Knowledge that Transforms

Stiglitz and Shapiro tackle equilibrium unemployment as a worker discipline device.

Between 1971 and 1978, wages of more than one-half million nonwhite laborers in the South African mines tripled in real terms. In the same period, the nonwhites employed in the mines switched from being 62 percent foreign to 62 percent domestic.' These changes followed a period-from 1911 to 1971 -during which real wages of black gold miners did not rise, and terminated almost a century of reliance on foreign labor reserves for the majority of such labor.2 These dramatic events are examined here in the context of an econometric model of the demand for labor by the South African mining sector from 1946 to 1980. This affords an unusual opportunity to study the demand side of a market for internal and international migrants, in a society where racial discrimination is formalized in the apartheid system, where powerful mining houses wield potential monopsony power, and where political factors in the region are major determinants of economic behavior. To comprehend the derived demand for workers in this sector, it is essential to outline at least certain aspects of the industry's organization and that of the market for labor; this is undertaken in Section I. Section II develops a stylized model, which is then estimated, from data described in Section III, for the gold, diamond, coal, and other minerals sectors separately in Section IV. I. Organization of the Mine Labor Market

One of the more startling demographic trends of the 1970's has been the sharp rise in lifespan in the United States. Between 1970 and 1980, the life expectancy of 25year-old men and women increased by at least two years. While longevity has been steadily rising for many decades, the recent spurt has hastened the trend; the increase for both men and women in the 1970's more than doubled the improvement of the 1960's.1 The 1970's were also characterized by a decline in labor force participation of older workers. Although rising income and increased disability payments may have explained some of the decline (Donald Parsons, 1980), it seems clear that expanding longevity has not been strongly associated with deferred retirement (Daniel Hamermesh, 1984).2 Thus the retirement years have assumed greater importance in financial planning for the consumer as he or she retires earlier and lives longer. The traditional life cycle model would predict greater savings rates as consumers increase accumulation of assets for their retirement. Consumption while young would drop, and there might be an increase in labor supply as individuals substitute leisure while young for greater levels of life cycle consumption. Even under a Social Security system, as long as the current generation's tax liabilities reflect their longer lifespan, total savings (private plus public) should also rise.3 The predicted increase in savings is substantial. A simple example using the Modigliani-Ando-Brumberg life cycle model can illustrate the extent of the savings shift. Assume that earnings and consumption are constant over working years and lifespan, respectively, and that the interest rate is zero. Consumption is then a constant proportion k of earnings, where k is the ratio of working years to one's lifespan. If consumers made plans based on average lifespan, the representative 25-year-old would have saved 18 percent of earnings in 1970 (1-40/49 when he plans to retire at age 65 and expects to live until age 74). In 1980, after a rise in life expectancy of two years, the representative 25-year-old would have increased savings to 21 percent of earnings, or a net rise in the savings rate (per worker) of 17 percent. That is, this simple life cycle model implies that savings in the last decade should have risen (other things held constant) by approximately 17 percent, or about 80 billion dollars per year. The evidence from aggregate time-series data appears to contradict the life cycle model's prediction of rising savings rates. The net private savings rate in the United States declined from 7.8 percent of NNP in 1970 to 6.2 percent in 1980. While Alan Auerbach (1982) has suggested that some of the reported decline in savings may be caused by mismeasurement, it seems clear that there has been no substantial rise in the savings rate. There are numerous explanations for *Department of Economics, University of Virginia, Charlottesville, VA 22901. I am grateful to Maxim Engers, Daniel Hamermesh, William R. Johnson, Laurence J. Kotlikoff, John Strauss, and especially James Davies and John R. Wolfe for helpful suggestions and comments. 'Life table data can be found in Daniel Hamermesh (1985), the Life Insurance Fact Book (various years), or in unpublished form from the National Center for Health Statistics. 2John Wolfe (1983) provides an ingenious argument for why the Social Security benefit schedule has induced those with shorter life expectancies to retire early. We might therefore expect some rise in the average retirement age as those expecting to live longer defer retirement until age 65. 3If the Social Security taxes do not rise as expected lifespan increases (perhaps because current tax proceeds are paid out as current benefits, rather than retained in a trust fund), young workers will save more to offset the lower expected benefits in the future. 4In a two-period model, the percentage change in savings by the young is equal to the percentage change in aggregate savings.

Most recent studies of macroeconomic behavior fall into one of two categories. The first, often called the equilibrium business cycle approach, stems from the fundamental contribution of Robert E. Lucas (1972), and related work by Thomas Sargent and Neil Wallace (1975) and many others. These studies espouse the view that a positive correlation between output and the stock of paper assets can arise if households are unable to identify the source and, therefore, the permanence of price movements. Employment and output responses in this view are driven by the intertemporal substitution effect, especially the substitution of current leisure for future consumption. Since cyclical fluctuations in employment are large relative to the corresponding real wage movements, substantial wage elasticity of labor supply is required to validate these models. The equilibrium approach to business cycles implies certain restrictions on the conduct of monetary policy. In particular, rational expectations undermine the ability of the monetary authority to influence economic activity in a systematic manner; see Sargent-Wallace for an example. Sticky wages and prices are the cornerstone of an alternative description of macroeconomic behavior. Rooted vaguely in Keynes, and more firmly in the dual decision hypothesis of Robert Clower (1984), this approach studies equilibria with quantity rationing; see Edmond Malinvaud (1977). The rationing story lacks a precise specification of the source of price stickiness, offering very little guidance about the eventual causes of price change. Considerable efforts were made in the 1970's to fill this lacuna in Keynesian macroeconomics. Beginning with work by Martin N. Baily (1974) and others, the implicit contracts literature focused on the incomplete insurability of human capital. Unable to find insurance against fluctuations in labor income elsewhere, workers demand insurance from those best placed to observe labor income-their own employers. Both wage inflexibility and layoffs, then, can be viewed as an outcome of a joint insurance-employment relationship between workers and firms; see Azariadis (1975). Critics like George Akerlof and Hajime Miyazaki (1980) soon discovered that the original contracting models could not produce layoffs without prohibiting severance pay or otherwise limiting the terms of the contract. Others pointed out that these models were determinedly microeconomic, offering few insights into the stickiness of nominal wages or the effectiveness of stabilization policy. Two quite distinct lines of research developed out of the original implicit contract ideas. One focuses on asymmetric information and implementability (see the QJE 1983 Symposium for original work and the review article by Oliver Hart, 1983) as a means of driving a wedge between the ex post marginal rates of substitution of the contractants. Under some technical assumptions, the outcome is involuntary underemployment or unemployment. We are most concerned here with the other line of research, which sought to fit labor or tDiscussants: Guillermo Calvo, Columbia University: Jo Anna Gray, Washington State University.