Knowledge that Transforms

Martin Feldstein and Otto Eckstein (1970) have set out and estimated a model of interest rate determination which they claim represents an integration of Keynes's liquidity preference theory with Irving Fisher's theory of the impact of expected inflation on interest rates. They achieve their integration by first inverting a Keynesian demand function for real balances, solving it for the nominal rate of interest. Then to incorporate Fisher's effect, they simply add to the right side of this equation a distributed lag in current and past actual rates of inflation, the same proxy for expected inflation that Irving used. Feldstein and Eckstein interpret sizable and statistically significant estimated coefficients on the elements of that distributed lag as confirming the presence of an effect of anticipated inflation on the interest rate. It is questionable whether Keynes's and Fisher's theories stand in need of any integration at all, since they are in principle compatible in the first place. The two theories are on very different footings. Keynes's liquidity preference theory is a theory about one particular structural equation relating real money balances, income, and the nominal rate of interest. On the other hand, Fisher's theory is one about how the whole economy is put together; that is, it is a statement about the reduced form equation for the nominal interest rate. In Fisher's theory, an exogenous increase in the anticipated rate of inflation is asserted to work its way through the economy in such a fashion that the nominal interest rate rises by the amount of the increase in anticipated inflation.1 In this note, I suggest that Feldstein and Eckstein's equation does not successfully synthesize Keynes and Fisher. Furthermore, I suggest that Feldstein and Eckstein's econometric procedure is not a good one for estimating the dimensions of the Fisher effect. In particular, the effect may be present in full force but still not be detected by Feldstein and Eckstein's procedure. On the other hand, it is possible to construct examples of economies in which there is really no effect but in which Feldstein and Eckstein's test would point to the presence of one. Finally, I show that in a model that includes both Keynes's liquidity preference schedule and a reduced form for the interest rate like the one posited by Fisher, Feldstein and Eckstein's equation is not statistically identifiable.

T HE response of government policy variables to the targets or objectives of policy has received increasing attention in the literature. All of the published studies have limited their investigations to the behavior of the monetary authority.1 This concern with the reaction function of the monetary authority can be traced to three factors: (1) Since the monetary authority is independent (or semiindependent) of the government, investigators have been concerned with whether the central bank has responded to the appropriate social objectives such as price stability and full employment; (2) Central banks have often expressed concern with short-run objectives (such as interest rate stability) and investigators have examined whether the pursuit of such objectives has hindered the implementation of the social goals; and (3) Econometric modelbuilders have become concerned with the statistical problems arising from the endogeneity of policy.2 This paper reports on an attempt at estimating the reaction function of the Federal Home Loan Bank System (FHLBS) and the Federal National Mortgage Association (FNMA). Both FNMA and the FHLBS are governmentsponsored agencies (the FHLBS being under more government control than FNMA) whose major objective concerns the mortgage market and housing activity. FHLBS makes loans (advances) to savings and loan associations which, in turn, use the funds for mortgage loans. FNMA buys (or sells) mortgages in the secondary market. During the 1960's these operations were aimed primarily at stabilizing the volume of mortgage credit with the implicit view towards stabilizing housing activity.3 There were a number of factors during the 1960's which might have prevented the FHLBS and FNMA from pursuing their objective of minimizing the variability in mortgage flows and housing starts. First, before 1968, FNMA purchases and sales of mortgages appeared in the United States budget, hence mortgage purchases, in particular, were carefully examined since such purchases added to the deficit. After 1968 this was no longer true since FNMA was made a private corporation and taken out of the federal budget. During the credit squeeze of 1966 the Johnson Administration placed severe restrictions on the debt issues of various government agencies, including the FHLBS. In effect, this meant that the FHLBS was not free to decide to raise funds to lend to savings and loan associations without severe scrutiny. In the next section, two alternative models of FHLBS behavior are derived within a utility function framework. Empirical estimates of the alternative reaction functions are presented. same is done for FNMA behavior in a subsequent section. remainder of the paper presents the implications of the results for a number of issues, including: (1) whether FNMA and the FHLBS have, indeed, used Received for publication June 29, 1972. Revision accepted for publication December 13, 1972. * author wishes to acknowledge the financial support of the Federal Home Loan Bank System. views expressed are not necessarily shared by the FHLBS. He also thanks M. J. Hamburger, S. M. Goldfeld, and an anonymous referee for their comments. 1See, for example, H. G. Johnson and W. G. Dewald, Analysis of the Objectives of in D. Carson (ed.), Banking and Studies, Irwin Inc., 1963; and J. H. Wood, Model of Federal Reserve Behavior, in George Horwich (ed.), Process and Policy, Irwin Inc., 1967. 2See F. de Leeuw and J. Kalchbrenner, Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization-Comment, in Review, Federal Reserve Bank of St. Louis, April, 1969, and the Reply by L. C. Anderson and J. L. Jordon in same issue. 3See Harry Schwartz, The Role of Government-Sponsored Intermediaries in the Mortgage Market, in Housing and Policy, Federal Reserve Bank of Boston, Boston, 1970; and Ernest Bloch, The Federal Home Loan Bank System, in Federal Credit Agencies, Commission on Money and Credit, Prentice-Hall, Englewood Cliffs, N.J., 1963.

