A Key to Coase s Thought: The Notion of Cost

Transcription

1 Colloque de l ACGEPE, Nice, juin 2012 A Key to Coase s Thought: The Notion of Cost 1 The use of Ronald Coase s early reflections on cost during the 1930s as a key to read some of his main works brings to light one neglected aspect of the nature and origin of his novel approach. I shall show what his theoretical insights owe to his taking into account the subjectivity of decision-making and to his definition of the cost associated to choice as an opportunity cost. First, I shall bring to light two dimensions of the notion of cost that Coase developed in 1938: a cost is an opportunity cost and it is subjective. Then I shall show that both these dimensions pervade his main theoretical insights, and do so in two directions. On the one hand, his insistence on the subjective aspects of the choice of the businessman fuels his works on the theory of the firm, especially on the monopolist s choice (1937a) and the choice between make and buy (1937b). On the other hand, the importance of reasoning in terms of opportunity cost shapes his criticisms of standard economic policies, such as marginal cost pricing for natural monopolies (1946b) or policies aimed at equalizing private cost and social cost (1960). Coase progressively abandoned the subjective dimension of his analysis, an evolution for which I provide some elements of explanation. Keywords: Ronald H. Coase, opportunity cost, subjectivity, firm, social cost, marginal cost pricing, accounting 1. Introduction During the 1930s, as described by Buchanan (1969) in his reconstruction of the notion of subjective opportunity cost, Ronald H. Coase participated to the development of a subjectivist tradition in economics, specific to the London School of Economics and integrating Hayek s subjectivism to Arnold Plant s concern for practical applications. In particular, Coase wrote a series of articles titled Business Organization and the Accountant (Coase 1938) in which he explained to accountants recent developments of economic theory on the notion of opportunity cost, insisting on the subjective dimension of this cost. Some authors had the intuition that this article on cost had influenced others of his works, or at least brought up some themes that he raised elsewhere. Buchanan, without developing his idea, asserted that Coase s developments on cost (Coase 1938) influenced his works on the marginal cost controversy (Coase 1946b) (Buchanan 1969, ch 2, fn 36 2 ) and social cost (Coase 1960) (Buchanan 1973, 1.18). However, criticizing Coase s presumably objectively-measurable approach in this last article, Buchanan (1984, 11) will later turn Coase (1938) against Coase (1960). Medema (1994) established a link between Coase s reflections on cost (still Coase 1938) with, on the one hand, his work on monopoly price (Coase 1937a) and, on the other, the theory of the firm (Coase 1990). Finally, Arena (1999) showed that the subjective dimension of entrepreneurial decisions can be found in Coase s works of the 1930s on duopoly (Coase 1935), monopoly pricing (Coase 1937a), expectations (Coase and Fowler 1935; 1937; 1940) and accounting (Coase 1938). I wish here to deepen and enlarge these intuitions. First, I shall bring to light two dimensions of the notion of cost that Coase develops in the 1938 text: a cost is an opportunity cost and it is 1 CNRS (PHARE University Paris Pantheon-Sorbonne). Elodie.Bertrand@univ-paris1.fr. I thank Steven Medema for his precious comments and suggestions on earlier versions of this paper, presented at the 2010 HES Conference, 2011 ESHET Conference, and the ERMES-PHARE seminar (March 2011); this paper also benefited from remarks made by the participants to these seminars. Errors and omissions remain mine. 2 Our references to Buchanan (1969; 1973) and Coase (1938 [1973]) are to the electronic version available at and therefore contain no indication of pages, but refer to chapters and paragraphs.

2 2 subjective. Then I shall show that both these dimensions pervade his main theoretical insights, and do so in two directions. On the one hand, his insistence on the subjective aspects of the choice of the businessman fuels his works on the theory of the firm: this will be seen in his analyses of the monopolist s choice (1937a) and of the choice between make and buy (1937b). On the other hand, the importance of reasoning in terms of opportunity cost shapes his criticisms of standard economic policies, such as marginal cost pricing for natural monopolies (1946b) or policies aimed at equalizing private cost and social cost (1960). This time, what is examined is public decision, and the subjective dimension of the decision is lost. Therefore Buchanan s claims are made consistent: the subjective dimension of Coase s 1938 article disappears in the Problem of Social cost, but not its other dimension, that of opportunity cost. This new reading of Coase s main theoretical works through his reflection on the notion of cost will not only allow me to characterize the origin and novelty of Coase s approach, but also raises other issues. First, although it is usually accepted that the revolutionary feature of Coase s approach comes from the fact that he was not trained as an economist (Medema 1994, 3) 3, my interpretation complementarily argues that Coase owed a part of his originality to a concept of cost that is derived from highly theoretical debates among economists. Second, while Buchanan (1969, 2.25) and Medema (1994, 58) have regretted that Coase s 1938 article had no influence on economists, my conclusion leads to reconsider this influence : it had existed if we consider its two different aspects, the subjectivity of entrepreneurs decision through The Nature of the Firm and the opportunity cost reasoning through The Problem of Social Cost. Finally, if Coase s articles of the 1930s examined here stress the subjectivity of producers decisions, this subjectivity is forgotten in his criticism of standard policies written later: these criticisms, and the method he suggests for the design of policy, are based on the idea that costs are objective and measurable; which poses the question of Coase s abandonment of subjectivity. Section 2 details the formation of Coase s view on cost during the 1930s, and brings to light two dimensions in this view: a cost is an opportunity cost and it is subjective. Section 3 and 4 show how the latter dimension is to be found in Coase s analysis of the choice of the businessman and the former in his analysis of the choice of policy. Section 5 concludes and suggests some elements to explain why Coase progressively abandoned the subjective dimension of his analysis. 2. Linking cost with choice: from economics to accounting Three notions of cost will be here distinguished (cf. Buchanan 1969): The classical notion of real cost, independent from utilities, adopted by Marshall (1890), who follows Smith and Ricardo. 4 The objective opportunity cost (OOC below) notion developed by the Austrian marginalists (Wieser 1884 ; 1889, but also Menger and Böhm-Bawerk), who define the cost of a product as the objective market value of the alternative product that could be produced with the same resources in alternative uses. This opportunity cost is therefore equal to a market price determined by marginal utilities; it eventually depends on subjective evaluations: this is the subjective theory of value, also developed by Jevons and Walras. But the opportunity cost, 3 Which is confirmed by Coase (1990, 3) himself: I did not take any course in economics while I was a student at the London School of Economics, a circumstance which I believe gave me a freedom in thinking about economic problems which I might not otherwise have had. 4 But see Buchanan (1969, ).

