Measuring the effect of husband's health on wife's labor supply.

A sizable proportion of women remain married well into late life and an increasing proportion of them participate in the labor force. Since women tend to marry men older than themselves and men tend to experience serious illnesses at younger ages than women, women frequently witness declining health in their husbands. This is likely to affect a wife's labor-leisure trade-off in offsetting ways. Prior studies have not sought to disentangle the effect of a husband's poor health on his wife's reservation wage from the income effect of his ill health. We argue that, if we control for husband's earnings, the coefficient of husband's health in models of his wife's labor force participation (and hours of work) will reflect, in part, her preference over whether to decrease her labor supply to provide health care for her husband or whether to instead increase it to purchase this care in the market. However, husband's earnings are likely to be endogenous in these models due to unobserved characteristics common to husbands and wives. We find that the estimated effect of husband's health depends on whether we instrument for husband's earnings and on the health measure used. This is indicative of the importance of using a variety of health measures and controlling for husband's earnings, and their endogeneity, in future research on the effect of husband's health on wife's labor supply.