N this paper, I will generalize the modified Newton method previously applied in Chow (1968) to the computation of full-information maximum likelihood estimates of parameters of a system of linear structural equations to the case of a system of nonlinear structural equations. The success of that method for linear systems 1 has stimulated my present attempt to generalize it for nonlinear systems. The subject of maximum likelihood estimation of nonlinear simultaneous equation systems has been studied by Eisenpress and Greenstadt (1966). There are three main differences between their approach and ours. First, their basic formulation is more general, assuming that all parameters in the system may appear in every equation,2 whereas we assume as the basic setup that there is a distinct set of parameters belonging to each equation. Second, partly because of the first, we are able to obtain simpler and more explicit expressions for the derivatives of likelihood function required in the calculations. Third, and also partly because of the first, we can conveniently deal with the important problem of linear restrictions on the parameters in the same equation or in different equations. A fourth feature of this paper, and a feature which has partly motivated it, is the contrast of the linear with the nonlinear case. As it will be shown, there are many similarities in the computations of both. This demonstration can enhance our understanding of the nature of the estimation equations. Two additional features of this paper are the treatments of identities in the system and of residuals which may follow an autoregressive scheme. We will derive in section II the estimation equations for nonlinear systems, under the assumptions that each structural equation contains a distinct set of parameters, that the parameters are not subject to any linear restrictions, and that the (additive) residuals are serially uncorrelated. Section III treats the special case when some equations are linear, and contrasts this case with the nonlinear case. Section IV deals with identities and linear restrictions on the parameters. Section V is concerned with the problem of autoregressive residuals.

T HE basic purpose of this paper is to show that an improved understanding of longswing mechanisms in economic-demographic interactions may be attained with cross-spectral analysis even though spectral analysis has, to date, cast only doubt on the existence of such long swings. The superiority of crossspectral analysis to simple estimation of power spectra stems from its emphasis upon examining relationships among economic and other variables (Granger, 1966). For exploration of the long-swing hypothesis, spectral analysis suffers from a further disadvantage its results are sensitive to the form of the data analyzed (levels, rates of growth, or deviations from trend), while cross-spectral analysis is not (see section II). Application of both spectral and cross-spectral analysis to economic and demographic data in Sweden, the United Kingdom, and the United States indicates (1) results of spectral analysis are generally similar across the three countries and negative with respect to the long swing hypothesis; and (2) cross-spectral analysis shows quite different long-swing mechanisms at work while giving some positive confirmation to the Easterlin model of long swings for the United States.

M UCH work has been done on the many aspects of the size distribution of income. This includes a large body of literature purporting to explain variations in this phenomenon both temporally and spatially for large geographic units in terms of different and/or changing functional distributions of income (Kuznets, 1953), differential growth rates (Kravis, 1960; Kuznets, 1955), random or stochastic processes (Champernowne, 1963; Gibrat, 1931; Rutherford, 1955), and combinations of economic, social and demographic factors (Aigner and Heins, 1967; Al-Samarrie and Miller, 1967). Conspicuously absent from the literature, however, is any empirical work done on variations in the size distribution of income in small areas, either over time or over space. The omission of time-series analysis of variations among small area income distributions is due largely to the unavailability of small area personal income statistics for all but decennial census years. No such excuse may be offered for the absence of cross-sectional analysis, inasmuch as the last three decennial censuses (1950, 1960 and 1970) have provided data for regions and small areas on a comparable basis. It would appear, however, that published empirical research on size distribution of income in small areas has tended to concentrate more on the nonsubstantive statistical questions such as the identification of ''proper area definitions and less on inter-area variations than is true of inter-state and international income distribution studies which by and large have tended to avoid facing the former issue squarely, or merely assumed that it was irrelevant. This paper will draw together these heretofore unjoined aspects of the study of income distribution. In it we shall construct and test an analytical model for explaining variations in the degree of income concentration among small spatial units in terms of various economic, social and demographic variables which are thought for a priori reasons to be influential. We shall draw upon the approaches of similar studies while modifying them for our own ends, keeping in mind also that our aim is not to verify any single hypothesis such as economic growth or development vis-a-vis income distribution, but to isolate and quantify the influence of a broad group of causal factors which operating simultaneously may determine the degree of concentration of family incomes within a spatial area. Not only will it be shown that the economic and social factors most highly associated with spatial variations in income inequality can be positively identified for small areas, but what is more important, these will be shown to vary over space in relative impact as well as varying in importance according to level of aggregation of the observation unit.