3 3 defined as a price, is objective in the sense that it is measurable by an external observer. At equilibrium, this OOC is equal to money outlays since the price of the alternative product is reflected in the price of the resources (Buchanan 1969, 3.10). In his criticism of Marshall, Wicksteed (1910; 1914), also influenced by Jevons, asserted the superiority of the Austrian conception of cost and diffused their theory in the English language, alongside with Davenport (1908; 1913) and his student Knight (1921; 1924; 1928). The conception of cost as OOC became dominant, named by Buchanan (1969) as orthodox, or neoclassical. The subjective opportunity cost (SOC below) defined by Buchanan (1969), who criticizes the notion of OOC as not implying an actual choice: if the individual just minimizes her cost, she is an automaton (ibid., 3.11). On the contrary, if choice is introduced, argues Buchanan: Cost becomes the negative side of any decision, the obstacle that must be got over before one alternative is selected. Cost is that which the decision-taker sacrifices or gives up when he makes a choice. It consists in his own evaluation of the enjoyment or utility that he anticipates having to forego as a result of selection among alternative courses of action (ibid., 3.12). Six features of the SOC follow: it is exclusively borne by the decision-maker; it is subjective in the sense that it exists in the mind of the decision-maker and nowhere else ; it cannot be measured or observed by someone else than the decision-maker; it is forwardlooking, or ex ante; it is never realized since the alternative not taken is not by definition; it can only be dated at the moment of choice (ibid., 3.13). Buchanan details on several occasions (ibid., 2.21; ; ; ch 6) the conditions under which the SOC is equal to the remuneration of resources: equilibrium, no alternative, no uncertainty, economic behavior (that is, in Buchanan s thought, absence of non monetary consideration and of consideration for someone else than self). Buchanan names his concept as London-Austrian since he built it under the influence of, on the one hand, the subjectivist philosophy of Mises and Hayek and, on the other, the works developed at the London School of Economics during the 1940s and 50s The LSE tradition on cost The thought on the notion of cost during the 1920s and 30s was the chance for economists to deepen the consequences of the marginalist revolution and involved several important debates, among which those on planning and on the cost function of the entrepreneur (which will lead to the theory of the firm). The London School of Economics was at the center of these debates. It was founded in 1895 by personalities closed to the English Historical School and in opposition to the domination of Marshall and Cambridge, but the arrival of Robbins in 1929 transformed its economics department in highly deductive and theoretical, thus far from the historicists, albeit still opposed to Marshall. 5 Buchanan (1969) argues that LSE developed a specific notion of cost, at the crossroad of different traditions. The first is made of these authors who progressively introduced the dimension of choice when speaking of cost, thus introducing a subjective dimension, but not taking into account its full consequences on the notion of opportunity cost and the differences that it could make with orthodox OOC: Wicksteed (1910, 380; 1914), who occasionally lectured at LSE in the 1910s, Davenport (1913, 60) and Knight (1921, 63; 1924, 592; 1934; 1935) in the United States and finally, at LSE, Robbins 5 On the history of the LSE, see the symposium published in the Atlantic Journal in 1982, articles and books by former professors, as Coase (1982), Hayek (1946) and Robbins (1971), or directors, as Beveridge (1960), Caine (1963) and Dahrendorf (1995).

