mise and result- the historicallygiven capitalist relations of production?Now, we will considerRoemer's starting point. El- sterhas alreadytold us: "differentlyendowed individuals." But, let Roemerexplain morefully. Responding to the criticismby Nadvi (1985),he indicatesthat his model"has 'explained'some phenomena,in derivingthem from logically prior data. In GTEC [Roemer,1982], the data are: differentialownership of themeans of production,preferences and technology.Everything is driven by thesedata; class and exploitationare explainedto be a conse- quence of initial propertyrelations" (Roemer, 1986b, 138). We see,not surprisingly,that Roemer also startsfrom "logi- cally priordata" which are not the subjectof his analysis(i.e. "unsubstantiatedpostulates"). It happensto be thesame logically prior data with which neoclassicaleconomics (in particular, neoclassicalgeneral equilibriumtheory) begins. And Roemer proposesthat, on thebasis of thosesame neoclassical premises, he has succeededin demonstratingthe of exploitationand class - a classic case of hoistingneoclassical economics by its own petard. Let us think,though, about theselogically prior data. (This particularsuccess may be a poisondraught for Marxism.) Where do theycome from?Roemer answers: history. "The historical processwhich gives rise to the initial endowmentswhere my modelbegins is not a subjectof myanalysis. That is a topicfor an historian"(Roemer, 1986b, 138). History thus has yieldeda set of individualswho, withgiven preferences and technology,have differentialproperty endowments. Is that it? Has historypres- entedus witha groupof atomistic individuals who haveno prior connections,no priorinteractions - individualswho are onto- logicallyprior to the society? Obviouslynot. Whatwe have,rather, is thatan analysthas decidedto modelthe individuals as if theywere initially outside societyand thenentered into societyto exchange.The starting point,then, is nothistory, but historymediated by an ideological assumption,one identifiedby Marxas earlyas 1843(Marx, "On theJewish Question," 1975). Now, it is easy to understandsuch an operationwhen conducted by a neoclassicaleconomist - but a Marxist? 204 SCIENCE & SOCIETY

Roemer'sinstrumentalist response, however, would be thatif the model succeedsin explainingthe desiredphenomena, then clearlythe "modelhas made theright asbtractions: it has ignored thingswhich are not crucial to its topic and has focusedour attentioncorrectly" (Roemer, 1986b). Methodologically,this is not objectionablepractice; Marx similarly engages in abstraction and puts aside questionspertaining to themembers of theset or coalition. Many Marxists,however, will find the idea that "society"is an appropriatevictim of Occam's Razor rather troub- lesome. Nevertheless,rather than debatingthis issue, it is more pertinentto considerwhether the model has indeedsucceeded in its object- whether,in short,Roemer's model makes "the right abstractions."

