 This panel is entitled, Revising Traditional Notions of Competition and Monopoly. And the first paper or the first talk to be given is called, Reexamination of the Rules Against Horizontal Collusive Arrangements. And the lecturer is Dr. Mario Rizzo, received his PhD from University of Chicago. He's now a professor in economics department at New York University. And dear to many of our hearts, he's a faculty member in the Austrian economics program there. His specialization is in law and economics. And Mario will be a fellow in civil liberty at the Yale law school in the spring semester this year. Mario? Thank you, Richie. Can everyone hear me? Okay. After listening to Jim Buchanan's speech this morning, it was clear to me that his role here was to hold up the extreme left-wing position on antitrust economics. And I hope to at least be one of the participants of this conference who will redress the balance and hold up the extreme right-wing position on antitrust. I thought of calling my paper an ode to collusion, but then I thought it might be considered not sufficiently scientific or not sufficiently value-free, so I quickly dropped that idea, though I decide that this paper should be published. That might be a good subtitle, anyway. My previous work in the area of law and economics has really been in torts, evidence, and contracts, and other areas of law that at least to me make sense. This is my first effort, my virgin effort in antitrust economics, and I hope that after I finish, the lack of sense of antitrust will be even more evident than it was before. What I want to speak to you about today is, as Richie said, a re-examination of the rules against horizontal collusive arrangements. Now, the title in the programs is a little broader than that, and I don't intend to talk about vertical price restrictions because I think the economic literature in that area is sufficiently good. We haven't been successful in convincing the courts yet, but I don't think that that's really the challenge. I think the challenge for economists is to debunk some of the myths that have been perpetrated by the perfectly competitive model on horizontal collusive arrangements, and that's what I want to address in today's talk. The market process, as I see it, is far broader than that implied by the traditional theory of competition and especially by its formalization, the Arrow de Bruy general equilibrium construct. Of course, that construct has been given a Nobel Prize, so it's even more important, I think, to combat some of its improper implications. Even those economists who claim to follow an essentially process view have often focused solely on one dimension that is price competition. In this talk, I will emphasize the multi-dimensionality of the market process. In particular, I will show that specific institutional or organizational structures are extremely important in facilitating market adjustments. More importantly, and perhaps more controversially, sometimes the market quite properly develops institutions whose purpose it is to modify or even repress both the free movement of prices and the independence of decision making with regard to non-price variables, such as plant expansion, division of territories, etc. From the perspective of the traditional competitive model, price fixing and other forms of horizontal cooperation are unambiguously inefficient. Indeed, the economic rationale for the per se rule of this price fixing, especially in a horizontal context, is that little of a redeeming nature can be said in its defense. It is clear, however, that there are many areas in which the market represses the free movement of prices to apparently good effect. The most obvious example is simply a contract that specifies a fixed price over a period of time in which the spot price may fluctuate. Examples of this are employment contracts, futures contracts in which an individual may agree to buy X units of 1984 soybeans at Y dollars, despite what happens to the spot price. The purpose of these contracts is either to shift uncertainty onto those who can bear it more cheaply or to reduce uncertainty as in the case of futures markets. Another and perhaps more pertinent example for our purposes is that of the firm itself. Coase argued many years ago that sometimes it is inefficient to work through price relations. Here, firms or little islands of socialism, as it were, develop in which behavior is directed through non-price administrative means or, more generally, through cooperative behavior. There may, however, be diseconomies associated with certain forms of cooperative behavior and not with other forms. When, for example, it is efficient to cooperate in all respects, two formerly independent firms will merge. But if only partial cooperation is warranted, they may collude. Collusion then may be seen as a point on the continuum between outright merger and atomistic independence. If we apply a rule of reason to mergers, why not to collusion of all types? In what follows, I wish to examine two forms of collusive practices. First, those practices, the object of which is to reduce uncertainty. And second, those the object of which is to compensate for deficient incentive structures and thus to reduce costs. Let me begin with collusion that reduces uncertainty. Here, I think it's important to differentiate between two types of uncertainty. And I'm only speaking really about one. The first type of uncertainty is perhaps the most traditional type that we think about in economics, and that's what I want to call exogenous uncertainty. That is uncertainty with respect to the underlying data, consumer taste, resource availability, technology, and so that sort. Sometimes trade associations arise to exchange this kind of information. And I think Posner and others have convincingly argued that in unconcentrated markets, these exchanges of information are indeed pro-competitive. While these arguments, Posner and others are quite interesting, this is not my primary concern here. The second form of uncertainty that I wish to discuss in greater detail is what I will call endogenous uncertainty. That is, uncertainty about what competitors will do, what prices they will charge, or how they will invest or disinvest in its response to profit opportunities in that industry. This type of uncertainty is endogenous to the process of adjustment within certain institutional arrangements. In principle, this form of uncertainty can be eradicated or reduced by changes in the organizational setup. And this is what I want to discuss. Consider first the endogenous uncertainty with respect to price. I think James Mead, who in many respects uses this notion of uncertainty with respect to price to argue for essentially government planning, government coordination of industry. Nevertheless, I think that many of the points that Mead raises are relevant in my context, that is to say the context of voluntary price agreements. Let me quote briefly from Mead's book, The Controlled Economy. And Mead says the following, the oligopolist in his turn has to consider not simply how his customers will react directly to a change in the prices he charges. He must also consider how his closest competitor or competitors will react in their prices. Which depends, of course, upon how they think he will react to their reactions. We're in the realm of the theory of gains, of bargaining, of conventional good behavior, of tacit or open collusion, and so on. End quote. Now this is very much the kind of uncertainty that Keynes described in his famous beauty contest story, where the object was to guess what average opinion thought the most attractive faces were in this beauty contest. The object wasn't to decide whom, any given individual thought the most attractive people were, but rather the object was to determine what average opinion thought the most attractive people were. So the object in the Keynes example and in the Mead discussion was not to guess something that stays put or at least is independent of the guessing process, but rather like taste or resource availability would be. But rather the firms and the individuals in the Keynesian example must form expectations about the expectations of other firms and the two are not independent of one another, as we have seen. A similar story can be told about the endogenous uncertainty with respect to competitive investment decisions. G.B. Richardson in his neglected but brilliant work, Information Investment, summarizes this phenomenon in the following way. Quote. The existence of a general profit potential cannot automatically be assumed to create particular profit opportunities for individual entrepreneurs. Before any particular entrepreneur is prepared to invest in the production of the commodity, he will have to be assured that the volume of supply planned by competing producers who are also aware of the opportunity will not be so large as to overstock the market, thus converting the expectation of profit into the realization of loss. Without this assurance, entrepreneurs would not invest and supply would not be expanded. A general profit potential which is known to all and equally exploitable by all is, for this reason, available to no one in particular. This again is another example of the Keynesian beauty contest style of uncertainty, the endogenous uncertainty. Each firm must form expectations about the expectations of other firms and once more these two are not independent of one another. Now, on the basis of what Mead and Richardson say, it might occur to you, well, how then do firms ever, in the absence of any type of explicit coordination, make decisions? Now, it is true there are certain factors that may exist in certain market contexts that mitigate this endogenous uncertainty. Firms do not always face the pure unadulterated form of endogenous uncertainty depicted here for several reasons that I think are important to discuss in order to see why in their absence collusive arrangements, explicit coordination might be beneficial. Some of the reasons that this endogenous uncertainty is not present to extreme degrees in a number of industries can be looked at and can be seen in the following way. In some instances, where firms tend to be of uniform efficiency and where costs can be unambiguously allocated, this is kind of ideal type, average costs can place a limit on the variance of price differences. So that firms in looking to their own average costs, to the extent that they can unambiguously allocate their costs, will see more or less the range in which they can expect if other firms are equal efficiency, the range in which they can expect price fluctuations to take place in other firms. It puts kind of a limit on what they can expect their competitors to do, though it doesn't by any means eliminate this form of uncertainty entirely. Organized futures markets are another instance of ensuring greater stability or greater certainty into the price structure in a particular industry. And where these exist, will make planning in the part of individual firms easier and less characterized by endogenous uncertainty. In addition, with respect now to say the issue that Richardson raised on the profit opportunities, the danger that in response to generally known profit opportunities, the market might over respond, and hence that very danger preventing people from adequately responding. That kind of situation can be ameliorated where, for example, knowledge of a potential profit opportunity is not widespread. Then, in those cases, the threat of oversupply of competitive investments will be less. Similarly, where