Caveat Emptor: To and Fro Over the Coase Theorem in Economics and Law During the 1970s. Steven G. Medema* Version 1.8.

Transcription

1 Caveat Emptor: To and Fro Over the Coase Theorem in Economics and Law During the 1970s Steven G. Medema* Version 1.8 June 29, 2013 * Department of Economics, CB 181, University of Colorado Denver, Denver, CO , USA. steven.medema@ucdenver.edu. The research support provided by the Institute for New Economic Thinking and the National Endowment for the Humanities is gratefully acknowledged.

2 Introduction It is well established that Ronald Coase s article, The Problem of Social Cost (1960), and, in particular, the negotiation result that came to be known as the Coase theorem, played an important role in stimulating the development of an economic analysis of law a subject that evolved into both a way of doing legal analysis and a sub-field of the discipline of economics. Less well understood, however, are the ways in which this process played out during the formative period of this new approach to thinking about legal issues. The present paper attempts to shed some light on this history via two cases studies from the 1970s the symposium on Products Liability: Economic Analysis and the Law, published in the University of Chicago Law Review in 1970, and the controversy over the Coase theorem that played out in the Australian journal, The Economic Record, over the course of that decade. It is very easy to fall into the trap of assuming that the road from The Problem of Social Cost to the economic analysis of law was straightforward, involving little more than the generalization of Coase s discussion of crops and cattle, a handful of British common law cases, and Pigou s railway sparks example all foisted upon an unwilling legal profession by a group of market-oriented Chicagoans. But the reality is much more complex than this, as we have discussed at some length elsewhere (Medema 2013c). The fact is that the development of the economic analysis of law and the role of the Coase theorem in it was much more messy and much less Chicago-centric than the potted histories would suggest. Coase is now suitably famous in economics and law for having offered the insight that his soon-to-be colleague at Chicago, George Stigler, later christened the Coase theorem (Stigler 1966, p. 113). 1 Instancing a hypothetical situation in which a rancher s roaming cattle destroy a neighboring farmer s crops, Coase argued that, in a world characterized by costless transacting (a fundamental, if unstated, assumption of neoclassical economics at the time), it would not matter whether the rancher 1 This is by no means the only, or even the most important, insight to be found in Coase s article, as is now reasonably well accepted. But see Medema (1996; 2013e), Veljanovski (1977), and Coase (1988). 2

3 or the farmer was made liable for the damage, as any inefficiencies would be cured by bargains among parties to the externality: It is necessary to know whether the damaging business is liable or not for damage caused since without the establishment of this initial delimitation of rights there can be no market transactions to transfer and recombine them. But the ultimate result (which maximises the value of production) is independent of the legal position if the pricing system is assumed to work without cost. (1960, p. 8) Coase went on to apply this insight to the analysis of several legal cases before moving on to a lengthy discussion of how best to analyze problems of potential market failure in a world in which the costs of transacting are significant and, as likely would be the case, prohibitive. In spite of the relatively minor role that Coase s negotiation result played in The Problem of Social Cost, economists and others latched onto it almost from the start, 2 giving it a measure of realworld credence that seemed to go beyond what Coase had contemplated when penning his own analysis. The almost wholly uncritical acceptance of Coase s result during the 1960s, though, quickly gave way in the 1970s, as economists began to debate both validity and applicability. The two case studies taken up in the present paper reflect two such episodes and are illustrative of the issues that emerged in attempts to use the Coase theorem as a basis for legal reasoning. The cases are also indicative of a more general set of issues that pervade early scholarship on the Coase theorem those of coming to grips with (or working out) what the theorem said and did not say, the content given to its underlying assumptions, and determining its potential range of real-world applicability. The next section of the paper discusses the role of the Coase theorem in the University of Chicago Law Review symposium on products liability from 1970 and the varying attitudes toward the theorem found among the economists and legal scholars participating in the conference from which it emanated. The third section takes up the controversy over the theorem that played out in The Economic Record over the course of the 1970s a controversy that turned on the incentives for 2 See Medema (2013f; 2013g; 2013c). 3

4 blackmail provided by alternative legal rules in nuisance situations. The paper concludes with some thoughts about what these cases tell us about this important period in Coase theorem history. Case I: The Coase Theorem as a Potentially Defective Product In its Autumn issue for 1970, the University of Chicago Law Review published a symposium on the subject, Products Liability: Economic Analysis and the Law. This symposium was the product of a conference held at Stanford University in March 1969, organized under the auspices of the Joint Committee of the Association of American Law Schools and the American Economic Association. This committee was established to explore the possibility of cultivating increased interaction between economics and law, and the 1969 conference was the first significant effort undertaken by the committee. 3 The conference featured a major paper, prepared by Roland McKean (1970) of the University of Virginia, that examined the history of and current issues in products liability law from the perspective of economics. Four prominent economists and legal scholars James Buchanan (Virginia Polytechnic Institute), Robert Dorfman (Harvard), Guido Calabresi (Yale), and Grant Gilmore (Chicago) were enlisted to provide responses to McKean s analysis, and these presentations were followed by a lengthy discussion among the panelists and the audience, an edited transcript of which was published in the Review along with the five papers. There are any number of features of this symposium that stand out, but the one which will concern us here is the role of the Coase theorem in the various commentaries. Coase s negotiation result which at this time was only beginning to be known by the name hung on it by Stigler was still a very new element of the legal vocabulary and in its relative infancy in the economics literature as well. This result had slowly begun to work its way into the legal literature in the mid-1960s at the hands of scholars associated with the University of Chicago and, especially, Yale, 3 The Joint Committee and its work represent an important moment in the history of the economic analysis of law. See Medema (2013a) for a discussion. The present discussion focuses on one small slice of that history. 4