RoemerianExploitation

In discussingRoemer's success in generatingas theorems both classes and exploitationwithin capitalism, we mustlimit ourselvesto selectedaspects of his theoryas developedin his book and subsequentarticles. (Some other issues are raised in my reviewof the book; Lebowitz,1984.) We will not concernour- selves,for example, with the exploitation that Roemer discovers in his linear productionmodel of an economyof simplecom- modityproducers with differentialasset ownershipsince the inequalitythat Roemerfinds here is manifestly"rent," and its designationas "exploitationin theMarxian sense" would neces- sitateholiday pay forhis words;nor, for similar reasons, will we considerRoemer's "socialist exploitation." The core argumentin Roemer(1982) occurswhen Roemer introducesa labormarket into his modelof individualswith dif- ferentialendowments of productiveassets. He demonstratesthat, as a consequenceof optimizingbehavior, those individuals with low endowmentswill end up selling labor-powerand will be exploited(perform surplus labor) whereasthose who have high endowmentswill hire labor-powerand will be exploiters.The argument,generating the classical Marxistproposition, appears quite powerful. Roemer proceeds,however, to introducea creditmarket (ratherthan a labormarket) and revealsnow a functionallyequi- valentresult: those with low endowmentshire capital and are ANALYTICALMARXISM 205 exploited(perform surplus labor) while thosewith high endow- mentsrent capital and are exploiters.Exploitation is, indeed,the samein bothcases. Accordingly, Roemer offers his "isomorphism theorem"that "truly it does not matterwhether labor hires capi- tal or capital hires labor: the poor are exploitedand the rich exploit in eithercase" (Roemer,1982, 93). Now, thistheorem (whose clauses successively skewer Marx- ian and neoclassicaleconomics) is centralto all thatfollows. Roe- mer himselfdraws the robustinference that "the fundamental featureof capitalistexploitation is notwhat happens in thelabor process,but the differentialownership of productiveassets" (94-5). Yet,precisely the wrong conclusion has beendrawn from the isomorphismtheorem: rather than revealingthe power of Roemer'sanalysis, it exposes its weakness. Considerwhat has occurred.Logical priorityhas shifted fromspecific relations of productionto propertyrelations; their connectionhas beeninverted. Rather than seeing capitalist prop- ertyrelations (KP) as theproduct of capitalist relations of produc- tion (KRP), Roemer argues that differentialownership of productiveassets necessarily yields capitalist relations of produc- tion,exploitation and class. Since,further, this can be demon- stratedto occurwith either a labor or creditmarket, it follows thatunequal propertyendowment plus a factormarket are suffi- cientto generate"class relationsand thecapital relationship" (as theorems). Let us, however,stress what Marx saw as criticalelements in capitalism.These are:(1) thesale of theproperty right over labor- powerby theperson who ownsno meansof production;and, (2) thepurchase of thisproperty right by an ownerof meansof pro- ductionwhose goal is valorization(M-C-M). The two elements hereclearly presuppose capitalist property relations (KP) - the specificinequality in propertyownership. However, KP is not sufficientto yieldthese two elements - since(as Roemerhimself demonstrates)it is obviousthat KP can also support:(la) thehir- ing of meansof productionby someonewho owns only labor- powerand (2a) therenting of thesame by theowner of meansof production.KP is a necessarybut not sufficient condition for cap- italism(KRP). In short,two quite distinctregimes can be generatedon the basis of Roemer'slogically prior data, the initialproperty rela- 206 SCIENCE & SOCIETY tions. A simple questionreveals that : who owns the productof labor} In 1/2,property rights in theproduct of labor belongto theowner of meansof production(who also purchased theproperty rights over the disposition of labor-power); in la/2a, on theother hand, it is theowner of labor-powerwho possesses theproperty right over the product. It is notdifficult to establish that Marx's analysisof capitalismrefers to 1/2 - but not to la/2a. For Marx, the situationin which the purchaseof labor- powerdid not occurwas explicitlypre-capitalist. Where there is formalsubsumption of labor by capital (the initial formof the capital relation),"the relationsof capital are essentiallycon- cernedwith controlling production and . . . thereforethe worker constantlyappears in themarket as a sellerand thecapitalist as a buyer"(Marx, 1977, 1011). In contrastto formalsubsumption, on the otherhand, was the case wherecapital is to be foundbut "whereit has notemerged as thedirect purchaser of laborand as theimmediate owner of theprocess of production"(as with,e.g., usuryand merchantcapital). "Here we have not yetreached the stage of formalsubsumption of labour under capital" (1023). Characteristicof pre-capitalistrelations was preciselythe credit-marketcase thatRoemer presents. Thus, in theGrundrisse, Marx commentedthat the relationin whichthe producer,still independent,faces means of productionwhich are independent "formingthe property of a particularclass of usurers. . . neces- sarilydevelops in all modesof productionresting more or lesson exchange"(Marx, 1973,853). Here, the worker"is not yetsub- sumedinto theprocess of capital.The modeof production there- foredoes not yetundergo essential change." There is, of course, exploitation- indeed,the "mostodious exploitationof labor." In the mode of productionitself, capital is still "materiallysub- sumedunder the individual workers or thefamily of workers. . . . What takesplace is exploitationby capital withoutthe mode of productionof capital. . . . This formof usury,in whichcapital does not seizepossession of production,hence is capitalonly for- mally,presupposes the predominanceof pre-bourgeoisforms of production"(853). Marx similarlyobserved that "capital arises only wheretrade has seizedpossession of productionitself, and wherethe merchant becomes producer, or theproducer mere mer- chant" (859). ANALYTICALMARXISM 207

