In The Aggregate Effects of Health Insurance: Evidence from the Introduction of Medicare (NBER Working Paper No. 11619), NBER researcher Amy Finkelstein challenges the belief that the spread of health insurance played only a small role in contributing to the dramatic rise in health care spending over the last half century. In a related study prepared with colleague Robin McKnight (NBER Working Paper No. 11609), Finkelstein asks: What Did Medicare Do (And Was It Worth It)?

At an annual cost of $260 billion, Medicare is one of the largest health insurance programs in the world. Providing nearly universal health insurance to the elderly as well as many disabled, Medicare accounts for about 17 percent of U.S. health expenditures, one-eighth of the federal budget, and 2 percent of gross domestic production. Medicare's introduction in 1965 was, and remains to date, the single largest change in health insurance coverage in U.S. history.

Finkelstein estimates that the introduction of Medicare was associated with a 23 percent increase in total hospital expenditures (for all ages) between 1965 and 1970, with even larger effects if her analysis is extended through 1975. Extrapolating from these estimates, Finkelstein speculates that the overall spread of health insurance between 1950 and 1990 may be able to explain at least 40 percent of that period's dramatic rise in real per capita health spending.

This conclusion differs markedly from the conventional thinking among economists that the spread of health insurance can explain only a small portion of the rise in health spending. This belief is based on the results of the

Rand Health Insurance Experiment (HIE), one of the largest randomized, individual-level social experiments ever conducted in the United States. The HIE compared the spending of individuals randomly assigned to different health insurance plans. Based on these comparisons, the estimated impact of health insurance on hospital spending was at least five times smaller than Finkelstein's estimates of the impact of Medicare on hospital spending.

Finkelstein suggests that the reason for the apparent discrepancy is that market-wide changes in health insurance - such as the introduction of Medicare - may alter the nature and practice of medical care in ways that experiments affecting the health insurance of isolated individuals will not. As a result, the impact on health spending of market-wide changes in health insurance may be disproportionately larger than what the estimates from individuals' changes in health insurance would suggest. For example, unlike an isolated individual's change in health insurance, market wide changes in health insurance may increase market demand for health care enough to make it worthwhile for hospitals to incur the fixed cost of adopting a new technology. Consistent with this, Finkelstein presents suggestive evidence that the introduction of Medicare was associated with faster adoption of then-new cardiac technologies.

Such evidence of the considerable impact of Medicare on the health care sector naturally raises the question of what benefits Medicare produced for health care consumers. Finkelstein and McKnight investigate this question, noting. two potential benefits that public health insurance might provide to the elderly:: better health and risk- reduction.

The period after Medicare's introduction, for example, was one of declining elderly mortality. However, using several different empirical strategies, the authors estimate that the introduction of Medicare had no discernible impact on elderly mortality in its first ten years in operation. They present evidence suggesting instead that, prior to Medicare, elderly individuals with life- threatening, treatable health conditions (such as pneumonia) sought care even if they lacked insurance, as long as they had legal access to hospitals.

Even absent measurable health benefits, Medicare's introduction of Medicare may still may have benefited the elderly by reducing their risk of large out-of-pocket medical expenditures. The authors document that prior to the introduction of Medicare, the elderly faced a risk of very large out- of- pocket medical expenditures. Tthe introduction of Medicare was associated with a substantial (about 40 percent) reduction in out-of-pocket spending for those who had been in the top quarter of the out- of- pocket spending distribution, the authors estimate.

Finkelstein and McKnight conduct a cost-benefit analysis comparing the insurance value of the reduction in the risk of large out- of- pocket medical expenditures provided by Medicare with the costs of the program. They estimate that even in the apparent absence of health benefits, the insurance value of Medicare alone is enough to cover between 45 percent and 75 percent of the its costs. In addition, the authors caution that Medicare may well have had health benefits that their analysis cannot detect, such as improvements in health status, even without mortality improvements. Moreover, given the evidence that the introduction of Medicare was associated with more rapid adoption of new cardiac technologies, in the long run Medicare's impact on elderly mortality may be much larger than the ten-year impact they examine.

V. Conclusion

By studying the introduction of Medicare, this paper has examined the impact of market-wide changes in health insurance on the health care sector. My central estimate is that Medicare is associated with a 37 percent increase in real hospital expenditures (for all ages) between 1965 and 1970. This estimate is over six times larger than what evidence from the impact of an individual’s health insurance on health spending would suggest. About half of the impact of Medicare on spending appears due to the induced entry of new hospitals, while the rest is due to growth in existing hospitals. This induced hospital entry helps explain the disproportionately larger impact on health spending of market-wide changes in
health insurance relative to individual-level changes. The paper also presents suggestive evidence that market-wide changes in health insurance may fundamentally alter the character of medical care both for individuals who experience a change in insurance coverage, and for those who do not as well.

A back of the envelope calculation that extrapolates from the estimated impact of Medicare to the impact of the spread of health insurance more generally suggests that the spread of health insurance between 1950 and 1990 may be able to explain about half of the six-fold rise in real per capita health spending over this time period. This raises the natural question of whether a similar mechanism can
explain why most other OECD countries have also experienced sustained growth in the health care sector over the last half-century [OECD 2004]. Interestingly, like the United States, many of these countries also established their national health insurance systems in the 1960s and 1970s [Cutler 2002]. An important question for further work is whether other health insurance systems had a similar impact on health
spending, or whether idiosyncratic features of the Medicare system resulted in a uniquely high impact. In addition, if Medicare’s impact on the practice of medicine in the United States influenced treatment practices or coverage decisions in other countries’ national health care systems, it is also possible that the effect of Medicare on health spending may substantially exceed its impact within the United States. This is also an interesting avenue to explore in future research.