26th October 2023
Reinterpreting the externality problem – Harold Demsetz
This essay, from his 2008 book, From Economic Man to Economic System, holds the key to properly understanding the concept of externalities. Demsetz then explained it all in his 2011 paper.
I’m making annotated notes (in blue colour) to assist my study of the topic.
Extract from the essay: Ownership and exchange – Harold Demsetz
Markets and the price system lie at the center of economic theory, but they rest on an institution that this theory hardly touches: private ownership of resources. If people create markets in which they expect to exchange assets, they must have title to these assets. Ownership entitlement is simply presumed in much of what we call economic theory. This often is true even in discussions in which ownership is explicitly discussed. R. H. Coase, whose ideas are discussed below and in the next essay, discusses the difference, if any, in the uses made of resources if the identities of the persons who own these resources are “shuffled” in hypothetical comparisons. His discussion of this issue explicitly involves ownership. Yet, it takes the existence and nature of a private ownership system as a known given. The issue he raises would make no sense if ownership were not already an operable institutional arrangement.
Thus, there are two components to the social arrangement used to resolve competing interests in a market-based economy. One is the institution of private ownership; the second is a legal system that makes exchange of owned assets a reliable activity. Out of this come markets that resolve differences in the way people would like scarce resources to be used.
What difference, if any, does identity of owner make? R. H. Coase, in his important article “The Problem of Social Cost” (1960), brought this question to the attention of economists when he explained why the profession had mischaracterized the problem of externalities. His explanation deals with this problem under two alternative conditions, one in which the cost of using the price system is assumed to be zero and the other in which this cost is assumed to be positive. The positive cost case, with which I have some disagreement. In the zero cost case, I fully agree with Coase. My intent in discussing the zero cost case is to prepare the reader who has not yet become familiar with Coase’s work to understand the more complex case in which the cost of using the price system is positive.
The problem discussed by Coase is exemplified by his discussion of Sturges v. Bridgman, decided by an English court in 1879. A doctor had taken occupancy of a premise located next to one in which a confectioner conducted his business. Eight years after moving in, the doctor added a consulting and treating room to the original structure. This brought his work with patients into closer proximity to the confectioner’s machinery. As a result, the noise coming from the confectioner’s quarters interfered with the doctor’s ability to diagnose illnesses of the chest and, the doctor claimed, made it difficult from him to think clearly about the medical problems brought to him by his patients. The court decided for the doctor, entitling him to more quiet than was allowed by the confectioner’s equipment. [Sanjeev: This was obviously an incorrect decision. Based on Donald Boudreaux’s 2019 reasoning, with which I agree, sequencing matters. The doctor should have known and therefore planned for the noise from the confectioner’s equipment – the externality was expected to be internalised by the doctor.]
Coase then inquires as to the resource-use consequence of a court’s decision in such cases, doing so by carefully thinking through a comparison of the consequences that would flow if the court favored one of the petitioners with the consequences that would flow if, instead, the court favored the second petitioner. The decision choice is not to be thought of as taking place sequentially through time but as a comparison of substitute decisions. Will resources be used in different ways depending on which of the two is favored? The intuitive answer to this question is “yes,” and so was the answer implied by the externality doctrine that prevailed at the time Coase wrote. Coase’s subtle reasoning shows this intuition to be wrong if the price system can be used as freely as neoclassical price theory assumes.
Both alternatives are consistent with a private property system. The process is not one of regulation, as neither decision insists on a court-defined specific decibel-level outcome. Given the court’s decision, the parties can negotiate with each other to achieve some mutually acceptable decibel level. Who pays and who receives payment to achieve this level is determined by which alternative assignment of ownership rights the court has chosen. If the doctor is favored by the decision, the confectioner will need to pay the doctor if he wishes to make use of a noisy candy-making machine. Had the confectioner been favored by the court, the doctor would need to pay the confectioner to obtain a lowering of the decibel level. Hence, the court’s decision has an effect on which way wealth flows between the two parties. The wealth distribution issue, however, is a diversion from the question Coase sought to answer; this was the effect of the court’s decision on resource allocation. In this particular case, does the court’s decision have an effect on the machinery used by the confectioner?
