NBER Reporter: Research Summary Summer 2006
Since the early 1990s, prominent proposals for health insurance reform have focused on increasing consumer choice and competition among integrated health plans. Under "managed competition" models, consumers choose from a menu of health plans on the basis of price and quality. Proponents of these market-oriented plans argue that, in such a system, consumers will sort themselves into lower cost, higher quality plans; this pressure by consumers will provide strong incentives to health plans and their affiliated providers to control costs and increase quality in order to compete for enrollment. The Clinton administration's Health Security Act and the "premium support" proposals for reforming Medicare are variants of the managed competition approach. Although a "managed competition" model has yet to be adopted as national policy, many large employers organize their health benefits programs according to the same basic principles. Research on the behavior of employees and retirees in these employer programs provi des a useful laboratory for the role of price and quality in consumer health insurance decisions.
One distinct problem for market-oriented solutions to health insurance is that, when consumers are offered a choice of health insurance options, the healthy (less risky) consumers may sort themselves into certain plans and the more risky consumers into others. Consequently, some plans will attract a disproportionate share of less costly low risk consumers, while others will attractive older, sicker consumers who are more costly to insure. This "risk selection," in turn, is influenced in part by the rules concerning how insurers are allowed to vary premiums according to subscriber characteristics. State reforms that tightened these rules provide good case studies for understanding the relationship between pricing and risk selection.
The Effect of Premiums on Consumer Health Plan Choices
Two notable experiments in "managed competition" took place in the mid-1990s: the University of California (UC) and Harvard University both offered a menu of plans that varied in generosity, but adopted a "fixed dollar contribution" policy. The plans also varied significantly in cost, so employees had a greater incentive to consider price when selecting a health plan. Because out-of-pocket premiums increased for some employees but not for others, these changes provide a natural experiment for estimating the impact of price on employee health insurance decisions. Studies that I have conducted with colleagues at the University of California, Irvine,(1) and by David M. Cutler and Sarah J. Reber,(2) analyze the effect of these policy changes on employee plan choices, total spending, and risk selection.
The results for UC and Harvard are strikingly similar. In both cases, employees were quite sensitive to price, and were willing to switch plans to save as little as $5 per month in out-of-pocket premiums. Cutler and Reber estimate a short-run premium elasticity of -2. In addition to this demand response, participating insurers lowered their premiums in order to compete for enrollment. At Harvard, the combined effect of employees shifting to lower cost plans and the premium reductions was a 10 percent reduction in total spending in one year. Over a three-year period, total spending in the UC program fell by over 25 percent. This was at a time when increased competition among managed care health plans was causing premiums to decline throughout the country, so these savings cannot be attributed entirely to the adoption of a fixed dollar contribution policy. However, the reduction in premiums charged to Harvard and UC were larger than those observed in the general market, suggesting that the pricing reforms en acted by each university did result in a one-time savings.
In both cases, however, this also came at the expense of the number of choices employees had. The most generous indemnity insurance - which covered care from the doctor of your choice - was subject to an "adverse selection death spiral." Faced with an initial increase in price for this coverage, the healthiest dropped out of indemnity insurance into lower cost plans. Those who remained in the plan were, therefore, sicker on average. To cover their costs, the price of the coverage was raised, which led to more dropouts until, after a few years, no one was covered by the indemnity plan.
One possibly important impediment to market based solutions is that consumers often face substantial "switching costs" when they try to change their health insurance plan. Under managed care, switching insurers often means having to change providers, and even when that is not necessary, individuals may be reluctant to switch plans for fear of suffering an interruption of treatment. "Status quo bias" in decisionmaking is another potential source of persistence. Combining the UC enrollment files with hospital discharge and cancer registry data, Bruce Strombom, Paul Feldstein, and I estimate separate premium elasticities for groups of employees whom we hypothesize to face different switching costs. Specifically, we define 18 distinct groups based on age, job tenure, and health risk, where "high risk" individuals were defined as those who had recently been hospitalized or diagnosed with cancer. Consistent with the switching cost hypothesis, we find that young, low-risk employees who had recently joined the uni versity were the most price-sensitive; older, high-risk employees with long job tenure were the least price-sensitive.
