Knowledge that Transforms

Professor of Economies, Columbia University, and Senior Research Staff, National Bureau of Economic Research. This paper is based, in large part, upon ongoing research in human capital at the National Bureau. This research is financed by grants from the Carnegie Corporation and from the Economic Development Administration of the United States Department of Commerce. I am grateful to Gary Becker, Barry Chiswick, and Victor Fuchs for helpful comments, and to Masanori Hashimoto and Sara Paroush for most competent assistance.

the probability of an equal percentage change occurring is far smaller for R. In terms of equally probable changes, income has a greater impact on interest rates than unborrowed reserves. The partial correlation coefficients present less clear evidence as to the relative contributions of R. and Y. The partials are greater for R. than for Y in the equations for 1953-1960, but greater for Y than for R,2 in the 1961-1965 period. For the overall period, the partials are greater for RX when the equations are specified in the H-S form for 19531960 and for Y when the equations are specified in the H-S form for 1961-1965. Thus, the partial correlation coefficients also provide little evidence in support of H-S's conclusion that RX has a greater impact on interest rates than Y. , i the 1961-1965 period. For the over-

Empirical studies of the relationship between money and expenditures frequently seek to explain movements in aggregate expenditures by movements in some appropriate money total. For example, in the most notable recent such study, Friedman and Meiselman 1 conclude that the appropriate money total is currency plus all commercial bank deposits. Past and current changes in this variable better explain expenditure changes than do changes in each of two alternative money totals.2 The assumption implicit in such tests of the money-expenditures relationship is that the behavior of the components of any money total does not matter. Specifically, Friedman and Meiselman estimate:

None of the added coefficients was significant at the 0.05 level, so there is no strong evidence of substantial cyclical variability in the relation between ANHWI and AU. The most one can detect is some indication of a shift during the expansion of early 1955. We take this opportunity to report the performance of our model during the past few years as a predictor of NHWI.1 The predicted and actual values of A1NHWI are shown below. Averaging quarterly predictions smooths the results. Nevertheless, the equation does not seem to have drifted systematically off course since the sample period. It over-predicted ANHWI in 1966, 1967, and 1969, and under-predicted in 1965 and 1968. However, the cumulative under-prediction slightly exceeded the over-prediction. In summary, we continue to believe that the relationship between the change in normalized help-wanted advertising and the change in unem-

Edwin Burmeister, Winston W. Chang, Rodney Dobell, The Role of Saving in a Growth Model with Induced Inventions: A Correction, The Review of Economics and Statistics, Vol. 52, No. 4 (Nov., 1970), pp. 446-447

any point in this region will converge to a quasiequilibrium state with a zero rate of growth of capital [2, p. 66]. The present note indicates that the steady-state growth equilibrium with a positive rate of growth of capital in this case is, in fact, globally stable. The erroneous result in [2] comes from the assertion that K(0) -0 implies K(t) 0 O for all t (p. 65). Instead this assertion should be stated as

' . when p 0 (i.e., the elasticity of substitution is less than unity), the isoquants have asymptotes which do not coincide with the axes. The exact formula for the asymptotes is given by K val/p K =

Although the validity of quoted prices is an important issue and must not be slighted, it is suggested here that there are also other important problems in the steel industry which need to be considered. These problems relate to the various ways the industry can respond to changes in external conditions which are made possible by very close relationships among the large number of products it sells. The purpose of the present article is to indicate the importance of these responses for steel price change analysis and then to develop and test a new kind of price change index which attempts to measure these responses. While the orientation of the present article is perhaps somewhat unfamiliar, it is not new. For example, some of the following can be traced to a suggestive

where G is the growth rate of aggregate demand, and gi and Ei are the actual growth rate and the income elasticity of output for the ith sector respectively. GEi is defined as the expected growth rate of the ilh sector.2 Measure (2) is preferred to (1) as the latter is unnecessarily sensitive to extreme deviations in sectoral growth rates. Swamy then correlated both measures of imbalance with the aggregate growth rate, G. Coefficients were positive and statistically significant for all periods except 1938-1948. He concludes that the 'statistical evidence does not corroborate the balanced growth theory.' 3 Swamy's results have to be interpreted carefully. In the first place there are weaknesses in the statistical techniques he adopts. These result from his reliance on the correlation between G and V. This correlation is questionable on two counts. Firstly, where the k sectors form a large proportion of aggregate output, we would expect high sectoral growth rates to be accompanied by a high growth of overall output. The association between V and G merely reflects a mutual component in both (i.e., gi). However, since Swamy disaggregated into 13 manufacturing sectors, this spurious element may not be important. Secondly, variations in G automatically affect V since one of the elements of the latter is GEi. This will generate a positive correlation between G and V when the sign of (gi GEj) is negative (and a negative correlation when the sign is positive). Take the limiting case,