FOR the economy to work well, resources should flow freely from one industry to another in response to changing demands and costs. In textbooks, this process is via the capital market and entry and exit of firms. But entry can be by new firms or existing ones through diversification. Williamson (1970) and Weston (1970) stress advantages of internal capital transfers over the market.' Gort (1962, p. 4) and Rumelt (1974, p. 2) state multimarket operation of firms will speed redeployment of resources in response to profitable opportunities. They would be right if multimarket operation were not coincident with multimarket contact. While diversified companies may have advantages that would facilitate the movement of capital, they have enhanced opportunity for coordination if they meet in several markets. I term multimarket grouping the phenomenon of groups of diversified firms whose activities span to a significant extent the same markets. Multimarket grouping of sellers could reduce the flow of resources, thereby inhibiting a socially desirable competitive process, if it proceeded until mutual dependence among diversified sellers was recognized and reduction in competition coordinated, tacitly or otherwise. In short, where sellers have grown large through diversification, resources may not be efficiently reallocated among markets in response to changing conditions because interdependent groups of sellers recognize that reduces their profits. Even then, in a frictionless world, sellers other than those in the multimarket group could move resources into profitable areas. I reject such a frictionless world given evidence such as Mueller's (1977b) and the general theory of barriers to mobility (Caves and Porter, 1977). Those sellers having grown interdependent across markets are those who would have been most likely to enter given new profitable opportunities. Capacity expansion given such opportunities should be less rapid than if the multimarket interdependence did not exist. Is it in fact that just as with market concentration, not only the philosophy embodied in the Jeffersonian ideal, but economic efficiency provides grounds for concern about aggregate concentration when it coincides with multimarket grouping? This paper introduces methodology for measuring the significance of grouping and shows that when significant and coincident with high seller concentration, multimarket grouping does have economic implications. That coincidence, ceteris paribus, leads to higher profits. The question is whether those profits result from coordinated behavior or lower costs or both. The evidence suggests they are the result of economies of multimarket operation and barriers to the mobility of resources from outside the interdependent groups of sellers. Profits are lower for lines of business where multimarket contact is high but seller concentration is low, but higher when both contact and concentration are high than when concentration alone is high.

A barrier to entry is a structural trait of a market implying that incumbent sellers can earn more than a normal rate of return without attracting entry (Bain, 1956; 1968, chapter 8). As regards capital costs, incumbents may have an absolute cost advantage because barriers to entry resulting from product differentiation, economies of scale, or impacted technological information increase the uncertainty enveloping the potential entrant's investment decision (Caves and Porter, 1977). This paper (1) advances two causes (other than fixed costs of entering markets for funds) for a pure capital-cost barrier to entry, (2) shows that barrier implies both (a) the coincidence of seller concentration with long-run excess profits and (b) a direct relation between concentration and excess profits even if the ability to coordinate price and nonprice rivalry were held constant over observed levels of seller concentration, and (3) provides evidence supporting the existence of such a barrier. In a market with multi-plant economies of in noncapital-raising aspects of production,' product differentiation, afsolute cost advantages other than those associated with capital costs, and financial markets which priced assets to provide an expected rate of return commensurate with risk, incumbent firms can elevate price above average costs without inducing entry if the post-entry capital costs for all firms in the industry exceed the pre-entry costs. Even in the absence of fixed costs of entering markets for funds, there are reasons for such a difference in costs. First, Sullivan (1978) has found that the systematic risk resulting from the stock market's revaluation of assets in response to new information is less for firms in concentrated industries than for those operating in more competitive markets. He suggests that may be because of the ability of large firms in concentrated industries to mitigate the adverse effects of new bearish information. His conjecture might be supported and extended by building oii the attempt of Salamon and Siegfried (1977) to establish links among elements of market structure and measures of political clout. For example, political influence might enable a firm to avoid the full impact of an economy-wide increase in taxes. Second, a different hypothesis is advanced in Scott (1977, 1980). Sullivan's work is based on the capital asset pricing model (CAPM), for which investors' general expected utility maximization problem has been reduced to a choice over two parameters of subjectively-evaluated probability distributions of returns. But, ceteris paribus,the probability distributions of returns in concentrated industries offer downside protection and upside potential relative to unconcentrated industries. That skewness could imply lower capital costs for firms in concentrated industries. To develop the skewness possibility, suppose that the stochastic process generating market demand is in general a Markov chain with large diagonal elements in the transition matrix.2 While that stochastic process together with existing capacity can imply the probability of capacity shutdown at any future time, adding market structure to the picture implies the probability thatfirms will exit during various periods of time. The probability that any given number of firms will exit varies inversely with seller concentration. Even if investors could diversify costlessly, the extra flexibility given market power to choose a profitable price and output in the face of a downturn in demand and the difference in costs-resulting from bankruptcy costs-for capacity shutdown versus firm exit can imply that capital costs fall as seller concentration increases. Thus, this paper suggests two reasons (other than Sullivan's political power speculation or fixed costs of transacting in financial markets) why the cost of capital of firms in concentrated industries could, ceteris paribus, be lower than for firms in unconcentrated industries. The two reasons follow from the fact that in an industry with fixed capacity and variable demand, a reduction in demand at a point in time may force firms to exit. But the probability of forced exit of a given firm is directly related to the number of firms in the industry. Hence, increased market concentration results in the reduced probability of forced exit. The reduction in that probability reduces capital costs by Received for publication January 23, 1980. Revision accepted for publication November 7, 1980. * Dartmouth College. I thank two anonymous referees of this REVIEW for very helpful comments on an earlier version of this paper. ' Thus by no economies of scale I do not mean costless and instantaneous adjustment to any point on a horizontal average cost curve. Capacity does come in discrete chunks (plants) for which we have the usual U-shaped cost curves. However, regardless of the number of these plants producing at minimum average cost that are combined to make a firm, unit cost is the same. Further, at the current levels of market demand, these units of capacity need not be large relative to the output of the market. That is, there need not be a scaleeconomies barrier in the sense of Bain. 2 Market demand has a high probability (but less than one) of staying where it currently is. If demand should fall, the fact that it is unlikely to rebound will force a reduction in capacity.

