State provision of vocational rehabilitation services for those with a severe and persistent mental illness (SMI) depends in part on state resources and a state’s inclination to provide such support services. Clearer answers as to how employment affects both the mental health status and use of mental health services by those with a SMI will help states determine if and how such vocational services should be offered. Results may also inform the psychiatric treatment community of the significance (if any) that working has in one’s overall treatment and recovery process.

Evidence suggests employment may improve symptoms and reduce mental health service use. However, this evidence comes from project-specific evaluation data that a) fails to control for endogeneity of employment status and b) may not generalize to broader populations. This study estimates the effect that employment has on two outcome variables, 1) mental health status and 2) total mental health costs, controlling for endogeneity. Analyses are performed for the total population and separately for persons eligible for public mental health coverage through their beneficiary status in the Federal program Supplemental Security Income (SSI).

The two primary data sources are 1) the Outcomes Measurement Survey (OMS) for those using Maryland’s Public Mental Health System (PMHS) and 2) Maryland PMHS claims data for the same persons and time period. The design is quasi-experimental, with a “treatment” group (n=794) and a “control” group (n=4,264) defined by employment status (currently working or not) in the 2nd of 3 consecutive OMS interviews over a one-year period. Mental health status is measured by a 24-question instrument that produces an overall quantitative score. Total mental health costs are calculated from claims records from the 6-month period between the 2nd and 3rd interviews. Full information maximum likelihood (FIML) estimations simultaneously model the potentially endogenous employment status and the outcome variable.

The study population is persons with an SMI, covered through Maryland’s PMHS between September 2006 and August 2009. Primary diagnosis groups are schizophrenia (26.6%), bipolar (23.7%), depression (36.6%) and other (13.1%). Approximately half of the population is eligible for PMHS coverage through SSI.

Results for the total population are that mental health status is not improved, but total mental health costs are significantly lower for those in the treatment group. The SSI population estimations, however, show worsening effects on mental health status and increased total mental health costs.  Heterogeneity in treatment effects is also observed for different baseline mental health scores, with persons who have poorer prior mental health scores showing more improvement on both outcome variables.

Policy implications are that vocational rehabilitation programs may have cost offsets in the form of reduced mental health costs. This study estimates the reduction in total mental health costs to be $270 per person in the 6 months following the measure of employment status. Differences in treatment effects suggest that employment may have aggravating effects on mental health for some individuals. This is motivation for better integration between mental health providers and vocational rehabilitation providers.