A Framework for Medicare Reform

Studies | Health

No. 315
Friday, September 12, 2008
by John C. Goodman

APPENDIX A — Estimating Supply-Side Effects

by Andrew J. Rettenmaier

The demand-side effects of higher deductibles on Medicare spending are derived from estimates based on the RAND Health Insurance Experiment (HIE).41 The total deductible is compared to total per-capita spending on Medicare-covered services by age.42 This ratio is compared to similar ratios from the RAND simulation results to stratify the effects of different deductible amounts. Today, a base deductible of $2,500 is equal to about 17 percent of estimated total average spending on services covered by Medicare. At the time of the RAND experiment, a deductible of $100 represented about 12 percent of average spending by individuals who had access to free care, while a deductible of $200 was equal to 25 percent. The estimate of the effect of a $2,500 on a retiree’s demand for health care would fall between the 12 and 25 percent reduction suggested by the RAND results. The reduction in health care spending due to total deductible amounts that rise relative to average spending are estimated in the same way. Such a rise will occur with the intermediate case and the Rettenmaier–Saving plan.

Estimating possible supply-side responses due to higher cost sharing is less straightforward, though recent work has hinted that the supply-side responses may be substantial. Amy Finkelstein finds that as much as 50 percent of the real per capita health care expenditure growth between 1950 and 1990 can be attributed to the reduction in out-of-pocket spending.

Previous estimates of the degree to which declining out-of-pocket spending explains expenditure growth suggested a smaller effect than those obtained by Finkelstein. Joseph Newhouse used the RAND HIE to determine the percentage of the growth in real per capita health care spending between 1950 and 1980 that is due to the change in out-of-pocket spending over that time period. He notes that the RAND HIE results indicated that moving from an average coinsurance rate of 33 percent to a zero rate at a point in time produced a 40 to 50 percent increase in demand. Given that the average realized coinsurance rate dropped from 67 to 27 percent between 1950 and 1980, or 40 percentage points, he estimates that demand should have increased about 50 percent as a result of the change in the coinsurance rate. However, there was a 400 percent increase in real spending over that time period. He therefore concludes that the increase in real spending was eight times as large as one would expect from the change in the coinsurance rate.

Finkelstein estimated that spending increased 37 percent during the first five years of the Medicare program. Medicare decreased the population average coinsurance rate by 7 percentage points. Thus, the 50 percentage point decrease in the coinsurance rate between 1950 and 1990 would increase spending by 264 percent which accounts for about half of the total real per capita spending increase over this period; this is in contrast to the one-eighth estimated by Newhouse. Finkelstein notes that Medicare provided incentives that allowed health care suppliers, particularly hospitals, to invest in capital.

A reform that potentially moves cost sharing in the opposite direction of the cost sharing percentages between 1950 and 1990 will likely produce supply-side responses in addition to the demand-side responses that will both lower the level of spending as well as the growth rate in spending. In the present study, the adjustment due to the higher deductibles is estimated in two stages. The first stage is the demand-side adjustment due to the rising total deductible as described above.43 The second stage of the adjustment attempts to account for effects of the change in the realized percentage of out-of-pocket spending. In the core case this percentage stays the same through time, so no adjustment is made to the per capita growth rate used by Medicare. In both the intermediate case and the case of the Rettenmaier-Saving plan, however, realized out-of-pocket spending as a percent of total spending would rise. The estimates in Newhouse and Finkelstein are both based on realized out-of-pocket spending as a percent of total spending, not deductibles and copays as a percent of spending. In the case of the Rettenmaier-Saving plan, the realized out-of-pocket spending as a percent of total spending would rise by an estimated 15 percentage points by 2050. The second-stage adjustment is estimated by assuming that this rise in out-of-pocket spending reduces the real growth in per capita spending by 2050 by 15 percent.


Amy Finkelstein, “The Aggregate Effects of Health Insurance: Evidence from the Introduction of Medicare.” Quarterly Journal of Economics, February 2007.

Joseph Newhouse, “Medical Care Costs: How Much Welfare Loss?” Journal of Economic Perspectives, Vol. 6, No. 3, summer 1992.


Read Article as PDF