The author analyzes data from the National Medical Expenditure Survey
of 1987 to measure the importance of ''job lock'' - the reduction in j
ob mobility due to the non-portability of employer-provided health ins
urance. Refining the approach commonly used by other researchers inves
tigating the same question, the author finds insignificant estimates o
f job lock; moreover, the confidence intervals of these estimates excl
ude large levels of job lock. A replication of an influential previous
study that used the same data source shows large and significant job
lock, as did that study, but when methodological problems are correcte
d and improved data are used to construct the job lock variables job l
ock is found to be small and statistically insignificant.