Effects of psychiatric disorders on labor market outcomes: A latent variable approach using multiple indicators

Using a combined sample of about 7500 individuals from the National Comorbidity Survey Replication (NCS-R) and the National Latino and Asian American Study of Mental Health (NLAAS), we examine the effect of psychiatric disorders (depression, panic attack, social phobia and generalized anxiety disorder (GAD)) on a host of labor market outcomes. We estimate a series of structural equation models with latent indices for mental disorders. The latent continuous measures are generated from the model using multiple indicators (symptoms) and multiple causes (determinants) of the disorder and the different indicators are linked to the labor market outcome variable. The latent factor loadings reflect the fact that not all symptoms of a disorder are equally important in explaining the labor market behavior of individuals and disability. Further, we address possible reverse feedback effects of labor market participation on mental illness using new instrumental variable approaches. We find substantial impact of mental illness on employment and labor force participation for both males and females, but the effect on weeks worked and missed work days are significant only for males. We also examine differential effects for different race/ethnic populations. Based on concordance analysis between the conventional binary indicator for a mental disorder and our continuous latent measure, we find that current approaches understate the effect of mental distress on labor market outcomes. We also conduct counterfactual simulation exercises to demonstrate significant labor market benefits from improved treatment due to the newer generation of drugs, and suggest appropriate rehabilitation approaches for the mentally disabled.