Labor market conditions and employment of the mentally ill.

BACKGROUND: The mental health services literature includes assertions that workers with mental illness are at earlier risk of unemployment than other workers when the economy contracts. This possibility is important for several reasons. One is that such a phenomenon would support the argument that the lives of mentally ill persons are made unnecessarily stressful by the stigma of mental illness. Another is that the phenomenon could distort comparisons of the effectiveness of programs designed to prepare persons with severe mental illness for work. Despite its importance, the assertion that severely mentally ill workers are at early risk of unemployment has never been empirically tested. AIMS OF THE STUDY: We aim to test the hypothesis that unemployment among persons with severe mental illness (SMI) increases before job loss among other workers. METHODS: We test the hypothesis by applying Granger causality methods to time-series data collected in two communities in the United States (i.e., Concord and Manchester, NH) over 131 weeks beginning on 12 May 1991. RESULTS: We find no relationship between job loss in the labor market and the likelihood that persons with SMI will be unemployed. DISCUSSION: We speculate that persons with SMI participate in the secondary labor market and that their employment status is unlikely to be well described by data gathered in the primary labor market. This implies that widely available measures of labor market status, which are designed to describe the primary labor market, cannot be used to improve the evaluation of programs intended to prepare the mentally ill for work. We also discuss the possibility that persons with SMI may have needs that are better met by the secondary than by the primary labor market. CONCLUSIONS: The intuition that workers with severe mental illness are affected earlier than other workers by labor market contraction may not be correct. We infer that persons with severe mental illness may participate in the secondary labor market about which we know relatively little. We cannot, therefore, easily adjust program evaluations to disentangle intervention effects from those, if any, of the labor market.