In short,specifically capitalist relations of production (KRP) as examinedby Marxrequire more than unequal distributionof propertyin meansof production (KP); theyalso requirethat capi- tal has "seizedpossession of production"(true for 1/2 but not la/2a), thatcapital directs the process of production, that produc- tionis subordinatedto thegoals ofcapital. Only with this second elementdo we necessarilyhave twoessential characteristics of the capitalistlabor process: that "the worker works under the control of thecapitalist to whomhis labourbelongs" and "theproduct is theproperty of thecapitalist and notthat of theworker, its imme- diateproducer" (Marx, 1977, 291-2). Only here is it characterisitc that,rather than the workeremploying , meansof productionemploy the worker(a metaphorthat cap- turesMarx's conception). Thus, Roemer's"logically prior data" cannotselect between capitalismand pre-capitalism.Does it matter?Consider what fol- lows from1/2 whichdoes not followfrom la/2a. The perfor- manceof surpluslabor will be compelled(given by M-C-Mand thesale of theproperty right over labor-power); i.e., therewill be exploitationspecific to capitalistrelations. The capitalist- but not theworker - will gain by increasingthe intensity of labor; thecapitalist - butnot theworker - will be thedirect recipient of gains resultingfrom increased productivity and thushas an incentiveto alterthe production process. Capitalist exploitation will be thebasis of ; KRP will be a sufficient conditionfor the reproductionof KP, forcapitalist relations. By contrast,under la/2a (the creditmarket case), it is the producerwho gains fromincreased labor and productivityand who decidesover the process of expanded . (Consider how thiscollective game would differfrom the capitalist game.) Potentially,this producer may succeed in securingmeans of pro- ductionfor self as theresult of intensive efforts. Unlike the case of 1/2,the creditmarket case is, in fact,a "transitional"relation- ship.The dynamicproperties, the laws ofmotion, inherent in the two structuresclearly differ. What"phenomena," then, have beenderived from Roemer's logicallyprior data, unequallyendowed atomistic individuals? Whattheorems have been successfullydemonstrated by means of this primeexample of methodologicalindividualism? We find 208 SCIENCE fi?SOCIETY that thereis no distinctionbetween a capitalistand a pre- capitalistrelation, no distinctionbetween specifically capitalist exploitationbased on capitalistrelations of productionand pre- capitalistexploitation based merely on unequal propertyendow- ments.Roemer, of course,is entitledto call anythinghe wants capitalism(as is MiltonFriedman) - but it should not be con- fused for a momentwith Marx's (and a Marxist)concept of capitalism. Thus, Roemerdoes not arriveat the same destinationas Marx. Ratherthan strengtheningthe case thatwe can proceed fromindividuals with differing property endowments to generate capitalistrelations of productionand capitalistexploitation, the veryindeterminacy apparent in his model(the isomorphism theo- rem)undermines his argument.Still, it mightbe respondedthat all this simply proves that Marx was wrong to distinguish between capitalist exploitation(where 1/2 holds) and pre- capitalistexploitation based upon unequal propertyendowments (la/2a) since exploitationis the same in both cases. To answer this argument,we must brieflyconsider Roemer's model. One of the critical problemsin Roemer's model is his assumptionof a commonproduction function for all regimes. Precluded,then, by definition is anyeffect of particularrelations of productionon the productionfunction. By assuming,for example,in his linear productionmodels that a unit of labor- power exudes a certainquantity of labor (i.e., the qualityand intensityof labor are presumablygiven technically), he notonly effectivelyassumes away the contentof the Marxiandistinction betweenlabor-power and labor,but also leavesus withproduc- tion consideredmerely as a technicalprocess transforming inert inputs into final products.Thus, the distinctionRoemer once acknowledgedbetween the neoclassical and Marxistapproach - that the Marxistasks "how hard are the workerslaboring?" - fades away (Roemer,1981, 143-5). Whatis theimplication? Consider the difference between the creditmarket model and thelabor marketmodel. In theformer, the producerssecure the fruitsof theirown labor (i.e., own the product).They choose whether or not to selectleisure on thejob. Presumably,there are no inherentproblems of shirking, no neces- sarycosts of surveillanceand monitoring,etc., which would be reflectedin theproduction function. To assumethe same produc- ANALYTICALMARXISM 209 tioncoefficients in thecase of thelabor market model, however, is to presupposethat workers who have no propertyrights in the productsof theirlabor will behavein thesame way as thosewho do. (Nor shouldwe ignorethe likelihood, in thelatter case, that thechoice of techniqueand the divisionof labor will be deter- minednot solelyby technicalefficiency but also by theneed to monitoreasily and to reducethe ability of producers to engagein coalitions.) AlthoughRoemer concludes that capitalist exploitation does not requiredomination at thepoint of productionbecause "the class and exploitationrelations of a capitalisteconomy using labourmarkets can be preciselyreplicated with a capitalistecon- omyusing credit markets," his modelsgenerate these results only because of his hiddenassumptions (Roemer, 1986a, 268). In a model in which producerswish to maximizeleisure (which includesleisure in the"pores" of theworkday), the assumption of unchangedtechnical coefficients in the two cases amountsto assumingthe existenceof an efficient(and costless)capitalist monitoringprocess - withoutacknowledging the assumption! One can onlyabstract from the requirement of capitalist domina- tionby assuming(as Roemerdoes explicitly)that the delivery of labor forthe wage is "as simpleand enforceablea transactionas the deliveryof an apple for a dime" (269). In short,Roemer's discovery that the labor marketcase and thecredit market case yieldequivalent solutions and that,accord- ingly,capitalist domination is not necessaryreflects merely the ideologicalassumption he has imposedupon his model.Having assumedthat productive relations do not matter,Roemer finds littledifficulty in then"proving" that they don't matter; he is of coursenot thefirst to believethat he has provenwhat is merely embeddedin his assumptions.

