Knowledge that Transforms

The prevalent but incorrect belief that the United States now has a natural unemployment rate around 6 percent fosters an unwillingness among the nation's leaders to promote economic growth sufficient to cut joblessness below that rate. Earlier, at least through the 1960's, a consensus held that economic growth could bring the nation's unemployment rate down to 4 percent, and even lower when combined with policies to address structural problems. In the last half of the 1960's, unemployment remained below 4.0 percent for four consecutive yearsone of the best sustained performances in the post-World War II period. But in the 1970's and 1980's, joblessness climbed to higher levels at each peak and trough and, even though the 1978 Humphrey-Hawkins Act required the achievement of 4 percent unemployment, the consensus to do so fell apart. Several reasons have been advanced to explain the deterioration and justify an upward redefinition of the minimum unemployment rate achievable through policies that raise aggregate demand. If examined carefully, however, the arguments just don't stand up. Among such ideas are the following: 1) women and youth entered the labor force in great numbers, and allegedly have naturally higher rates of unemployment; 2) joblessness cuts, in particular below the socalled normal rate, supposedly generate wage and price increases which cause recessions and rising joblessness; and 3) social programs and labor legislation ostensibly reduce beneficiaries' needs and desires to work and also make firms hesitant to add workers. The proposed reasons at most account for a minor part of the climb in joblessness, and do not prevent the attainment of 4 percent unemployment through policies that promote strong economic growth. Nonetheless, structural unemployment is a serious problem, and its alleviation calls for industrial and trade policies, job and training programs, antidiscrimination enforcement, and other special initiatives. Such efforts will add momentum as economic growth propels the nation towards full employment.

In 1976, Robert published a devastating critique of the then current practice for quantifying the effects of alternative policies. He argued that, in formulating plans, economic agents necessarily look into the future, and thus the decision rules guiding their actions depend on parameters describing the expectations of future variables, as well as parameters of taste and technology. viewed economic policy as the selection of rules that generate the values of policy variables, rather than the selection of arbitrary sequences of policy variables. Thus, any change in policy will systematically alter the structure of econometric (1981, p. 126), and the estimated coefficients in (the then current) consumption, wageprice, or investment models could not be considered structural, that is, invariant to alternative policy regimes. The important and damning implication for policy analysis is that these econometric relationships will prove unstable in precisely those situations in which they are called upon to analyze proposed policies. This paper seeks to exploit the substantial variation in business tax policy during the 1980's to shed some light on the quantitative importance of the Lucas Critique (LC) and to evaluate the ability of explicit theory to enhance econometric analysis. In August of 1981, the economic agenda of the Revolution began to be put in place with the passage of the Economic Recovery Tax Act (ERTA). This was followed by the Tax Equity and Fiscal Responsibility Act of 1982, the Deficit Reduction Act of 1984, and the Tax Reform Act of 1986. Other changes -such as the increase in defense spending and growth of and restrictions on the federal deficit-indicate that expectations of the current and future course of fiscal policy were altered radically during the Reagan Administration. Maintained in this paper is the assumption that these changes in tax and other aspects of fiscal policy can be viewed as an unanticipated change in policy regime. In regard to the 1981 legislation, the Economic Report of the President claimed that [T]his tax policy is a sharp break from the policies of the recent past. It reflects a different understanding of the way tax policy affects the U.S. economy (1982, p. 109). Under the assumption that the 1980's witnessed a fundamental change in the stochastic process characterizing fiscal policy, econometric equations susceptible to the LC should become temporally unstable. Those models that successfully address the LC by isolating expectation parameters should remain stable. The quantitative importance of the LC is evaluated by examining the stability of four different econometric models of business investment that to varying degrees are vulnerable to the LC: the Neoclassical model studied extensively by Robert Hall and Dale Jorgenson (1967), Robert Eisner and M. Ishaq Nadiri (1968), and their numerous collaborators, the Q model associated with James Tobin, the Vector Autoregressive model (VAR-S) -advanced by Christopher Sims (1982), and a Hybrid VAR model (VAR-H) used to study investment by Robert Gordon and John Veitch (1986). Since the four models differ in the extent to which they are guided by explicit theoretical frameworks, these results will also be useful in assessing the impact of economic theory on econometric performance. *University of Chicago, Chicago, IL 60637. I thank Charles Hulten, Robert Lucas, and Allen Sinai for critical comments on a preliminary draft. All errors, omissions, and conclusions remain my sole responsibility.

Over the past few decades there has been a phenomenal increase in the number of households that are headed by women. This fundamental change in family structure has been accompanied by increases in the number of women in poverty, relative to the number of men in poverty. For instance, in 1959 women were 23 percent more likely to be poor than were men, but by 1985 they were 51 percent more likely to be poor.1 At the same time, the ratio of children's poverty, relative to that of men's has skyrocketed. While children were 53 percent more likely to be poor than were men in 1959, by 1985 they were 121 percent more likely to be poor. The feminization of has become shorthand for describing the disproportionate percentage of poverty that is borne by women living alone or with their children. This paper analyzes the forces behind these changes. A framework is developed that allows changes in poverty rates to be decomposed into component parts. Using a timeseries of independent cross sections covering the period from 1967 to 1985, this framework is then used to examine the empirical significance of a variety of factors that contribute to poverty. The first set of factors examined includes women's wage rates and hours of work, men's earnings, and AFDC-guarantee levels. The effects that these variables have had on poverty rates are twofold; they have affected poverty directly (holding marital status constant), and have affected poverty indirectly through their effects on marital status. Both of these effects are considered. In addition, the direct (but not the indirect) poverty effects of changes in the number of children, and changes in income from child support and alimony, are examined.