It was Coase’s insight to see that freely entered negotiations between the parties would yield the same allocation of resources no matter which of the two possible decisions is chosen by the court. If the court favors the doctor, as it in fact did, the confectioner could nonetheless obtain permission from the doctor to continue using noisy candy-making machinery, and he would be able to secure this permission if the value he attaches to the use of noisy machinery exceeds the cost this noise imposes on the doctor. Given that this is the case, a mutually agreeable bargain can be struck that allows for a noisy environment. [Sanjeev: This begs the question why the doctor paid money to go to the court. That money was wasted. If he was to live with the noise, anyway, it would have been better to be paid by the candy maker in the first instance]. However, this would also be true if the court had favored the confectioner. In this case, the doctor would need to pay the confectioner to reduce the noise level. If we keep the measures of costs and benefits just used, we know the doctor would not be willing to pay enough to the confectioner to achieve this result because the noisy environment is less costly to the doctor than is the cost borne by the confectioner if the noisy candymaking machinery is not used. Either ruling the court can make will result in continued use of the noisy candy-making machine. The values assumed in these calculations can be changed so that a quiet environment results, but then the new values would result in a quiet environment no matter which of the alternative rulings the court chooses. Although the distribution of wealth is affected by the court’s decision, the allocation of resources is not.
In retrospect this is all very clear. The private ownership system allocates resources to their highest value uses. And this use is the same no matter who owns the resources if ownership assignment has no significant wealth effects on the demands for medical services and candy.
The difference between this method of resolving conflicts and the approach customarily taken if regulation is used to influence the use of machinery may be noted. Regulation involves a law that makes noise levels above a certain amount illegal or that bars the use of noisy candy-making machines. It is illegal under this approach for the parties to the conflict to negotiate a solution that would yield high noise or that would allow for noisy machinery. [Sanjeev: typically such regulation would have a grandfathering clause to minimise harm to the candymaker] Regulation also has an effect on wealth distribution, since the confectioner must forgo the use of cheap noisy candy-making machinery in favor of more expensive quiet machinery or the doctor must forgo building an extension that brings his office closer than “x” feet to the candy maker’s place of business; here, the doctor, who would presumably need to build a second story at greater cost if he is to gain space, suffers a potential loss of wealth. Regulation has wealth consequences, but more than this, and unlike the common-law solution to a conflict over resource use, it also has resource allocation consequences. One regulation yields an outward extension of the doctor’s premises coupled to quiet candy-making equipment, while the other regulation results in upward expansion of the doctor’s premises coupled to noisy candy-making equipment. Resource allocation and wealth distribution both result from the regulatory approach.
The regulatory approach might by chance result in an efficient allocation of resources, but the common-law court approach certainly will if there is no cost to negotiating an agreement between contending parties. The level of noise that obtains after such negotiations is that which results in the highest possible joint value of the two activities; noisy equipment is abandoned only if its value to the confectioner is less than the value of quiet to the doctor, and noisy equipment is retained only if its value to the confectioner exceeds the value of quiet to the doctor. And, in terms of wealth distribution, the resource allocation that results from the court’s decision holds whichever party is favored by the court.
The view of economists before Coase wrote on this topic was that the noise emanating from the confectioner’s machinery imposed a cost on the neighboring doctor but not one that would be taken account of by the confectioner when choosing which machinery to use to make candy. The total social cost in this “two-person mini-world” consisted of the cost to the confectioner of purchasing and operating the candy-making machine and the cost to the doctor of noise emanating from the confectioner’s machine. On the then-accepted view of this type of problem, it would not be total social cost that influences the candy maker’s equipment-purchasing decision but only the cost to him of making the purchase and buying the electricity used to power it. Since these costs take no account of the harm caused by the neighboring doctor, they are only part of the total cost borne by society from the use of noisy candy-making machinery. As a result, the confectioner would be inclined to purchase a noisy machine even if this resulted in an increase in total social cost. The cost borne by the doctor, which would in fact influence the confectioner’s decision through negotiations between the two, was ignored by the then-existing doctrine of externalities. It was treated as external to the decision-making process, and this doctrine made the presumed difference between total social cost and total private cost the test of the presence of an externality problem.
