Need to replicate Tables 3, 6,7 and 8 using Stata software using the data given. So you can just send me the do-files from Stata.
The Impact of Economic Conditions on Participation in Disability Programs: Evidence from the Coal Boom and Bust The Impact of Economic Conditions on Participation in Disability Programs: Evidence from the Coal Boom and Bust By DAN BLACK, KERMIT DANIEL, AND SETH SANDERS* We examine the impact of the coal boom of the 1970’s and the coal bust of the 1980’s on disability program participation. These shocks provide clear evidence that as the value of labor-market participation increases, disability program par- ticipation falls. For the Disability Insurance program, the elasticity of payments with respect to local earnings is between !0.3 and !0.4 and for Supplemental Security Income the elasticity is between !0.4 and !0.7. Consistent with a model where qualifying for disability programs is costly, the relationship between eco- nomic conditions and program participation is much stronger for permanent than for transitory economic shocks. (JEL J0, H0) In 1993, 3.7 million former workers in the United States were receiving Disability Insur- ance (DI) payments from Social Security. At the same time, 4.4 million disabled Americans were receiving income from Supplemental Security Income (SSI). These two programs accounted for payments to the disabled of nearly $51 bil- lion. Expenditures for disability programs far exceed expenditures on other cash assistance programs. For example, despite the attention it received from politicians and the media, Aid to Families with Dependent Children (AFDC) ac- counted for payments of less than $23 billion to about 14 million recipients in 1993. These high levels of disability payments are the result of a 30-year expansion of disability programs expenditures in the United States. To- gether SSI and DI real expenditures increased 400 percent from 1970 to 1993, and the number of recipients increased 400 percent as well. By comparison, during the same period AFDC pay- ments increased only 50 percent in real dollar terms, and the number of welfare recipients doubled. Because of the enormous growth in disability programs over the last 30 years and their current importance as a source of public aid, understanding why these programs have grown so rapidly is an important public policy question.1 * Black: Center for Policy Research, 426 Eggers Hall, Syracuse University, Syracuse, NY 13244 (e-mail:
[email protected]); Daniel: The Monitor Com- pany, 650 Madison Avenue, 9th Floor, New York, NY 10022 (e-mail:
[email protected]); Sanders: Department of Economics, 3105 Tydings Hall, University of Maryland, College Park, MD 20742 (e-mail:
[email protected]). We thank Susan Black, Amitabh Chandra, Noreen Connery, Leigh Ann Leshock, Philip Mc- Givney, Hoda Makar, Victoria Moyer, Reginald Roberts, Suzanne Plourde, and George Tita for their research assis- tance. Rona Blumenthal of the Office of Research and Statistics of the Social Security Administration provided us with much of the data on OASDI’s disability program. Gerry Faulhaber, Matt Kahn, Janice Madden, Terra Mc- Kinnish, and seminar participants at the Regional Science Association International meetings, the Australian National University, Carnegie Mellon University, the University of Chicago, the University of California-Berkeley, the Institute for Research on Poverty, Johns Hopkins University, the University of Maryland, Northwestern University, the Uni- versity of Pennsylvania, RAND Corporation, Syracuse Uni- versity, Virginia Polytechnic University, and the University of Western Ontario provided helpful comments. The com- ments of two referees greatly improved the paper. We acknowledge the Center for Business and Economic Re- search at the University of Kentucky for research support. Daniel received financial support from the University of Pennsylvania in the form of a University Research Founda- tion Grant. Sanders received support in the form of a Re- search Initiation Grant from Carnegie Mellon University. A part of this research was completed while Seth Sanders was a National Fellow at the Hoover Institution. Black and Sanders acknowledge financial support from the National Science Foundation. 1 Statistical Abstract of the United States, Tables 582, 597, and 598 (U.S. Department of Commerce, Bureau of the Census, 1996), provide the DI and (1980 and 1990) SSI expenditures and recipient counts. Committee on Ways and Means (1994) Table 8.1 provides the AFDC expenditures, and we use Tables 4-20 and 4-24 to impute the approximate 27 The rise in expenditures on DI and SSI coin- cided with a sharp reduction in the relative earnings of low-skilled men. McKinley L. Blackburn et al. (1990), John Bound and George Johnson (1992), Lawrence F. Katz and Kevin M. Murphy (1992), Murphy and Finis Welch (1992), and Chinhui Juhn et al. (1993), to name but a few, examine the fall in earnings of low-skilled workers during the 1970’s and 1980’s. A decline in the labor-force participa- tion of low-skilled men accompanied their de- cline in wages. Juhn (1992), analyzing Current Population Survey (CPS) March Supplement data from 1968 to 1988, finds the decline in labor-force participation particularly severe among less educated men. To what extent has the decline in the earnings of low-skilled men contributed to the increase in disability payments? We address this question in our paper. Donald O. Parsons (1980, 1982) addresses a related but distinct issue. He argues that growth in disability programs accounts for a large share of the post-World War II decline in labor-force participation. His central piece of evidence is that labor-market dropout appears larger for in- dividuals who face more generous DI pay- ments.2 Specifically, the larger