This thesis studies the labor supply effects of health shocks for aging Americans. To
address the mismeasured binary health variables used in the labor supply equation, this
thesis develops a new theoretical approach to the non-classical measurement error. In
addition to Chapter 1 which provides an overview of the thesis, there are three primary
chapters explaining the theoretical development and the empirical studies.
Chapter 2 theoretically addresses the estimation bias due to the misclassification
of a binary regressor in treatment models. Different from the assumption of a valid
instrument in the literature, this paper allows the potential instrument to be correlated
with the measurement error. In such a general setting, I propose an estimator relying
only on those extreme observations that are free of misclassification. This proposed
estimator is proven to have large sample properties and much better performance than
OLS and traditional IV estimates in finite samples.
Chapter 3 uses the method proposed in Chapter 2 to handle the binary, misclassified
health variable in studying the labor supply effects of health shocks. Extracting information on true health from objective health measures, for example functional limitations
and doctors’ diagnoses, this new method relies on such information to dynamically select observations that are free of misclassification. Using the 2012 wave of the Health
and Retirement Study (HRS), this paper primarily examines the labor supply effects
of health shocks for men and women aged 45-61. This study finds that individuals in
middle age will greatly reduce their labor supply when experiencing health shocks and
that the estimation results are very sensitive to the health measures used.
In Chapter 4, I examine how an individual’s labor supply responds in the short- and
long-run to a negative shock to her spouse’s health. I propose an optimal instrument
strategy with fixed effects to study labor supply effects of spousal health shocks. Analysis of the 1996-2012 data from the Health and Retirement Study (HRS) indicates that
in the short run, both husbands and wives change their labor supply very little when
their spouses become ill. However, in the long run, husbands adjust their labor supply
in response to their wives’ health problems. As the duration of wives’ health problems
increases by two years, husbands work 165 fewer hours per year.