Knowledge that Transforms

Abstrac-t-This paper gathers evidence on the contribution of various techniques of exchange rate management to the output stability of twelve industrial countries. We estimate the distribution of unanticipated disturbances in outputs and the payments balances under pegged exchange rates from the 1955-1971 experience. We use this distribution to characterize the foreign exchange market intervention procedures which simultaneously minimize the output variances of the sample countries. The efficient procedures and the alternatives of managed floats, basket pegs, and the European currency area are compared according to structure and efficacy; and several implications for I.M.F. surveillance of exchange rates are drawn.

The Box-Cox technique has been frequently em- ployed in the estimation of economic models for which theory suggests no a priori appropriate functional form. Typically, an iterative OLS procedure is used to find the maximum likelihood estimates of the regression coefficients. However, Spitzer has demonstrated that biased estimates of the coefficient variances result from the use of the OLS covariance matrix. The purpose of this paper is to investigate the magnitude of the bias for two cases typical of the kind encountered in many empirical studies that employ the iterative OLS approach. The paper finds that hypothesis testing based upon the OLS covariance matrix may be quite misleading since the magnitude of the bias is quite large in one of the cases analyzed.

Conventional empirical studies of money neutrality have focussed on the response of aggregate economic measures to anticipated and unanticipated money supply shocks. The present paper uses data from the U.S. pork industry to test for money neutrality at the microeconomic level. The appeal of the pork industry stems largely from the homogeneity of the product we consider and the fact that the product is traded in what are essentially auction markets. We find that current unanticipated, but not anticipated, money supply shocks have real effects in the industry.

A bstract-This article proposes an explanation for the fact that while wages of married women contribute to equalizing the distnrbution of family wages the equalizing effect declines during the early stages of the married life cycle. The explanation is based on the interaction between on-the-job accumulation of human capital and labor supply behavior. Empirical results from the NLS panel data suggest that the explanation is plausible and also show that in contrast to the results of previous cross-section studies there is no decline over time in the equalizing effect of wives' earnings on the distribution of family earnings.

Steven Husted, Tryphon Kollintzas, Import Demand with Rational Expectations: Estimates for Bauxite, Cocoa, Coffee, and Petroleum, The Review of Economics and Statistics, Vol. 66, No. 4 (Nov., 1984), pp. 608-618

A model is developed in this paper which allows the growth rate of public sector productivity to be estimated without an explicit measure of sectoral output. In this model, communities are regarded as generalized households, and the goods and services provided by the community's government are interpreted as internally produced household goods. The model is then fitted to quarterly data from the U. S. National Income and Product Accounts, and the rate of productivity change is found to average 0.50% per year for the period 1959-1979. THE sources-of-growth approach to productivity measurement distinguishes between the growth in real output due to the growth in factor inputs-capital, labor, materials-and the growth in output due to advances in technology, managerial practice, and other factors related to productive efficiency.' The former are collectively associated with the movement along an aggregate or sectoral production function, while the latter are associated with the shift in that function.2 In practice, the shift in the production function is measured by the residual real output not attributed to the growth of the share-weighted inputs, and termed total factor The sources-of-growth model is based on the underlying assumption that real output and real input can be measured with a reasonable degree of accuracy. While this presumption is correct for many types of economic activity, it is essentially incorrect for those sectors which produce services. As Kendrick (1982) has recently noted, approximately one-third of the real gross product originating in the service sector of U.S. private business economy is estimated using employment, manhours, or labor earnings. Such procedures systematically misstate the true growth of real output, and thus systematically bias the estimated growth in total factor productivity. The situation is even worse in the public sector: in the U.S. National Income and Product Accounts, the output of general government is defined to be equal to labor input, implying that the growth of total factor productivity exactly offsets the effect of nonlabor inputs. A model is developed in this paper which allows the growth rate of public sector productivity to be estimated without an explicit measure of sectoral output. In this model, communities are regarded as generalized households, and the goods and services provided by the community's government are interpreted as internally produced household goods. This interpretation allows us to appropriate the analytical apparatus of the household production model, and thereby permits us to estimate the key parameters of the model-including a government total factor productivity parameter-without first having to measure the output of the govern-

A bstratAcross industry and country comparisons have found the speed of adjustment in short-run employment functions to be positively related to the rate of unemployment. The present paper develops a time-series model in which the reaction coefficient varies with unemployment, and estimates it for the manufacturing sectors of the eight OECD countries for which suitable data are available. The new model is non-linear and is estimated using full information maximum-likelihood procedures. Application of the log-likelihood ratio test finds that the new vanable reaction coefficient model is superior to the fixed reaction coefficient model for the United States, Canada, Australia, Austria, Ireland and the United Kingdom.