products are differentiated and consumer loyalty is high, there will be greater certainty of demand. Hence, firms will be less anxious about the effects of competitive increases in capacity. Because the increases in capacity by other firms in the industry will not be quite as relevant since they can rely on their demanders, their consumers, to continue to purchase from them since products are differentiated. In a sense, what all of these arguments do, I might add as a footnote, is convert, as Richardson shows, I think very brilliantly, convert the so-called defects of the market into the very virtues of the necessary for adequate adjustment. That is to say, the absence of certain, excuse me, the presence of certain frictions, which would be absent in the perfectly competitive equilibrium, are precisely those things which prevent mal-adjustment in the competitive process, in which enable firms to plan in a way which is characterized by the certainty of the environment. Now, nevertheless, when these and other so-called frictions are not present to the requisite degree, collusion may be an effective means to reduce uncertainty. So, I guess, in a sense, I'm viewing frictions and collusion as being alternative ways of adaptation. To the extent that we have a lot of frictions of the sort that I have discussed, collusion will be less necessary. To the extent, in some sense, that the market is more perfect paradoxically, collusion will be more necessary. Direct coordination of prices and investment decisions will produce more social benefits as the following characteristics are present. One, the more difficult it is to allocate costs, say, in multi-product firms. The more difficult it is to allocate your costs, the more you really don't know what the average costs are in any precise way. And to the extent that you can assume that other firms have similar cost structures, the more you're not quite sure about what those firms will find in their interest to charge, because you yourself don't exactly know what your costs are. Your costs are, in fact, uncertain. So, the more uncertain your costs are, the more difficult it is to allocate your costs, as it would be in multi-product firms. The more it might be necessary, the more direct coordination because the less the un-collusive market, the non-collusive market, will be able to instill a degree of certainty. The second factor, which would make collusion more likely to yield social benefits, is the greater the differences in efficiency among firms. In firms of differential efficiency, the more difficult it is to predict the future of other firms, of competitive firms. The third factor is the fewer the opportunities to engage in futures trading. We see that the presence of futures markets creates a certain degree of certainty in the price structure. Obviously, the fewer the opportunities to engage in futures trading, the less of this certainty will be in the price structure of the particular market. The fourth characteristic that makes for beneficial social consequences of collusion are the more widespread the knowledge of potential profit opportunities, which in a sense, again, paradoxical because one would think this is one of the characteristics of the perfectly competitive model, widespread knowledge of profit opportunities and in perfect competition there's no collusion. But if you concentrate on the adjustment process, the more widespread this knowledge is, the more the danger of overreaction, hence the more the need for explicit collusion to prevent this kind of uncertainty. Five, the less differentiated the product. With undifferentiated products, the expansion of competitive capacity will be more of a threat, and hence, again, the more the uncertainty of the environment. Six, the less marginal costs rise with an increase in the rate of capacity expansion. That means the competitive firms can increase their investment rather rapidly, hence threatening over supply. Seven, the longer the transmission interval between when investments are made and competing firms find out about them. If, of course, people could find out rather rapidly what their competitors have done or intend to do through other means other than explicit collusion, then there would be less need for the direct coordination. Eight, the greater the amount of durable and highly specific fixed equipment in an industry. If there's a lot of non-specific equipment, then, of course, there's costless exit. And when there's costless exit, then if you make a mistake and there's too much capacity, there's no overreaction, no big deal, because exit is costless. But to the extent that capacity is specialized and fixed and highly durable, mistakes will be more costly, and hence the greater the social benefits of direct coordination. Okay, these, I think, are a number of indicia of when explicit coordination tends to produce social benefits. Now, the reduction of this form of endogenous uncertainty will have at least two beneficial effects. One, it will facilitate more effective adjustments in coordination. And two, it will tend to increase industry output to the extent that firms exhibit risk-averse behavior. And Donald Dewey in an AER article a few years ago emphasizes this aspect of the reduction of uncertainty could actually have an output increasing effect to the extent that firms are risk-averse. Now, many of you may be thinking, well, you've mentioned all the possible good things about collusion, but don't I recognize the negative aspects of collusion? Well, I do. For reasons familiar to economists, there does exist a tendency in collusive arrangements to raise prices beyond what they would be in under non-collusive circumstances with the identical core structure. So the question obviously is, do the beneficial features of collusion outweigh the negative features? And there are two ways of approaching this question. One way which will be more more congenial to those who want to maintain the antitrust laws would be to say let's examine the market for the presence of these eight factors listed that I've listed. The greater their number and quantitative significance, the higher the probability, Ceteris paribus, that collusion is on net beneficial. We could do some kind of argument like that. And that would keep the antitrust of division on the FTC business because they could go around measuring these things. The second way to approach the question which I find more congenial since I'm always looking to save money and we could save money on the FTC and the Justice Department, the Antitrust Division of the Justice Department would be to apply a market test. My contention here, which I'm going to have to go through rather rapidly is that if entry is free and I don't mean costless, in fact my requirement is that entry be free but not costless, that is to say it be not blockaded by the government but at the same time not free in the costless sense, then actual potential competition will eliminate inefficient collusive arrangements. It will not however eliminate efficient collusive arrangements. My argument is as follows. If entry costs reducing features dominate, then a cartel or let's say organization will be a superior form of institutional arrangement. The cartel will be able to out-compete non-collusive new entrants because in effect its costs will be lower, the uncertainty aspect of its costs will be lower and it will then survive in the market process. Therefore the market provides an automatic check or test on the relative efficiency of different forms of organization. Now let me briefly mention my way of a footnote why I say that the requirement is that entry be free that is to say not blockaded by government but not costless. If entry were costless then the uncertainty reducing features of cartels would be dissipated and no relative efficiency test would be possible. Costless entry means that many firms could enter at once thus making price and investment coordination amongst the cartel members impossible. Costly entry on the other hand implies slow entry and hence less disruption of the co-ordinative capacity of the cartel and hence a greater capacity for the market to test in effect the balance of cost reducing and price raising features of the cartel. I think I've done here to the extent I'm correct much in the spirit of Richardson's work is to say convert the supposed disadvantage of costly entry into a positive advantage as far as adjustment processes are concerned. Now in the few minutes that remain let me briefly go over the second form of collusion and perhaps somewhat less controversial claim that collusion could be beneficial in the context where it offsets deficient incentive structures. While the first form of collusion improved the adjustment of firms to consumer demands by reducing endogenous uncertainty the second form improves adjustment by modifying the structure of incentives. Now an excellent example of this is the recent Supreme Court case Arizona versus Maricopa County Medical Society Society of 1982. In this case the defendants were two foundations for medical care, FMC's for short and the local county medical society with which one of the foundations was closely affiliated. And FMC is a non-profit corporation of physicians which integrated physician care and health insurance. The FMC entered into contractual relations with sellers of health insurance. These insurers agreed to cover for the purchases of their policies 100% of the cost of care administered by only physicians were members of the FMC. Member physicians in return agreed to among other things the following two things. One, to abide by maximum prices set by the FMC for each type of care. Two, to submit to utilization review of the treatment they provided patients. If the FMC concluded that the treatment was unnecessary the physician would not be paid. This was a very clear example of horizontal price fixing. Indeed, the FMC determined its maximum fees in conjunction with the local medical society. It is not relevant that this is a maximum price fixing because the Supreme Court has firmly placed all price agreements under the per se rule whether they raise low or stabilize prices. The role of price fixing the price fixing agreement in this case was I think to overcome the well-known problem of moral hazard. Under 100% insurance we have two consequences at least. One, consumers have no incentive to search for the lowest price because the marginal price is zero. And secondly, physicians have no incentive to limit the quantity and quality of care provided in a cost effective way. Both will treat the marginal price of care as zero when from the social perspective it is clearly not zero. The maximum price is a substitute for consumer search. Presumably the FMC had a great deal of price information. And secondly, the utilization review is a substitute for a positive marginal price and will discourage physicians from administering too much care. This form of organization is more efficient than insurance without price quantity regulation. Clearly, those insurers who participate in the plan will have lower costs and this can be reflected in lower insurance premiums. Those who fail to participate will therefore be unable to compete. Similarly, physicians who participate will attract more patients while those who do not will lose patients. Finally, consumers will be better off because they can now get 100% coverage at a lower price. The price fixing agreement thus provides an offset to the imperfections inherent in the market for insurance. This is perhaps a clear illustration of the familiar