4 4 (1934, 2), who recognized the influence of Wieser, Wicksteed, Davenport and Knight. The second tradition is Austrian, Robbins having invited Hayek at LSE in 1931 and having him appointed as Professor until Hayek (1937), following Mises (1933), gave the LSE tradition the subjectivist underlying methodological basis (Buchanan 1969, 2.16): data are individual perceptions unknowable by anyone else. All these ingredients precipitated thanks to Arnold Plant, Professor of Commerce and head of the new Department of Business Administration at LSE since 1930, who, as a former student of Edwin Cannan (the first chief teacher of economics at LSE), kept his commonsense approach (Coase 1982, 33). 7 Plant wanted to treat actual problems, specifically those really faced by entrepreneurs. His research group aimed at applying theoretical research in economics to the management of firms, and was composed, among others, of Ronald Coase, Ronald Fowler, Ronald Edwards and George Thirlby. Specifically, they were mainly working on introducing the opportunity cost concept into practical management. Coase (1938) synthesized their reflection on how this economic concept has to thoroughly modify the practice of accounting so that it can actually help the entrepreneur s decision. Buchanan commended him for his link between cost and choice, but regretted that he did not fully incorporate the subjectivist economics of Hayek and Mises into his analysis, nor did he draw the distinction between his concept and that embodied in neoclassical orthodoxy (Buchanan 1969, 2.31). Thirlby (1946a; 1946b) further integrated subjectivist economics and described a SOC: inspired by Wicksteed and Hayek, he asserted that the opportunity cost exists in the mind of the decision-maker and is ephemeral (1946a, 33 and 34). 8 This tradition, pursued by Thirlby (1952; 1960) and Wiseman (1953; 1957), helped to recognize that if cost is introduced in a logic of choice, it is obviously subjective (Buchanan 1969, 3.22), and developed a notion close to Buchanan s SOC. This tradition is the LSE tradition as Buchanan puts it; he and Thirlby edited its main essays in LSE Essays on Cost (Buchanan and Thirlby 1973). This LSE tradition did not last: The concept of opportunity cost which emerged from both the subjectivist-austrian and the common-sense approaches the concept that blossomed for two decades at LSE seems to have lost in its struggle for a place among the paradigms of modern economics (Buchanan 1969, 2.43). 9 Buchanan provides several reasons that could explain why this LSE tradition did not fully draw the implications of the subjectivist perspective in terms of cost, and specifically the differences that it made with OOC. First, as long as economists focused on individual decisions and interactions on the market, the distinction between subjective and objective opportunity cost was not so important (ibid., 2.19). Second, usual theoretical assumptions are the conditions under which SOC is equal to OOC, and to money outlays (choice in equilibrium, between monetary alternatives, no uncertainty, etc.) (ibid., 2.21). Third, the debate on socialism (to which Hayek, Mises and Robbins participated) focused attention on the impossibility of socialism due to problems of information rather than on the fundamental criticism based on subjectivism (ibid., 2.22 and 6.26). Finally, Buchanan (1973, ) 6 Coats (1982, 25) writes: During the 1920 s, Robbins arrived at Austrian economics via Gustav Cassel, Irving Fisher, and Frank A. Fetter, and, as in Plant's case, [via] meetings with Ludwig von Mises in Vienna On Robbins's invitation, Hayek gave a series of lectures at LSE early in 1931, and accepted a permanent chair later the same year. From that time, Robbins's seminar became the Vienna seminar of Menger and Mises So swift was the translation of the LSE ethos in economics that Plant was not surprised, on visiting Kiel in 1933, to find LSE described as a suburb of Vienna. 7 On Plant, see Coase (1986; 1987). Robbins was also a student of Cannan, but of course became more of a theorist. 8 Mises (1949) developed a conception of SOC close to this one (see Buchanan 1969, ). 9 Coase (1990, 8) seems to accept the genealogy that Buchanan (1969) suggests: The opportunity cost concept developed at LSE was of course derived from Knight and Wicksteed as expounded by Lionel Robbins and was also no doubt influenced by Hayek who would have added an Austrian flavour.

