Knowledge that Transforms

E CONOMISTS today are well aware of the many problems in statistical investigations of demand functions and other economic relationships. The small number of observations normally available, the unreliability of the usual tests of significance when applied to correlations obtained from highly autocorrelated time-series data, and other difficulties have led to an increasing number of attempts in recent years to combine time-series and cross-section information and, most recently, to make use of successive cross sections. This paper presents results of a study of the demand for refrigerators in the United States which relies primarily on crosssection data at the state level for I930, I939, and I949.' It offers estimates of the influence of income, other non-price variables, and, less satisfactorily, price on aggregate demand for refrigerators.2

EXCHANGE rate depreciation can be used by a country either to improve its balance of payments or to stimulate domestic employment and output. But there is a conflict between these two objectives. If expansion at home is allowed, the depreciation cannot be expected to result in the trade account improvement obtainable if domestic policies to stabilize output and employment were pursued. This article discusses the extent to which the income movements that are a consequence of devaluation conflict with the goal of improvement in the trade balance. analysis of devaluation was extended to incorporate income movements in the studies by Laursen and Metzler' and by Harberger.2 Subsequent criticisms and extensions have clarified many aspects of the problem,3 especially the money-illusion that was implicit in Harberger's analysis. Emphasis in their work has been placed upon the value that must be exceeded by the of elasticities of import demand in order to insure stability in the foreign exchange market. dominant conclusion that emerged was that this sum exceeds unity if employment and output are allowed to respond to the forces set in motion by depreciation. In this paper I assume that demand elasticities exceed the critical for exchange stability, so that depreciation becomes a feasible technique for improving the trade balance, and examine the nature and extent of the dampening effect of output and employment changes on the balance of payments. This is accomplished by investigating what I call the dampening coefficient associated with output changes. Consider the improvement in the trade account with variable outputs relative to the improvement if stabilization policies are pursued. This fraction subtracted from unity is an index of the extent to which income changes have reduced the improvement in the trade account. higher the value of this dampening coefficient, the greater the cost4 to the devaluing country of output changes. Two types of dampening coefficient must be distinguished. first relates to the comparison between (i) devaluation in which all countries' outputs vary as they will and (ii) devaluation in which all countries stabilize levels of output and employment, and is a measure of the comparative importance of effects5 and income effects. It is examined in the first section of this paper. second dampening coefficient is relevant to trade policy in the devaluing country. It measures the diminution in the gains in the trade account resulting from failure to control output in the depreciating country when some specific assumption is made as to stabilization policies abroad. That is, it is assumed that decisions in the devaluing country concerning stabilization policy do not influence other countries' stabilization policies. For simplicity, I assume, in the second section of this paper, that other countries decide not to permit any fall in employment. expressions for either dampening coefficient depend upon the period of time during which the of devaluation are considered. distinction made in this paper between the short-run and long-run of currency depreciation is based upon studies suggesting dif* This paper had its origin in my doctoral thesis submitted to M.I.T. in 1955-56. It has benefited from comments by Robert Solow and Lionel McKenzie. ' S. Laursen and L. Metzler, Flexible Exchange Rates and the Theory of Employment, this REVIEW, XXXII (November 1950). 'A. Harberger, Currency Depreciation, Income, and the Balance of Trade, Journal of Political Economy, LVm (February 1950). 'For example, cf. Spraos, Consumers' Behavior and the Conditions for Exchange Stability, Economica, xxII (May 1955); Pearce, A Note on Mr. Spraos' Paper, Economica, xxii (May 1955); H. Johnson, The Transfer Problem and Exchange Stability, Journal of Political Economy, Lxrv (June 1956). 4 Cost measured in sacrificed improvement in the trade account. 6 By price effects I mean the change in the trade balance resulting from the change in relative prices, neglecting the impact of changes in aggregate outputs.

PpT HE term underdeveloped is often used to refer either to countries with low incomes or to countries in which the level of per capita income is not rising, without clear discrimination between the two concepts. The purpose of this paper is the simple factual one of presenting two classifications of all the areas of the world, one according to their levels of per capita income and one according to whether continuing rise in per capita incomes seems to have begun, and of noting the degree of congruence between the two lists.'