second best rule. We ought not to judge price fixing by static first breast standards when other conditions of perfect competition are also absent. Adjustments under imperfect conditions adjustment under imperfect conditions is not necessarily promoted by trying to approximate a perfectly competitive equilibrium. In terms of the legal doctrines in this area, it's clear that arguing that price fixing arrangements have horizontal price fixing arrangements especially, have certain redeeming social value is not one which fits into the current legal system in the sense that the per se rule really requires us to eliminate consideration of possible advantages of a particular practice. But one thing which is not all that often recognized is that the rationale of the per se rule the underlying rationale of the per se rule is that it is addressed to those kinds of behavior which have an extremely high probability of being antisocial in consequence. And so that the per se rule is itself not something fixed in concrete. That is to say if a convincing case could be made that there are certain classes of activity in which the probability of antisocial behavior are far, far lower than that originally thought by economic and legal analysis. Then the rationale, the underlying rationale of the per se rule breaks down. So I don't think it's impossible by showing the social benefits of certain types of collusive arrangements under certain circumstances that the per se rule cannot be whittled away gradually in the courts or perhaps initially in the economic and legal literature. Thank you. Thank you, Mario. Next talk is entitled Perfect Competition and Economics of Information. And this is Dr. Jack Hyde who received his PhD from UCLA. He's now on leave from the economics department here at George Mason University and is visiting Cal State Long Beach for the year. And he's a sorely missed faculty member in the Center for the Study of Market Processes and his research interest revolves around reformulating microeconomic tools in the market process or disequilibrium framework. Jack. Thank you, Richie, and good afternoon. In his historical contemplation of Perfect Competition, George Stigler wrote, I quote, the complete theory of competition cannot be known because it is an open-ended theory. It is always possible that a new range of problems will be posed and then no matter how well the theory was developed with respect to the earlier range of problems, it may require extensive elaboration in respects which it previously glossed over or ignored. Hardly any important improvement in general economic theory can fail to affect the concept of competition. End quote. Stigler wrote that in 1957, a few years later, he himself was to introduce a new range of problems into economic theory. The problems that are connected with costly information and therefore, imperfectly informed traders. There is a large and growing literature in economics now that passes under the title of search theory. One of the interesting aspects of search theory is that there are a lot of references in it to competition. But there is no systematic treatment of the subject of competition within search theory. And it seems to me that systematic treatment should be of interest to economists. For one thing, one of the essential conditions to perfect competition is perfect knowledge. And that assumption, of course, has been dropped in search theory. Another reason that I think it should be of interest is that there are a lot of references in search theory or a lot of statements in search theory about competition that simply contradict one another. And yet a third reason that this should be of interest to economists is that when you see competition referred to in search theory very often you see competitive in quotes. You even find terms like quasi-competition, whatever that might mean. So I think it might be worthwhile for us to go through and see how perfect competition fits into search theory. My plan here is simply to take the conditions, the assumptions or conclusions that are usually associated with perfect competition theory and to see what extent they are compatible with search theory, to see to what extent they have to be modified, and especially to see what policy conclusions follow from the modification of perfect competition. To briefly anticipate the conclusions, I think that might make it easier to follow the paper. My main conclusions are number one that perfect competition with only very slight modification is compatible with search theory, in fact is essential to search theory. And just to give a name to this slightly revised version of perfect competition, I call it near perfect competition. My second conclusion is that although near perfect competition serves very well as a theoretical construct in search theory, it doesn't serve at all as a welfare norm. As far as I can tell, none of the policy conclusions that flow from perfect competition in general equilibrium theory follow for near perfect competition in antitrust policy. Well, let me start off first then with large numbers, large numbers of buyers and sellers trading. An identical product is usually thought of as an essential condition of perfect competition. And it is obviously compatible with search theory. More than that, large numbers are essential to a theory of search. That is unless you have large numbers, you won't have search. Maybe I should just mention briefly here, search theory is characterized by buyers and sellers maximizing the expected value of a random probability distribution. Now, without large numbers, except for the simplest kinds of distributions, there will be no distribution over which buyers can search. Furthermore, for firms, unless there are a large number of buyers on the other side of the market, the firms can't figure out what their demand functions are by experimenting with price. So large numbers are essential to the theory of search. This has the interesting implication that a monopolist, at least one who charges a single price to all of his customers, really has no place in search theory. If you have a firm out there who is just charging a single price, there's no reason for buyers to search. Consequently, any comparison now between competition, that is any comparison between large numbers and a single seller will be sort of an unfair or an incommensurate comparison. Because on the one hand, you're comparing a market in which information is imperfect, that is the market with large numbers. But on the other hand, you're comparing a situation in which knowledge is perfect. That is all buyers know what the lowest price is that they can find. Now, a policy implication follows from this unequal information of the two different market structures and that conclusion is this, that large numbers are not necessarily better than a single seller. Large numbers introduce costs into a market that a single seller does not introduce into a market. Moreover, large numbers of producers may if they face random distribution of prices, they may produce a smaller output even than a monopolist. Now, Mario mentioned this earlier, this was a somewhat surprising conclusion to me when I came across it. But the reason for it is this, that a monopolist has profits greater than that of a competitor and these expected profits of his will to some extent offset risk aversion that businessmen are assumed to have. Because of this, because a monopolist is more willing to undertake risk in a world of uncertainty, he actually may produce more output than would a competitor in this world. Another condition, this is usually a derivative, an implication of large numbers in perfect competition theory is price taking behavior. When knowledge is perfect, large numbers mean that no single firm can significantly affect price and consequently, at least for all practical purposes, the demand curve facing each firm is horizontal. This implication has been carried into models with imperfect information mostly as an assumption. Much of the work on the theory of the firm in search theory makes this assumption. But the soundness of it has been challenged and I think rightfully so Steve Salop who I understand spoke here at George Mason recently has asserted that when information is costly that firms will not be price takers. That they will in fact face downward sloping demand curves and that therefore imperfect competition is the appropriate model. Search models that take account of both buyers and sellers in a market simultaneously indicate that price taking behavior by firms is not consistent with profit maximization. That is, if there is uncertainty in a market then for various reasons any particular firm even if he forms only a small part of the market can increase his sales by lowering price. Well, if large numbers do not imply price taking then the slope of demand curve cannot be used as a basis for distinguishing between a single seller and many sellers. Both a competitive and a monopolistic market structure will exhibit downward sloping demand curves when information is costly. Now it might be possible I don't know of any work that has yet compared what happens to elasticity of demand curve when you say have 20 sellers as compared to when you have 30 to 40 and so on in search theory. That is, I don't think search theory models have yet examined what happens as you add more competitors to the market. But even if they can establish that as you add more sellers to a market that the demand curve becomes more elastic it won't have the same meaning. It won't carry the same significance in the theory of imperfect information than it as it does in the theory of perfect information. There are a number of reasons for this. I have already mentioned that even if elasticity is infinite the demand curve is flat it does not mean that large numbers will produce more output than a monopolist will. But another reason that elasticity at least it would seem to me won't have the same importance is that we're really comparing two different kinds of demand curves if we compare a say a monopolist to an industry with a lot of sellers. That is, when you're the demand curve of the monopolist is the typical demand curve in the textbooks. It's a demand curve under full knowledge. But the demand curve facing a competitor is a stochastic demand curve. What meaning can we attach to elasticity when the demand curve is a probability distribution? I don't know. I open that question up. But the policy implications of competitive markets and monopolistic markets facing downward sloping demand curves is I think crucial and also fairly obvious. It means that even with large numbers of firms we cannot conclude in a market with costly information that price will equal marginal cost. Consequently, any policy conclusions that are based on the superiority of price equaling marginal cost cannot be maintained. This has rather important ramifications especially for antitrust policy. Most antitrust policy today is predicated on antitrust enforcement achieving a greater equality between price and marginal cost. One of the peculiar things in antitrust today is that you see writers very able antitrust writers, Robert Bork, Oliver Williamson to mention too, who on the one hand want to have price equal they want to have the equality between price and marginal cost as a welfare standard. But on the other hand, they want to readily admit that information is not perfect in a market and antitrust law should take account of that. Well, what I'm suggesting is that if antitrust law takes account of that properly, they have lost their