"Just"Exploitation

For othersin theAnalytical Marxism camp, Roemer's discov- erythat capitalist exploitation requires neither labor-power as a commoditynor domination in productionhas beenmost persua- sive. Wright,for example, initially resisted the argumentthat capitalistdomination was unnecessaryfor exploitation but then yielded - maintaining,however, the importanceof a link 210 SCIENCE & SOCIETY betweendomination in productionand class relations(Wright, 1982,331). Subsequently,he succumbedon thislatter point as well (acceptingRoemer's theory as the frameworkfor his own empiricalwork) and announced,"I now thinkthat Roemer is correcton thispoint" (Wright,1985, 72). Elster,too, is unequivo- cal; afterpresenting Roemer's conclusion, Elster characteristically declares,"I believe that Roemer'sargument is an irrefutable objectionto the 'fundamentalist'view thatexploitation must be mediatedby domination in thelabour process" (Elster, 1985, 181). Once the specificcharacteristics of capitalismand capitalist exploitationhave beenobliterated (leaving only unequal endow- ments),however, can meditationson "just" exploitationbe far behind?Posing the question"Should Marxistsbe Interestedin Exploitation?",Roemer responds with his "verdict. . . that exploitationtheory is a domicilethat we need no longermain- tain: it has provideda home forraising a vigorousfamily who now mustmove on" (Roemer,1986a, 262). Having emptiedthe house of all its contents,Roemer's up-market move is to "the modernconcept" of exploitationas "an injusticein thedistribu- tion of income resultingfrom a distributionof endowments whichis unjust" (Roemer,1986a, 199). Exploitation,in short,is simply inequality- "the distributionalconsequences of an unjust inequalityin the distributionof productiveassets and resources"(Roemer, 1986a, 281). The obvious implicationis thatexploitation/inequality is not unjust if the original inequalityin propertyendowments itselfwas not unjust. While this point is indeed exploredby Roemer(1986a), it is Elsterwho mostclearly draws out thelogic of the AnalyticalMarxist argument.We judge exploitation unjust,he proposes,because "exploitation in historyhas almost alwayshad a thoroughlyunclean causal origin,in violence,coer- cion, or unequal opportunities"(Elster, 1986, 99). But, what if therewere a "clean path" of originalaccumulation? What if peo- ple differin theirtime preferences? What if some people choose to save and investrather than consume (thereby building up a capi- tal stock)?"Could anyoneobject if they induce others to workfor themby offeringthem a wage above what theycould earnelse- where?"(Elster, 1985, 226). Here, Elsternotes, is a "powerful objection,that must be takenseriously by anyone who setsout to defendMarx's theoryof exploitation"(227). ANALYTICAL MARXISM 2 11