M UCH attention has been devoted in recent decades to the statistical evidence relating to share in the national income of the United States. It is generally agreed that this evidence, though inadequate for the earlier years, clearly establishes that the share of wages and salaries in United States national income has been significantly higher since World War I than in the nineteenth century. Some economists have concluded that the share of wages and salaries continues to be characterized by an upward trend. In some studies this upward trend is viewed as evidence that labor unions have succeeded in obtaining a larger share of the pie for their members. Others have concluded that wages and salaries have tended to constitute a stable proportion of national income in recent decades, after allowing for cyclical fluctuations. The significance of the results obtained in such studies cannot be determined without consideration of the changing structure of the economy. The growth of corporate organizations and the displacement of individual proprietorships and partnerships have entailed the conversion of large numbers of self-employed persons to wage and salary workers; not only has the proportion of employees in the labor force been raised, but also many of those added to this category have been relatively highsalaried corporate officers, managers, research and similar personnel.' This process is a continuing one. In addition, the growth of public institutions at a more rapid pace than the private sector of the economy has had an important influence on the share of employee compensation in national income.Furthermore, a considerable part of the wage and salary payments made by government has gone in the last two decades to persons outside the civilian labor force, i.e., to military personnel. These and other changes in the structure of the economy call into question the contention that labor's share of national income is best measured . . . by the ratio of Employee Compensation to National Income, 3 as well as many of the conclusions that have been reached by studying these ratios. The main purpose of the present study is to take account of some of the more important changes in the structure of the American economy, as they bear upon the share of employee compensation, by comparing this share with the percentage of employees in the labor force.4 To

T HE main proposition of this paper is that the tremendous long-run upward tide of wages has created an illusion of rigidity by drowning out wage variations that are not only wider than most related phenomena, but are also sensitive to short and long cycles in the general economy. Almost three-fourths of that wage trend has been justified by productivity advances. Slightly over one-fourth of it has been associated with price inflation and is therefore open to the suspicion that the wage trend may have been the prime mover in a wage-price spiral and an integral part of the problem of wage rigidity. Any attempt to convict wages of such responsibility, however, must explain why year-to-year percentage changes in wages rarely led the expansion of business, prices, employment, and productivity and why those in unit labor costs have normally lagged such expansions. Few notions about wages are more widely and firmly held than the belief in their comparative unresponsiveness to economic downturns. The literature abounds with statements like the following: Moderate changes in employment are not associated with very great changes in money wages.' Money-wage levels, like individual wage-rates, are rather insensitive to cyclical forces and decline only in response to major depressions.2 Changes in British and American factory wages during I9 I9-39 lagged behind business activity by substantial numbers of months, and their fluctuations were smaller in amplitude than those of production, employment, and wholesale prices of raw and semi-finished goods.3 The level of wages in prewar Britain was fairly constant in the face of wide employment fluctuations.4 German wage rates have failed to show genuine cycles and have reacted only to major depressions and then with a lag.5 Little decline in Swedish wages could be found in depressions between I887 and I930.6 When there is a considerable increase in unemployment . . . do wages drop as they would in a competitive market? History answers, No.7 There has been no lack of explanation for this apparent wage rigidity in contractions. The most usual has been union resistance, but other explanations have included statutory wage minimums; insistence of the unorganized worker on maintaining his living standards; 8 reluctance of employers to invite popular disapproval, provoke unionization, or risk loss of valuable employees; 9 time required to ascertain that a recession is on; and finally the bureaucratic wage policies of big firms and unions, which can be altered only at substantial cost and difficulty.10 Opinion has been less unanimous concerning wage behavior in expansion. Some believe that wages initiate and aggravate inflation-at least under the recent drive of mass unionism.11 But

T HE study of the credit system is, traditionally, an important part of economic discipline. It is the purpose of this paper to introduce a new method, the method of the linear graph, for the study of this time-honored subject. Ostensibly, the contribution of this paper is pedagogical. However, this method enables us to identify and classify all transactions from properly constructed economic models of credit structures. Since all the operations of a credit system are formed of transactions, it is hoped that an axiomatic approach, which is conspicuous by its absence in the study of money and banking, can be initiated through this approach. In order to do this, certain elementary mathematical concepts of the linear graph must be introduced. For this reason, we shall proceed informally and omit all proofs in our presentation. However, the applicability of the mathematical theory of linear graph to the study of money and banking problems is by no mean limited to the elementary mathematical knowledge contained in the text. It is hoped that the method initiated in this paper can be further developed to yield more fruitful results in economic analysis. This is illustrated in the appendix by some examples of economic analysis using non-trivial theorems of linear graph. The text of this paper, however, can be read by readers with no mathematical background in linear graph theory.