Coase saw that if the court had decided in favor of the confectioner, it would have created an incentive for the neighboring doctor to offer a payment to the confectioner to lower the noise level (or to purchase a more expensive, quieter machine), such payment, in the limit, being equal to the cost borne by the doctor as a result of the noise. Since the confectioner would not receive this payment should he invest in and use noisy machinery, he would in fact bear a cost for doing so that reflects damages done to the doctor. This is brought into the confectioner’s profit calculations in the form of forgone revenue or as an implicit cost of using noisy machinery, and in this way it influences his decision about what type of machinery to use.
The discussion just concluded reveals a very important tendency of capitalism. On the assumption that people are in a better position to seek their interests than are people whom they have not employed to assist them, capitalism, by privatizing ownership of resources and allowing people to negotiate the uses to which these resources will be put, tends to guide resources into those uses that yield maximum marketplace values. I describe this conclusion as a strong tendency because we have not yet brought competition and monopoly into the discussion; the presence of monopoly can undermine this tendency.
The problem just discussed should be viewed from a forward-looking perspective in the sense that the locations of doctor and candy maker, and the equipment the candy maker will use, are yet to be determined. In this context it must be recognized that land owners, doctors, and candy makers will compete in the rental market or in the property-for-sale market. Since no doctor and no candy maker are yet neighbors, the locations they choose and the contracts they will write will be determined under competitive conditions. Supposing here that the cost of transacting is zero, we may conclude that competition between private owners of resources will yield an allocation of resources that maximizes the market value derived from these resources.
 The cost incurred by the doctor to extend the structure of his office becomes relevant if the analysis is forward-looking from the time the doctor merely considers whether to add a room. Implicitly taken by Coase is a cost incurred before the case arrives in court.
Essay: Reinterpreting the externality problem
Coase’s reasoning … reaches the conclusion that resource allocation is unaffected by the identity of the person who is assigned the right to control the use of a scarce resource if people can use markets and negotiations freely. Furthermore, he concluded that there can be no difference between social and private cost in such a world. It is a world without inefficiencies (and, by implication, without externalities). However, he reaches a different conclusion for a world if the cost of using the price system is positive. Inefficiency cannot be ruled out in this more realistic world. This contrasting conclusion is now solidly incorporated in economic doctrine. The present essay’s objective is to change doctrine in this respect.
Preliminary to doing this, some attention to terminology is needed. In a work of mine, “The Cost of Transacting on the New York Stock Exchange” (Demsetz, 1968), I empirically examined the cost of using the NYSE to execute orders to buy and sell equity shares. I called this cost “transaction cost,” which seemed quite natural for a market in which trading is so active. As my article explained, I meant this to represent the cost of using the price system in the particular case of the NYSE. I continue to use transaction cost to mean the cost of using the price system. Coase means by this cost the value of resources used to obtain information about prices and to engage in exchange at these prices. Coase describes the cost of using the price system in his 1937 article on the firm. He writes:
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. The most obvious cost . . . is that of discovering what the relevant prices are . . . The costs of negotiating and concluding a separate contract for each exchange transaction which takes place on a market must also be taken into account.
And in his 1960 article about externalities he gives a similar notion:
In order to carry out a market transaction it is necessary to discover who it is that one wishes to deal with, to inform people that one wishes to deal with and to what terms, to conduct negotiations leading up to a bargain, to draw up the contract and undertake the inspection needed to make sure that the terms of the contract are being observed, and so on.