5 but it was far from the legal-economic phenomenon that it would become in the ensuing years. 4 Yet, it became the axis around which much of the discussion in the products liability symposium revolved thanks to the pivotal role that it played in McKean s analysis. The Coase Theorem and the Case for Caveat Emptor McKean had been trained at Chicago and spent more than a decade at RAND and at UCLA before taking up his position at Virginia in His paper for the symposium, however, was his first foray into the economic analysis of law. McKean, quite sensibly, introduced his lengthy analysis of the state of products liability law and the potential future directions for it with a brief tutorial in economics for the legal scholar not trained in the discipline. Perhaps the most noteworthy aspect of this discussion, for present purposes, was his presentation of exchange and efficiency. The prospect of gains from exchange, he pointed out, provides an incentive for individuals to come together to work out a deal that will allow these potential gains to be realized. If the gains from exchange exceed the associated transaction costs, he said, the parties will bargain to an efficient outcome. While this was a standard piece of welfare economics theorizing, the same cannot be said of McKean s further conclusion that, where the costs of negotiation exceed the gains from exchange, the status quo may be the most efficient attainable solution (1970, pp ). The generally accepted conclusion from the (Pigovian) welfare economics of the day was that if the costs of exchange exceed the associated gains, we are confronted with a situation of market failure and its attendant inefficiencies, the appropriate remedy for which is some form of government intervention. The idea that such a market outcome could be deemed efficient was a relatively recent insight, one that owed to Coase and, especially, Harold Demsetz (1968). The relation of these principles to the law of products liability, according to McKean, came through Coase s analysis in The Problem of Social Cost. Specifically, he said, Coase had shown that, 4 See Medema (2013c) for a discussion of the treatment of Coase s negotiation result in the legal literature in the 1960s. 5

6 If there were zero transaction costs, and if people eschewed the use of coercion or interference with voluntary exchanges, it would not matter, as far as resource allocation is concerned, how the legal rights or liabilities were initially assigned, that is, whether the purchaser or the manufacturer was liable for injuries and damages. (1970, p. 30) McKean stressed that this result might appear counter-intuitive that a decision to impose costs on one party or another would impact the allocation of resources and he was cognizant of the controversy regarding this very issue that had been bubbling around Coase s result. But McKean sided with those who supported the theoretical validity of Coase s result and informed his readers that There now seems to be general agreement that Coase is right if one accepts the basic assumptions (1970, p. 31). McKean allowed that these assumptions are fairly heroic, but he believed that Coase s negotiation result offered a useful framework for simplifying one s thinking about products liability, and he defended the practice which he considered a standard one in economics of attempting to understand over-simplified situations before attempting to wrestle with the more complex ones (p. 31). To illustrate the utility of the Coase theorem framework, McKean referenced the very products liability-relevant concepts of risk-bearing and injury prevention. Both, he asserted, can be conceptualized as outputs, and Coase s analysis informs us that these outputs can be analyzed within a market framework with attendant efficiency implications just as can traditional good and services (pp ). In each case, agents have an incentive to negotiate an efficient placement of the output. And If transaction costs are zero, McKean said, voluntary exchange could lead to an efficient point, depending upon the initial wealth distribution (p. 34). Although McKean considered the assumption of zero transaction costs fairly heroic, he argued that existing product markets, such as between automobile manufacturers and automobile purchasers, may closely approximate these scenarios, thereby making the Coase theorem s insights relevant for the analysis of such contexts. In other situations though, he noted, transaction costs will be positive and significant even potentially prohibitive meaning that the assignment of liability will indeed have efficiency consequences (p. 34). 6

