Business Organization as a Coordination Problem: Toward a Dynamic Theory of the Boundaries of the Firm Richand N. Langlois The Universityode Connecticut Paul L. Robertson Universityof New South Wales,ADFA Introduction Many writershave noted that, since 1873, the main thrustof mainstream theoryhas shifted away from the concernsof Adam Smithand the classicals [12]. This "marginalist"or neoclassicaltheory was designed not to understand the springsof economicgrowth and the sourcesof wealthbut ratherto analyze the allocationof known and given resources. In his Theory of Political Economy,William StanleyJevons [11, p. 267] put the matterthis way. "The problemof economics,"he wrote, "may,as it seemsto me, be statedthus: -- Given, a certainpopulation, with variousneeds and powersof production,in possessionof certainlands and other sourcesof material:required, the mode of employingtheir labourwhich will maximize the utility of the produce." Even at the start,however, marginalist theory was far from homogeneousin its concerns[10], and both the Austrian and Marshallian streamsretained, albeit in slightlydifferent ways, many of the classicalpreoccupations. In any case,neoclassical economics grew to encompassa numberof distinctvariations, each arguably pointing at a differentset of concerns.The Walrasian system poses an answer to an abstractlogical problem whose generalrelevance one may question.But "MarshallJan"comparative statics -- itself only one aspectof Marshall's opus-- was designedto answersome quite importantquestions: how in the short run do exogenouschanges in boundaryconditions affect the directionof changein price and quantity (suppliedand demanded)in relativelyisolated markets [26]? Is this the set of questionsthat interestbusiness historians? Arguably, business historians are at least as interested in the sorts of issues that animated Smith: what are the sources of economic growth and industrial competitiveness?How do the organizationof productionand the institutions of societyaffect economicgrowth and competitiveness?To the extentthat businesshistorians have been interestedin suchquestions, then, mainstream BUSINESS AND ECONOMIC HISTORY, VolumeTwenty-two, no. 1, Fall 1993. Copyright¸ 1993 by the BusinessHistory Conference. ISSN 0849-6825. 31 32 theory has been of limited usefulness. Indeed, in the areas in which the concernsof businesshistorians and economictheorists have overlapped-- namely so-calledIndustrial Organization theory --neoclassical modelshave beenstrained well beyondtheir limits, leading to inappropriateapplications of theoryand, as in the areaof antitrustpolicy, often absurd and harmful policy conclusions. Attackson mainstreamtheory are in abundantsupply these days, and sucha blanketcriticism is not our goalhere. Rather,we wantto suggestthat the appropriatetheory for businesshistorians ought to be animatedby the questionswith whichbusiness historians concern themselves. These are surely historicalquestions. More abstractly,however, they arequestions, not onlyof allocationand welfare, but alsovery importantlyof growthand development: how is new value created? They are, moreover,institutional questions: how do socialinstitutions and forms of businessorganization lead to growth and competitiveness?And how arethese institutions shaped in turnby growthand competition? Confronting such questionsanalytically may not mean abandoningmainstream ideas so much as abandoningthose assumptions that were designedfor other purposesand are inappropriatefor the historian's questions. The "why" of Organizations. The formal neoclassicaltheory of the firm takes "the firm" as a fundamentalbuilding block in the constructionof a theory of the industry. This building block is a simplifiedand anthropomorphizedideal type -- a "monobrain,"as Fritz Machlup [30] put it. Especiallyin its true Marshallian formulation,such an approachhas proven extremely valuable for the questions of partial-equilibriumcomparative statics for which it was intended. But, not surprisingly,that theory has not proven very usefulin analyzingwhat goeson insidethe firm or, more importantly,how productionis actually organized within the economy.In Axel Leijonhufvud'sirreverent image [29, p. 203], the formal neoclassicalconception of the firm "is more like a recipe for bouillabaissewhere all the ingredientsare dumpedin a pot, (K, L), heatedup, f(o), and the output,X, is ready." It providesno insightinto organizational structureor the sequencingof tasks. More generally,economists tend to centertheir theoriesaround the premisethat firms existto provideprofits for their owners. This assumption holdsboth for the relationshipof the firm to its externalenvironment -- what the firm choosesto do itself and what it purchasesand sells to others-- and for the