I use transaction cost to mean no more and no less than what Coase describes as the cost of using the price system. The stipulation is necessary because later writers broadened the meaning of transaction to include the costs of information and of cooperating between parties whether these costs are incurred in exchange across markets or in any other setting, such as in managing workers within a firm. Coase clearly meant to distinguish costs incurred to manage resources within the firm from costs incurred to interact across markets at market-determined prices, and I wish to preserve this distinction.
The main source of concern among economists about externalities is found in the writings of A. C. Pigou, especially in his Economics of Welfare (1920). Pigou viewed his work as a criticism of conclusions held by mainline economists of his day about the ability of a private-ownership, competitive economy to allocate resources efficiently. His central point was that the neoclassical model of this economic system is wrong in presuming that private decision makers take all costs and benefits into account when deciding how to use their resources. Because of this, we cannot conclude that resources will be allocated efficiently.
However, if an effect of using a resource is a cost or a benefit, there must be someone who experiences this effect. If so, this effect must be taken into account by someone, even if this is not the person who owns the resource whose use produces this effect. This means, if the cost of using the price system is zero, that there will emerge prices measuring each and every cost and/or benefit that arises in the course of interactions between persons. As discussed in essay 6 [Ownership and exchange], Coase showed that, in this case, all costs and benefits are taken into account when resources are allocated by way of a market-based price system. Efficient resource allocation results, or, as Coase puts it, the value derived from the use of resources is maximized. Coase also observed that if costs are incurred to use the price system, these effects, even though they are borne by someone, will not be entirely borne by the person whose use of resources gives rise to these effects. [Sanjeev: this is the case that Demsetz disagrees with]
A steel mill puts soot in the air in order to produce steel at the lowest possible cost, which, we may assume, requires the mill to use soft coal. The soot descends on neighboring laundries and increases the cost to them of laundering cloths. The owner of the steel mill takes the price of coal into account and will take the increased cost borne by launderers into account if the cost of using the price system is zero, for then a launderer will offer payments to the owner of the steel mill if he will reduce the amount of soft coal he uses, such payments being determined by the cost increases borne by a launderer as a result of soot. However, these payments might not be forthcoming in sufficient amounts to accomplish this if a launderer must incur a cost, call it a transaction cost, to negotiate with the owners of the steel mill. This suggests that markets may not bring all costs and benefits borne by some people to bear on those whose use of resources brings these into existence. Coase concludes because of this incomplete reckoning that resources may not be allocated efficiently.
Coase was critical of neoclassical theory for neglecting the cost of using the price system, but I think he is wrong about this. Neoclassical theory is written as if there is no cost of using the price system, but the theory could be rephrased to take account of transaction cost if it shared Coase’s objective. Coase seeks to explain when a price system will be used and when it will not. Neoclassical theory seeks to explain the allocation of resources if there is a price system that transmits to all the costs and benefits that arise from the way in which resources are allocated. The two tasks differ. Even so, neoclassical theory could be turned to the task of understanding resource allocation if prices are available only at a cost. This is what I intend to do.
It is of course true that positive transaction cost will keep negotiations between interacting parties from being as finely tuned as they would be if transactions could be executed freely; indeed, if high enough, transaction cost may block negotiations completely. However, it is incorrect to infer from this that resources are allocated inefficiently. Missing from Coase’s and from the contemporary treatment of this situation is a recognition that efficiency itself requires foregoing fine-tuning of the sort that would be appropriate (efficient) in a world in which transaction cost were zero. This point, made some time ago, has not penetrated discussions of externalities.
Transaction cost is no different from other costs in regard to determining which good or service is to be produced. [Sanjeev: This is one of the most important points made by Demsetz] If the cost of producing a hydrogen-fueled automobile exceeds the price that people are willing to pay for the vehicle, efficient resource allocation requires that this vehicle not be produced. Similarly, efficient resource allocation requires that a transaction not take place if the cost of producing the transaction exceeds the price that people are willing to pay to engage in exchange. We do not shout “inefficiency!” if the vehicle is not produced. Why proclaim inefficiency if a transaction is not produced?