5 5 explains the failure of the LSE tradition to build a subjectivist school by the fact, on the one hand, that Mises and Hayek were too isolated at LSE and, on the other, that Plant s group was too focused on specific applications, while the economics department of LSE was increasingly devoted to high theory The exportation of the notion of cost to accounting (Coase 1938) Coase enrolled in a commerce degree at the London School of Economics in There, he was very much influenced by his professor Plant, who initiated him to economic reasoning and the benefits of a competitive system. Even when Coase was a teacher at the Dundee School of Economics ( ) and at the University of Liverpool ( ), his association with LSE never ceased (Coase 1982, 31) and he came back to teach at LSE from 1935 until 1951 (when he left for the United States). Though he was not directly working with economists, he was under Robbins influence, thanks to whom he had read in 1933 Wicksteed (1910) and Knight (1921), two books whose close study gave him and his colleagues such a firm hold on cost theory (Coase 1982, 33). Coase was working within Plant s research group on industrial organization. They were interested in the new developments of economic theory at LSE, but even more so in the use of these analyses to understand the operation of the real economic system (ibid., 33) 11, specifically in the application of the economic notion of cost to business problems, and its implications for accounting. Coase wrote a series of twelve short articles published in The Accountant in 1938, and titled Business Organization and the Accountant, which were the results of this collective work (with Edwards and Fowler in particular) 12 ; a reprint was published in Buchanan and Thirlby s collection in 1973, with minor modifications. Coase thought of these essays as just an exposition of views which were generally accepted by economists (Coase 1938 [1973], 5.2), or at least views which were the common property of the economists at LSE or, at any rate, of those with whom [he] was associated, such as Edwards, C. L. Paine and David Solomons (Coase 1990, 7). 13 This 1938 work aims at modifying accounting so that it can help business decisions. Entrepreneurs, Coase argues, have to know how costs and revenues will vary if a specific decision is taken (for example a variation of output), therefore business decisions depend on estimates of the future while cost accounts report past operations (Coase 1938 [1973], 5.4). Coase defines production cost as 10 For a biography of Coase, see Coase (1995) and Medema (1994). 11 Coase (1970, 114-5) describes the coexistence of two distinct groups at LSE (those of Robbins and Plant): Lerner was a member of that group which consisted of Robbins, Hayek, Hicks, Allen, Kaldor and others, who, at the London School of Economics, were making great strides at that time in the development of economic theory. I was, however, associated with another group of economists who, on the commerce side, were working under the leadership of Professor Arnold Plant. The Plant group was keenly interested in cost analysis and pricing problems,.. and so on, and therefore we paid great attention to Lerner's work [on marginal cost pricing] and to the other theoretical work being done on cost and pricing. But we thought of it in practical terms. We were influenced but not converted by work such as Lerner's. In our discussions we stressed the practical aspects of these theoretical developments. We treated them seriously, but we treated them as ideas to be applied in the real world. 12 On the collective work of Coase, Edwards and Fowler on accounting, see Coase (1990). They wrote two books together (1938; 1939) that used published accounts as a source of statistical information on firm behavior to show economists that the accounts provided valuable source material for economic research (Coase 1990, 5). This is the first aspect of their work, the second (to which pertains the 1938 series of article) being to persuade accountants to change their practices so as to make the accounts more valuable for this purpose (ibid., 5). 13 And Coase argues that this is these series of articles being a result of a collective work that gives them some importance: Perhaps because the outbreak of the war diverted economists from their academic studies, these articles came to represent the only extended statement in print of the approach to costs, particularly as applied to business problems, which was the common property of economists at L.S.E. in the 1930s. If Professor Buchanan's thesis about the special character of the L.S.E. approach to costs is correct, it is the fact that these articles do not represent a personal view which gives them their historical significance (Coase 1938 [1973], 5.2).

6 6 an avoidable cost: The businessman might produce nothing. If he produces some output certain additional costs will be incurred. These we may term the avoidable costs of that output because they can be avoided by not producing it (ibid., 5.6). It can be noted that this avoidable cost is an opportunity cost if the only alternative is to do nothing at all. The assessment of costs and revenues encounters three analytical difficulties (ibid., 5.10). First, the non-pecuniary elements of the decision (such as the ethical preferences of the businessman) are not reflected in accounts, whereas without a knowledge of the preferences of the businessman no decisions on questions of business policy can be reached (ibid., 5.11). Second, uncertainty is unavoidable, some events may not be assigned a probability (ibid., ); and businessmen differ in their risk-aversion, which is purely subjective (ibid., 5.15). Third, revenues and costs will be paid in the future, and therefore have to be actualized, subjectivity coming from the preceding elements: uncertainty and attitude to risk-taking (ibid., ). Consequently, the decision is unpredictable by an external observer, who, in other words, cannot measure the cost: Since no method of accounting can reproduce on paper the mental processes of a businessman, the decision to be taken is one which no mechanical process of discounting can disclose (ibid., 5.15). The subjectivity of cost here depends on uncertainty and non monetary elements of the producer s choice, which a thought in practical terms cannot neglect. 14 Coase then details the very nature of cost as lost opportunity: The notion of costs which will be used is that of 'opportunity' or 'alternative' cost. The cost of doing anything consists of the receipts which could have been obtained if that particular decision had not been taken This particular concept of costs would seem to be the only one which is of use in the solution of business problems, since it concentrates attention on the alternative courses of action which are open to the businessman (ibid., 5.16). Cost is here linked to the process of choice and differences with the accountant s vision of cost as expenditures (ibid., 5.17) are stressed through the consequences of this definition. This notion of cost first encompasses non-monetary elements (ibid., 5.16). Second, it implies, by definition, that the entrepreneur maximizes his profit: Costs will only be covered if [the businessman] chooses, out of the various courses of action which seem open to him, that one which maximizes his profits. To cover costs and to maximize profits are essentially two ways of expressing the same phenomenon (ibid., 5.16). Buchanan (1969, 2.27) makes this statement Coase s most significant contribution to the history of the construction of the subjectivist notion of cost: Any profit opportunity that is within the realm of possibility but which is rejected becomes a cost of undertaking the preferred course of action. Given its context, Coase s assertion can only refer to a subjective profit in the sense that the businessman chooses his subjectively preferred option among the several alternatives that he considers. Third, the opportunity cost concept is forward-looking : It is useless to look back at the past, except as an object lesson (Coase 1938 [1973], 5.17). Coase here means that the cost influencing the decision is the ex ante opportunity cost estimated before the decision is taken, not after: Of course one can say that one might have made a wiser decision and that in this sense costs were not covered. But to employ the term in this way does not seem to be 14 When Coase writes on accountants costs and the actual decision process of businessmen, he has no doubt in the back of his mind the results of his observations made during his trip in the United States in , during which he was investigating integration and costs functions (see Coase 1988b).