7 When McKean turned his attention to the specific options for structuring products liability law, the Coase theorem remained at center stage. While the fashionable trend in the products liability literature at the time was in the direction of strict manufacturer liability, McKean began his discussion with its antithesis: complete caveat emptor, a system which would make consumers liable for all harm incurred while using a product. Caveat emptor, he noted, would generate an efficient allocation of resources in a world of zero transaction costs, 5 in that each output, including product safety, would be produced by those having a comparative advantage in that activity: If manufacturers had the comparative advantage in safety, consumers would bribe them to produce safer products, and if the comparative advantage lay with consumers, they would take the precautionary steps necessary to efficiently minimize the potential for harm. An identical logic was applied to third parties e.g., innocent bystanders who might be harmed as a result of the use of the product, where, McKean argued, If buyers of items were liable, they would modify their product selection and hire third parties to be careful as long as the gains outweighed the sacrifices (p. 43). In short, the logic behind McKean s defense of caveat emptor here was straightforward Coase theorem stuff. The end result, McKean pointed out, would be efficiency all round: bargaining would lead to efficiency in producing safety features, warnings and instructions to users, warnings to bystanders, care in using products, caution in standing or walking nearby, and so on. This result would obtain even if bystanders were made liable for their own injuries, since, if there were no transaction costs, the third parties could costlessly get together and hire manufacturers to make safer products or users to handle them more carefully. Whoever could prevent accidents at the lowest cost would again be hired to produce the accident prevention (p. 43). In short, caveat emptor would guarantee an efficient outcome in a world of zero transaction costs, though, as McKean realized, some would 5 McKean defined zero transaction costs as a situation in which there are zero costs of negotiating and enforcing agreements, and zero costs of information about products characteristics. Such a world, he said, is not one of complete certainty in this analysis, for such a world would have no accidents, unforeseen events, or doubts about liability (1970, p. 43n.108). 7

8 object on fairness grounds to the very idea of making customers liable for product related harms, efficiency or no (p. 43n.109). McKean acknowledged that, in the real world, one must allow for the existence of transaction costs. But his sense was that customer liability would make certain transaction costs relatively low, the reason being that the existence of established market relationships for products facilitates consumers registration of their preferences for safety features, safety information, and the like in the marketplace via their willingness to pay for these things (p. 44). If consumer preferences on this score are frustrated, they will substitute other versions of the product or accept the higher risk only in return for a lower price. In a competitive system, market forces will respond and the result will be the creation of more products of the type that consumers prefer. Though McKean was quick to note that Coase-theorem-like processes could not be expected to operate smoothly in all instances under a system of complete caveat emptor, he believed that the alternatives various possible forms of producer liability, ranging from fault-based to strict liability were still less efficient. The Coase theorem, of course, tells us that efficiency will result under these producer liability schemes as well, so long as the costs of transacting are zero, and McKean said as much (p. 51). In the real world, however, he believed that the effect of these systems was to reduce the incentives for consumers to take precaution and to negotiate with producers to achieve safer products, the stimulation of significant amounts of litigation (due in large part to the liability insurance system that would result and the attendant incentive for insurance companies to litigate insurance claims), increased goods prices, and a smaller range of product options in the long run (pp ). Worse still, McKean said, would be a government compensation scheme, which, along with many of the aforementioned problems, would involve significant bureaucratic costs (p. 52). While the foregoing discussion reflects only a small slice of McKean s much larger commentary, it makes clear his sentiment that economics had much to offer to the law in the way of thinking about the consequences of alternative legal rules, and the Coase theorem was a useful tool for shedding light on these consequences. But it was more than just this; the theorem was also 8

9 suggestive of possibilities for the market or market-like systems that could manifest themselves in the real world under appropriate conditions conditions the existence of which McKean believed the legal system could help facilitate. Contested Ground: The Responses to McKean s Analysis When it came time for the respondents to offer their opinions of McKean s analysis, the most remarked upon aspect of it, aside from his more general advocacy of caveat emptor, was the Coase theorem and McKean s use of it. Indeed, it is fair to say that the remarks of the four designated commentators and the audience discussion that followed turned into something of a referendum on the Coase theorem. The first respondent, Coase s former University of Virginia colleague James Buchanan (1970), 6 defended caveat emptor more vociferously than had McKean, who at least had hedged his bets somewhat on the normative front. Unlike McKean, however, argued that caveat emptor was more efficient than other systems even in a world of zero transaction costs 7 a sentiment that would seem to run afoul of the Coase theorem. His response, though, was not to criticize the validity of the theorem but, instead, to suggest that the theorem may not be applicable to situations in which two different qualities of a product are being marketed. The Coase theorem will only be fully applicable, he said, if it includes the assumption that there be no prohibitions on any mutually advantageous exchanges that may be made as between potential buyers and potential sellers. Strict manufacturer liability, though, violates this condition by imposing just such a prohibition: Potential buyers of the low quality product are effectively prevented from buying risk, making the transaction costs associated with such exchanges prohibitively high (1970, p. 69). The significant generality that he had previously ascribed to Coase s result, 8 then, hit a roadblock when it came to products liability, forcing him to (depending on one s perspective) modify/supplement/redefine the 6 Coase was on the Virginia faculty from , a period that includes his writing of The Problem of Social Cost. Buchanan was a member of the University of Virginia faculty until 1968, at which time he departed for UCLA. He moved from UCLA to VPI in The Coase theorem, of course, suggests that there would be no differential efficiency effects across alternative liability systems in such a world. 8 See Medema (2013f) for a discussion. 9