standard by which to prosecute firms. The next condition of perfect competition that I would like to consider in this new setting is the mobility of resources. Costless resource mobility has been a traditional assumption of a perfectly competitive theory. It's costless mobility of resources among different industries is required to show that there are no profits to be made in any industry. It's required to show that the return to each resource will be equal in all different industries and therefore of course it's necessary to conclude that in perfect competition price equals marginal cost. Well, costless resource mobility obviously cannot be carried over into search theory, into theory with costly information without making some kind of change. Just the cost of acquiring information means that it's going to be costly to acquire resources. So we'd have to modify that assumption to mean something like this, that resources are costlessly mobile except for information costs. That's probably as close as we can come to the costless resource mobility assumption in search theory. Well, if we do modify the assumption that much, here's one implication that falls out of this is that we cannot then assume that firms don't do not earn profits. As Grossman and Stiglitz have argued recently costly information and zero profits are incompatible with one another. If it is costly to gather information about profit opportunities but there are no profits to be earned then no one is going to gather any information. They obviously have no incentive to. But if no one gathers any information then how are the factors of production going to be channeled into those areas where they have the highest return and consequently reach an equilibrium state? Well, if costly information then is extended to include information about profit opportunities it means that markets must be characterized by profit rates that are great enough to provide an incentive to gather the information. One direct policy implication of this is that profitability in an industry is not necessarily a sign of monopoly. It is simply a manifestation that information is not free. Another policy implication that follows from this is that any attempt by the courts to reduce profitability any attempt runs the risk of impeding the flow of valuable information through the economy. Now this is a particularly I think it's going to this particular implication of search theory is going to cause particular problems for enforcing antitrust law. Contrary to the assumptions of search theory, marginal costs of gathering information can rarely be specified with any accuracy. And there's no reason for these costs to be the same in all industries. They could very well be higher in some industries than in others. Well, what this means for the courts is that they're going to have a hard time deciding which firm should be allowed the higher profits and which firm should be allowed the lower profits. Not only that, there's a complication to all of this. A firm who discovers a profit opportunity and enters a market and bids for resources does other people in the economy a service? They perform a service. What they do is they bid up the price of resources to more accurately reflect the true value of those resources in the economy. And that information is passed along to everyone else in the market free, so to speak through the operation of the price system. Well, what value should a court place on this? Well, here is a firm that is earning profits that is at the same time providing a valuable service to other people in the economy. What value should be placed on that service? This also has implications for the evaluation of collusion. See, a price system may not be always the most efficient way to transmit information. It may be that your efficient way of transmitting information is for firms to collude. Mario has already covered this in some detail, but one of the implications that falls out of search theory is that in general collusion can be an efficient method of gathering, transmitting, and using information. Even therefore if firms collude and even if that collusion yields a profit to them it cannot be condemned per se. That profit may be the result, may have the beneficial result of generating and transmitting valuable information. And finally the last condition of perfect competition that I want to consider here is perfect knowledge. Perfect knowledge is obviously incompatible with search theory. In fact, the modification of this assumption is a very reason for the existence of the theory of search. But even if knowledge is not perfect in search theory it is still very extensive and it still plays a crucial role in the logic of the theory. Here is why that is. Search theory retains for the retains as the sole method of decision making economics maximizing. Buyers maximize the expected returns from searching over a distribution for a low price. Sellers maximize the expected profits by maximizing their expected profits by searching for the optimal price to charge customers. Here is what we want to ask. What kind of information is required that buyers and sellers can perform this kind of maximization? Here is what buyers have to know. They have to know all the prices that are being charged by the sellers and they have to know the frequency with which each price occurs. They also have to know the cost of search and they have to know where to search. Without all of that information the computation of a reservation price and the strategy of searching is not necessarily optimal. What firms have to know? Firms have to know enough to compute their mathematically expected profit functions. What firms have to know to compute this varies depending on which search model you read but generally speaking here is what firms have to know. They have to know a great deal about the reservation prices of consumers. They have to know a lot about the search strategies that consumers are going to follow and they have to know a great deal about what other firms are charging or are going to charge in the market. The knowledge that the business firms have to have in order to maximize their expected profit function is also very great. If we extend our horizon to models with more than one good information that consumers and firms have to know just in a partial equilibrium model if we extend our horizon to include a more general model buyers would have to know additional information. Here is what they would have to know. They would have to know every good that was available for them to purchase. In other words they would have to have the Sears catalog for example memorized. They would have to know the utility that each unit of all of these goods would yield them. They would have to know the price distribution of every good in the market. They would have to know where to go to search for the low prices in every good in the market and they would have to know the cost of searching. If we look on the other side of the market firms well firms have to know then not only in a general model firms have to know not only the expected profit function in their own industry they have to know the expected profit function in every other industry as well. Otherwise they may commit resources to an industry that really does not have the highest expected profits for them. So while the knowledge required for agents to maximize probability distributions isn't perfect it's not perfect it's very nearly so. I call this knowledge requirement near perfect knowledge and consequently the idea of competition that's based on near perfect knowledge I call near perfect competition. Near perfect knowledge and near perfect competition are as essential to search theory as perfect knowledge and perfect competition are to neoclassical theory. Well to briefly summarize and to make of all this here's what we conclude we can conclude that with only slight modification the conditions necessary for perfect competition are also applicable that is also compatible with economics with costly information. Even more important we can conclude that conditions very similar to those required for perfect competition and equilibrium theory are also required for search theory and consequently for near perfect competition unless there are large numbers there's no random probability distribution to maximize unless resources are mobile at fairly reasonable costs there is no incentive to gather information on profit opportunities unless there is near perfect knowledge maximizing probability distributions does not lead to optimal decisions. Now I might mention here that large numbers in near perfect knowledge are used in all in every search model that I have come across and assumptions about resource mobility I think will become more important as the theory is extended from partial equilibrium to general equilibrium. So as far as positive analysis is concerned that is as far as analysis of economic markets with costly information that is all captured in probability distributions we can say that near perfect competition performs the same function for that theory that perfect competition does for equilibrium theory but it doesn't come anywhere near to providing the same normative functions. Large numbers do not imply price taking by firms we don't even know for sure if larger numbers imply greater elasticity of demand and even if they do there is no assurance that this will lead to greater output. Therefore any conclusion based on equality between price and marginal cost is out the window when there is costly information about profit opportunities is incompatible with a no profit equilibrium and the external benefits of having information transmitted through the price system make it almost impossible to formulate a workable antitrust policy. I think the main conclusion to be learned from all from this is the following that once you introduce just a little bit of incomplete knowledge just a little bit of uncertainty in the markets and believe me search theory doesn't introduce very much it's not a realistic theory by any stretch of the imagination but all you have to do is introduce just a little bit and all the welfare conclusions of perfect competition simply fall to the ground. When information is costly we have no economic rationale for antitrust policy. It appears that Stigler in the same article that appeared in 1957 was prophetic when he wrote and here's another quote from Stigler the time may welcome when the competitive concept suitable to positive analysis is not suitable to normative analysis. Well I think for economics of costly information that day has arrived. Now we'll hear from Dick Langua in a talk titled non-conventional approaches to the theory of competition. Dick received his Ph.D. from Stanford in engineering he was associated with the Austrian economics program at NYU for three years where he also taught in the economics department in the graduate school business. His current research interests revolve around the role of information economic theory and the internal organization of firms I think since Jim Buchanan plugged his book this morning we may want to plug a book for Dick. I understand he's editing a book on economics as a process. Thanks. Thanks Rich. I get the feeling listening to the conference participants today that if not the title at least the theme of the conference is what's wrong with antitrust and I think there are at least two ways you could think about one of them is to say as I think Professor Brozen at least implied if not actually said this noon is that the problem with antitrust is that people haven't