Different methods are used by the NYSE/Amex and the Nasdaq to accomodate limit orders received from investors. This accounts for at least part of the excess of Nasdaq spreads over NYSE spreads, adjusted for trading volume, and is a factor in determining this excess that is independent of collusion on the Nasdaq. The spread-comparison evidence given by others to support their belief that there is collusion among market makers on the Nasdaq therefore overstates the probability of collusion and its significance if it exists.

Introduction, 33. — The definition and measurement of transaction cost on the New York stock exchange, 35. — The determination of the ask-bid spread, 40. — The determination of the transaction rate, 45. — Statistical results, 46. — Summary and comments, 50. — Appendix I, 52. — Appendix II, 53.

Although the notion of to entry plays an important role in economic theory and antitrust litigation, the substantial problems inherent in it are not fully appreciated. Much existing discussion of barriers hardly pauses to recognize the difficulties, and even more careful treatments of the subject proceed as if the definition of barriers can be tied quite easily to some purely objective measure of the cost of doing business.' The burden of this paper is to demonstrate that this is not so. The origin of the barriers concept is in the research custom of industrial organization economists during the post-World War II period. That custom was to seek monopoly explanations for data not obviously or directly implied by the perfect competition model. The perceived persistence of higher rates of return in some industries than in others was suggestive of barriers to entry, especially for industries exhibiting high levels of structural concentration. The equalization of profit rates through competition, however, is a proposition logically valid only with respect to investment on the margin of alternative economic activities. Only if all inputs are available in perfectly elastic supply does this imply equality between average profit rates. In addition, accounting profits are likely to be biased by the presence of uncapitalized assets, especially assets associated with advertising, research, and goodwill2 (see Lester Telser; Leonard Weiss; my earlier article). But my purpose here is not to probe these measurement problems nor to compare the merits of these explanations, but to examine the usefulness of the concept of barriers. Accordingly, I will ignore the possibility that factors other than barriers may partly or wholly explain the phenomena that the barriers notion seeks to explain.