10 assumptions underlying Coase s result in order to justify his defense of caveat emptor. This, then, suggested a potential chink in the Coase theorem armor and that there could be some issues when it came to mapping economic concepts onto law in straightforward fashion. The first respondent from the legal side, Guido Calabresi of the Yale Law School, came to this issue from an interesting place. 9 He was trained in economics at Yale and Oxford before completing his law degree at Yale, had been at the forefront of the (at that point relatively sight) efforts to infuse law with economic thinking during the 1960s, and had brought Coase s analysis into his own work at several points in his various discussions of accident law during this period. 10 Perhaps not surprisingly, then, Calabresi was very much on board with McKean s analytical framework. Indeed, he pointed out that his own work, like McKean s, took as its starting point for the analysis of liability issues the Coase theorem and the attendant notion that outcomes would be invariant under alternative legal rules if transaction costs were zero. 11 Calabresi proclaimed Coase s result crucial and argued that it deserves constant attention when contemplating issues in accident law (Calabresi and Bass III 1970, p. 77). But he then quickly distanced himself from McKean, emphasizing that a primary reason for the Coase theorem s importance lies in its converse that when transaction costs are positive, liability decisions impact efficiency. The normative implication of this, for Calabresi, was that the existence of transaction costs makes it imperative that liability be placed upon the leastcoast avoider if efficiency is the goal (p. 77). The problem with McKean s analysis, Calabresi argued, is that the least-cost avoider criterion does not point to caveat emptor because consumers tend not to be the least-cost avoiders of accidents related to the use of products, and transaction costs will get in the way of reallocations that would generate the cost-minimizing result. The third respondent, Harvard economist Robert Dorfman, freely professed his ignorance of the law of and legal issues related to products liability and so resolved to devote [him]self entirely to the economic question (1970, p. 94). In doing so, however, Dorfman noted that it will become 9 Calabresi s contribution to the symposium was co-authored with Kenneth Bass. However, we speak in terms of Calabresi alone here because the perspective on display is unmistakably his own. 10 See, e.g., Calabresi (1961; 1965a; 1965b; 1968; 1970). Medema (2013d) provides a discussion of Calabresi s various engagements with the Coase theorem throughout his career. 11 See, for example, Calabresi (1970). 10

11 amply evident in the next few minutes that Professor McKean and I adhere to different schools of economics, and that these differences are fundamental and are at bottom methodological (p. 94). The differences, for Dorfman, were brought to the fore in his and McKean s respective attitudes toward the application of the Coase theorem. McKean, he said, believes that you can get to the economic essence of a problem by simplifying it in a particular way. I believe in simplification, too, because I recognize that the human mind can deal with only very simple problems, but I am very suspicious of the particular kind of simplification that Professor McKean regards as being most illuminating (p. 94). Dorfman s suspicion went to the potential problems that would arise from trying to lay the Coase theorem framework onto real-world situations. The frictionless world of the theorem, he said, is a very useful conceptual device for some economic problems, such as analyzing the consequences that may result from the development of a less costly source of power (p. 94, emphasis added). But for other purposes, he argued, this device can be grossly distorting : For example, it would not be at all helpful to make an economic analysis of the insurance industry in a world in which all sources of ignorance and uncertainty have been assumed away, for the need for insurance would be assumed away also. So, I tend to regard this methodological device with some suspicion. And, in particular, it seems to me that the analysis of products liability is dangerously akin to the analysis of insurance. (p. 94) In short, abstraction puts one on very shaky ground when the things from which the analyst is abstracting go to the heart of the issue being studied. The problems here, for Dorfman, were put on full display when one pursued McKean s assumptions to their logical conclusions (p. 94) something that he felt McKean had failed to do. The assumptions placed before us, he said, are that the costs of information, of transacting, and of enforcement are zero. Given this, he continued, I feel obliged to draw a conclusion from them namely, that in such a world, the problem of products liability would hardly exist (p. 95, emphasis added). Manufacturing defects would not exist at all in a world of free information, since the existence and nature of the defect would be known to everyone from producers to end users. If such 11

12 goods were sold at all, they would be sold as defective products, and at a discount, and there thus would be no associated liability issues. A related logic would apply to design defects in such a context. If you knowingly purchase a product that is defective, and at a price that fully accounts for this, you get to live with the consequences (pp ). The costless information environment that McKean had assumed, then, would put us in a world in which, as Dorfman put it, the significance and occurrence of products liability would be so wildly different from what they are in the real world that the assumptions are hardly a first approximation to the actual problem that confronts us at this meeting (p. 98). 12 Because issues of products liability arise out of a context of costly and thus inadequate information, said Dorfman, We cannot abstract from these costs without changing the problem in an essential manner (1970, p. 98, emphasis added). 13 The relevant question then, said Dorfman, is what are the economic consequences of products liability in a world of imperfect markets (p. 101), and for such analysis he saw the Coase theorem as of little use. The final commentator, Grant Gilmore (1970) of the University of Chicago Law School, 14 came to the subject from a background in commercial law, the law of contracts, and admiralty law. And, like each of the other commentators, Gilmore latched onto McKean s repeated invocations of the Coase theorem, an idea with which he was certainly familiar from his time at Chicago: I take it that everyone even, after some initial hesitations, Professor Calabresi 15 accepts as true the theorem ascribed to my colleague, Professor Coase (p. 105). Gilmore seemed to have no truck with the Coase theorem per se; rather, he, like Dorfman, took issue with McKean s simplicity-based defense of the theorem as a relevant starting point for legal-economic analysis (1970, p. 105). Not only are there always transaction costs, said Gilmore, it will generally be impossible to predict how large they may be in any given situation or to ascertain how large the were even after the fact. Moreover, he said, truly voluntary that is, non-coerced exchanges have become almost an 12 Ralph d Arge (1973) made a similar argument related to externalities in general three years later, suggesting that externalities would not exist in the first place in a Coase-theorem world. See Medema (2013b) for discussion. 13 Dorfman also took McKean to task for failing to follow through on a second of his assumptions that markets are perfectly competitive (1970, p. 98). 14 Gilmore had previously moved to Chicago from Yale Law School, and he returned to Yale in the early 1970s. 15 This is a reference to Calabresi s 1965 critique of the correctness of Coase s result, which he retracted a few years later. See Calabresi (1968). 12