the replacement rate—the fraction of pre-disability earnings re- placed by the DI program—the lower is the probability of working among prime-aged men. Parsons’ work, however, has been vigorously challenged. Robert H. Haveman and Barbara L. Wolfe (1984a, b), Bound (1989), and Bound and Timothy Waidmann (1992), while not de- nying that disability programs have had some impact on labor-force participation, dispute the magnitude of the impact. They point out the potential endogeneity of Parsons’s measure of DI generosity.3 Specifically, because the DI sys- tem is progressive, workers with chronic health problems might have lower earnings that would qualify them to have a higher fraction of their earnings replaced by the DI system. In this case, a high replacement rate may simply reflect a worker’s chronic health problems and it is not surprising that dropping out of the labor market is positively correlated with health problems (and hence the replacement rate). Bound and Waidmann (1992) suggest that early diagnosis and treatment now allow many men who would have died to survive, but not to recover enough to work. They argue that this accounts for up to 80 percent of the decline in labor-force par- ticipation for men aged 45 to 54 during the 1970’s. Instead of relying on the replacement rate, and the problems involved with its use, we use varia- tion in local earnings growth within states to see how changes in the local economy affect partici- pation in the DI program. We use this approach because, for a specific geographic region in the United States, there is an exogenous shock to the value of labor-force participation. Because of reg- ulatory changes and the OPEC oil embargo there were a series of sharp increases in the price of coal during the 1970’s. The Appalachian regions of Kentucky, Ohio, Pennsylvania, and West Vir- ginia, in addition to other areas with coal, experi- enced rising employment and earnings while other areas within relatively short distances suffered the declines in economic activity experienced by the overall American economy. A subsequent bust in the coal market in the 1980’s reversed these gains. We use this variation in local economic conditions to identify the impact of labor-market conditions on DI program participation. Moreover, because the coal boom persisted for a long period, and the payments and number of disabled recipients for 1970 SSI. In 1970, the DI program paid $2.4 billion to 1.3 million recipients. Because SSI was not a federal program in 1970, corresponding estimates for the disability portion of SSI are not available. Approximately 1 million of the 3.1 million SSI recipients in 1970, however, were blind or disabled, and total expenditures were about $2.9 billion, which provides us with approximate expenditures of $1 billion. 2 Similarly, Frederic P. Slade (1984), using data from the 1969 Longitudinal Retirement History Survey, estimates that a 10-percent increase in the generosity of DI payments would reduce labor-force participation of men between the ages of 58 and 63 by 8.1 percent. 3 Haveman and Wolfe (1984a, b) note the potential en- dogeneity of the replacement rate and correct for this endo- geneity by predicting DI income. As Bound (1991) argues, however, exclusion restrictions of questionable legitimacy are required in order to generate the instruments required for this strategy. See Parsons (1984) for his defense to these challenges. Bound’s (1989) approach is to use DI applicants who are denied disability insurance as a control group for those who received disability insurance. See Parsons (1991) for a critique of Bound’s approach and Bound (1991) for his reply. Gruber (2000), using a 36-percent increase in disabil- ity payments in Quebec, finds an estimated labor-force nonparticipation elasticity of about 0.3. 28 THE AMERICAN ECONOMIC REVIEW MARCH 2002 subsequent coal bust persists through today, these economic shocks represent long-term changes in local labor-market conditions rather than transi- tory ones. Our results suggest that permanent job creation or destruction has a much larger effect on disability program use than more transitory changes in local labor markets. Establishing the relationship between local economic conditions and DI program partici- pation may help inform the debate between Parsons and Bound as well. The central issue in both our question and theirs is the degree to which disability participation and labor-force participation are substitutes. A finding, for ex- ample, of no relationship between economic opportunities and disability program use would support the contention that the effects of the rise of disability programs on labor-force participa- tion are small. Furthermore, the larger the ef- fects of economic opportunities on disability program enrollment, the larger is the substitu- tion potential between work and disability. This suggests a larger role for disability insurance in explaining the decline in prime-aged male labor-force participation. The remainder of the paper is structured as follows. In the next section, we describe fea- tures of the two major U.S. disability programs. In Section II we describe the coal boom and bust, documenting the impact that this cycle had on local economic conditions and on the use of disability programs. In Section III we discuss our econometric specification, and in Section IV we present the results of our estimation. In Section V we offer some concluding comments. The Appendix describes the data we use. I. Program Description4 The DI program covers all participants in the Social Security system. In order to qualify for DI benefits an individual must be deemed dis- abled by the Social Security Administration (SSA). According to SSA rules, an individual is deemed disabled if he is unable “to