just as proponents of a new theoretical approach speak in enthusiastic terms, so those who reject their approach often appear ready to condemn it on inadequate evidence. The purpose of this note is to discuss Tobin's statistical findings rather than to analyze differences in the theoretical approach. The reinterviews incorporated in the I952-53 Surveys of Consumer Finances, which Tobin uses, are adequate to warrant his conclusion that in that year expressed buying intentions did have predictive value (even though the buying intentions questions in those surveys are far too brief). But those data are insufficient and irrelevant for any test of the predictive value of attitudinal questions other than buying intentions. Therefore Tobin's assertion denying such predictive value (page iO) has no bearing on the Survey Research Center's position regarding the relation of consumer expectations to consumer demand for durable goods. Tobin tries to achieve the impossible by constructing an attitude index out of four questions, (i) the evaluation of past changes in personal financial conditions, (2) information about past income changes, (3) the evaluation of current marketing conditions, and (4) income expectations for the next year. Two of the questions used by Tobin (i and 3) are also contained in the six-question index constructed from the Center's Periodic Surveys. By adding question 2, Tobin loads his index with material about past changes in personal finances. Regarding Tobin's fourth question, Katona and Mueller showed six years ago (in Consumer Attitudes and Demand, pages 69-70), that it is realityoriented and insensitive rather than reflecting people's hopes and fears; the Center has therefore substituted other questions to measure personal financial expectations in its Periodic Surveys. There is not a single question in Tobin's index which reflects the more volatile expectations (the Center's six-question index contains four such questions). It is not at all surprising that Tobin's index duplicates to a large extent the predictive information contained in financial data and has no independent forecasting value. Our position, as it will be set forth soon again in detail, is that both buying intentions and other attitudes should be used to supplement (not to supplant) financial information. The tests of the predictive value of certain consumer expectations are being continued. It may be mentioned that in 1957 consumer expectations other than buying intentions became pessimistic earlier than buying intentions and in I958 they turned optimistic well before buying intentions. In spite of this evidence, Tobin is right in arguing that more observations are needed before conclusions can be based on rigorous statistical tests. There are many other points of disagreement, for instance, regarding the conclusions drawn by Tobin from Eva Mueller's article in the I957 American Economic Review, or regarding Tobin's notion that adding to liquid assets is a sign of pessimism. I wish to contradict particularly Tobin's statement that our treatment of price expectations is arbitrary (see his footnote 7). This treatment has been derived from the theory of psychological economics; additional empirical data supporting the treatment are being published in the May I959 issue of the Quarterly Journal of Economics. * James Tobin, On the Predictive Value of Consumer Intentions and Attitudes, this REVIEW, xui (February I959),