The source of confusion about this is the fact that Coase has embedded the externality problem in a hypothetical experiment involving alternative assignments of ownership rights. The counterexample pointed to in the preceding paragraph regarding a hydrogen-fueled automobile is not directed at the issues raised if we ask what difference it makes if X has the right to put a car on the street or if Y has this right. This issue, which involves who owns which rights, simply does not come to the surface when dealing with standard production problems. Coase embeds the externality problem in a model in which the assignment of rights is involved. Although this difference leads to somewhat different conclusions, these do not bear on the question of efficiency. This claim needs to be examined in a bit more detail.
Coase sees the possibility of inefficiency arising in a situation in which there are competing claimants for control of a scarce resource. They take their dispute to court, which assigns ownership rights to one of the claimants. In this case, however, Coase assumes that the cost for these parties to negotiate (i.e., transact) with each other after the court has made its decision is prohibitively high. Suppose the court assigns ownership to the party who is unable to put the involved resource to its highest value use. That is, the product this party produces with this resource is less valuable (in the marketplace) than is the product that would have been produced by the other party had he or she been favored by the court. One may wonder why either party is unable to put the resource to the same uses as the other party, but this objection is not in the spirit of Coase’s discussion and we may set it aside. Setting this aside, we can have no objection to Coase’s conclusion that the resource may not be put to its highest value use. This is because high transaction cost prevents the party whom the court favored from selling the entitlement to the party whom the court did not favor but who can use the resource to create goods that are more valuable. If we stay with the framework in which he has cast this problem, we must agree with Coase that more value could have been secured from the resource if the court had made the opposite decision. The error is in identifying this as an inefficiency of the economic system that is brought about by the cost of transacting. Coase may be quoted on this (1960, p. 16):
In these conditions [of positive transaction cost] the initial delimitation of legal rights does have an effect on the efficiency with which the economic system operates. One arrangement of rights may bring about a greater value of production than any other. But unless this is the arrangement of rights established by the legal system, the costs of reaching the same result … through the market may be so great that [this arrangement of rights] may never be achieved.
I suspect that Coase may not have carefully reviewed the phraseology he uses to describe this conclusion, a rare event indeed. Nonetheless, his statement has brought economists to the conclusion that positive transaction cost can make the competitive economic system function inefficiently, since a lesser valued mix of goods is produced than would have been produced if transaction cost had not hampered the market’s ability to reassign ownership rights. Coase puts his conclusion thus: “the initial delimitation of legal rights does have an effect on the efficiency with which the economic system operates.” The profession has accepted this phrasing and, based on it, the prevailing doctrine is that positive transaction cost can undermine the efficiency of the economic system. [Sanjeev: Demsetz proves that this is not true]
But wait! With each alternative assignment of ownership, the economic system does the best that can possibly be done. [Sanjeev: this is the key point] The private-ownership, competitive economic system does allocate resources efficiently given the court’s decision. If the court has errantly chosen ownership, well then, the economic system at least minimizes the loss borne by society because of the court’s error. Under the stated conditions, this loss is minimized by forgoing market transactions whose purpose would be to reshuffle ownership entitlements, since, by assumption, the cost of reshuffling more than exhausts the increment in the value of goods that will result. Legal error has caused the problem, not positive transaction cost. There is no inefficiency in the way the market accommodates to the court’s mistake.
How does the situation described by Coase differ from a government policy that affects the distribution of wealth? Special interest groups petition the state for a tax-and-spend policy that is expected to yield benefits. The state responds, and the mix of goods and the distribution of wealth that results will differ depending on just how the state responds. One response may make more valuable uses of resources than other responses. We do not conclude that the economic system is inefficient if it produces the most valuable mix of goods that is possible given the response chosen by the government. Coase’s assignment of entitlements by the court constitutes a distribution-of-wealth decision and, because of the difference in the assumed production capabilities of the plaintiffs, it also influences the mix of goods that is produced if transaction cost is positive. No matter how the court decides and no matter if transaction cost is zero or positive, the competitive decentralized economy is efficient in that it makes the assigned distribution of wealth yield the most valuable mix of output that is possible given the cost of using the market to reverse the court’s decision.