7 7 very helpful, for as Jevons reminded us, 'Bygones are forever bygones'. The only course which is open to a businessman is to make the best choice given the knowledge at his disposal, and in this task I hope to show that the concept of opportunity cost can be of considerable assistance (ibid., 5.17). Finally, Coase insists that costs are calculated in reference to a specific decision and cannot, therefore, be classified, as fixed and variable for example: One can discuss the meaning of the term 'avoidable costs' but what costs are avoidable and their actual measurement can only be determined with reference to a particular decision This linking of cost analysis to particular decisions makes any mechanical classification of costs almost impossible. The costs whose variations are of significance for one decision will be of no significance for others (ibid., , his emphasis). 15 The notion described by Coase is not exactly that of SOC developed by Buchanan, but we can draw from it two dimensions: the opportunity cost dimension, and the subjective dimension, both coming from a concern for realism in the analysis of producers decisions. 16 The subjective dimension can be found in other theoretical works on the producer s decisions written by Coase in the 1930 s, and particularly on the decisions regarding vertical integration and the price of a monopolist. 3. The choices of the businessman: applications of the thought on cost to the theory of the firm Make or buy: the decision of vertical integration (Coase 1937b) In two occurrences, Coase himself related his 1938 reflections on accounting to his approach of vertical integration, but both links he put forward are not based on the subjective dimension I wish to bring to light. The first link is to be found in Business organization and the accountant (Coase 1938), about which Coase will later assert that understanding cost accounting and opportunity costs within a firm was tied to understanding the organization of firms (Coase 1990, 3). The first draft of this article precisely dealt with the choice between make or buy, but Edwards complained bitterly that the accountants for whose benefit the articles were supposedly written would not understand what [he] was talking about, as the concepts and terminology [he] used would be completely foreign to them. Coase therefore decided to write an introductory section in which [he] explained the character of [his] approach. However, the introductory section came to occupy the whole of the twelve articles, and business problems were not discussed except as illustrations of the value of [his] approach (Coase 1990, 7). In this illustration with the vertical integration decision, Coase examines how an electricitysupply business owning a coal mine chooses the amount to produce itself, by comparing the (total and marginal) avoidable cost of producing to the cost of buying. He admittedly envisages these costs 15 The remaining of Coase s article consists in applying this cost concept to the businessman s decision of accepting a specific task, thus calculating the opportunity costs of materials, assets and capital: The 'opportunity' cost of using materials in stock we found to be either the price if sold minus the cost of selling, or the expense that would be avoided if the material were used on some other job. Depreciation considered as an 'opportunity' cost could be taken to be depreciation through use or the present value of the future profits lost through use. Interest on capital, if it is to be interpreted as 'opportunity' cost, must be regarded as the alternative net receipts that could be obtained by the use of the machinery (Coase 1938 [1973], 5.39). 16 For general studies on his views of the nature of economics, see Medema (1994, ch 6); Medema and Zerbe (1998); Mäki (1998b) and Wang (2003). Some specific issues have been raised: his realism (Mäki 1998a; 1998c; Pratten 2004; Bertrand 2012), his views on the roles of market and regulation (Medema and Samuels 1997; 1998; Pratten 2001; Campbell and Klaes 2005), or the tension between orthodoxy and heterodoxy (Foss 1994).

8 8 as avoidable costs, but they are measurable by an external observer (Coase 1938 [1973], ). The richness of the analysis regarding the link between cost and choice disappears, likely because it does not add anything essential to the understanding of the choice between make or buy. Coase linked a second time accounting and vertical integration, in Accounting and the Theory of the Firm (Coase 1990), a lecture addressed to accountants on the occasion of the fiftieth anniversary of the publication of The nature of the firm. In this lecture, Coase stresses that the theory of the accounting system is part of the theory of the firm (ibid., 12) in the following sense. When most of the production is operated within firms, most of the transactions arise inside firms and not on the market, therefore the institutional structure of production depends more on the relative costs of different firms in organizing particular activities than on transaction costs (ibid., 11). Now, these are the accounting systems of the different firms that (mis)calculate these (opportunity) costs of the factors and therefore determine the efficiency of the firms: The costs of organizing clearly depend on the efficiency of the accounting system (ibid., 11). Ultimately, the activities of the firms, and therefore the institutional structure of production, partly depend on accounting. Here again, the subjective dimension of cost emphasized by Coase in his 1938 article does not permeate his approach of vertical integration; this permeation appears in The nature of the firm. Coase s thought on vertical integration was nurtured by his investigation led in the United States during the year He found the explanation of integration by transaction costs in the summer of 1932 and wrote a draft of the article on the firm in Right from the start, Coase formulated the question of the existence of the firm in terms of a choice of the producer, as in this letter to Fowler written in May 1932: Assume a producer of some finished product finds he needs a special part. Then, he has two alternatives. One, to produce it himself and two, to let a supplier produce it (quoted in Coase 1988b, 15). This idea of choice appears also in the final article: We have to explain the basis on which, in practice, this choice between alternatives is effected (Coase 1937b, 389). And this way of thinking is confirmed in the 1988 lectures: I thought about the firm in terms of a choice of contractual arrangements (Coase 1988b, 29). However, Coase (1937b) does not insist on the sacrifice dimension of the choice as a renunciation to other alternatives; he insists on the contrary on the cost, somewhat real, of each institutional arrangement (firm or market). In other words, it does not seem that because there is choice, there is cost, but on the opposite that because there is cost then there is choice. Resorting to the firm is thus explained by the greater cost of its alternative: The main reason why it is profitable to establish a firm would seem to be that there is a cost of using the price mechanism (ibid., 390). But if the market alone were costly, there would be no choice; this implies that the firm is also costly: there are diminishing returns to management (ibid., 395). Coase s introduction of the notion of cost allows him to use standard economic tools, namely Marshallian tools of substitution at the margin (ibid., 386-7), in order to induce formal relations which are capable of being conceived exactly (Robbins 1932, 66 quoted by Coase 1937b, 387, Robbin s emphasis). As with factors of production, at the equilibrium, marginal costs of the market and of the firm are equal, which determine the size of firms (Coase 1937b, 404). And this equilibrium is optimal: Firms arise voluntarily because they represent a more efficient method of organising production. In a competitive system, there is an optimum amount of planning! (ibid., 389, fn 3) See also Coase (1992, 715-6). Another formulation is this one: This results in the institutional structure of production being that which minimizes total costs for the output produced (Coase 1988b, 39). The idea of substitution implies that, for

9 9 Thanks to the notion of cost, the whole of the structure of competitive industry becomes tractable by the ordinary technique of economic analysis (ibid., 398). It is in this sense that Coase was already writing to Fowler in October 1932 that he had certainly succeeded in linking up organization with cost (quoted in Coase 1988b, 4). 18 This straightforward formulation, used four times in his lectures on The Nature of the Firm (also pp. 17, 24 and 26), translates Coase s concern of thinking about a new subject organization in a way that could nevertheless involve traditional economic analysis. Coase s study of the choice between make and buy therefore refers to a notion of cost traditionally used by economists. The scientific meaning of the firm s size (1937b, 393), or the efficiency of the entrepreneur s decision, bring to light the orthodox dimension of Coasean analysis. However, as noted by Foss (1994), to this orthodox dimension of The Nature of the Firm is added a more heterodox dimension in which uncertainty plays a major role. Now, this heterodox dimension of Coase s work falls within the subjectivist tradition and it appears in Coase s discussion on the origin of transaction and organization costs. First, transaction costs result, at least partly, from uncertainty (Foss 1994, 47), which is a major source of the subjectivity of decisions. Coase (1937b, 391) writes, for example, that uncertainty makes impossible to completely specify a contract: Owing to the difficulty of forecasting, the longer the period of the contract is for the supply of the commodity or service, the less possible, and indeed, the less desirable it is for the person purchasing to specify what the other contracting party is expected to do. He infers that: It seems improbable that a firm would emerge without the existence of uncertainty (ibid., 392). Second, organization costs are also linked to the subjective dimension, on the one hand, of the entrepreneur s decisions, partly due to his cognitive limits 19, and, on the other hand, of the preferences of workers in favor of small firms: It is sometimes said that the supply price of organising ability increases as the size of the firm increases because men prefer to be the heads of small independent businesses rather than the heads of departments in a large business (ibid., 395, fn 1). 20 It seems, consequently, that if transaction and organization costs are considered as traditional costs by Coase, their origin nevertheless lies in the subjectivity of the decisions of entrepreneurs and workers in an uncertain context. The explanation of the existence of firms and of their limits therefore depends on the subjective dimension, even if this dimension is somewhat forgotten by Coase in his analysis of the decision between firm and market: the trade-off between costs of organization and costs of transaction is for its part assumed to be objective, efficient, predictable and subject to formalization. This is a new way of bringing to light the internal tension of this article, between orthodoxy and heterodoxy in Foss s language (1994), or between neoclassical and institutionalist elements in Medema (1996). Coase, the market and the firm play the same role of allocation of resources and that they are perfectly substitutable (cf. Coase 1937b, 392). 18 In saying that my concept of the firm was manageable, what I had in mind was that, looking at the firm in this way, we could analyze its activities using standard economic theory. This is what I meant when I said in the letter to Fowler in which I described my Dundee lecture that I had succeeded in linking up organization with cost (Coase 1988b, 24). 19 Coase (1937b, 394-5) writes: As the transactions which are organised increase, the entrepreneur fails to make the best use of the factors of production. 20 Coase (1988b, 32) later confirmed this argument: This rising supply price has nothing to do with the rising supply price of factors to an industry as normally thought of in economic theory, but relates to the fact that people working in a large firm may find the conditions of work less attractive than in a small firm and therefore will require a higher remuneration to compensate them for this.

