In this paper we model health as a latent variable to predict effects of health on wag
es and labor force participation. We jointly model the health, wage and labor force participation accounting for potential endogeneity of health and sample selection bias. Health is modeled as a latent variable of which multiple discrete indicators are observed. Multiple measures on wages allow us to control for measurement error in wage. We find that for women, after controlling for education and age, no effects of latent health are found on wages. However, the results do suggest that the estimates are sensitive to measurement of health. In participation equations a strong effect of health is observed when we do not control for wage rates. After controlling for wage rates, the effects of health are roughly halved but continue to remain strong predictors of
participation. We compare results for women with those for men, for whom we find strong impacts of
latent health on both wages and labor force participation.