In contrast, Royalty and Hagens (2003) used a unique dataset in which the variation in
findings, they estimated a price elasticity of insurance take up essentially equal to zero, much
lower than the price elasticity of demand for other fringe benefits. Gruber and Washington
(2005) capitalized on a natural experiment in 1994, when postal employees were allowed to pay
the employee share of their health insurance premiums using pre-tax dollars. This study also
estimated a very low take-up elasticity, approximately –0.02.
None of these studies accounted for the availability of coverage from alternative sources,
such as from a spouse’s firm. However, some recent analyses have considered elasticity in the
context of choice. Polsky et al. (2005) analyzed the choice for employees with an insurance
offer from their own-employer, an alternative source of coverage, and no coverage. Using a
nationally representative household survey (the 1996 and 1999 Community Tracking Study
[CTS]), they found that paying a larger share of premiums significantly increased the odds of
that a worker would decline an employer offer of coverage and remain uninsured.5 Simulating
the effect of reducing the employee share of the premium to zero, they found a smaller price
effect compared to other studies, and attributed the difference to the fact that their model
considered available alternative sources of coverage.
Honig and Dushi (2005) also considered employee take up in the context of choice, and
estimated the elasticity of the take up of coverage from a worker’s own employer with respect to
out-of-pocket premiums separately for wives and husbands. They estimated a price elasticity of
5 Polsky et al. (2005) estimated an odds ratio of 1.023 for married workers and 1.035 for single workers— meaning that when the employee share of premiums increased by 1 percentage point, the probability that the worker declined an employer offer of coverage and remained uninsured was 2.3 percent higher (P x 1.023).