13 extinct species in the all too real world (p. 105). The issue that this raises, he said, is that we cannot move in straightforward fashion from the simplified world of the Coase theorem to the complexity of reality. While such a move might be possible if the real world was recognizably of the same family as the simple model, one cannot make such a leap if, as he believed to be the case, there is simply no relationship that [can] be perceived between the real world and the model. Conclusions drawn for the real world based upon the simple model will either be false or will be true only by accident (pp ). All of this, for Gilmore, pointed to the dangers for law inherent in setting law s course based on the Coase theorem: Insofar as we put any store by or allow our thinking to be influenced by the Coase theorem, it predisposes us (emotionally or, perhaps, spiritually) to feel that the secular movement which has seen a notable expansion of the liability of manufacturers toward the users of their goods has been a great waste of everyone s time. If we make certain assumptions (the assumptions do not hold true in the real world but let s make them anyway), we can prove that it makes no difference, from an economic point of view, whether manufacturer or purchaser bears the liability. (p. 106) The danger that Gilmore saw in the use of the abstractions of economics was that our theoretical demonstration conditions our thinking about the real world (p. 106). 16 The ultimate counsel of McKean s Coase theorem-inspired analysis, for Gilmore that products liability should cease to evolve further in the direction of producer liability, as it at least does no good to proceed further in that direction was emblematic of exactly this (p. 106). At the heart of the matter, for Gilmore, lay serious doubts that abstract analysis of the Coasetheorem type gives us, or is meant to give us, guidance in handling real problems in the real world (p. 116). Given the wide gulf that he saw between the world of the Coase theorem ( in which there are no transaction costs and all exchanges are voluntary ) and the world in which we live (where there are always transaction costs and few, if any, exchanges are voluntary ), Gilmore was 16 On this propensity to view the world through the lens of one s model, see Morgan (2012). 13

14 not willing to accept the conclusion that the evolution of products liability law should cease, or that it would be a good thing to turn back the clock (p. 107). More importantly for our purposes, however, he was not convinced that economics, at least of the Coase theorem variety, had anything significant to add to law. Debating Nirvana The appropriateness of grounding the analysis of products liability in the Coase theorem continued to occupy center stage in the ensuing discussion among the conference participants. 17 Though McKean opened this discussion by noting that the commentators had read too much into his invocation of the Coase theorem, he nonetheless continued to defend the it as an important point of departure from which to go forward into the real world. While the theorem is far from having assumptions that correspond to the real world, he said, it is terribly important in thinking about liability or property rights assignments and things of that sort (Manne 1970, p. 118, 119). 18 Dorfman and Gilmore, though, remained dissatisfied with McKean s defense of the utility of the Coase theorem, even as a starting point for thinking about legal issues, and saw the theorem as simply one symptom of a much larger issue with applying economic analysis to legal problems. As Dorfman put it, The central issue of the whole discussion goes in part to the way that economists on the one hand and lawyers on the other hand seem to look at the problem. We are faced with the difficulty that economics is like the law of gravity in this sense it works best in a vacuum. The law of gravity works just fine in dealing with objects in orbit way out there. It is not helpful at all in predicting more terrestrial projectories. The problem both for ballistics experts and for economists is that we really are not concerned with problems in a vacuum. (p. 122) 17 It should be noted that the published version of the discussion does not constitute a complete record of the conversations that took place during this part of the conference. A search is underway for the original tape recordings and/or an unedited transcript. 18 Unless otherwise noted, all page number references in this sub-section are to Manne (1970). 14