The legal system is the institution that resolves the conflict over ownership entitlement. However, neoclassical economic theory implicitly assumes the existence of well-defined ownership rights. Care must be taken not to confuse the two situations. Coase’s 1937 discussion of the make-or-buy decision faced by the firm (discussed later) does not mix these two problems. He imaginatively reveals the way in which the cost of using the price system affects the internal structure of the firm. In this application, however, ownership entitlements are assigned and in place; there is no contesting of these in courts. And in that article, Coase does not at all speak of transaction cost as a source of inefficiency in the way a firm resolves the make-or-buy decision. In his article on externalities, however, he brings the court into the picture by posing an ownership entitlement problem. This allows the court to make a mistake.
The quality of the court’s decision, however, has nothing to do with how well the market works to bring social and private costs into equality. The court is not part of the economic system. The legal system, after all, is designed to be publicly provided and to be insulated from the price system, just as is the government when it puts a wealth distribution policy into effect. We do not know what motivates a court other than some earlier court decision with which it hopes to be consistent. If the court were brought into the economic system, say by allowing contesting parties to bid for the court’s decision, the court would not make the mistake that is essential to Coase’s logic, since the party who can make the most valuable use of the entitlement will be able to offer more to the court. I do not necessarily recommend making the court part of the economic system, but consideration of this option fully reveals that the court in Coase’s discussion is not part of the economic system. We cannot conclude from the consequences of an error made by the court that the economic system is responsible for the lower value mix of output that results.
Given that the court lies outside the economic system, what really is the “inefficiency” that Coase (mistakenly) attributes to the economic system? It is that the market does not tolerate the bearing of transaction cost to correct for the court’s error if the cost of doing so exceeds the gain in the value that is to be expected from realigning the ownership of the entitlement. The market is accused of being inefficient because it is efficient!
Coase goes on to write:
It is clear that an alternative form of economic organisation which could achieve the same result at less cost than would be incurred by using the market [to realign ownership entitlements] would enable the value of production to be raised. As I explained many years ago, the firm represents such an alternative. . . . Another alternative . . . is direct Government regulation. (1960, p. 16)
However, he should have rewritten:
It is clear that an organization alternative to the court such, perhaps, as the government or the firm, that could assign control of resources with less error, would enable the value of production to be raised.
He does not explain how this might be accomplished, nor does he note that the source of the problem is a nonmarket institution, namely, the court. It is unclear to me how the firm accomplishes the task. Once the court has made its decision, the party who has received the entitlement, if it is a firm, might merge with the party who has not received it, another firm, thus conveying control of the entitlement to the party who can put it to best use. However, this requires the equivalent of an exchange of assets, and so, it requires an encounter with the same high level of transaction cost that barred a market-based exchange. Alternatively, the contesting parties, prior to going to court, might settle their dispute by merging, or by exchanging, assets. However, this again incurs market transaction cost. The only advantage I see for this method as compared to a market-based purchase and sale is that it avoids a trip to the court.
Coase must have had an entirely different role for the firm in mind. The firm’s owner listens to two middle managers’ plea for control of some of the firm’s resources. The owner chooses one pleader over the other. The favored pleader then uses the resources he or she has been assigned and produces an output that has some market value. If this market value is judged by the owner to be greater than it would have been if the other pleader had been favored, all is well and good. If it is judged to be less than expected, the owner reassigns control to the pleader who, before, had been disfavored. Resources therefore end up generating the highest possible value. How does this differ from a situation in which the court has assigned an entitlement and then is asked, at some future time, to reconsider the case? The difference is in the fact that the firm’s activities are part of the economic system in as much as the firm’s owner is interested in maximizing the market value of the goods the firm makes, while the court is organized in ways that discourage the plaintiffs from bidding for the court’s decision. The difference does not lie in transaction cost, since the owner of the firm incurs the equivalent of transaction cost when reconsidering a prior decision and acting to remedy it; if this reconsideration cost is too high, the owner, like the court, will not reconsider. In which case, if we followed the terminology adopted by Coase when discussing the court, we would (mistakenly) claim that the firm (or is it the market?) is inefficient.