Asch, Peter, and Joseph J. Seneca, Is Collusion Profitable? this REVIEW 58 (Feb. 1976), 1-12. Block, M. K., F. C. Nold and J. G. Sidak, The Deterrent Effect of Antitrust Enforcement: A Theoretical and Empirical Analysis, Technical Report ISDDE-1-78, Hoover Institution, Dec. 1978. Clabault, James M., and John F. Burton, Jr., Sherman Act Indictments, 1955-1965 (New York: Federal Legal Publications, 1966). Feinberg, Robert M., Structure and Employment Instability, this REVIEW 61 (Nov. 1979), 497-505. , The Lerner Index, Concentration, and the Measurement of Market Power, Southern Economic Journal 46 (Apr. 1980), 1180-1186. Guth, Louis A., Robert A. Schwartz, and David K. Whitcomb, The Use of Buyer Concentration Ratios in Tests of Oligopoly Models, this REVIEW 58 (Nov. 1976), 488-492. Hay, George A., and Daniel Kelley, An Empirical Survey of Price-Fixing Conspiracies, Journal of Law and Economics 17 (Apr. 1974), 13-38. Lustgarten, Steven H., The Impact of Buyer Concentration in Manufacturing Industries, this REVIEW 57 (May 1975), 125-132. , The Use of Buyer Concentration Ratios in Tests of Oligopoly Models: Reply, this REVIEW 58 (Nov. 1976), 492-494. Omstein, Stanley I., Industrial Concentration and Advertising Intensity (Washington, D.C.: American Enterprise Institute, 1977). Palmer, John, The Profit-Performance Effects of the Separation of Ownership from Control in Large U.S. Industrial Corporations, Bell Journal of Economics 4 (Spring 1973), 299-302. Trade Regulation Reporter, Vol. 4, New U.S. Antitrust Cases (Chicago: Commerce Clearing House, 1979).

The Federal Trade Commission's Line of Business data imply that the effect of an industry's seller concentration on profitability has been misinterpreted because it has not been conditioned on capital intensity. Conclusions that market share rather than seller concentration is the primary structural determinant of profitability, and that mutual dependence recognized among oligopolistic sellers is less important than superiority effects of large-share firms, appear unwarranted. Significant firm effects exist, and explanatory power for a conventional model of structure and performance is found to be small relative to that of the general linear model within which the conventional model is nested.

Universities are a key institution in the U.S. innovation system, and an important aspect of their involvement is the role they play in public-private partnerships. This note offers insights into the performance of industry-university research partnerships, using a survey of precommercial research projects funded by the Advanced Technology Program. Although results must be interpreted cautiously because of the small size of the sample, the study finds that projects with university involvement tend to be in areas involving new science and therefore experience more difficulty and delay, yet are more likely not to be aborted prematurely. Our interpretation is that universities are contributing to basic research awareness and insight among the partners in ATP-funded projects.