15 Instead, Dorfman continued, economic policy often designed for no other purpose than to correct real-world imperfections. If one assumes away the very frictions that cause these imperfections, it would be difficult to conclude otherwise than that government policy actions will be at best ineffective. But such a world is not the world in which we live, based upon which he concluded that, the split in our camp is a difference in judgment as to how far we think the inspection of perfect markets can take you in dealing with our specific issue (pp. 122, 123). While McKean continued to insist on the utility of the Coase theorem as a point of departure for the analysis of liability rules, George Stigler went further than this in defending his intellectual progeny: I should like to start by complaining against the anti-theoretical attitude of all of the speakers. They are playing a game, popular in fields like labor-economics, but which I hope would not get a foothold here, of erecting a very formal theoretical model and then saying once you depart from it a couple of inches nothing is left. The theory has enormous real vitality (p. 124) The theorem tells us, Stigler said, that agents will exploit all potential gains from trade regardless of the initial positions. This is true whether trade is costless (as in the theorem s zero transaction costs world) or costly. The end result will be efficient because all cost-effective moves will have been undertaken. Thus, while the theorem proper assumes zero transaction costs, There is no sense in which the theory loses its validity as you move away from that extreme (p. 124). This insight, Stigler argued, holds equally well for the validity of the Coase theorem in the realm of products liability: [A]lthough everybody says that once we put transaction costs in, we are playing in a different ballpark that is not so. There are very strong incentives in the economic system to minimize transaction costs, to make the appropriate amount of transactions and not to embark upon unnecessary ones. But one of the clever things the economic system does is not engage in those transactions which are never going to occur. It is not expensive to have one unsafe car identified by having it hit somebody and then have the appropriate redress by 15

16 legal action. So an immense number of transactions never have to be engaged in. When people say that the theory no longer holds once we have departed from the zero transaction case, they are making what I think is an unfounded conjecture. (p. 124) Harold Demsetz, himself no stranger to the idea of applying the Coase theorem s insights to the world of positive transaction costs, 19 took a slightly different approach in defense of the theorem: There is a lot of naivete about what happens when the courts decide that one person or another is liable. The Coase theorem makes you reconsider simplistic notions about what the effect of the law is going to be. I think that it is a tribute to the Coase theorem that everybody here is now aware of the fact that, until you start playing this game of transaction costs, the law may not have any effect at all. I do not think that people realized that until the Coase proposition was launched. (pp ) For Demsetz, the Coase theorem established something of a baseline for thinking about the impact of alternative legal rules, informing us that, in a frictionless system, the allocative impact difference is nil. The operative question and the one that preoccupied Demsetz during this time is what, if any, significant impact arises as we move away from these conditions. In many cases, Demsetz suggested, the answer is, not a lot. For virtually all of the roughly twenty participants in this symposium not associated with The Chicago economics tradition, though, it seems that the answer was, a great deal, for much critical discussion was devoted to problems of transaction costs what they are and the ambiguity in the term, information problems, the barriers that all of this poses to reliance on Coase theorem-type mechanisms for real-world legal problems, and so forth. The implications drawn went both to the relevance of the theorem for legal thinking and the implications for caveat emptor as presented by McKean and defended by those supporting his stance. Although there was some sentiment expressed, even among certain of the participants suspicious of the Coase theorem, that there are circumstances when the direction of liability becomes irrelevant, the clear sense of the discussion was that the relevance of the Coase theorem to legal 19 See, e.g., Demsetz (1964; 1966; 1967; 1968). 16

17 analysis would be at best sporadic and that it was not an appropriate starting point for or guide to the analysis of the impact of alternative systems of liability. Indeed, to the extent that the theorem was taken as an emblem of what economics could add to legal thinking, the focus on it may have detracted from the larger enterprise of bringing economic analysis to bear on law. Case II: The Coase Theorem Down Under While the products liability symposium was organized to promote the application of economic analysis to legal questions and became part of a slowly growing body of literature suggestive of the Coase theorem s potential relevance for law, we must bear in mind that the context in which Coase s negotiation result was initially developed was the economic theory of externalities, an area in which the Pigovian approach had held sway for roughly a half-century and to which the theorem posed a fundamental challenge. It was not until the 1970s, though, that the Pigovian backlash against the Coase theorem and the larger Chicago-Virginia critique of the Pigovian system began in earnest, both among welfare economists and among those working in the field emerging field of environmental economics. This backlash, in turn, spawned a number of debates over the theorem s validity between Coase theorem critics and supporters, one of the first of which was instigated by a newly minted PhD from the University of Sydney, Yew-Kwang Ng, who at that time held a position at the University of New England. Ng s contribution, which was published in The Economic Record in 1971, set off a debate which ran on and off in that journal for the next six years. Ng s Challenge: Legal Rules and Future Decisions The 1960s, in Ng s estimation, had seen the Pigovian approach vigorously attacked by Coase and others. These attacks were not without some validity, he said, but also not without their shortcomings and excesses (Ng 1971, p. 169). Ng s attempt to rehabilitate the Pigovian approach was made on multiple fronts, but the aspect that concerns us here is his response to two recent developments in the concepts of externality : 17

