Are we Overdiagnosing Mental Illnesses? Evidence from Randomly Assigned Doctors

Almost two in 10 adults the U.S. and Europe are, at any moment time, diagnosed with a mental illness. This paper asks whether illness is over- (or under-) diagnosed, by looking its causal effect on individuals margin of diagnosis. We follow all Swedish men born between 1971 1983 matched to administrative panel data health, labor market, wealth family outcomes estimate impact diagnosis subsequent outcomes. Exploiting random assignment 18-year-old doctors during military conscription, we find that for people increases future likelihood death, hospital admittance, being sick from work, unemployment, while lowering probability married. Using separate identification strategy, measure service same set rule out our setting primarily mediated altering serving. Our findings are consistent potential over-diagnosis