applied haven't as he put it examined antitrust through the lens of price theory enough that is we haven't applied economic theory enough to antitrust in many ways that was what I heard coming through in some of the morning sessions but there's another and indeed a positive way we could think about what's wrong with antitrust and that's in some way the theme of this session and that is maybe what's wrong with antitrust is precisely that we've applied too much economics to antitrust policy that there may be something wrong with the theory itself and something we should concern ourselves about and maybe there might indeed be something in the very way we think about the economics of competition that leads to discussions about antitrust policy I want to explore that second approach with you right now before doing that let me let me make a distinction that I think is very important and that's a distinction between what I would call what other economists have called formal theory and appreciative theory that is to say there is on the one hand the theory that the formal theory that economists who call themselves theorists write to one another in journals and there is on the other hand appreciative theory that is the theory that economists use to explain things to themselves to explain things to people who aren't economists and perhaps to do antitrust policy and there's a case to be made that while the conventional appreciative theory is very subtle and very subtle it's not and it is adapted to the diverse problems of the antitrust law that the formal theory is left behind somewhere that when a good economist talks about what's wrong with antitrust examines it to the lens of price theory it's not the lens of formal price theory but the lens of a more down-to-earth appreciative theory it's formal theory that I want to talk about today and what I want to do is compare what you might call the conventional wisdom price theory or the theory of competition as it applies to antitrust with potential alternatives a couple of these view themselves as genuine alternatives and at least one of them I will claim is best view perhaps is not an alternative so much as a pre-emptive counter-revolution attempting to save them more conventional theory while these four are first of all the conventional theory which sometimes goes under the heading in industrial organization courses of the structure-conduct performance paradigm but I mean mostly straight ordinary everyday neoclassical price theory. I also want to talk briefly about the Austrian theory about so-called evolutionary or neo-shampaterian views of economics and finally about the new theory of contestable markets and what I want to do is describe those to you briefly and perhaps a little too carelessly, a little too carelessly both from an economic methodology and philosophy of science point of view and perhaps also a little too carelessly from the economic point of view but there's not much time and I hope I can at least give you the flavor of these things and then discuss what implications for theory, excuse me what implications for policy one might want to draw and sort of the intuitive draw sort of intuitively from these formal theories well I think I can be fairly brief on neoclassical theory with the so-called structure-conduct performance paradigm appended to it Mario and Jack have already talked a little bit I think about some of its some of its shortcomings you really have to go back to the 19th century to understand what's going on in neoclassical price theory it really started with Cournot and others in the 19th century and they had a vision of how economics should work and what economics should be about and that vision was in some ways to make economics a little more like the natural sciences of the day that's often been remarked upon but more particularly they wanted as well Rob put it, they wanted a science that was independent of human will that is they wanted the conclusions of this science not to depend on how people behave or how people act in some sense so what they did is they transformed a concept of competition that Adam Smith had possessed in which competition was what competition means essentially to the businessman it's an active verb it's a behavioral notion of competition when something is competitive when there are people competing against one another in some behavioral sense and the early neoclassical economists translated that into competition as a state of affairs competition in neoclassical theory is not defined as anything you do as a state of affairs that is a state of affairs in which all economic agents are price takers basically there are adjoint assumptions to that but basically in which no agent can affect price and the way we examine the welfare implications of such a theory is at least as Corneau had hoped look only at the structure of the economy that is count the number of firms and in fact Corneau wrote down a function that gave you the welfare implications of a market structure of the number of firms he could tell you what proportion of the competitive output it's N over N plus 1 if you're interested times the competitive output is generated by any structure of firms in the economy and while some of the formal theory and most of the appreciative theory of neoclassical economics has progressed beyond that the basic presumptions of the theory are that are very much as Corneau left them it's a static and timeless view of the world which market structure is predominant now if you believe this and then you go out and look at the world you begin to have this gap between formal theory and appreciative theory or as the philosophers of science might put it you begin to see some anomalies in the theory but as you say well this can't possibly be right a number of people have remarked well you can have any number of structures and still have any give me any structure you want and I can come up with any set of any along the spectrum of welfare implications for that structure under certain assumptions a duopoly might be just as good as a perfectly competitive firm well that's that's only one of the anomalies that came up and people dealt with that by adding to the pure structuralist view the conduct part that is the structure conduct performance paradigm says we'll assess performance by looking not just at structure but by looking also at conduct okay well that anomaly in a sense generated another one because then you once you put behavior into the equation then your whole theory of imperfect competition your whole theory of collusion of oligopoly falls apart because there you can't do with what I wanted you can't take out human will you've got a game theory problem if you've got only three firms you have this endogenous uncertainty that Mario talked about you can't predict anything it could given one set of conjectural variations as the game theorists say come up with one answer given another one you could come up with another so there's really no as many economists would concede there exists no neoclassical theory of oligopoly not one that's consistent with the philosophical foundations of neoclassical economics another problem with it is I think perhaps the most glaring problem to which Professor Brozen also alluded in his talk today is that it's very difficult to reconcile the microeconomic welfare implications of the neoclassical model with sort of macroeconomic performance of firms that is to say the neoclassical model in its pure form would lead you to suspect that the best structure from a welfare point of view is perfect competition that means a zillion little firms or as the hyper mathematical theorists would now put it firms all of which have a measure zero they're actually more than infinitely small this is the optimum arrangement yet we see that where all the progress lies in society as Schumpeter pointed out in the 1940s we look where all the progress is in societies where the cost is screaming down where all of the markets are being filled it's by big firms it's not by little firms we look at where the little firms are and their mom and pop grocery stores and people who grow rice and stuff like that and they're not where the engine as Schumpeter would put it in the sense neoclassical welfare implications with the more macroeconomic ones there are other problems for example just the neoclassical model simply assumes the structure of the market whereas we see markets progressing over time the neoclassical model sort of sees structure as if not entirely exogenous something that's given as indeed a control variable it's something we can play with and it's not something that's we're part of the process itself well other theories have tried to address some of these anomalies if you'll let me call them anomalies one of them is the Austrian theory which to some extent I think Mario and Jack have hinted at that goes back certainly to Carl Manger and includes names like Hayek and Kersner in some sense an attempt to do is resurrect Adam Smith's notion of competition the idea that competition is a behavioral activity and competing is a process that's something you do it's not a state of affairs they also have in some ways a different view of rationality that is they're not obsessed as Wallra or Cournot were with getting free will out of the system in the particular way that the neoclassical economists were so they're inclined to define rationality in a different way and they see rationality to do more with learning over time rather than making an optimal decision given a static set of given set of of resources Kersner's version of this will stress something called entrepreneurship that is the fact or the possibility that there are always people ready to jump into the system with new ideas as it were to seize on profit opportunities to snatch quasi rents if you will and this is really to Kersner this is sort of the engine of the system it's a process that's moved along by these entrepreneurs in Kersner's view in a sense the market is inherently competitive because if you postulate as an engine these entrepreneurs who are always out there competing in a behavioral sense then by your definition of competition the market is sort of inherently competitive this does better on these anomalies. You can reconcile in a sense larger firms or firms that are taking active price making active price moves in the market with good welfare implications it tends to the problem of oligopoly being indeterminate goes away because you've reformulated the problem in such a way that this isn't a problem basically and you allow for an endogenous market structure that is you allow market structure to be generated by your model itself it's not something that's outside of the model or control variable okay there's certainly some differences if not problems with this view especially it's not very specific about how all of this is supposed to work and never gets for example if we're worried about large firms there really is no Austrian theory of large firms but the theory is at such a level that those problems in some sense don't arise now the third view that I want to talk about briefly is in many ways closely related that is the evolutionary the evolutionary approach and this I associate mostly with names of Nelson and Winter who've recently come out with the book on the subject but in many ways this might also be called an Austrian theory in the sense that the book is motivated by an attempt to capture the spirit of the writings of Schumpeter who was of course also an Austrian this evolutionary model focuses attention on some different