Jonathan Hall and Dr. Mark Showalter, Economics
The presence and severity of adverse selection in the market for health insurance has profound implications for public policy. Adverse selection occurs when individuals who are higher risk than observationally equivalent individuals purchase more insurance. Adverse selection can be used to argue in favor of nationalized health care, and at the same time suggests it that regulation simply makes problems worse. Due to the importance of this question, researchers from a wide variety of fields—public health, economics, insurance, law, human resources and medicine, to name a few—have contributed to the literature on adverse selection. Unfortunately, most researchers seldom cross disciplinary boundaries, making it difficult to fully evaluate the evidence for or against adverse selection. Differences across fields in terminology and methodology further hinder the sharing of ideas and results. My paper aims to survey this diverse literature by explaining the differences in terminology, outlining various approaches for testing, and taking a closer look at a selection of papers from a handful of disciplines.
Almost any discussion of health care policy is incomplete without considering adverse selection. In a discussion of public policy regarding HMO’s, Pauly and Nicholson point out that “the source of this difficult problem, as with so many other problems in health insurance, is adverse selection.” (1999, p. 921). Additionally, George Priest asserts that “adverse selection is a problem central to every insurance context, and it dominates the insurance function” (1987, p. 1541). There are a variety of areas where adverse selection has a profound influence on policy.
The most extreme policy implication that comes from adverse selection is that the United States should have a single-payer national health insurance program. The basic argument is that adverse selection prevents the market from operating efficiently. As a result of adverse selection, people are uninsured/underinsured, and the government needs to step in and provide insurance centrally, thus overcoming the problem of adverse selection.
However, this is not the only implication of adverse selection. It is used to justify mandated mental health provisions, since adverse selection is so severe in that market that it doesn’t even exist (Siegelman, 2004, p. 1230). In an effort to mitigate the severity of adverse selection, some suggest that employee insurance plans should limit choices and not offer a wide variety of insurance plans to choose from (Ginsburg, 1985, p. 279; Frech, 1996, p. 134). Furthermore, adverse selection concerns lead to opposition to regulation that would limit insurers’ ability to use certain information in setting rates. Markets have collapsed for pollution coverage (Abraham, 1988, p. 953) as well as directors’ and officers’ liability insurance (Romano, 1989) due to adverse selection problems that resulted from “expansive judicial interpretations of insurance contracts” (Siegelman, 2004, p. 1233). As Siegelman points out, “there is no necessary ideological or political bias to adverse selection arguments: They can be used to justify intervention or to condemn it; they can support state programs that amount to progressive taxation or undermine the goals of civil rights legislation” (2004, p. 1231). (See Siegelman, 2004, for a discussion of how adverse selection has been used in court cases covering topics such as annuities, civil rights, regulation and antitrust.)
The importance of health to an individual’s welfare means that understanding how adverse selection affects the market for health insurance is even more important. As a result of this, researchers from a plethora of fields have approached this question; however, they tend to be unaware of each other.
Adverse selection in health care is a serious issue, with profound policy implications. This has led to a wide number of researchers to look into this issue; however, most are unaware of the research done outside their discipline. As a result, many believe there has not been much work done with this problem. My paper hopes to fill an important hole by helping researchers and policy makers become more aware of the other work done on this issue and its implications.
I find that while the evidence is mixed, it leans towards concluding that there is no essential adverse selection. Those papers that explicitly focus on essential adverse selection tend to not find evidence of it, but a significant number of other papers do find adverse selection. One thing that is certain is that when insurers are not allowed to use basic demographic information adverse selection becomes a problem.
Perhaps all that is certain is that more research needed. Hopefully my paper can provide insight into how others have approached the topic, methods of testing, and sources of data. Those working in this area should be clear what they mean by adverse selection. This survey has highlighted several data sets well-suited to this task and has explained ways to test for adverse selection.