Working Paper: NBER ID: w2433
Authors: Richard Frank; Paul Gertler
Abstract: We employ a unique data set from a community based survey to assess the effect of mental distress on earnings. The main advantage of the data is that detailed measurements of mental health status were made on all subjects in the study. This means that our population-based measure of mental distress does not rely on a patient having had contact with the health care system and obtaining a diagnosis from a provider. The use of diagnosis-based measures may introduce measurement-error bias into the estimates. Our results show that the presence of mental distress reduces earnings by approximately 21% to 33%. To assess the magnitude of any measurement-error bias we present a estimates of models using measures of mental health both on a population-wide basis and on a diagnosis basis. The estimated impact of mental illness on earning is only 9% lower using the using the diagnosis-based measure. The conclusion drawn from this is that little bias is introduced by using the diagnosis-based measure.
Keywords: Mental Health; Earnings; Measurement Error
JEL Codes: I10; J30
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| mental distress (I31) | reduced earnings (J31) |
| diagnosis-based measures (C52) | estimated impact of mental distress on earnings (J17) |
| mental distress (I31) | earnings (using population-based measures) (J39) |
| demographic variables (J10) | reduced earnings (J31) |