parameters of the system for example it's concerned with diversity in the economy Mario talked about frictions okay well as an evolutionary theory very much like biological evolution this evolutionary theory accepts the idea and depends on the idea just as Mario suggested that you have to have these frictions in the economy you have to have diversity that it's something that doesn't make any sense it won't work if everybody has the same cost function you got to have diversity because there can't be selection without diversity there's also got to be some frictions in the sense that there has to be a habit in convention in the system that people aren't constantly running around optimizing things that there are just too many things for them consciously to optimize so they can be a little resistance in the system and that's how you get the genetic part of the selection process they also have a behavioral view of competition they also have a learning view of rationality that rationality is more learning in the process than it is making optimal decisions in any sense differentiates this a little bit not only the explicitly biological aspect but also the Schumpeterian aspect differentiates this from what I've called the Austrian theory that is these writers are more interested in the problem of innovation following Schumpeter as I say Schumpeter was concerned with the question of the issue that big firms tend to do well but more than that he was concerned with his idea of entrepreneurship which if we could call Kersner's idea of entrepreneurship entrepreneurship in the small Schumpeter's idea was entrepreneurship in the very large that is he was concerned with massive changes in the system massive innovations in the system how things change not how one industry takes a given set of resources a given set of knowledge and allocates it in some optimal fashion but how industries change what we think of as knowledge change the cost structures of industries change the markets invent new products and so on okay so these writers try to build that idea of what Schumpeter called creative destruction into the market now this this theory also does reconcile some of the is able to deal with some of the anomalies okay it also can reconcile the welfare implications of large firms with the fact with sort of its view of market structure and it does this in some ways in a more detailed way than the Austrian theory does because they are concerned with the theory of the big firm how firms grow by what causes big firms what are the ramifications of innovation versus imitation for example again the problem of oligopoly goes away because the whole premise of the system is changed in such a way that this is not a problem that you have firms instead of assuming that there are three big firms and that they are fighting with each other in some sense and trying to out guess one another in the way that Mario described you have a system in which firms the size of firms is the result of the process and not the input to the process that is oligopoly is not something you start with it may be something you do or do not end up with as you watch the process evolve over time and this is again much closer I think to our appreciative theory of how things work when we talk about the micro electronics firms and how the shakeout is going to work in the small computer industry for example we are doing this what they would call a reverse shampaterian model that is where firms start off small and then end up big as innovative firms beat out one another or try to excel try to get better than one another in a behavioral behavioral sense and again this one also in a very real way stresses market structure as endogenous and it is concerned with the solution of market structure the fourth cluster of models or theory that I want to talk about is called the contestable markets theory that is recently been this is something that has come to the fore in the last couple of years it is associated with especially the name of Will Beaumont a former colleague at NYU and a number of other people what contestable markets seeks to do and let me say it basically stays I think the key thing to understand about contestable markets theory is that it basically stays within the premises and framework of the theory set down by Cornot and Wal-Rah in the 19th century but it seeks to replace the standard of perfect competition of many little producers with perfect information free entry who can't affect price with the idea of perfect contestability and what they mean by perfect contestability is a market in which it is perfectly costless to enter and perfectly costless to exit okay but they mean costless not in the sense that Mario meant it but in precisely the opposite sense that is I mean free entry in precisely the opposite sense it has to be costless in the sense that Stigler wants to find that is someone can enter the market without any cost disadvantage over the existing firms or under the existing firms that they have this if somebody jumps into the market and tries to attack IBM and personal computers that they can produce personal computers at the same cost as IBM and it makes all of the it takes in all of the standard assumptions about what one means by a cost curve and so on this does deal with I mean in some ways I understand this theory as precisely an attempt to deal with those three anomalies that I talked about but keep it within the traditional framework okay that's one way to understand what contestable markets is for example in the second anomaly the one about oligopolies in this problem of indeterminacy in the theory of oligopoly they get around that completely because if you have free entry in this sense people always running into quasi-rents then the conjectural variation of the incumbents is perfectly determinate you don't have to worry about what it is okay and they prove this with mathematical theorems and so on notice that this is very similar they've done this in a way that's very similar to Kirchner's idea of entrepreneurship the people what disciplines the system is not the structure so much as it is the possibility of people running in and snatching quasi-rents but since his cohorts insist on retaining the neoclassical conception of rationality they have to define this as free exit free entry and exit in the neoclassical sense because all actors are rational and have perfect information and no rational actor would enter a market if he had to commit fixed costs that were higher than the cost of the incumbent and so on Kirchner doesn't have to worry about that because he doesn't necessarily adhere to that definition of rationality that it may be possible for entrepreneurs to run in for all sorts of reasons to get quasi-rents there can be divergences in people's estimates we can have more radical kinds of uncertainty in which people will enter anyway even though it's quote irrational if you know all the facts but you don't know all the facts and even if you did know all the facts you might not interpret them all the same way so they handle that in some ways the same way also they get around the market is that you're not worried about market structure anymore what they show is that you don't need perfect competition in the sense of atomism to have good welfare implication again these are good welfare implications in the neoclassical sense of no dead weight loss triangles and all that sort of thing you can have any kind of market structure monopoly, a single monopoly seller might be a problem but duopoly and on up that's fine any number of players is consistent with the Pareto Optima first best indeed Pareto Optima now this is in some ways responding to the anomaly that the Chicago school and other people have pointed out saying that if you have a pure structuralist view it doesn't make any sense because I can give you plausible cases why any market structure would reach an optimum well they're sort of conceding that point market structure can concede an optimum as long as it's perfectly, perfectly contestable the contestable market writers also make the claim that they're superior to the standard theory in the sense that they also can generate market structure endogenously that unlike neoclassical theory they don't take market structure as exogenous well I think careful reading will suggest that this is a somewhat overstated claim in fact market structure is endogenous in contestable markets theory in precisely the same way that it's endogenous or exogenous in standard theory that is if you give me the assumptions and the cost functions then I can calculate a quote optimal market structure okay maybe that I get a different optimal market structure under contestable markets but to say that I've generated because I can calculate an optimal market structure to say that I've therefore generated a market structure endogenously as I think stretching it at least in the sense that it could certainly also be held true of the standard theory okay there were some other implications of contestable markets that I won't go into I'll probably leave those to the Marxist whether the right wing Marxist or the left wing Marxist since I'm neither kind I won't go into but these have to do with what we might call the AT&T problem that is contestable markets have the following implications if a market is perfectly contestable that is a monopoly can be good a monopoly can be under certain restricted assumptions a monopoly can be second best optimal maybe even first best optimal a monopoly like say AT&T and this is especially true of a multi-product or at least is also true of a multi-product monopoly like AT&T but while it may be optimal in the sense that it perfectly allocates resources in a given period over time that is if free entry is permitted free entry in Mario's sense is permitted okay over time then you'll get the waste of competition and inefficiency so what this says at least or it might say to a left or right wing Marxist is that these writers all of whom were in the employ of AT&T when they came up with these theories have come up with a theory that says that what's good is to have a regulated monopoly that doesn't we don't allow anybody to have free entry to compete with that monopoly but as I say I'm not going to pursue whether these conclusions are right or wrong okay what are the some of the problems with contestable markets theory well one of the problems as I hinted is that it retains all of the assumptions basically of neoclassical theory a lot of the assumptions that Jack and other people have attacked is being not very not very useful however you like that I won't draw conclusions but it's important to notice that it's and sometimes contestable markets theory is used to try to generate dynamic conclusions about things okay there was an interchange in the American economic review a few months ago in which some writers pointed out some problems with trying to apply contestable markets as a dynamic theory and Beaumont responded that while you misunderstand it's purely a static theory it has no dynamic implications at all which is certainly a weakness for a theory that sort of in some sense advertises itself as a way to get around some of the static problems of neoclassical theory a more important problem is that it's not at all clear what happens if you have any deviation what the welfare implications are of even slight deviations from perfect contestability okay maybe the theory needs to be developed to work these out and so on but in perfect competition there's sort of a spectrum at least in the pure corno version of it if you