Knowledge that Transforms

IN recent years a number of economists, including several who have been actively concerned with questions of public policy, have expressed the point of view that economists working in this field should concentrate their attention on economic considerations and not to take into account political considerations. purpose of this paper is to examine the validity of this point of view. attitude that economists engaged in analyzing or forming public policy should ignore political factors has been vigorously supported by such well-known men as Milton Friedman, the late E. A. Goldenweiser, and Edwin G. Nourse. Friedman, for example, has written: The role of the economists in discussions of public policy seems to me to be to prescribe what should be done in the light of what can be done, politics aside, not to predict what is 'politically feasible' and then to recommend it. 1 Goldenweiser has stated, I agree emphatically with Friedman that economists must think things out on their merits without regard to political feasibility. 2 To the layman or the neophyte student of economics it may appear that statements such as these represent an attempt to maintain a tradition of long-standing in the field of economics, and thus to preserve the integrity of economics and economists against recent incursions or defections. To those who are familiar with the history of economic thought and the development of economics as a social science, however, it is obvious that the attitude expressed by Friedman and Goldenweiser is not firmly based in tradition. McCulloch, a leading exponent and synthesizer of traditional economic theory in the first half of the nineteenth century, went so far as to write in the preface to the third edition of his Principles of Political Economy, . . .the truth is that Political Economy and Politics are so very closely allied, and run into and mix with each other in so many ways, that they cannot always be separately considered. 3

LIABILITIES of business failures move id in cycles which frequently lead general business, and for this reason have been designated as a leading series by the National Bureau of Economic Research.' The natural explanation of such lead lies in the profit experience of business concerns, particularly of submarginal concerns, since the lack of profits is undoubtedly the most important factor in causing business failures. The purpose of this paper is to examine this idea. The suggestion is made that profit margins per dollar of sales may lead business cycles, and since such margins are critical to the survival of submarginal firms, they are instrumental in causing liabilities of business failures to lead the cycle also.

COMPARATIVE dynamics is a relatively unworked but highly significant area of economic analysis. It is important to appreciate the implications of changes in the structure of a dynamic system and, thus, the extent to which the analytical results depend on the particular, restricted forms employed. These are the problems with which comparative dynamics deals in general 1 and with which this paper is centrally concerned, with particular respect to the stability of difference-equation models of income determination. Dynamic difference-equation models of economic fluctuations have been useful tools for the economist. Though not intended to provide, in themselves, a complete description of aggregate economic activity, such models have led to a better appreciation of cycle-producing forces and the character of cyclical movements. These models have been developed in a variety of forms based on different assumptions about time sequences and using different types of expenditure functions as components. Inevitably, however, each of the models represents a special case whose analysis provides conclusions of only limited applicability. There has not, I believe, been adequate recognition of the important qualifications which must as a result be applied to many of the existing dynamic models of economic fluctuations. In this paper greater generality will be sought in two ways: first, by the construction of a number of alternative systems based on different assumptions about the lag sequences involved; second, by detailing somewhat more than has been done the expenditure functions which are the components of the models. I believe that it will be possible in this way to provide further insights which are important in themselves and in qualifying previous dynamic analyses. The variety of results which can be obtained will be illustrated rather than treated exhaustively. The procedure of this paper will be to construct and compare a series of alternative models. In section I the general method of analysis will be illustrated in presenting models based on the Robertsonian lag of expenditures behind the receipt of income.2 Section II will begin to break new ground in the analysis of models which use the Lundberg lag of output behind sales.3 Finally, in section III models will be developed using both types of lag. The investigation is confined to models of effective demand in a closed economy. The expenditure functions used are, in general, adaptations of a simple consumption function and the acceleration principle -admittedly gross, but useful, simplifications of reality. Even though thus confined to simple lag relations and spending functions, the analysis becomes quite complicated. After constructing each model, the types of movements which it may generate and the conditions which it must satisfy for stability will be investigated.4 The systems constructed will be subject to the well-known limitations of linear models. It would not be difficult, however, to apply simple types of nonlinear restraints to the systems, and the effects of such restraints will be noted at several places. The mathematical techniques of stability analysis which are the basis of this paper have been introduced into economics by Professor P. A. Samuelson. Their usefulness has, perhaps, not been fully appreciated, and this paper provides an example of their application. A general comment on the use of stability conditions may be in order here. We want to investigate the economic implications of a number of alternative dynamic systems; we know that such systems are able to produce a variety of motions consisting of monotonic and

T HERE are many dangers in viewing federal expenditure distribution among the states as more than a by-product of the pursuit of national objectives and in establishing at this stage of research a benchmark for testing the actual distribution against some goal. First, there is no correct way of distributing or allocating total federal expenditures among the states. While federal tax allocations derive from an extensive body of economic theory, no comparable theoretical framework supports expenditure allocations. The field of state distributions of federal expenditures has not been as well ploughed by others as has that of state distributions of federal tax levies. The statistical processes are more complex on the expenditure side than on the tax side. Expenditures are more varied in type, and individual classes of expenditures are more numerous. Moreover, the statistical information needed for allocation is much less adequate for expenditure items than for tax types. The illustrative estimates presented here are patently based on sets of assumptions. Other assumptions could be applied and recomputations made accordingly. With the work carried to the present point of analysis some changes in framework and procedures appear desirable for purposes of greater clarity and wider usefulness, if federal expenditures are allocated for a year more recent than I952. Far more important than the statistical inadequacies are the limitations implicit in a detailing of federal expenditures directed to the achievement of national purposes and program objectives as a series of state-by-state figures. Ease of movement across state lines, the dependence of industries in one state on raw materials, machinery, and semifinished goods in others, the frequency of absentee ownership of property in the state -all contribute to an emphasis on national objectives and purposes. In historical perspective, sectional interests have influenced national policies on tariffs, railroad rate regulations, resources development, minimum wages, and many other national programs. While discussions of legislative proposals often stress the divergent economic interests of the different sections of the nation, special sectional as well as national interests are essentially parts of an over-all common concern with national prosperity and economic growth. The long-run economic interests of various sections of the nation patently coincide.