18 1. The problem of externality is a two-way one. To avoid harm to B would inflict harm on A: in general it does not matter who has to compensate whom in the determination of the optimum output. 2. The classical tax-subsidy policy is not the only way of solving the problem of externality: central directives, mergers, or bargaining may also be feasible or even better than the classical treatment. (1971, p. 169, quoting Mishan (1967, p. 29)) The first of these developments identified by Ng, of course, presents us with one version of Coase s negotiation result. The second is suggestive of the notion that this result was considered, by some at least, to offer an alternative and perhaps a superior one to traditional Pigovian tax/subsidy schemes. Though Ng s characterization of these two developments did not make explicit the assumption of zero transaction costs underlying Coase s result, he was well aware that the presence of such costs may affect allocational efficiency. Because of this, he intentionally assumed away both these costs and relevant institutional obstacles to negotiated solutions for the purposes of his analysis (1971, p. 169n.3). Having done so, Ng was only too willing to grant the validity of Coase s result in the static contexts contemplated by Coase and most of the early commentators on his result. But when we situate the analysis in a dynamic context that allows one to consider the impact of legal rules on future decisions, he said, the character of the problem is altered in ways that cause the assignment of liability to take on allocational significance in short, in ways that invalidate the Coase theorem. To illustrate his point, Ng constructed a simple hypothetical, one more simple, in fact, than the crops and cattle illustration utilized by Coase: Imagine that a factory is already in existence in a certain location and that the working of this factory necessarily emits smoke. Now consider A who is going to build a house by the side of the factory. Suppose the cost of the house is $10, the benefit without the smoke is $12, and the smoke nuisance reduces the benefit to $7. [T]he social net benefit of the house is 18

19 clearly negative. The cost of the house is $10, while the benefit of the house is only $7. Hence the net benefit is -$3. (p. 171) Efficiency thus dictates that the house not be built, and if homeowners are liable for damages this is precisely the result that will obtain. The net private benefit to the homeowner, -$3, is identical to the net social benefit, meaning that costs are efficiently internalized. If the factory is liable for damages, however, the house will be built. Because the homeowner would be compensated by the factory for damages incurred, he will realize the full net private benefit of $2 ( ) associated with the house. As such, he has no incentive to refrain from building. Thus, said Ng, the law that requires compensation from the damaging party may not achieve optimum output (p. 171). But this potential inefficiency accompanying polluter liability was not, for Ng, the end of the story. The problem is compounded, he said, by the possibility of blackmailing behavior, since even if A does not have the intention of building a house, he can still threaten B that he is going to do so unless B agrees to pay him, say, $4 (p. 171). As Ng pointed out, the possibility of blackmail had been raised previously in the literature evaluating Coase s result. Stanislaw Wellisz had argued in a 1964 critique that the Coasean bargaining situation opens up magnificent business prospects: any activity can be turned to profit as long as it is sufficiently annoying to someone else (1964, p. 353). But while Wellisz was concerned about blackmail by, e.g., polluters, Ng pointed out that the blackmail issue cuts both ways that either party may have an incentive to blackmail the other (1971, p. 171). And because blackmail is likely to absorb some resources (p. 170), allocation will be impacted by the direction of liability assignment, contrary to what Coase had claimed. 20 The conclusions that Ng drew from this analysis were two. First, the assignment of liability does indeed affect resource allocation when one considers the impact on future decisions, to use Ng s terminology. Second, efficiency can only be assured if the party who was present first, so to speak, is the one designated to receive compensation, for it is only under such a scheme that the party who comes to the harm (or threatens to do so) has all relevant costs internalized to it and so faces the 20 It bears noting that Coase had ruled out such rent-seeking actions when he first stated the negotiation result in The Federal Communications Commission (1959, p. 27n.54), the paper that was the forerunner to The Problem of Social Cost. See also Medema (1997). 19

20 incentives necessary to make proper (that is, efficient) decisions (p. 183). Though Ng was convinced that efficiency considerations point to the first party having prior claim to receive compensation, he was not arguing that economic considerations should dominate legal decisions. Rather, he allowed that other factors such as justice and provisions of the common law also have to be taken into consideration in the legal decision-making process. And because economists have little to say about such matters, he did not head down that road (p. 171). 21 Neither Efficiency nor Equity: Walsh Against Ng and Coase Ng s analysis attracted the attention of Cliff Walsh, then a young Lecturer in Economics at Australia National University. Walsh, whose interests lay in the area of public economics, was well aware of the recent surge of interest in the theory of efficient internalization of externalities and of the discussion of Coase s result within this literature, a phenomenon that he attributed to recent public concern over the problem of pollution and the quality of the environment in general (1972, p. 254). But he was also of the mind that Ng s analysis missed its mark, and his critique appeared in The Economic Record for Walsh s attack was two-pronged. First, he utilized a numerical example parallel to Ng s to demonstrate that Ng s first-party priority rule could easily lead to an inefficient outcome (pp ). In doing so, he joined Ng is arguing the falsity of the Coase theorem; their disagreement was over the structure of legal rules to generate efficient resolutions of externality problems. 22 More interesting for our purposes is a second issue raised by Walsh that of distribution. Ng, as we have already seen, had suggested that extra-efficiency concerns such as justice were beyond the purview of the economist. Walsh, however, took a very different stance. He pointed out, as had several previous commentators on Coase s negotiation result, that any allocative decision has distributional 21 Though Ng seems not to have realized it, his analysis had further legal-economic import. His hypothetical involved the not uncommon legal issue of coming to the nuisance, and his analysis, if correct, it suggested an efficiency rationale for the first-in-time rule that has occupied a prominent place in common law history. But unlike McKean, Ng was not pursing the development of an economic analysis of law and so, having demonstrated the superiority of first-party priority to his satisfaction, proceeded to turn his attention to other criticisms leveled against Pigovian remedies. 22 Walsh also noted that Ng was not the first to consider the possible allocative impact of who pays whom in a longrun context that accounts for the impact of alternative legal rules on future decisions that is, their long-run impact. Though Walsh did not provide citations, these include, in addition to Ng, Kamien, Schwartz, and Dolbear (1966), Bramhall and Mills (1966), Nutter (1968), and Calabresi (1968). 20