get a little bit away from perfect competition well that's only a little bit worse than perfect competition what happens when you get away from perfect contestability nobody knows it could be terrible okay so this is something that they need to be worried about how much time do I have okay well I probably don't need that much let me sum up by they're giving you a view of what the sort of heuristic implications are on some level of each of these theories for antitrust policy that if again if you take seriously the formal theory what is it how does it affect your intuitions about what you should look for an antitrust policy okay well if you take the traditional formal theory what you want to look out for is structure you want to look out for how many firms there are in the market come up with Herfindahl indexes and things like that if you're pushed a little bit and get away from the purely formal theory and throw in a little game theory then maybe you want to worry about what some people call behavior and other people call conduct that was whether there's collusion but you interpret that not in the subtle way that Mario did but in the simple minded sense that only thing collusion can do is raise is raise prices okay so you want to watch out for collusion also you want to worry about cost curves okay now if you're a any kind of a sensible economist you'll recognize what's called the Williamson some people call the Williamson trade off that is sure having a smaller number of firms might be might be bad in the sense that these firms will collude and raise their price above marginal cost but these bigger firms may have lower cost to start out with in the smaller firms precisely because they're big and can do economies of scale so there's a trade off there a trade off which by the way is not permitted in the current antitrust laws that efficiency is not at least my understanding I may be stand corrected is not a defense in an antitrust case so if you say well sure we're big and we're colluding and so on but we're much more efficient than all those guys who weren't colluding anyway that's still not a defense okay so those are the things you want to look out for if you an Austrian theorist well here you have this idea that the market is by our definitions of what a market process is in some sense inherently competitive that is unlike the unlike the neoclassical case you don't necessarily suspect that there is a natural tendency towards monopoly or that monopolies can arise sort of arbitrarily and without cause sort of monopolies that sort of spring up sort of spring up there's always competitive forces in some sense these entrepreneurs waiting around to snatch quasi-rents that are going to discipline the system this makes you somewhat skeptical especially if you add in the subjectivity, the subjectivism view of cost that Jim Buchanan alluded to this morning it makes you skeptical of one's ability to judge the welfare properties of various market structures okay what it might lead you to do but again the theory is it's such a high level of concern to the details you sort of rule out even looking at some of the details of whether this Williamson trade-off or the shrimpaterian trade-off which is the Williamson trade-off and other guys are operating you don't care you're less concerned perhaps with whether there are temporary raises temporary rises of price above marginal cost you're more concerned with the long run view of the process but what it might lead you to look for is resistance to competition okay and that would mean certainly government intervention but it might mean other things you might be able to make statements about how an organization of the firm or an organization of markets might or might not speed up this inherently competitive process indeed you might want to link this whole theory with the theory of property rights and a property rights approach to monopoly that other Austrian writers are trying to do if you have the evolutionary or the shrimpaterian view you have some of the same intuitions as the Austrians but in fact a lot of these writers are perhaps less unwilling to plunge into the details of competition they would say maybe we can make statements about antitrust and maybe antitrust isn't a fruitless thing but we want to look for entirely different things for under neoclassical economics for one thing you want to be concerned with innovation you want to think that innovation is far more important or at least as important as a static allocation of resources that is what structures in the market or what market structures if we have to talk about market structures will lead firms to be more innovative which what set of property rights among in terms of patents, appropriability of information and so on will be most conducive to an efficient use of R&D and to a faster innovative innovative process you'd also be concerned with institutions and how they might generate diversity in the economy and you might be concerned with questions of gaining and losing diversity okay as I say you might be concerned with the problem of appropriating knowledge and how not only strict property rights but de facto property rights and knowledge are set up and used in firms and you might want to judge firms by a different standard you might want to judge the economy by a slightly different standard instead of worrying about an optimal in some sense if you've got a world that's always innovating and changing you might be more interested in looking at things like flexibility and adaptiveness of the system rather than whether it's efficient at any point of time you might want to worry about what some writers call the Schumpeterian tradeoff that is how do we trade off or how do firms trade off between static efficiency allocations in a given period and innovativeness over time do we have to give up a few dead weight loss triangles in the current period in order to get firms that are more innovative in the long run this is essentially the gist of Schumpeter's discussion in capitalism, socialism and democracy that we have sure firms big firms don't set price equal to marginal cost they're not price takers they do all this nasty stuff that looks like bad static allocation but they're dynamic the process that they're dealing with is a much more important process in fact Schumpeter said that the process of innovation is to static efficiency views of competition as a frontal assault is to break in at the door it's like trying to blow the whole building down instead of forcing a door and Schumpeter at least believed that we should worry much more about that side of things if you have a contestable market's view it's not entirely clear to me perhaps I'm not as familiar with the literature as I ought to be but it's not entirely clear to me what your views should be you're still going to be concerned with cost you may not be as concerned with structure as you were in the past because we've agreed that it's a perfectly contestable ideal structure itself doesn't matter any structure could give you good results but you're still interested in cost curves because you're worried that you've got this new classical version of rationality which everybody knows everything already and so nobody's going to jump in until they're sure that they're not going to give up they're not going to commit themselves to any fixed cost so you have to know about cost curves in order to on the one hand know whether the market is contestable, whether people are going to jump in or not and on the other hand to deduce from calculations what the optimal market structure will be to know that you have to have cost curves you're still worried about that and again there are these AT&T implications that I'm not going to speculate on what is the benefit of looking at alternative theories okay well again one of the benefits I think is that the formal theory however far away our appreciative theory gets from the formal theory the formal theory still tugs at us it tugs at us with metaphors it tugs at us with words and if antitrust is as professor Brozen suggested at lunch a matter of accusation by labeling then it matters how our theories label things if our theory says that to go out and actively adjust change prices and bring new products to the market is defined as competition that's a very different label from saying that that's predatory pricing or that's inconsistent with our definition of pricing which we admit is just a technical definition but the words have resonance and the words are important so if there's no and I think there are a lot of other reasons for wanting to think about theories but if there are no other reasons there's at least that reason I have 15 minutes of questions and answers and I'd like to ask the audience to speak up and I'm going to ask the speakers to repeat the questions when they get up to the podium so that everyone in the audience can hear Any questions? Yes? The question is why did I leave the consumers out of the picture of endogenous uncertainty I'm not sure that I did so first of all I'm not sure I understand the import of the question why the questions being asked exactly but I didn't I didn't leave the consumers out with any purpose in mind other than to focus in on a particular aspect of the problem which was I thought important and there were time limitations the well could you make more explicit what you mean? For example sure I think that's an important point I'm confused by your use of the word collusion to a lawyer which I am and in the dictionary collusion necessarily has as an element secrecy, concealment or fraud I assume none of you meant that when you use the word collusion could you tell me what you do mean? I think for purposes of the antitrust law whether the agreement say to fix prices not to use the word collusion again an agreement to fix prices per say illegal whether it's secretive or not I mean if I in fact it's almost necessarily going to be secretive in a regime in which it's illegal to do so so it's self fulfilling in that sense but I don't as far as what I'm discussing whether the agreement is secretive or open is not relevant the hesitancy of entrepreneurs to invest they don't know the intentions of their competitors if I understand you if this were a serious widespread problem in the economy we should look at the economy and we should expect to see a relatively large number of cases where there was unfulfilled demand opportunities for investment and say standardized products such as petrochemicals and prices above the level necessary to draw in capital and shortages of those types of products and I would like so that if this type of problem were at all widespread in the economy would you not expect that type of to be able to look at the economy and see those types of cases fairly often and in fact do we it's not my experience it's my experience that that is an extremely rare occurrence in our economy I suggested to Richie think that Dick Languaugh and Jack High answer questions on my paper and I will answer questions on their papers but I don't know if that's permitted procedure why if what I discuss in terms of the question is if what I discuss in terms of endogenous uncertainty is at all important why you would expect to observe presumably unexploitable profit opportunities in the number of industries and so we would see fairly high profit rates in those industries over periods of time first of all I think in discussing sort of the eight factors which make for beneficial collusion the other way of looking at those eight factors is when they're not present or when their