Although industrialized countries continue to have business cycles, such cycles have changed significantly in character. In what follows I shall describe some of these changes and point to their possible implications for research and policy. Perhaps the most obvious change is that business recessions-periods of actual decline in economic activity -have become less frequent, shorter and milder. Interruptions to a steady rate of growth are more often simply slowdowns rather than actual declines in aggregate economic activity. This kind of shift can be observed in the business recessions identified by the National Bureau of Economic Research. On the whole, the five recessions of 194870 were shorter than the five recessions of 1920-38, produced smaller declines in output, income and employment, and were less widespread in impact. But recent recessions have been accompanied by higher rates of unemployment than might have been expected in view of other evidence attesting to their mildness. One of the factors underlying this shift toward recessions of lesser severity, and one reason why it may be expected to persist, is the trend in the industrial composition of employment. Industries that normally experience larger percentage reductions in employment when recession hits are less important in the overall economic picture nowadays, while industries that often continue to expand right through recession have become more important. Of the eleven major industrial sectors that account for total employment, seven experienced reductions averaging three percent or more during the five recessions of 1948-70 (Table 1). These seven sectors include manufacturing of durable goods like autos and appliances, with an average drop of 12 percent; mining, with an average drop of 10 percent; transportation and utilities, with an average drop of 5 percent; and farming, manufacturing of nondurable goods like textiles, construction, and federal employment, with drops of 3 to 4 percent. Employment in these seven sectors constituted more than half of total employment in 1955, but by 1972 their share had declined to about two-fifths. The other four major sectors--wholesale and retail trade; services; finance, insurance, and real estate; and state and local government -experienced much smaller declines or actual increases in employment during the five most recent recessions. They accounted for slightly less than half of total employment in 1955; by 1972 they accounted for three-fifths of the total. In short, the industries that have coiitributed most to reduced employment during recession have shown little or no growth during the past fifteen years or so, while those that have contributed least to recession have grown much faster. The added stability has reduced the impact of recession upon total employment by something like one-third. If the 1955 distribution of employment among the eleven sectors had prevailed in all five recessions of * Vice-President/Research, National Bureau of Economic Research, Inc., and Senior Research Fellow, Hoover Institution, Stanford University.

During the last two decades the issue of equality according to sex and race has become one of increasing importance to economists. An extant literature on the economics of discrimination began with the pioneering work of Becker (1957). This paper is an attempt to shed further light on the extent of employer discrimination by sex and race by comparing the earnings of self-employed workers to their wage and salary counterparts. In short, if employer discrimination is a principal source of discrimination against blacks and women, then we would expect the black/white and female/male earnings ratios to be higher for the self-employed compared to their wage and salary counterparts. No discrimination of this type is applicable to self-employed workers. In addition, blacks and women should be relatively overrepresented among the self-employed compared to wage and salary workers in the economy. Section II of this paper further elaborates on this indirect method to estimate the extent of employer discrimination in the labor market for blacks and women. In section III, the results of this method are presented using the 1978 Current Population Survey as the data source. Results indicate the black/white and female/ male earnings ratios are no larger for the self-employed compared to their wage and salary counterparts, even after making various attempts to adjust for differences in other variables that affect earnings, and to limit the influence of consumer discrimination on the results.