Thus, as counterexamplesto the view that exploitation is inherentlyunjust, Elster and Roemer each presenta two-person case where,as the resultof the patternsof capital asset ownership ' and leisure preferences,the asset-poorperson 'exploits the rich person" (Roemer, 1986a, 274-7; Elster, 1986, 98). Elster'sconclu- sion from this example is that "it demonstrates,I think conclu- sively, that exploitation is not inherentlywrong" (98). In a second example ("more relevant for real-lifeproblems"), Elster posits two people with the same skills and capital but a different orientation toward present consumption. One postpones con- sumption and thus accumulates capital - enough ultimatelyto pay the other to work for her - at a wage that exceeds what he could gain by himself."True, he will be exploited - but who cares?" From this, Elster concludes that "the example suggests that exploitation is legitimatewhen the unequal capital endow- ments have a 'clean* causal history" (99). All thatis leftto exploitation,thus, is the contingentcharac- terof original accumulation. Since exploitation has been severed fromany connection to the capitalist process of production and rests solely upon the pre-existenceof unequal endowments,all that remains is the question as to whetherproperty rights were violated in the formationof those differentialendowments. Hav- ing begun by invertingthe connection between property relations and relations of production in an organic system,Elster and Roemer find fromtheir apocryphal storiesof original accumula- tion that "exploitation is not a fundamentalmoral concept" (Els- ter, 1986, 99). Sadly, it is as if the distinctionbetween "original capital" and that which emergesfrom specifically capitalist exploitation had never been made; as if Marx neverpointed out that even if capital were originally acquired by a person's own labor (the cleanest possible path to accumulation), "it sooner or later becomes value appropriatedwithout an equivalent" (Marx, 1977, 715, 728). The inherentinjustice of exploitation is just one more elementthat disappears in the course of making "sense" of Marx.

Conclusion

It would be quite easy (but, at the same time,quite wrong) to conclude fromthe above discussion thatAnalytical Marxism has 212 SCIENCE & SOCIETY littleto offerMarxists. In fact,these writers pose importantques- tionsand challenges.They reject,in particular,teleological rea- soning in Marx; it should be rejected.Similarly, functional explanationsare viewedas suspect.They are - and, wherethey appear, theyshould be scrutinized.Analytical Marxism, in this respect,can keep us on our toes. Further,there are aspectsof thework of thesewriters which can be incorporatedeasily into Marxist analysis. Roemer's exami- nation of exploitationas an implicitcounterfactual proposition (Roemer,1982) points to a wayaround the neoclassical objection that the verysale of labor-powerproves that the wage-laborer benefitsfrom the exchange (compared to theexisting alternative of non-sale).Elster's early discussion (Elster, 1978) of the"fallacy of composition"(what is possiblefor one memberof a setis not necessarilytrue for all memberssimultaneously) strikes directly at attemptsto reason fromthe positionof the isolatedindividual (and all such Robinsonadesof neoclassicaleconomics). And, Cohen's "lockedroom" parable,in his "The Structureof Prole- tarianUnfreedom," dramatically poses the contrastbetween the individualworker's ability to escape the statusof wage-laborer and the structuralinability of the class as a whole to do so (Cohen, 1986). These last two examples,in particular,provide powerful argumentsagainst neo-classicalconventions. I regularlyintro- duce themin thefirst week of myclass in Marxianeconomics - as an introductionto the question as to why Marx saw the necessityto begin fromthe considerationof an organicwhole (which is, of course,precisely contrary to the methodological imperativeof AnalyticalMarxism). Nevertheless,not only is therenot much of Marxleft in Ana- lyticalMarxism, but its essentialthrust (as tracedabove) is anti- Marxist.So whydo thesewriters wish to retaintheir connection to any kindof Marxism?The answer,it appears,is thatthey con- siderthemselves socialists and thatthe Marxist "label doesconvey at least thatcertain fundamental insights are viewedas coming fromMarx" (Roemer,1986, 2). As Elsterputs it, "if, by a Marxist, you mean someonewho can tracethe ancestry of his mostimpor- tantbeliefs back to Marx,then I am indeeda Marxist"(Elster, 1986, 4). Yet, if selectedbeliefs and insightsdetached from a Marxian ANALYTICAL MARXISM 2 13 frameworkwere sufficientfor designation as Marxism, then the termwould lose all integralmeaning. For, situatedin an alterna- tiveframework, those selected beliefs acquire quite differentprop- erties.The transformation,within the neoclassical framework,of Marx's theoryof exploitation (one of Elster's "most important beliefs") into a conception of distributivejustice which accepts the possibility of just exploitation illustratesthis fundamental dialectical principle quite well. What makes AnalyticalMarxism anti-Marxistis that the beliefs and insightsonce absorbed from Marx have been incorporatedwithin an anti-Marxistframework, and the parts have acquired propertiesfrom that whole.

