This paper describes changes in hours of work and income between 1959 and 1979 of women and men ages 25-64. It includes attempts to measure and value nonmarket production and leisure as well as market work, to take account of possible income sharing within households, and to allow for economies of scale in household production. The most important empirical result is that, relative to men, women's access to goods and services and leisure was lower in 1979 than in 1959. Changes in hourly earnings, hours of work, and household structure contributed to this result. The sex differential in hourly earnings is explored in detail.

Introduction, 278. — I. Comparisons of sizes and operating characteristics within industries among different types of establishments, 279. — II. Inter-industry differentials in multi-unit operations; multi-unit operations and other industry characteristics, 283.

SINCE I929 there has been a substantial change in the location of manufacturing in the United States. The direction of redistribution has fairly consistently been from north to south and from east to west. In I929 the South and West together accounted for less than one out of every four United States manufacturing employees and for only one-fifth of value added by manufacture. By I958 1 their share of United States manufacturing had increased to one-third as measured by either variable. This change has had a significant impact on the political, social, and economic life of the nation. The consequences for businessmen, labor leaders, and government officials have been strong, direct, and sometimes painful. Inevitably, the question that arises is why? Why did this change occur? What were the historical developments and economic forces that determined its extent and direction? The purpose of this paper is to delineate the outlines of an answer to this question. In particular, we shall examine and reject the hypothesis that regional shifts in demand or markets were the major determinant of locational change. Area differentials in the rate of growth of manufacturing are the result of area differentials in the rate of growth of individual industries and of area differences in industrial structure. The latter refers to the extent to which an area's industries are those which, at the national level, are experiencing rapid or slow growth. If an area has predominantly fast-growing industries, we say it has a structure. If it has predominantly slow-growing industries, we refer to its structure as unfavorable. 2 The first point to be made is that the redistribution of manufacturing since I929 has been primarily the result of area differentials in the growth of individual industries rather than the result of differences in industrial structure. This need not always be true. The shift of industry to the Southeast in England in the inter-war decades was attributed more to differences in industrial structure than to area differentials in the growth of specific industries.3 In the United States, I929-I954, the comparative growth of manufacturing employment by state, unadjusted, was highly correlated with comparative growth, adjusted for industrial structure. (Spearman's coefficient of rank correlation equals +.8 ii.) Comparative growth, unadjusted, showed no correlation with the extent to which a state had a comparatively favorable or unfavorable industrial structure. (Coefficient of rank correlation equals + .003.) 4 Table i presents the comparative gain or loss of manufacturing in each Census division between I929 and I954, and shows it divided into two parts. The first represents the sum of the comparative gains or losses of individual industries. The second, or remainder, is a rough measure of the influence of industrial structure.5

This paper discusses the recent behavior of the female-male earnings differential and asks how it is likely to change in the future.' After adjusting for hours, age and schooling, we find that, contrary to results obtained in other studies from unadjusted data, the female-male earnings ratio among whites increased between 1959 and 1969 by 4.8 percent despite an unprecedented increase in the female-male employment ratio among whites of almost 20 percent in the same decade-a favorable augury for future trends in the earnings ratio. The data used below come from the 1/1000 sample of the 1960 and 1970 Censuses of Population.2 Due to lack of time and space, I focus on the sex differential among whites. However, the color differential between blacks and whites narrowed appreciably between 1959 and 1969. Black female hourly earnings, adjusted for age and schooling, rose 82 percent compared with 68 percent for black males and 53 percent for white females. By 1969, hourly earnings of black females were only 15 percent less than those of white females of comparable age and schooling, while for women with more than twelve years of schooling the adjusted color differential had practically disappeared. Total earnings in 1959 (1969) are calculated for workers classified for a variety of characteristics; total annual hours are estimated by multiplying weeks worked in 1959 (1969) by hours worked in the census week in 1960 (1970) for each worker and summing across all the workers in a classification. Finally, total earnings are divided by total hours to obtain, for each classification, average hourly earnings-or, equivalently, the mean of individual average hourly earnings weighted by annual hours. I focus on hourly earnings as the best mea* Professor of economics, City University of New York, and Vice-President, Research, National Bureau of Economic Research (NBER). Work on this paper was begun while I was a Fellow at the Center for Advanced Study in the Behavioral Sciences, 1972-73, with partial support provided by the Russell Sage Foundation. It was continued at the NBER under a grant from the Rockefeller Foundation. Carol Breckner, Phyllis Goldberg, Jan Platt, and Christy Wilson provided research assistance and many useful suggestions; Charlotte Boschan wrote the earnings program; Robert Michael and Yoram Weiss made helpful comments on an earlier draft. I am grateful to all of the above; they bear no responsibility for the views expressed. In particular, this paper has not been submitted to the NBER Board of Directors for approval and is not an official NBER publication. I I do not consider determinants of the general level of earnings (female and male). 2 See Fuchs (1968) for a discussion of the strengths and shortcomings of this source, which provides much useful information on employed persons' sex, schooling, age, race, marital status, occupation, and industry. All persons who were at work during the census week and had earnings in the year preceding the census are included in the sample except agricultural, religious, and unpaid family workers, since, as is well known, the estimation of earnings and work hours for such workers is difficult.

A HIGH birth rate, a declining demand for labor in agriculture, and poor educational facilities have traditionally provided the South with a relatively large supply of untrained labor for manufacturing industries. This condition suggests the following hypotheses: i. Average wages in manufacturing will be relatively low in the South because the southern industrial structure will tend to be heavily weighted with low-wage (low-skilled) industries. 2. Because labor and capital are not perfectly mobile, southern wages will tend to be lower than elsewhere for identical work. 3. For the same reasons of excess supply and incomplete mobility, southern wages will be particularly low for work requiring little skill and training. 4. Because of the wage differential, the South should be gaining in manufacturing relative to the rest of the country. 5. Because the differential is greater in lowwage employment, the South should be gaining most in those industries which make the greatest use of low-wage labor. The purpose of this paper is to test these hypotheses and to draw conclusions regarding the balance of forces which tend, on the one hand, to eliminate regional wage differentials, and on the other hand, to perpetuate them. In addition to examining shifts in the location of manufacturing, we will consider the influence of population change and minimum wage legislation on the southern wage position.

Specialists in labor economics and public economics at 40 leading research universities provided opinions of policy proposals, quantitative best estimates and 95-percent confidence intervals for economic parameters, and answers to values questions regarding income redistribution, efficiency versus equity, and individual versus social responsibility. Their positions on policy are more closely related to their values than to their estimates of relevant economic parameters. Average best estimates of the economic parameters agree well with the relevant literature, but individual best estimates are usually widely dispersed. The individual 95-percent confidence intervals are much narrower than the substantial cross-respondent variation in estimates would warrant.