Knowledge that Transforms

In a recent paper in this Review, Duncan MacRae and Elizabeth MacRae perceived an important aspect of the labor-supply effects of the payroll tax namely, that the effect is crucially dependent upon the individual response to a kinked budget constraint. However, their actual conclusions on the direction of the response need major qualification. In particular, they assumed that all responses would be marginal, when in fact the possibility of nonmarginal response is always present when non-linear budget constraints shift. This latter possibility makes the labor-supply response much more ambiguous. This is illustrated in Figure 1 (modeled after Figure 1 of MacRae and MacRae). The budget constraint before the imposition of the tax is A B, giving the utility maximizer a choice of any income-leisure combination along AB. After the imposition of the tax composed of a marginal tax rate on earnings up to a maximum level of earnings E-the (disposable income) budget constraint is BDC. (Yn is the amount of nonlabor income.) At point D, the taxable maximum is reached. MacRae and MacRae assumed that an individual initially located above the maximum level (say, point II) would relocate along CD, resulting in an increase in labor supply if leisure is a normal good (the income effect); and that an individual below the maximum level (say, point I2) would relocate along BD, resulting in a decrease in labor supply if the substitution effect dominates the income effect. However, although this would occur for marginal changes, nonmarginal changes are also possible: an individual initially at II could relocate along BD, reducing labor supply; and an individual initially at I2 could relocate along CD, increasing labor supply. Both are possible and not in conflict with any of the assumptions (that leisure is a normal good or that the substitution effect dominates the income effect). Heuristically, imagine a shift from AB to CH, causing the individual at I, to definitely move to CD; a subsequent of DH to DB could easily induce the person to relocate along DB. Or, imagine a shift from AB to BG, causing the individual at I2 to definitely move to DB (under the assumptions); a subsequent pivot of DG to DC could easily induce the person to relocate along DC. Thus, the possibility of nonmarginal movements introduces more ambiguity than realized before.' As a sidelight, note that the labor-supply effects of an increase in the taxable maximum and of an increase in the tax ratepolicy alternatives that are currently being considered to raise revenue are both ambiguous for the same reasons. Figure 2 shows an increase in the tax rate as causing a shift from BDC to BD'C'. Those initially below the maximum level may decrease labor supply (marginally) or increase it (nonmarginally), while those above the maximum level may increase labor supply (marginally) or decrease it (nonmarginally). Figure 3 shows an increase in the taxable earnings maximum from E to E' as causing a shift from BDC to BD'C'. Although there is no response by those below the maximum (by revealed preference), those above may either increase labor supply (marginally) or decrease it (nonmarginally). The latter effect is likely to be stronger for this group than

Dale Henderson and Thomas Sargent (hereafter H-S) and Y. C. Park have analyzed the effectiveness of monetary and fiscal policy in the two-sector analogue of James Tobin's dynamic aggregative model. Contrary to the assumptions of Tobin's model, fiscal policy can affect real income in the H-S model. However, the sign of the effect of fiscal policy on real income depends on a number of conditions relating to the parameters of the money demand function and the capital intensities in the two sectors. A somewhat more disturbing result is that the sign of the effect of fiscal policy changes as one changes, ceteris paribus, the assumption of which sector is the more capital intensive. The H-S results seem to imply that the analysis of the effectiveness of fiscal policy in the traditional ISLM analysis is very sensitive to the assumption of a one-sector production technology. It is important to realize, however, that the H-S model differs from IS-LM analysis in two ways. Besides assuming a two-sector production technology, the H-S model also assumes a perfect capital market, where the asset value of capital is always equal to reproduction cost. Sargent and Neil Wallace have analyzed a one-sector model with and without the perfect capital market assumption (without and with a disequilibrium investment demand function). They find that many of the strange results of Tobin's model are eliminated if there are costs of adjusting the capital stock. For example, if the costs of adjustment are large enough, expansionary monetary policy lowers the interest rate as in IS-LM analysis, contrary to Tobin's results. Also, the stability conditions of the model imply that with costs of adjusting the capital stock, fiscal policy affects income and interest rates in much the same way as in IS-LM analysis. This paper alters the H-S model to allow for an investment demand function based on costs of adjustment.' In analyzing the comparative static results of this model one can determine whether the strange results of the H-S model are due to the assumption of a two-sector production technology or the assumption of a perfect market in existing capital goods. The profit-maximizing subsystem in the H-S model can be solved to yield the price level and the marginal product of capital as functions of the relative price of investment (see the Appendix). A general equilibrium occurs when the consumption good, money,2 and investment good markets are in equilibrium.

A recent article in this Review by Joan Haworth, James Gwartney, and Charles Haworth (hereafter H-G-H) presented some significant findings on the source and structure of improvements in the relative economic status of nonwhite males during the 1960's. Specifically, H-G-H concluded that approximately one-half of the increase in the nonwhite/white earnings ratio (NWER) during the 1960's was simply . . attributable to the exiting of older nonwhite workers with low relative earnings combined with the entry of younger, better-prepared nonwhites who have high relative earnings' (p. 167). The balance of the gain in relative nonwhite earnings was the result of a decline in employment discrimination against nonwhites and improvements in the relative productivity of nonwhites. If correct, these findings have several important implications for the prospect of black and white earnings equality. First, they suggest that the effects of past discriminatory practices in both employment and the acquisition of human capital continue to reduce the earnings power of older black males still in the labor force. These past practices are an important source of current differences in the average earnings of blacks and whites in aggregate. Hence, they constrain the success of policies to achieve racial earnings equality. This brief note presents some additional evidence consistent with the H-G-H conclusions. We are mainly concerned with their data in Table 3 (p. 164) on changes in the NWER within age cohorts between 1959 and 1969. The H-G-H hypothetical identical productivity' ' NWER measures the nonwhite-white earnings gap caused by factors other than measured productivity variables, such as racial differences in occupational structure which are unrelated to productivity differences. Since the relative occupational distributions indirectly revealed by these hypothetical NWER underlie some of the major H-G-H conclusions, a more direct examination of changes in the occupational distribution of blacks and whites during the 1960's may prove a useful check on their findings. A group's index of occupational status can be calculated by weighting the proportion of the group employed in an occupation by the mean earnings for the occupation and summing across major occupational categories.' The higher (lower) a group's index, the greater the proportion of the group in higher (lower) paying occupations. Therefore, the ratio of nonwhite to white occupational status (NWOS) will measure solely racial differences in the distribution of workers among occupations. The higher (lower) the NWOS, the more (less) favorable the occupational structure of blacks relative to whites, ceteris paribus. Estimates of the male NWOS for age cohorts in 1959 and 1969 are presented in Table 1, along with the corresponding NWER.2 In aggregate, the NWOS increased by 10.2 percent during the