Most experts believe that health care demand is fairly inelastic. If you are sick, you will not be very price sensitive. There are exceptions to this rule (e.g., elective surgery such as plastic surgery, purchases of eyeglasses) but most studies find that patients are fairly insensitive to changes in health care prices. For instance, the RAND Health Insurance Experiment found that the price elasticity of medical expenditures is -0.2.
An working paper by Amanda Kowalski claims that medical care and prices have an elastic relationship. “My main results show that the price elasticity of expenditure on medical care is -2.3 across the .65 to .95 quantiles of the expenditure distribution, with a point-wise 95% confidence interval at the .80 quantile of -2.5 to -2.0. Although I allow the price elasticity estimate to vary with expenditure, I find a fairly stable elasticity across the estimated quantiles. This estimate is an order of magnitude larger than the RAND estimate of the mean elasticity of -0.2.”
Kowalski uses claims and patient level data from a large employer’s database. Since price and quantity are often correlated, one needs a random shock to quantity in order to identify this elasticity. For an instrument, Kowalski uses whether or not a family member has an injury. When a family member has an injury, this will not affect the medical expenditures of other family members (assuming they are not also injured). However, an injury will use up a large portion of a family’s deductible and thus lower coinsurance rates from 100% (during the deductible) to 20% (after the deductible is used up).
One may worry that sickness risk is correlated among family members. For instance, if you investigated a family of extreme snowboarders, the probability any one person is injured is high. It is possible that we can observe one person’s injury in the data which will be correlated with a high probability of injury for their spouse and child. If the other family members are covered under an employed spouse’s health plan, the injury may not show up in the data, but some medical expenditures will.
To check whether or not sorting in mating along the health risk dimension occurs, Kowalski looks at couples who each have their own deductible. Thus, the injury of a partner will not affect their spouse’s coinsurance rate. Kowalski finds that price elasticity in this case is not statistically different from zero and, as predicted, the instrument has little power.
One possible explanation for the large elasticity is that partners may leave the insurance coverage after their spouse gets injured. If their spouse is seriously injured, they may have to stay home to take care of them. However, before leaving their coverage, they may decide to have all of their major medical procedures done. Because the data are only from 2003 and 2004, intertemporal price elasticity could be a problem. Kowalski does find that some evidence that inter-temporal shifting is not driving her results, however.
Recent comments