21 consequences (Medema 2013f; 2013g). More importantly, though, he was convinced economists must be prepared to say something about the distributional consequences of policy proposals (1972, p. 256, emphasis added). Of particular import here, for Walsh, was the potential conflict in a social or political sense between efficiency and equity. While economic theory posits a resolution of distribution issues through costless and non-distortionary lump-sum transfers, in reality, he said, such transfers are almost certainly infeasible. In light of this, the relevant allocative alternatives are likely to be either (i) efficiency cum an unacceptable distributional outcome or (ii) an acceptable distributional outcome cum inefficiency. Moreover, achieving the distribution of income that society requires i.e., through demands registered via the political process may mandate living with a measure, perhaps significant, of inefficiency and even possibilities, such as blackmail, that some economists, including Ng, deemed unacceptable. All of this led Walsh to conclude that appropriate policy remedies, or legal rules, can only be determined on a case-by-case basis. Like Ng, then, Walsh accepted that the Coase theorem, defined to include both efficiency and allocative invariance, may fail when the impact of legal rules on future decisions is taken into account. But he parted company with Ng in his belief that the Coase theorem framework could form the basis for a normative economic assessment of legal rules, insisting that economists must be willing to eschew model-directed pronouncements and instead consider the full range of effects engendered by alternative legal rules. Rescuing Coase: Swan and the Grand Tautology While this initial skirmish went largely to questions of what the economist qua economist should be willing to bring to the legal-economic policy table, the discussion began to get into the meat of what the Coase theorem is really all about when, in 1975, Peter Swan, then a Lecturer in Economics at Australia National University, resolved to take issue with both Ng and Walsh, Swan had only recently returned from a period spent as a Visiting Assistant Professor at the University of Chicago s Graduate School of Business and, though his Chicago experience was not the stimulus for his critique of Ng 21

22 and Walsh, the fact that he found the Chicago view of economic theory and policy rather congenial to his way of thinking 23 no doubt played a role in the perspective that informed his analysis here. Though the various authors examined here tended to speak of the Coase theorem as if it were a well-defined concept, in reality it was anything but. Witness Swan s statement of the theorem, which can be usefully juxtaposed with that of McKean (1970, p. 30), quoted above: The Coase Theorem states that as long as property rights are well-defined and in the absence of transactions costs, it is irrelevant whether or not the damaging party, A, is liable for the harm that he inflicts on some other party, B. In either case bargaining between the parties can achieve a Pareto optimum solution, although there is no need for the solution in which A compensates B to be the same as the one in which B bribes A so as to modify A s harmful behaviour. Notice that, in contrast to McKean, Swan s statement of the theorem overtly excludes in fact, rejects Coase s invariance proposition, the idea that the final allocative outcome will be identical regardless of how rights are initially assigned. Coase, of course, had made the invariance claim explicit and early commentators had followed his lead. But concerns about income effects and the move by some to encapsulate the theorem in a Paretian/Edgeworthian framework had raised questions about the validity of the invariance thesis. 24 That said, the certitude evidenced in Swan s statement masks the fact that the invariance claim was still very much in play in the larger Coase theorem literature at this time Correspondence with the author, June 7, Medema (2013g). Because all Pareto efficient outcomes are equally efficient, these Coasean solutions are identically efficient but not necessarily identical in terms of prices and output levels. 25 Swan was aware of the potential tension here, acknowledging that it is not clear whether the theorem, as he had characterized it, was consistent with Coase s own analysis, since, on Swan s reading, Coase assumes that there is only one Pareto optimum solution and it is not clear from Coase whether he was willing to contemplate the existence of an infinite number of such solutions (1975, p. 268n.2). It is also noteworthy that Swan explicitly couched the theorem in a Paretian efficiency context. While he was by no means the first to do so, neither Coase nor Stigler had analyzed the efficiency of negotiated solutions through the Paretian lens. Interestingly, both Swan and Walsh acknowledged having had helpful comments from and conversations with Geoff Brennan, who was himself a product of the Pareto-infused Virginia school tradition. See Walsh (1972, p. 254n) and Swan (1975, p. 268). It is also noteworthy that Swan explicitly couched the theorem in a Paretian efficiency context. While he was by no means the first to do so, we must bear in mind that neither Coase nor Stigler had analyzed the efficiency of negotiated solutions through the Paretian lens. Interestingly, both Swan and Walsh acknowledged having had helpful comments from and conversations with Geoff Brennan, who was himself a product of the Pareto-infused Virginia school tradition. See Walsh (1972, p. 254n) and Swan (1975, p. 268). 22