internal organizationof the firm -- how it goes about producing whatever goodsor servicesit has decidedbelong within its proper sphereof activity. From the perspectiveof businesshistory, this approachbegs some of the most importantquestions. It essentiallyrelegates to secondplace, or even assumesaway altogether,the activitiesthat the peopleworking for the firm are actually engagedin. These include decidingwhat to produceand how to produceit and then actuallyproducing it in the way that best rewardsthe firm's owners. Hence, successderives from the firm providing goodsand 33 servicesthat meet the needsof potentialcustomers in a way that generatesthe highestpossible returns to its owners. In short,firms, and thosewho would understandthem, must keep an eye on bothblades of the Marshallianscissors [27]. If this is true (and we think it is), then we should conceiveof firms as organizationsthat need to tackle a variety of goals. These goals are interdependentin the sensethat, while all of them are important,many may be in conflict if they are not coordinatedor "managed."This holdseven if we leave asidequestions of opportunismand shirking. If the firm is to survive, let aloneprosper, it mustmake surethat it producessomething that customers desire,which meansthat it must acquireand accuratelyuse information,or knowledge,about products and productionprocesses. In fact, a firm mustbe organizedto undertakeone, several, or all of the following activitiesassociated with the profitableproduction of a good or service: conception,design and development,manufacturing, provision of inputs, marketing and distribution,and many others. A design that is outstandingin the sensethat it meetsthe performanceattributes [15] that potentialpurchasers desire, however, may for that very reasoncost more to producethan those same purchasers are willing to pay; or it may be impossible to produceat all. Thus,one reason for organizingsuch diverse activities is to providecoordination between aspects of productionso that a plausibleoutcome resultsin the form of a goodor servicethat can be produced(a) with non-cost attributesattractive to potentialbuyers; (b) at a price that is also acceptableto those buyers;and (c) that allows for an acceptablereturn on the productive resources involved. Whether the proper vehicle for generatingsuch a plausibleoutcome is a verticallyintegrated firm, or severalfirms specializingin differentlinks in the productivetrain, or a groupof independentand unattachedworkers is a separatequestion. The answerdepends, inter alia, on levels of transaction costsand the relativestrengths and relevanceof the capabilitiesof the possible participantsin the productionprocess. Transaction-cost Approaches Recently, of course,a new set of theorieshas emergedto focus more clearly on issuesof organization.In one way or another,these strands have takentheir inspirationfrom the work of RonaldCoase [6]. Here the firm is by no meansthe sort of black box it is in traditionalIO theory. Moreover, these approachesdo not "take the firm as a unit of analysis."In Oliver Williamson'formulation of transaction-costanalysis [44], for example,it is the "transaction"-- which may occur within the firm or acrossmarkets -- that becomesthe fundamentalunit of analysis. This approachis able to ask questionsthat are somewhatdifferent from those of traditionalIO theory. Principalamong these is the matterof the boundaryof the firm: why are some activitiesorganized across markets and someorganized within firms. As a consequence,the transaction-costapproach has also proven able to provide 34 someanswers that are not only differentfrom but arguablyricher and more sensible than those of traditional IO. 1 Salutaryas theseinnovations have been, it nonethelessremains the case thattransaction-cost analysis retains fundamentally the neoclassical conceptual apparatus.This is so in a coupleof respects.First, transaction-cost analysis in all its principalforms is concernedwith the allocationof knownand given resources.Now, it is certainlytrue that imperfectionsin informationfigure prominentlyin this tradition. We might even say that the assumptionof imperfectionsin informationlie at the very heart of all transaction-cost approaches.But it remainsthe casethat the "imperfections"allowed are of a ratherparticular and limited sort. Imperfectinformation or knowledgein these approachesis alwaysof a "parametric"or purely quantitativekind [16]. For example,it may be costlyto monitorlevels of effort, or it may be costlyfor one party to know whethera productis a "lemon." But there is never any "structural"or qualitativeuncertainty. There is never any disagreement betweenparties aboutthe fundamentalcategories of action:all know what it would mean to providea certainlevel of effort; all know what