Knowledge that Transforms

In a recent paper in this Review, Dale Osborne poses four internal faced by cartels. The (of the contract surface) and together pertain to the determination of an optimal collusive policy; the and pertain to firms' incentives to unilaterally cheat on that policy. Claiming Standard theory teaches us that cartels are inherently unstable, mainly because of the sharing and deterring problems (p. 835), Osborne proceeds to argue that these are readily solved so that only the locating and detecting remain as obstacles to internal cartel stability. Quite apart from his treatment and purported resolution of the sharing and deterring problems, which provide the focus for this comment, there is some question as to whether or not Osborne correctly interprets theory. It would appear he does not. In particular his argument that economic orthodoxy stresses the deterring problem is misplaced; in fact it is the detecting problem that is generally emphasized. For example in his treatise on monopoly, Donald Dewey (p. 19) indicates the importance to a cartel's success of reliable information on the behavior of member firms. The information is necessary, Dewey claims, to detect cheaters. In a recent volume on antitrust law, Richard Posner writes, Cheating is presumably least likely when detection is prompt and certain... (p. 53, emphasis added). George Stigler certainly emphasizes the detecting problem over the problem of deterrence; indeed he does not view deterring per se as a problem: Once detected, the deviations [i.e., cheating] will tend to disappear because they are no longer secret and will be matched by fellow conspirators if they are not withdrawn (p. 42). Osborne's exegesis of standard theory is incomplete. Detection is a serious and widely recognized problem that has been emphasized more than deterrence.' Questions of economic literature aside, even if deterring is to be granted equal status with detecting as a cartel its importance depends crucially on the extent of the detecting problem. As we shall show, Osborne's purported resolution of the deterring problem is either ineffectual or unnecessary depending on the status of the detecting problem. If the cartel has a detecting problem, then his resolution of the deterring problem breaks down. If detecting is not a problem for the cartel, then following Stigler, neither is deterring, since detection alone is sufficient to bring an end to cheating. Moreover if deterring persists as a problem, as Osborne seems to aver, even when there is no detecting problem, his resolution of it is not strictly superior to an infinite number of alternative resolutions. We also argue that Osborne's resolution of the sharing problem, although correct in the case of a single joint profit-maximizing point, is unconvincing for the case of multiple joint profit-maximizing points. Finally, in spite of its failure as a final and complete resolution of the deterring problem, Osborne's quota rule has some interesting implications for still another (domestic) cartel problem: the avoidance of antitrust prosecution and conviction.

Perhaps the most striking feature of the New England economy is that it is different-not only from the rest of the nation, but from the rest of the northeastern United States as well. New England's main departure from the national norm is, of course, relatively slow growth; it is conventional to describe New England as -economically, industrially, and maybe even demographically. This maturity manifests itself in many ways. While aggregate personal income in the United States expanded at an average annual rate of 4. 1 percent between 1960 and 1975, New England expanded at a rate of 3.6 percent per annum. Further, total manufacturing employment in Massachusetts and Rhode Island is only slightly higher today than in 1914. In the recent recovery from recession, New England lagged well behind the rest of the United States in expansion of total employment but nevertheless recorded some of the sharpest declines in unemployment rates so that New England unemployment is now near the national average even though it was much higher at the depth of the 1975 recession. The secret, of course, to New England's relatively rapid unemployment decline is slow workforce growth, as expected in a mature economy. New England's differences from the rest of the Northeast are perhaps less obvious and certainly less well known. It is fashionable today to speak in very broad terms of vs. sunbelt and to suggest that public policy should modulate differences in growth among the different sections of the country. The reality, though, is that aggregate figures for large regions of the country hide a good deal of internal diversity. Thus, New England not only seems to be doing better than conventional frostbelt wisdom would suggest, but its immediate prospects also appear more favorable than those of the mid-Atlantic states and probably much of the Midwest as well (see Benjamin Stevens and Glinnis Trainer). Even in the recent past, as between 1960 and 1976, when New England's aggregate personal income was growing 3.6 percent per year, the states of New York, New Jersey, Pennsylvania, Maryland, and Delaware had a combined average annual compound growth rate of only 3.3 percent. Similarly, a shift-share analysis has indicated that the entire Northeast (by virtue of a favorable industry mix) should have been in a position to gain in share of U.S. jobs throughout the 1960's. The New England states (except for extreme northern Maine) have indeed done as expected. Large areas of the remaining Northeast, however, have experienced significant competitive shifts or losses (see Richard Olsen). To a considerable extent, in fact, any New England success in recent years may have been at the expense of its immediate neighbors. New England production costs perhaps have not been as low as in much of the Southeast in recent years, but they apparently have been competitive with the Middle Atlantic, and especially New York City. In fact, total manufacturing costs in several industries (for example, ordnance, primary metals, fabricated metals, nonelectrical machinery, transportation equipment, paper and printing) have been lower recently in Massachusetts (probably the highest cost New England Harvard University. The Economic Development Administration of the U.S. Department of Commerce and the 1907 Foundation provided financial support for this research.