THERE may seem little excuse for writing still more about the problems of exporting British goods to the United States.' But while many aspects of the problem have received thorough individual treatment, for instance the invisible tariff or the technicalities of marketing, less attention has been paid to the more general question of the relative importance of all the alleged difficulties as factors currently limiting sales of British goods in the American market. The problem is fraught with interesting but unanswerable questions, and impressions have often to take the place of statistics. As a starting point a volume and average value series was calculated for United Kingdom exports to the United States for the years I94855 2 The result is shown in Table i together for comparative purposes with volume indexes of total United Kingdom exports and total United States imports and with the United States consumer price index. The detailed average value index by commodity groups is shown in Table

THE purpose of the investigation reported here is to examine factors associated with the amount of money which consumer units spend on life insurance premiums. Life insurance premiums represent an extremely large and stable component of personal saving and one which, unlike mortgage and other debt payments, has no offset in the purchase of a consumer investment item. Their long range future is thus important, and we hope to throw some light on this by seeing what factors are currently associated with high or low proportions of income paid as premiums. In addition to making available the results of this study, this report demonstrates one type of solution to the problems of statistical analysis of data from a sample survey. The survey which is the source of the data used in this study is the Survey of Consumer Finances. This survey has been conducted annually since I945 by the Survey Research Center of the University of Michigan in cooperation with the Board of Governors of the Federal Reserve System. It includes about 3,000 interviews each year with a cross-section of the heads of consumer spending units in the United States. Interviews last about one hour and cover a variety of information about the unit's income, assets, and expenditures as well as about the demographic characteristics of the family and the attitudes of the respondent. The sample design involves selection of individual addresses with known probability. The sample is complex since it involves selection of successively smaller geographical areas in stages, with clustering and stratification at each stage, and the procedures differ in communities of different sizes.' The questions asked about life insurance in the I954 Survey were as follows:

V IRTUALLY the full employment of available labor and resources was tacitly assumed by most of the nineteenth-century wage theorists. Actual full employment has been a rather rare condition in the western world during the last two centuries. However, the economies of both Great Britain and the United States moved decisively toward full employment in the period between the years just before World War II and the subsequent wartime and postwar years. In what ways and to what extent has this relatively new condition of prolonged full employment affected the structure of wages? There are many ways of answering such a question, depending on what aspect of wage structure is considered. For example, it is now reasonably clear that a transition to full employment works toward the narrowing of wage differentials between workers of different grades of skill.' What happens to the relative levels of wages paid by different industries, as an economy moves toward full employment? Comparatively little attention has been devoted to this question. One outstanding study of interindustry wage structure has been the recent analysis by Donald Cullen.2 This painstaking study was mostly concerned with long-period relationships between the average wages paid by different American industries. Cullen found that the rank-order of industries was very stable, even over long periods of time, as regards the average annual earnings of their respective employees. Over a mere ten-year period, there was, naturally, even less change in interindustry wage structure than occurred over longer periods. Thus for the decade I939-49, Cullen's coefficient of rank correlation for seventy American industries was .92.3 These findings suggest that even a sharp change from very considerable unemployment (I939) to virtual full employment (I949) will have little effect on the structure of wages as between industries -at least that such a change will not alter materially the rank-order of the average wages of the various industries. In Great Britain, the full employment conditions of wartime and postwar years provide a similar contrast with the slack employment of the prewar period. Did the relative wages paid by different industries also remain stable in the face of this drastic change in labor market conditions? We shall see presently that the answer depends on how one chooses the method of measurement. Our first test of interindustry wage structure was selected to provide the greatest possible comparability between the British and American wage data and the method and data used by Cullen. We were able to find British and American wage information, prewar and postwar, for 28 industries which were reasonably comparable with the American industries selected by Cullen.4 We used i938 as a repre'E.g., see Harry Ober, Occupational Wage Differentials, I907-I947, U.S. Department of Labor, Monthly Labor Review, August I948; Louis R. Salkever, Toward a Theory of Wage Structure, Industrial and Labor Relations Review, April I953; K. G. Knowles and D. J. Robertson, Differences Between the Wages of Skilled and Unskilled Workers i8881950, Bulletin of the Oxford Institute of Statistics, xm (Apr1l I95I), I09-27 and Earnings in Engineering, I926I948, ibid. (June I95I), I79-200. Our limited purpose is to investigate interindustry differences in average earnings. We are acutely aware of the variability of wages between firms within an industry, between workers of different levels of skill, and the week-to-week variability of the earnings of individual workers. Compare Robert R. L. Raimon, The Indeterminateness of Wages of Unskilled Workers, Industrial and Labor Relations Review, vi (January 1953), I80-94; and K. G. J. C. Knowles and Ann Romanis, Dockworkers' Earnings, Bulletin of the Oxford University Institute of Statistics, xiv (September and Octo-