10 10 The monopolist s choice: The subjectivity of entrepreneurial decision (Coase 1937a) The subjectivity of entrepreneurial decision and the uncertainty of its context are more obvious in other works written by Coase in the 1930s, as stressed by Arena (1999). First, in his article on duopoly, Coase (1935) explains that a duopolist, realizing that his competitor reacts to a modification of his price, has then to anticipate her reactions, but the decisions would be unpredictable except, perhaps, by a psychologist! (ibid., 139, fn 2), which in itself adds to the already unpredictable uncertainty. Second, the series of papers on the cycle of the pork price, written with Fowler (Coase and Fowler 1935; 1937; 1940) progressively introduces subjective elements to explain the expectations of pork producers. 21 Finally, it is in Some Notes on Monopoly Price that Coase (1937a) clarifies the subjective elements of the producer s decision process. In this article, he criticizes Robinson s (1933) theory that a monopoly, in order to maximize its profit, chooses the price and output combination that equalizes its marginal cost and marginal revenue. It is worth noting that this is Coase s concern for realism that leads him to question this still-accepted part of industrial economics: This paper aims at making monopoly analysis more useful by introducing certain of the more important modifications which have to be made if Mrs. Robinson's theory is to be of use in increasing our understanding of the working of the actual economic system (Coase 1937a, 17). Coase s argument, which intends to take into account realistic elements of the monopolist s decision, stands at different levels. First, the monopolist does not know that equating marginal cost and marginal revenue maximizes its profit. An evidence is that his accounting books are often prepared in a form which makes it very difficult to discover marginal costs 22 ; a surprising fact if business men thought that marginal costs were such an important element in the determination of prices (ibid., 18). The monopolist cannot either find the point of maximum profit by tâtonnement, due to complexity and uncertainty: he cannot determine if the variations of profit he goes trough are due to variations of price, of output, or of anything else (ibid., 18). He hence contents himself with a satisfying solution: So long, therefore, as the fact that there is a divergence between the price he is charging and the monopoly marginal cost equals marginal revenue price does not become too obvious, the actual price that is being charged is likely to be accepted as right or reasonable (ibid., 19). Second, were monopolists willing to equate marginal cost and marginal revenue, they would not have the knowledge necessary to enable them to apply this rule (ibid., 19). On the one hand, necessary information on costs will not be practically available since there will be considerable difficulties in calculating the cost curve for each product (ibid., 19). Here again, it is in any case doubtful whether the cost accounts are prepared in a form which will enable the marginal cost curve to be obtained (ibid., 19). On the other hand, demand curves are even more difficult to know in an uncertain environment, which brings in the picture the producer s risk-attitude (ibid., 22). 21 Their first article insists on the uncertainty of demand (Coase and Fowler 1935, 160-2) whereas the second, demonstrating that a determinate relationship between expectations and past and present prices and costs cannot be established, introduces learning, which depends on psychological factors (Coase and Fowler 1937, 79). The third article explicitly eliminates from calculus a specific subjective cost as too difficult to measure: the psychological cost of change to the producer (Coase and Fowler 1940, 281), and argues that expected prices depend on the expected costs of production factors and an expected normal margin profit, which itself depends, among other factors, on the producer s attitude to risk (ibid., 283-7). 22 This argument can also be found in Coase s criticism of actual cost accounting: The figures provided relate to average rather than marginal cost (Coase 1938 [1973], 5.56, his emphasis).

11 11 Third, Coase goes back to one of Hicks (1935) 23 argument that brings to the fore two non-monetary costs faced by the monopolist. The first is the subjective cost of increasing the output (of working more): the monopolist will be prepared to sacrifice some income if he can produce less or, alternatively, if he is to produce more and have an equal psychic income his money income must increase (Coase 1937a, 29). The second non monetary cost is a subjective cost of decision (of searching the price/output that maximizes profit) 24, which implies that the best of all monopoly profits is a quiet life (Hicks 1935, 8 quoted by Coase 1937a, 30). Consequently, at these different levels, Coase underlines the subjectivity of costs and of producers decisions. This subjectivity raises two issues. On the one hand, Coase s argument that the monopolist s profit is not maximized could seem inconsistent with his 1938 argument that a good understanding of decision implies that (subjective) profit is maximized. One has therefore to recognize that this is the monetary, objective profit, as defined by Robinson, which is not maximized by the monopolist. 25 The latter does not know that equating his marginal OOC to his marginal revenue will maximize his objective profit, he does not even know how to calculate his marginal OOC. He is rather making his decision in a radical uncertainty, the process of decision is costly and his decisions imply non monetary costs; in this sense, he is considering his SOC and, by definition of choice, his subjective profit is maximized. If this interpretation is right, it implies, on the other hand, that Coase argues that the marginal OOC is not measurable by the producer. However, OOC is conceptually observable and measurable. We have to insist, hence, that Coase here stresses a practical problem faced by the producer, a modification that, as he will detail later, must be made to the assumptions of simple monopoly theory if it is to be used to explain monopoly pricing in practice (1946a, 278, fn 3, my emphasis). It is a practical problem that he sometimes neglect to study other aspects of imperfect competition theory: in his articles on duopoly (Coase 1935), multi-products monopoly (Coase 1946a) or monopoly of a durable good (Coase 1972a), he uses the rule of the equalization of marginal cost to marginal revenue, thus developing more internal