The fact that younger, healthier consumers are more willing to switch health plans in response to a change in prices contributes to adverse selection against plans that are favored by higher-risk consumers. Adverse selection reduces efficiency by distorting the prices of competing plans and, in the extreme, driving certain options from the market. In both the UC and Harvard examples, this dynamic caused the most expensive plan available at each university to experience an "adverse selection death spiral" and to be priced out of the market. Cutler and Reber estimate that after two years, the efficiency loss from adverse selection in the Harvard program was roughly 2 percent of premiums.
These and other similar studies(3) tell a consistent story about the price sensitivity of active employees. Proponents of managed competition Medicare reform proposals point to these results and the experiences of other large employers to explain how such reforms would work. However, the extent to which these results generalize to elderly adults in the Medicare program is unclear. In particular, the switching costs that we show to be important for active employees are likely to be even larger for elderly Medicare beneficiaries. In a recent paper, I use administrative data from a different employer-sponsored health benefits program to estimate premium elasticities for Medicare-eligible retirees.(4) This employer's contribution to retiree coverage depends on when a person retired and her years of service at that date. Therefore, two otherwise similar individuals who retired at different points in time, or at the same time with different years of s ervice, face very different out-of-pocket premiums for the same menu of health insurance options. Because this price variation is uncorrelated with the features of those options, or other retiree characteristics, it can be used to obtain unbiased estimates of elasticity. The results indicate a negative and statistically significant effect of price on the choice of a health plan, albeit one that is slightly smaller than the results from the literature on active employees. While this is consistent with the hypothesis that older consumers are less price-sensitive than younger ones, the price effects are large enough to suggest that if Medicare went from a system of administered pricing to competitive bidding, health plans would face strong incentives to compete on price.
From a policy perspective, the "near elderly" adults -- that is, those between the ages of 55 and 64 -- constitute another important population. Because many firms have cut back on retiree health benefits, early retirees in this age group are especially at risk of being uninsured. Some recent policy proposals, such as allowing individuals under age 65 to buy into Medicare, would address this problem directly. Other proposals, such as tax credits for the purchase of non-group insurance, do not explicitly target the "near elderly" but would be especially relevant for this group. A key parameter for evaluating the cost of such proposals is the price elasticity of take-up.
Sabina Ohri and I(5) estimate the effect of out-of-pocket premiums on the decision by early retirees between the ages of 55 and 64 to take up insurance coverage offered by their former employer. The data are from the same employer-sponsored program I used to model the health plan choices of Medicare-eligible retirees. We find a statistically significant, but small effect of price. The range of our elasticity estimates, from -0.10 to -0.16, is consistent with other studies that use different types of data and different research designs.(6)
Does Limiting Insurers' Discretion Help Consumers in Insurance Markets?
One factor contributing to adverse selection in the UC and Harvard cases is that, in each system, premium contributions faced by employees and premium payments to plans were "community rated" -- that is, they did not vary with the risk characteristics of those being insured. As discussed earlier, one result is thus that the most generous plan faced an adverse selection death spiral.
The relationship between the way premiums are rated and risk selection is a major issue in the regulation of private insurance markets, particularly the small group market (typically defined as employer-sponsored groups of 50 or fewer employees) and the individual (or non-group) market. In the early 1990s, nearly every state enacted reforms that targeted insurers' underwriting practices that were seen as discriminating against high-risk groups. The most extreme type of regulation, "pure community rating," mandates that the same premium must be charged for a given plan to all subscribers, regardless of age, gender, or any other risk characteristic. A main goal of these new regulations was to increase coverage, although critics argued that by raising premiums for low risk groups the laws may have reduced coverage, inducing low-risk consumers to drop it.
Although the Harvard and UC experiences suggest that these "adverse selection death spirals" could be important, the results don't tell us how adverse selection affects the overall level of insurance coverage in a market because the universities are closed systems. In both cases, employees facing higher premiums simply switched to other, less expensive plans. John DiNardo and I test for an effect of underwriting regulations on insurance coverage, focusing on reforms enacted by New York in 1993.(7) New York's reforms, which mandated pure community rating in both the small group and individual insurance markets, were the strongest, and hence most controversial, in the country. We compare trends in New York to those in two neighboring states: Pennsylvania, which was one of a handful of states that enacted no reform, and Connecticut, which enacted moderate reforms. We show that, while insurance coverage did fall in New York after the reforms took effect, coverage also was falling in Pennsylvania and Connecticut. One important prediction of the death spiral hypothesis is that coverage should have fallen most among younger consumers for whom the reforms caused the greatest premium increases. This should cause the average age of the insurance pool to increase. While we find some evidence of such changes in New York, the trends for the two comparison states are nearly identical, suggesting that they are driven by factors other than the reforms.