The only true alternative to the marketplace is an institution that can use a greater measure of coercion than is available to participants in the marketplace. This, in fact, may be a firm, but it surely is the state. The court’s error might be expeditiously rectified by the government through coercive means that, depending on how one measures the cost of using these means, might be less costly than using voluntarily struck agreements. Yet, the foundation of a private-ownership, market-based economy is its generally superior ability to work with resources as compared with central planners, and that it can do a better job while preserving a greater measure of freedom for individuals. I write “generally” because the state has a role to play. This will not generally be in activities the equivalent of producing food or steel unless the use of slave labor does reduce the real cost (including that part borne by the slaves) of production, and even in this case the state would be unneeded since workers would then agree voluntarily to work under slavelike conditions. Anyway, this possibility has nothing to do with externalities. The state might be able to organize transactions more cheaply than can markets, but I see no reason for this to be the case.
The case for the state is strongest in the presence of high costs of barring free riders, costs I would describe as high ownership costs, not as high transaction costs, since the two types of costs need not correlate strongly. Soot from a steel mill descends on a neighboring laundry and raises the cost of laundering. [Sanjeev: in this case sequencing is important, as Donald Boudreaux has explained in 2019. If the laundries came after the factory, they knew the cost of pollution in advance and factored any remedies that they might undertake, into their business plans. The rest of Demsetz’s comments on this topic make sense only if we assume that the laundries came earlier to, or simultaneously with, the factory.]
There is a transaction cost that must be borne if this cost is to play a role in decisions made by the owner of the steel mill (if the mill is entitled to use soft coal when producing steel), but there is no free-rider problem. The demand for soot reduction will become evident in the negotiations between the two parties if the transaction cost is reasonably low, and there is no reason to think that the state can bring the parties together at less cost than can the market.
Now, suppose that multiple laundries surround the steel mill. Each laundry owner has a strategic interest in underrevealing what he or she is willing to pay to have soot reduced, hoping that other laundry owners will purchase a desired amount of reduction in soot. A psychological preference for behaving strategically cannot reasonably be interpreted as a cost of using the price system as Coase describes this cost. If it is anything more than a psychological propensity, then it is a cost of having multiple laundries in the vicinity of a steel mill. In this case, the state can force a solution that is better than the market’s if we make the heroic assumptions that the state knows what it is doing and seeks an efficient mix of steel output and laundry services. The state applies an appropriate tax to the use of coal in the manufacture of steel, as Pigou suggested. [Sanjeev: under conditions of strategic beaviour, a tax is considered a viable potential option]
This is not just a definitional issue, one in which strategic behavior, if one chooses, can be called a cost of using the price system. Strategic behavior is entirely different from the cost of soliciting price information and concluding an exchange; it is a problem of strategic provision of misleading price information. But if the reader insists on confusing the two, so be it, but, then, please remember that situations in which strategic behavior is important are but a part, and a small part at that, of all situations that involve positive transaction cost.
 “For produced goods . . . optimality theorems require equalities among various marginal rates of substitution. . . . [but] do not . . . for goods and services that are not produced in the final efficient equilibrium; for these goods we have corner solutions involving inequalities . . . a premise of requiring equalities is that we are talking about goods which we require to be produced in positive quantities” (Demsetz 1964).
 The required inability may derive from purely personal differences in the utility derived from the way a resource is used, but then the value derived from personalizing the rewards of using the resource one way instead of another should be taken into account when addressing the efficiency question.
 See Demsetz (1964).