Both Philip Cook and Michael Jones-Lee criticize my assumption that the critical value above which the value of human life (VHL) exceeds is presumably at a level of reply, I develop an extension the original article, which eliminates this qualification, and reconstruct my in light of their comments. line with common usage, I had defined as expected discounted lifetime labor income (p. 47) without any subtraction of necessary expenditures as would be required in determining business income. However, such expenditures must be so subtracted properly define a meaningful income. As a leading accounting text emphasizes: In computing the (available dividends) a period all forms of expense incurred in the of such must be provided for (Eldon Hendriksen, p. 62). Put another way, all of business is a surplus and may be taxed or distributed without any loss of output (in the short run). The public finance literature has similarly recognized that should be deducted from [gross] income, regarding such outlays as a cost of production (Richard Musgrave, p. 171), and Adam Smith admonished that taxable should be defined as 'clear' income, or as above subsistence (Musgrave, p. 95). To define accurately, one must therefore subtract from gross receipts a level of expenditure which allows continued work and makes the individual indifferent between life and death. Only gross receipts above this level are potentially taxable or available safety expenditures. If a major share of gross receipts is taxed leaving funds less than (as measured by amount CO in Figure 2 of the original paper), the person will starve death, otherwise perish, or commit suicide. Thus any useful definition of net consistent with its inherent meaning as a measure of surplus must subtract from some measure of gross the expenditures necessary living. The same holds consumption. However, is difficult measure and possibly highly variable among individuals. I expect it is on average at a low level of income, as argued below. Given this change, my results are valid under all circumstances. That is, if C' = C CO and Y' = Y Y?, VHL > C'(= Y') in all cases. Graphically, the slope of the tangent any point on a curve, as shown in Figure 2, will always be less than the ray arising out of CO the intercept value on U. We have found that VHL = f(UIC')l (OU/OC')j * C', and the factor in brackets is the inverse of a'. Since it is also the slope of the ray divided by the tangent, from the previous line of reasoning, the quotient is always greater than one, a' is always less than one and the value of human life is always greater than consumption (where consumption and are defined of expenditures). Cook wishes to provide a relatively transparent derivation of Conley's major theoretical result, which avoids the complexity of his multiperiod model (p. 710). A simple of the main results is contained in my original footnote 16.' His last sentence is a reasonable caveat my origi-

The interest elasticity of investment is one of the oldest and most important unresolved issues in monetary theory. It is a key relationship in the most widely accepted theory of how Federal Reserve actions effect the economy.' Unfortunately, there is no consistent empirical verification of this relationship.2 In spite of this lack of empirical support, the theoretical arguments for the interest elasticity of investment are so strong that most experts of monetary theory continue to write books and advise the government on the basis of this relationship. Yet policy exercises are suspect in the absence of some knowledge of the slope of the investment function. The purpose of this article is to use an errorlearning model to investigate the modification of business investment decisions in response to errors in long-term interest rate expectations.