Most persons recognize that a free enterprise system will survive only as long as competition keeps business energies within socially desirable channels. Business mergers are a worrisome public problem because some mergers are destructive of competition, while others promote competition.' A large proportion are innocuous, yet whole industries have been transformed by the accretion of scores of small mergers that individually seemed to be only of slight consequence. For these reasons, and others also, it is difficult to assess the long-run significance of the stream of merger reports in the press. Some reported mergers are never consummated, while others more logically are viewed as divestitures of isolated assets or isolated segments of a large company's business. Still other announced mergers are not mergers at all, inasmuch as the companies involved already were under common control. One index of the impact of mergers on a particular sector of the economy is the percentage of the largest companies at some past date that subsequently disappeared as a result of mergers.2 For manufacturing, newspaper publication excepted, such a list is available for the i,ooi largest companies on December 3I, I950. With slight modifications, this is the list of i,ooo largest companies identified in the Federal Trade Commission's Report on Industrial Concentration and Product Diversification in the I,000 Largest Manufacturing Companies: I950.1 3 In restricting the scope of the Commission's list to exclude newspapers, for which adequate information is lacking, it is necessary to delete three companies for which newspaper publication was the principal business. Two other companies are deleted because they appear to have been part of another enterprise with which they subsequently merged.4 The five deletions are offset by the addition of six companies which were overlooked.5 After giving effect to these changes it is possible to report the experience of the I99 largest manufacturing companies, the next largest 302 manufacturing companies, and the next largest 500 manufacturing companies. The smallest company included on the list of the i,ooi largest manufacturing companies had I950 shipments (including interplant transfers) valued in excess of $I3,000,000, and assets as of December 31, I950 of at least $II,5oo,ooo. The companies ranking iggth and 5oist had shipments (including interplant transfers) of $I20,000,000 and $40,000,000 respectively. Although there were over 300,000 manufacturing companies in the United States in I95o exclusive of newspaper publishers, the first 199 companies as a group made 4I per cent of the shipments originating in all manufacturing plants in the continental United States, 'Every firm that acquires another business expects as a result to be a more effective competitor. Even though its expectations are realized, it does not necessarily follow that competition has been enhanced by the merger, for competition rests on a balance of the capabilities of many rivals. Provided there are no unfortunate side effects, a merger is salutary if it places productive facilities under more vigorous management, or if it enables the seller to transfer his resources to the exploitation of a more attractive business opportunity. A merger of two relatively small firms also promotes competition if the necessary scale of operations has doubled as a result of changes in technology or the size of the market, and the companies lack the resources for rapid expansion. But beyond a certain point it is not necessary to be larger to compete effectively with still larger companies. To be sure, the bigger and more diversified a company is the better it is able to bludgeon its rivals by long-sustained sales below cost, favoritism to full line patrons, and reciprocal deals. But such tactics are not to be condoned, because their effect is to eliminate (or intimidate) competitors, rather than to outdo them with more attractive offers to customers generally. 2 To result in a disappearance a merger must reduce the number of firms on the roll of largest firms. Two companies among the i,ooi which merged to form a firm operating under a new name (for example, the merger of NashKelvinator and Hudson Motor Car to form American Motors Corporation) are considered to have occasioned one disappearance. A merger between a company on the roll of largest companies and a company not on the roll is considered a disappearance only where control clearly passed to the latter. A prominent instance is the merger of Consolidated Vultee and Electric Boat Company. With the exception of companies having ties with alien concerns (for example, Lever Brothers) only independent enterprises are considered companies. Thus a company would be counted as a disappearance if it were acquired by another but continued to operate as a subsidiary corporation. A company that sold off its manufacturing business and continued to operate in some other business (for example, Willys Overland) also is treated as a disappearance. 'Washington, Government Printing Office, January I957, Appendix F. The difficulty in preparing a list of this sort for an unregulated sector of the economy is the identification of all large unregistered corporations. Most such corporations jealously guard information on the size of their business. The 'Commission included 200 such companies on its I950 list which do not appear on its List of I,ooo Large Corporations in Manufacturing and Mining . . . as of two years earlier. 'Clark Thread Company and Eddy Paper Company. 'The Humko Company, Minute Maid Corp., Park & Tilford Distillers Corp., H. K. Porter Co., Inc., Reliance Manufacturing Company (Illinois), and Rotary Electric Steel Company

PROFESSOR SAMUELSON's path-breaking article on Interaction Between Multiplier Analysis and Principle of Acceleration appeared in this REVIEW almost twenty years ago. A large literature has developed in which basic ideas of that article have been applied to both business cycle and economic growth problems. In a considerable portion of that literature, Samuelson's warning that the representation is strictly a marginal analysis to be applied to study of small oscillations has been overlooked.' Samuelson's warning can be interpreted as meaning that time series generated by any particular solution of model will determine actual income for only a short time. Given mathematical model, relevant particular solution can change due either to (i) accelerator or multiplier coefficients changing (as Samuelson suggests), or (2) imposition of new initial conditions. Goodwin2 has examined various models in which accelerator coefficient is a variable. These non-linear models are mathematically complex, and specific limit cycles that Goodwin derives obviously are due to special assumptions he makes about how path of income affects accelerator coefficient. Hicks 3 has investigated how an otherwise explosive accelerator model will be affected by floors and ceilings. In this paper such floors and ceilings will be interpreted as imposing new initial conditions, and therefore this paper can be considered a reinterpretation of Hicks's setup.4 We will work with a slightly modified version of Samuelson's model, and assume that

The following sections are included:The Leontief Paradox?Computational Procedures and Statistical DataStatistical Results and Implications