criticisms of imperfect competition theory. This practical difficulty of measuring marginal OOC should logically imply that a government cannot measure a marginal cost of a business and much less a social cost. Even more, that the producer takes into account non monetary elements in his decision in uncertainty makes it impossible for a government to perfectly determine the decisions of this producer by policies based on objective costs, be they individual or social. A subjective conception of cost entails not only the impossibility of applying policy rules that assume the existence and measure of objective costs, but also their inefficiency, as emphasized by Hayek, Mises or, in Coase s closer circle, Thirlby; and we could have expected, from Coase s part, a criticism of policies along these lines. His criticism, however, is based on another dimension of his thought on cost elaborated in the 1938 article, the opportunity cost dimension, and leaves aside the criticism based on the subjectivity of costs. 23 John Hicks was lecturer at LSE between 1928 and It is because Coase was given Hicks course on monopoly in 1935 that he became interested in this subject (Coase 1988b, 23). 24 Coase (1937a, 30) paraphrases Hicks (1935, 8) second argument that the variation in monopoly profit for some way on either side of the highest profit output may often be small ; and if this is so, the subjective costs involved in securing a close adaptation to the most profitable output may well outweigh the meagre [sic] gains offered. 25 This resolves the similar inconsistency that appears inside the 1938 article, in which Coase asserts: We may lay down as a general rule that it will pay to expand production so long as marginal revenue is expected to be greater than marginal cost and the avoidable costs of the total output less than the total receipts. It would be Utopian to imagine that a businessman, except by luck, could manage to attain this position of maximum profit (Coase 1938 [1973], 5.8-9).

12 12 4. The choices of the policy-maker: applications of the thought on cost to the criticisms of standard policies This section will focus on Coase s criticism of two standard policies: the marginal cost pricing of natural monopolies and the Pigovian policy that makes the polluter bear the social cost of an externality. Marginal cost pricing for natural monopolies (Coase 1946b) One of Coase s first criticisms of neoclassical remedies to market failures concerned natural monopoly: in conditions of decreasing average costs, the traditional solution consists in setting a price equal to marginal cost and compensating the firm s loss (the difference between marginal and average cost) by a subsidy financed out of taxation (Hotelling 1938; Lerner 1944 and Meade and Fleming 1944). In The Marginal Cost Controversy 26, Coase (1946b, 172) begins by reminding the reader that the pricing system has the advantage over the government, as a mode of resources allocation, of conveying information on preferences that a central planner cannot afford and, more often than not, at a lower cost. 27 He then details the features of an optimal system of prices: The price should be the one which equates supply and demand and it should be the same for all consumers and in all uses (ibid., 172). In these conditions, the price of the good reflects the value of the best alternative use of the resources employed to produce it, namely its opportunity cost. The preceding condition indeed implies that the amount paid for a product should be equal to the value of the factors used in its production in another use or to another user. But the value of the factors used in the production of a product in another use or to another user is the cost of the product. We thus arrive at the familiar but important conclusion that the amount paid for a product should be equal to its cost (ibid., 172-3). 28 Coase is here restating the usual definition of OOC as the value of the alternative product, value that, at equilibrium, is reflected in the price of the resources used to produce it, which allows measuring OOC by money outlays. Coase falls within the orthodox tradition of OOC, and even makes explicit that the cost faced by the producer is a cost for others: cost is not only the value that these resources could have for the producer in question if they were used in another manner, but also the value that they would have for another producer or consumer. 29 According to Coase, the price system is efficient 26 The arguments of this article are also present in Coase (1945; 1947; 1970). The last of these articles (Coase 1970, 114-5) precisely begins by mentioning the reflections on cost at LSE in the 1930s with Plant s group, which confirms that the two issues are tightly linked together. Coase asserts that their interest for multi-part pricing came from Plant, but that they missed Hotelling (1938) publication because of the war; this is therefore the publication of Meade and Fleming (1944) that initiated Coase s (1945) reflection on this issue. 27 No Government could distinguish in any detail between the varying tastes of individual consumers ; without a pricing system, a most useful guide to what consumers' preferences really are would be lacking; furthermore, although a pricing system puts additional marketing costs on to consumers and firms, these may in fact be less than the organising costs which would otherwise have to be incurred by the Government (Coase 1946b, 172). 28 The argument can be found again in 1990 (10), where Coase details the conditions of competition and zero transaction costs under which it holds: Leaving aside the effects of monopoly, the prices paid for resources must be equal or (slightly) greater than they would yield in another use or to another user, cost (the price of the resources) is opportunity cost, and resources will be employed in such a way as to maximize the value of production It assumes that the operation of the market is costless. 29 This falls within Buchanan s (1969, 3.19) description of the OOC notion: In the orthodox price-theory conception where cost is measured objectively by money outlays, it is helpful, for explanatory purposes, to equate these outlays to the values that members of society place on the alternate end products that might have been produced by the same outlays differently directed. In a certain ambiguous sense, therefore, cost here does reflect opportunities lost. But it is noteworthy that the opportunities lost in this context more accurately reflect the value of potential alternatives as judged by others rather than by the chooser himself (my emphasis).