PpT HE purpose of this paper is to test empirically two propositions which have been closely, although not exclusively, associated with Milton Friedman in recent years. The first is the hypothesis that changes in monetary or fiscal policy variables are frequently ineffective in stabilizing some target variable because they are poorly timed.1 The second hypothesis, which was presented by Friedman and Meiselman (1963), is that the money supply is a more important determinant of aggregate demand than autonomous expenditures.2 We test these hypotheses by considering the effects of changes in fiscal and monetary variables upon the movements in gross national product within the framework of a small, short-run econometric model of the United States economy. In our model both money supply and government expenditure are regarded as autonomous manipulative policy instruments. A special feature of the study is a quarter-by-quarter investigation of the effects of changes in each of the two policy variables upon the movement of GNP. The period of investigation dates from the end of the Korean War (1954-I) to the beginning of serious military involvement in Vietnam (1963-IV). The plan of the paper is as follows: In section II we specify and estimate the structural equations of the model. Section III is concerned with a dynamic analysis of the system. Here we derive our estimates of the dynamic multipliers and examine the system for stability. In section IV we utilize the preceding results to determine the relative importance of each of the two policy variables during the sample period. Simplified criteria are suggested and applied for evaluating the actual operation and relative effectiveness of the two types of policy. The final section contains a summary of the main results and some concluding remarks.

T HERE is a growing body of literature on the choice of technology for developing countries.' Much of this literature deals with the factor proportions problem,2 yet there has been little attention directed to the role multinational firms may play in the choice of technology. Yeoman (1968) in a cross-sectional study comparing the operating characteristics of the foreign subsidiaries of 13 United States firms, did find that there was little difference in the amounts of capital used per worker in plants located in advanced, as compared with developing countries. Where the cross-elasticity of demand is low and where manufacturing costs are low, relative to selling price, as in the pharmaceuticals industry, there is little incentive to adapt technology. Yeoman found very little technological adaptation of production processes on the part of pharmaceutical firms. On the other hand, in the home appliances field where cross-elasticities of demand are high and the share made up by production costs in final value, high adaptation was more extensive and more labor was combined into the production process in developing countries. In contrast to Yeoman, the particular concern of this paper is to compare the operating characteristics of multinational and local firms with respect to the ratios in which they combine capital and labor in final output. We address the question: Do multinational firms employ production techniques which are more capital using than those employed by local firms producing similar products? If they do, they could then be singled out as a major contributor to the factor proportions problem confronting developing countries. Two field studies were conducted in which detailed information was compiled for 14 United States subsidiaries and 14 closely matched local counterparts. Nine matched pairs of firms were studied in the Philippines and five matched pairs in Mexico. In four of the sectors studied in the Philippines it was infeasible to obtain closely matched pairs in the Mexican field study.3

Robert C. Blattberg, Evaluation of the Power of the Durbin-Watson Statistic for Non-First Order Serial Correlation Alternatives, The Review of Economics and Statistics, Vol. 55, No. 4 (Nov., 1973), pp. 508-515

PpTHE terms balanced and unbalanced appear in many and varied contexts within economic theory. One writer has even suggested that it is a cause of much confusion that 'balance' in economic could mean almost anything (Ohlin, 1959, p. 338). This lack of definition has also characterized attempts at statistical verification of the theories. Balanced in the Nurkse (1953) sense arose in the context of development economics and sought to provide some policy conclusions for less developed countries. Solow-Samuelson-Von Neuman balanced is of more theoretical interest and of broader application. Two recent papers (Swamy, 1967 and Yotopoulos, 1970) laboured under the disadvantage of mixing these two quite separate notions of balanced growth. We shall argue that consequently, the evidence already presented sheds little light on the development controversy. This paper examines evidence for balanced and unbalanced theories in the Nurkse-Hirschman sense (Haberler, 1961; Hirschman, 1958; Nurkse, 1953). Nurkse's balanced argument involves two propositions: firstly, the size of the domestic market is maximized only with a balanced of sectors because, by assumption, excess supply is wasted and excess demands are frustrated; secondly, the size of the market is the main determinant of investment in less developed countries. Therefore, Nurkse viewed balance as a means of stimulating growth. Hirschman's version of unbalanced growth, like Nurkse's theory, was concerned primarily with the inducement to invest. Only Hirschman argued that excess demand and supply are necessary in order to make investment decisions obvious. Excess demand induces investment in supplying industries (backward linkage) and excess supply leads to the establishment of using industries (forward linkage). It is clear that Hirschman's theory of unbalanced applies mainly to the intermediate goods sector. A number of studies (including Bhatt, 1965 and Mathur, 1966)) have made considerable ground in providing a synthesis of the theories, usually developing the fact that Nurkse and Hirschman exempt vertical and horizontal sectors, respectively. Consequently there may not arise a problem of policy choice between balanced and unbalanced growth. In the context of this paper, the most important feature of both theories is the implicit exclusion of the effects of international trade. If trade becomes the engine of growth in any less developed country, the theoretical controversy becomes less relevant for development policy. Furthermore, distortions in sectoral rates due to international trade, do not constitute unbalanced in the Hirschman sense. These considerations render difficult any empirical enquiry.