Simon Fraser University Burnaby, Canada

REFERENCES

Bradley,Ian and Ronald L. Meek. 1986. Matrices and Society. Middlesex: Pen- guin Books. Carling, Alan. 1986. "Rational Choice Marxism." Review, 160 (November/December). Clawson, Patrick. 1983. "A Comment on Van Parijs's Obituary." Review of Radical Political Economics, XV:2 (Summer). Cohen, G. A. 1986. "The Structureof Proletarian Unfreedom." In Roemer, 1986a. Elster, Jon. 1978. Logic and Society. Chichester: Wiley. Elster,Jon. 1985. Making Sense of Marx. Cambridge,England: CambridgeUni- versityPress. Elster, Jon. 1986. An Introductionto Karl Marx. Cambridge, England: Cam- bridge UniversityPress. Gerdes, Paulus. 1985. Marx DemystifiesCalculus. Minneapolis: Marxist Educa- tional Press. Gray, John. 1983. "The System of Ruins." Times Literary Supplement, December 30, 1983. Lebowitz, Michael A. 1984. Review of Roemer, 1982. Canadian Journal of Eco- nomics, XVII:2 (May). Lebowitz, Michael A. 1987a. "Labour Strategiesin a World of Capital." Paper Pesented to Canadian Political Science Association (), June, 1987. Lebowitz, Michael A. 1987b. "The Political Economy of Wage-Labor." Science b Society, 51:3 (Fall). 214 SCIENCE & SOCIETY

Levins, Richard and Richard Lewontin. 1985. The Dialectical Biologist. Cam- bridge, Mass.: Harvard UniversityPress. Maarek, Gerard. 1979. An Introduction to Karl Marx's . Oxford: Martin Robinson. Marx, Karl. 1968. Theories of Surplus Value. Vol. II. Moscow: ProgressPub- lishers. Marx, Karl. 1973. Grundrisse. New York: Vintage Books. Marx, Karl. 1975 (1843). "." Karl Marx and , Collected Works. Vol. 3. New York: International Publishers. Marx, Karl. 1977. Capital. Vol. I. New York: Vintage Books. Marx, Karl. 1981. Capital. Vol. III. New York: Vintage Books. Nadvi, Khalid, 1985. "Exploitation and : A Critique of Roemer's General Theory of Exploitation and Class." Economic and Poli- tical Weekly, XX:35 (August 31, 1985). Nadvi, Khalid. 1986. "Exploitation and Labour Theory of Value." Economic and Political Weekly, XXI:20 (May 17, 1986). Olson, Mancur. 1965. The Logic of Collective Action. Cambridge, Mass.: Har- vard UniversityPress. Roemer, John E. 1981. Analytical Foundations of Marxian Economic Theory. Cambridge, England: Cambridge UniversityPress. Roemer, John E. 1982. A General Theoryof Exploitation and Class. Cambridge, Mass.: Harvard UniversityPress. Roemer,John, ed. 1986a. AnalyticalMarxism. Cambridge,England: Cambridge UniversityPress. Roemer, John. 1986b. "Exploitation and Labour Theory of Value." Economic and Political Weekly, XXI:3 (January 18, 1986). Struik, Dirk. 1948. "Marx and Modern Mathematics."Science ¿r Society, 12:1 (Winter). Van Parijs, Phillipe. 1983. "Why MarxistEconomics Needs Microfoundations: PostscriptTo An Obituary." Review of Radical Political Economics, XV:2 (Summer). Wright,Erik Olin. 1982. "The Status of the Political in the Concept of Class Structure." Politics ¿r Society, 11:3. Wright, Erik Olin. 1985. Classes. London: Verso.