Mental Illness, Work, and Income Support Programs

Several studies document that 20% to 40% of rejected applicants for Social Security Disability Insurance later work (14 – 16) . This has been interpreted by Autor and Duggan (17) as clear evidence of moral hazard. We conclude, in contrast, that an equally compelling statistic is the 60% to 80% of rejected applicants who do not subsequently work. Of course, some of these individuals may choose not to work while they appeal their disability eligibility, and others may fail to work owing to low skill level rather than a health-related disability. However, existing evidence is no more likely to show that moral hazard exists than it is to document that most rejected applicants needed them.

Some analysts contend that increased disability awards based on mental illness are an indicator of moral hazard or a problem with the disability determination process. Specifically, Stapleton and colleagues (18) analyzed Supplemental Security Income caseload growth during the 1990s. They found that demographic and labor market factors could explain most of the rise in applications for most medical conditions but that far fewer of the increased applications based on mental disorders. They conclude that the Supplemental Security Income criteria for mental disorders had been expanded.

We do not accept this conclusion because the authors compare the 1990s growth in applications with application rates during the 1980s, when the Reagan administration disproportionately terminated people with mental illnesses from the rolls. A reduction in applications at that time could be expected independent of other factors, so it is not an appropriate baseline for comparison.

Some analysts suggest that the relationship between applications and unemployment rates indicates that many workers seek public benefits not because they have a disabling mental illness, but because they have been laid off (3) . An increase in applications does not equal a rise in awards.

Social Security Disability Insurance awards have increased owing to both mental illnesses and musculoskeletal disorders (3, 18) . A 2003 study found that Disability Insurance applications increased more with employment crises in the 1990s than in the mid-1980s (17) . In concluding that abuse of Social Security Disability Insurance has “reached unsustainable levels so that the Social Security Disability Insurance screening process is broken,” Autor and Duggan (17) invoke the cyclicality of Social Security Disability Insurance applications, the fall in application rejection rates, and the rise in awards for disabilities with low mortality rates, such as mental impairment and musculoskeletal disorders. Although Social Security Disability Insurance applications moved with unemployment rates over the 1990s, the award rate for applications is countercyclical over the entire 1970–2006 period Figure 1 ). In other words, stringent eligibility criteria closely mirrored the cyclicality in award rates.

In addition, the trends in Social Security Disability Insurance awards per population (versus application) do not show any clear relationship to the unemployment rate, especially after the 1970s. This is true for both mental and nonmental disorders, which drift upward slowly after the mid-1980s Figure 2 ). On the basis of these trends, we find no evidence for the contention that moral hazard is somehow more serious for disability awards related to mental illness.

The recent relationship between Supplemental Security Income award rates for adults with mental disorders and the unemployment rate is not consistent with a moral hazard explanation that posits that disability awards for mental illness are countercyclical, as they are for unemployment insurance. From 1992 to 2006, unemployment fluctuated widely, falling from 7% to 4% before rising and falling dramatically again. However, Figure 3 shows no consistent trend in Supplemental Security Income award rates for mental disorders in relation to rates of unemployment. Awards for nonmental disorders follow a similar trend.

The challenge of determining disability owing to mental illness means that some truly needy applicants are denied benefits. If this is the case, rejected applicants will have worse outcomes than those determined to be disabled. We gain some insight into the extent of moral hazard versus false negatives using data from about 500 women surveyed by the Women’s Employment Study (19) . We contrast the housing stability, work activity, and welfare experiences of Women’s Employment Study respondents who applied for and received Supplemental Security Income benefits to women who applied for but did not receive these benefits (data not shown).

In 1997, 36% of respondents met the diagnostic screening criteria for major depression, posttraumatic stress disorder, or generalized anxiety disorder. We did not have the statistical power to analyze the 36% of the sample with identified mental illness, but among the 12% of the overall Women’s Employment Study sample who received disability benefits at any time between 1997 and 2003, 30% had been evicted or had experienced homelessness at some point during the study, similar to the 28% of those who had never applied for or received benefits. By contrast, among the 26% of respondents whose Supplemental Security Income applications were not approved by the 2003 interview, 43% experienced housing instability, suggesting that they were in need of benefits. The rejected applicants were only half as likely to be working in 2003 as those who never applied for disability—43% versus 79%. They also worked much less over the 6-year period—in 26% of the 72 months versus 63%. These data are inconsistent with the basic moral hazard hypothesis.