converse is present okay when there is when some of these quote-unquote frictions are present a collusion will be less necessary less socially beneficial and so empirically one could argue that in many industries we don't really you know collusion isn't really necessary a social perspective to ensure coordination because enough frictions quote-unquote are present in order to do that even many fervent trust busters would agree that collusion is not the predominant form of arrangement or attempted collusion would not be the predominant form of arrangement in the American economy even if we abolish the antitrust laws that there are factors which make that more likely so I guess my roundabout answer to your question is simply that I would agree that probably in many industries the factors making for the benefits of social benefits of collusion are not present but I would also say that we probably do observe industries in which profit rates remain high over periods of time perhaps higher over longer periods of time than might otherwise be the case and I think that some of the explanation for that need not rest on monopoly or actual collusion but rather rest on the difficulties in exploiting profit opportunities but I agree with you ultimately that it's an empirical question and one has to look at the specific cases to determine whether these factors are important but I think the beginning ought to be made on the theoretical level to specify the conditions under which collusion might be socially beneficial and then once those things are carefully worked out to then go to the empirical level to see how it works out in practice and if it is indeed important. I think it may be legitimate to take a moderated prerogative to suggest the study by Rob Bradley who is working in Houston where he showed that the oil industry by colluding together was trying to solve the problem of free riders on common oil pools or private property. So they'd sink different oil wells on different pieces of land parts on the surface and of course for every gallon they took out they depleted their neighbors so they overused that resource and what they did was they got together and cartelized or colluded in order to conserve oil during the 1920s and that's the study I understand that the Kato Institute will be coming out in I think in terms of spring semester. And I would agree with that but it seemed to me that he suggested that this is some type of a political situation that is the case and I would disagree with it because I think those high information costs create opportunities for other opportunities to come in and reduce search costs for buyers a very good example that right now I'm thinking places that Peter saw for the market where that tens of thousands appropriate any other and now there are entrepreneurs coming in trying to set up computerized systems to in effect reduce search costs and I think that that is when they have several campaigns, number one it's a problem to reduce the number of sellers and come closer to equalization in that way and also break it down to a reduction in price to begin the reducing the cost of the buyer Do I need to repeat that? Did everybody hear it? Everyone heard it. I agree with you I'm talking about search theory and you're talking about the real world and those are two very different things there are no entrepreneurs in search theory they do prove equilibrium price distributions under very restrictive kind of assumptions one thing they rule out are the introduction of new products and the lowering of costs so they're just two different worlds maybe I should say something that I'm very critical of search theory although you would never know it from the talk I gave here today and the only purpose I had in going through an examining competition in search theory is that search theory introduces the smallest amount of ignorance possible into economic models and what I wanted to show is that even if you introduce just that much that we have no longer any economic rationale for antitrust laws that's a good question well it's not that you can't have market theory it's that you you can't justify the market on purely economic grounds now I see two possible avenues if we want to throw out the conventional price theory justification that is price equals marginal cost and all the optimality conditions I see two possible avenues if we want to throw out the conventional price theory and all the optimality conditions one is to try to develop a separate set of criteria by which to judge performance that are purely economic and the other which is the view I take is that we economists just ought to be open about the fact that we can't justify the market unless we want to make some kinds of ethical statements and ultimately the justification for the market economy rests on ethics not economics this is the office where Jack is saying he needs to make ethical statements for the market the welfare economists are making ethical statements but they call them scientific but you have to make some statement about something someone tries to call a marginal a price approximating marginal cost in a particular dynamic I have to repeat that one because I'm not sure what it is and you repeat it, I'm not we're talking about positive economics we tried to prevent all some regime which may or may not include if not precedent I trust policy some judges seems to be we're saying economy can't make a positive statement about whether markets are serving probably what you're saying but somehow we're making ethical judgments I'm not sure you have to send it away that way we can model there may be models in which something analogous to price approximating marginal cost is yielded by somebody's prior set using the word prior for a set of statements about the future and perhaps there's that sense in which positive economic conclusions can be reached that are not necessarily ethical conclusions I'm not sure how this sort of statement squares what the statement you just made it doesn't square with my personal opinion but I did say that economists really have two avenues open to them one is to try to develop a set of criteria which are quote positive non normative by which we can judge the market but the point is that criteria or that criteria those criteria cannot include price equaling marginal cost but it may well be I wouldn't want to discourage anyone from trying to develop a criterion by which to judge the market that is non-ethical just to my anything without any statements about the relevant well one in which price is always happening and one in which there's no such prescription there's no such requirement yeah I think he would be better off under the second system I know so that's a value judgment the 20th century he he said that you're making a value judgment in praising the 20th century one last question it strikes me from hearing Dr. Landau's comments as well some of the others that what we're leading to is a system of economic analysis that basically throws out most of the rules and that we come up with a judgmental ad hoc situation with each case and the legal system tries to create some certainty for folks who operate within it by creating rules are we going to a situation where the antitrust laws are basically irrelevant to the legal system and we can't use it anymore the question is best I understand was whether the implication of what we've all said here is that is that we've we have to in some sense throw out all the rules of economics and sort of judge every case on an individual basis without some fixed rules because of the turmoil in the theory and what does this say about antitrust is this the gist of what you were getting at it basically leads to so much uncertainty that lawyers have no way to apply what you're... what I was saying was in some sense just the opposite I was saying that is what the rules are of antitrust what you should look for and my contention was that because the formal theory was generated for well for a lot of a lot of reasons economists have gotten... have begun to use what I call the appreciative theory that's much different from the formal theory and this appreciative theory generally consists of analyzing special cases on their merits that we can't say in general that such and such a market structure is bad so let's look at the cost curves let's get in and get the data let's do this and see if this industry is competitive or not we'll make our own value judgments as good down to earth, nitty gritty industrial organization economists and that's what all good industrial organization economists do they go in there and get cost curves and they make judgments about industries based on their experience and so on and that's all case by case and the reason it's case by case it says we should do the following things about antitrust and what I was saying is exactly the opposite of what you were implying is that why we want better formal theory is exactly so we do have more predictability that we say maybe we'll want to have a theory that's based on property rights whether property rights allow entry or not that would be sort of an Austrian property rights version or maybe we want to have a theory that says we will have certain rules about market structures which market structures promote innovation and which ones don't but that's a different theory and that's very different from what we have now and I agree with you completely that what we need is more predictability in the theory lawyer is not the economist that because the lawyer is mostly on the defensive they're defending the case they're not the pros and cons that if they had to approach it like they're saying there is theories and the theory is it all depends and that's the best type of approach I would think of in the standpoint of the lawyer because he can't get this theory he can take this theory or this theory and end up winning his case the remark was that maybe he's better for the lawyer so there could be some indeterminacy in the theory so the lawyers can adapt the theory to their needs not so much the lawyers as it is the business man who need predictability I think this is an important question because in my research into the economic aspects of areas of the law which I think make more sense than antitrust like the tort contract evidence etc. areas the importance of rules of thumb is quite evident that courts are reluctant to have a rule which is fine tuned to each specific case even at the risk that they will do the wrong thing in a particular case for lack of fine tuning that they will still insist on a rule of thumb because it economizes on time and resources in ways which are economically very important but for example my recommendation that the per se rule against price fixing be shifted to a rule of reason it's just an interim suggestion ultimately what I would advocate is a strong presumption of legality in all cases of price fixing put the burden of proof heavily on those who claim that the price fixing is antisocial with the consequence that probably very few situations would be found to be illegal and the whole area of law would shrink down to almost zero which I think is almost the optimum I'd like to do two things so don't run yet I want to change the schedule slightly and I want to take a 20 minute break so we have a half an hour and let us get out of here 10 minutes early which I think since it's a Saturday afternoon we might be in favor of so there will be coffee I understand in the next room again and we can pursue the questions in that more informal setting and then I'd like to thank the panel