The Impact of Pre-existing Health Conditions on Job Mobility: A Measure of Job Lock

This paper uses the National Medical Expenditure Survey of 1987 to measure the importance of job lock as a deterrent to job mobility. I begin by using a simple difference-in-difference technique to estimate the differential effect of family sickness on the job mobility of individuals with and without employer-provided health insurance. I then amend my technique to take account of the possible bias of the difference-in-difference estimator when the control and experimental groups are dissimilar. In the empirical work in the paper, the coefficient on the job lock term is always insignificant. Using a medical expenses index that combines all the family sickness measures, I find that job lock accounts for a 0.33% increase in job mobility. Moreover, the confidence interval of this estimate excludes large levels of job lock. I then reanalyze previous work which uses the same data source and finds large and significant job lock. I find that after controlling for omitted variables, using better data to create the job lock measures, and constructing comparable control and experimental groups, job lock no longer has a significant effect on job mobility.