Designing Health Insurance for the Elderly

Featured in print Bulletin on Aging & Health

Expenditures on health care for the elderly are high and rising rapidly. In 2006, Medicare benefit payments totaled $374 Billion, accounting for 12 percent of federal government spending and 20 percent of the nation's total health care spending. Including supplemental insurance policies and out-of-pocket spending, the elderly consume more than one-third of all health care in the U.S., despite being only 13 percent of the population. Looking ahead, Medicare expenditures are projected to rise by 8 percent annually over the next ten years.

The federal government has employed several strategies to restrain the growth in Medicare spending, including prospective payment for hospitals and reduced payments for providers. Yet the rise in Medicare spending has continued unabated. As a result, there is increasing interest in using demand-side policies to control costs. These policies would pass more costs on to patients in the form of copayments and deductibles in order to induce them to use less health care.

Such policies could be effective in reducing Medicare expenditures, but there are two potential stumbling blocks. First, the elderly may not be sensitive to price in their consumption of health care. Second, increased cost sharing may lead patients to skip efficacious preventative care, resulting in expensive hospitalizations that negate the initial cost savings (the "offset" effect).

Surprisingly little is known about these two issues. The best evidence comes from the RAND health insurance experiments of the 1970s, which randomly assigned patients to plans with different coinsurance rates. The RAND study found that consumers are somewhat responsive to price - a 10 percent increase in price is associated with a 2 percent decrease in health care use - and that there is no offset effect. However, it is not clear if those results apply to the Medicare program today, as the elderly were excluded from the RAND study and there have been many changes in the health care landscape in the last thirty years.

In "Patient Cost-Sharing, Hospitalization Offsets, and the Design of Optimal Health Insurance for the Elderly" (NBER Working Paper 12972), researchers Amitabh Chandra, Jonathan Gruber, and Robin McKnight provide new evidence on these questions.

The authors use a recent policy change in the California Public Employees Retirement System (CalPERS), a program that provides insurance to 1.2 million workers, retirees, and their dependents. The policy change involved staggered copayment increases that affected different patient populations - those enrolled in a Health Maintenance Organization, or HMO, and those enrolled in a Preferred Provider Organization, or PPO - at different times. The authors examine the effect of copayment increases on health care utilization, using patients that did not experience the increase as a control group. The data for the analysis are medical utilization records for 70,000 persons enrolled in a CalPERS Medicare supplemental insur-ance ("Medigap") plan.

The authors find that elderly patients are quite price sensitive in their health care consumption. A $10 increase in the office visit copayment reduces utilization by 0.13 visits per member per month, a decline of nearly 20 percent. In their preferred estimate, the authors find that a 10 percent increase in price is associated with a 14 percent decline in utilization, a far greater effect than that found in the RAND study. However, the authors caution that the comparison may be misleading, since their study is based on copayments as opposed to coinsurance rates, and patients may be unaware of how one translates into the other.

The authors also examine the price sensitivity of prescription drug use. They find that a modest ($7 to $8) increase in the average drug copayment reduces drug utilization by 20% for HMO patients and by 6% for PPO patients. The stronger response among HMO patients may be due to the fact that their copayment increase affected both generic and branded drugs, while the increase for PPO patients affected branded drugs only.

Next, the authors ask whether the reduction in prescription drug utilization is due primarily to reduced use of drugs that affect quality of life, such as drugs to control allergies or acne, or whether there is also reduced use of drugs that control acute life-threatening conditions and chronic conditions. They find that the demand for all types of drugs is quite sensitive to price.

This finding highlights one possible pathway for the offset effect - when patients decrease their use of life-saving drugs in response to increased cost sharing, they may become sicker and need additional hospital care. Examining the effect of the policy change on hospitalizations, the authors find that higher office visit and drug copayments for HMO patients lead to a 6 percent increase in their probability of spending any days in the hospital. There is no significant effect for PPO patients.

Thus there is evidence of a modest offset effect, but how far do the cost increases from the additional hospitalizations go towards offsetting the cost savings from fewer office visits and prescriptions? The authors estimate that on average they offset 20 percent of the savings. However, the results are not uniform across patients. Patients with chronic illnesses have a very large offset effect, while the effect is near zero for those without them. This suggests that there are more severe health consequences from dissuading sicker patients from consuming office visits and prescription drugs.

It is important to note that CalPERS and Medicare are affected quite differently by the increase in cost sharing. Since the majority of CalPERS's expenditures are on office visits and prescription drugs and the majority of Medicare's expenditures are on hospital services, the savings from increased cost sharing accrue primarily to CalPERS, while the costs of increased hospitalizations accrue primarily to Medicare.

The study's results have several interesting policy implications. First, they suggest that the "doughnut hole" in the new Medicare prescription drug benefit could increase Medicare's costs - as chronically ill patients are faced with paying the full cost of their medications, some will reduce drug use and end up being hospitalized down the road. More generally, the results suggest that in the optimal health insurance plan, cost sharing would be tied to health status, with chronically ill patients facing lower cost sharing.

Gruber acknowledges financial support from the Kaiser Family Foundation and the National Institute on Aging. Chandra acknowledges financial support from the National Institute on Aging, the National Bureau of Economic Research, and the Nelson Rockefeller Center at Dartmouth College.