The fact that New York's community rating law did not immediately reduce the number of people with insurance does not mean it had no effect. It may have influenced the types of plans purchased by consumers. In fact, consistent with the simplest Rothschild-Stiglitz model and with the results from the UC and Harvard cases, we find that New York's reforms led to a shift in enrollment away from traditional indemnity plans to health maintenance organizations (HMOs). In a follow-up study, Su Liu and I use national data to test whether this result from New York generalizes to other states.(8) We find that HMO penetration among small employer-sponsored groups increased more in states that enacted relatively strong small group reforms than in states without such reforms.
Recent Developments and Future Research
Recent developments in public programs and private health insurance markets have resulted in significant changes in the options available to consumers and suggest fruitful areas for future research. January 2006 marked the introduction of Medicare Part D, which provides prescription drug coverage. Beneficiaries can obtain this coverage through the same HMOs that were already available in the program or through new stand-alone drug plans offered by private insurers. The resulting menu of options is quite different from what had been envisioned by most managed competition advocates, most notably in the large--some would say bewildering--set of options. Research on the choices made in this new environment is important not only for evaluating the prescription drug benefit, but also as it may inform policymakers concerning future reforms.
The legislation that created Medicare Part D also established Health Savings Accounts (HSAs), tax-free savings accounts that, when used in conjunction with a high deductible insurance plan, can be used to fund medical expenses. In the past, consumers have shown little enthusiasm for high deductible policies, although certain attractive features of HSAs combined with steady increases in premiums for other types of health insurance may change this. Some worry that these plans will be most attractive to low risk consumers who do not anticipate a great need for medical care, thus causing more comprehensive plans to experience adverse selection. If this occurs, it may be HMOs, which benefited in the 1990s from risk-based sorting in the small group market and within employer-sponsored programs like those of Harvard and the UC, that are adversely affected. How the introduction of HSAs affects consumer health insurance decisions and what these new products mean for the stability of insurance markets are interestin g areas for future research.
* Buchmueller is a Research Associate in the NBER's Programs in Health Care and Health Economics and a Professor of Economic and Public Policy at the Paul Merage School of Business, University of California, Irvine. His profile appears later in this issue.
1. T.C. Buchmueller and P.J. Feldstein, "The Effect of Price on Switching Among Health Plans," Journal of Health Economics, 16(2) (1997), pp. 231-47; T.C. Buchmueller, "Does a Fixed-Dollar Premium Contribution Lower Spending?" Health Affairs, 17(6) (1998), pp. 228-35; B.A. Strombom, T.C. Buchmueller, and P.J. Feldstein, "Switching Costs, Price Sensitivity and Health Plan Choice," Journal of Health Economics, 21(1) (2002), pp. 89-116.
2. D.M. Cutler and S. J. Reber, "Paying for Health Insurance: The Tradeoff Between Competition and Adverse Selection," NBER Working Paper No. 5796, October 1996, and Quarterly Journal of Economics, 113(2) (1998), pp. 433-66.
3. A.B. Royalty and N. Solomon, "Health Plan Choice: Price Elasticities in a Managed Competition Setting," Journal of Human Resources, 34(1) (1999), pp. 1-41.
6. M. Chernew, K. Frick, and C. McLaughlin, "The Demand for Health Insurance Coverage by Low-Income Workers: Can Reduced Premiums Achieve Full Coverage?" Health Services Research, 32(4) (1997), pp. 453-70; L. Blumberg, L. Nichols, and J. Banthin, "Worker Decisions to Purchase Health Insurance," International Journal of Health Care Finance and Economics, 1(3-4) (1997), pp. 305-25.J. Gruber and E. Washington, "Subsidies to Employee Health Insurance Premiums and the Health Insurance Market," NBER Working Paper No. 9567, March 2003, and Journal of Health Economics, 24(2) (2005), pp. 253-76.
7. T. Buchmueller and J. DiNardo, "Did Community Rating Induce an Adverse Selection Death Spiral? Evidence from New York, Pennsylvania and Connecticut," NBER Working Paper No. 6782, January 1999, and American Economic Review, 92(1) (2002), pp. 280-294.