Wednesday, 31 March 2004
113 Yale L.J. 1223 (2004)
The phrase "adverse selection" was originally coined by insurers to describe the process by which insureds utilize private knowledge of their own riskiness when deciding to buy or forgo insurance. If A knows he will die tomorrow (but his insurer does not), life insurance that is priced to reflect the average risk of death in the population as a whole will look like a very good deal to him. Conversely, if B knows she will live for much longer than the average person with her observable characteristics (age, gender, medical condition), insurance that is priced to reflect the average risk of death will seem like a bad deal to her, and she will be unlikely to buy it. When A buys lots of insurance and B buys none, insurers find themselves charging an average rate to a population that contains only the worst risks, and end up losing money by virtue of having their product selected only by high-risk individuals.
But informational asymmetry may not just be bad for insurers. When insurers cannot distinguish between good and bad risks, theory predicts that it is possible (although not necessary) to end up with no coverage for anyone: As the good risks begin to exit, the average quality of those insureds remaining falls and prices rise in a vicious circle, ending in a so-called "death spiral" where no one is covered. Even when insurance is available, it may be inefficiently distorted by the presence of adverse selection. Many theoretical models conclude that when adverse selection is a problem, good risks will be rationed: They will be allowed to purchase only limited coverage in an attempt to make such coverage less attractive to the bad risks, who would otherwise be eager to purchase it given its favorable price.
As we will see, courts, policymakers, and legal academics routinely--and often uncritically--discuss adverse selection as a major issue in the design and regulation of insurance markets. In addition, economists have devoted scores of articles to the subject over the last decade. But the thesis of this Essay is that although theory demonstrates that adverse selection can occur, and some instances have certainly been documented, neither the theoretical models nor the empirical studies provide much support for its widespread importance in insurance markets. The nature of selection pressures turns out to be vastly more complicated than the rhetoric of courts and academic commentators would suggest. And while the economic theory of adverse selection in insurance markets has become enormously sophisticated, much of it is devoted to rarified analysis of the nature and existence of equilibria. It has thus managed to obscure some essential features of insurance demand that may undercut or even reverse the typical adverse selection results. In short, while adverse selection in insurance markets is clearly a possibility, it is often not the serious problem that it is taken to be. Courts, policymakers, and legal academics need to do much more than trumpet a concern for adverse selection as a justification for their preferred course of action. And economists need to develop less obscure and more realistic models, and pay more attention to the empirical issues (as indeed they are beginning to do).
This Essay is organized as follows. Part I describes the importance ascribed to adverse selection in insurance markets by courts, regulators, and legal commentators. The common theme of these actors' analyses is that adverse selection is an extremely significant problem--one that justifies deference to longstanding common law doctrines in tort and contracts and a hands-off attitude with regard to insurance regulation.
In Part II, I briefly explain the theory of adverse selection as developed in the economics literature, and discuss its implications for the behavior and efficiency of insurance markets. Economic models suggest that adverse selection can cause the outright collapse of insurance markets and will always produce rationing and various other forms of inefficiency. But while enormously sophisticated, these economic theories are, I suggest, ill-suited for the (often rather casual) reliance that is placed on them by courts and commentators.
Part III considers the assumptions and predictions of the adverse selection model and compares them with the existing empirical evidence. After some preliminary questions, I focus on three issues: First, can insureds actually outpredict their insurers, as adverse selection theory requires, and does this lead the worst risks to buy more insurance? Second, are adverse selection "death spirals" a serious real-world phenomenon? And third, are good risks typically rationed in the amount of insurance they can buy, as adverse selection theory predicts? I answer all three questions largely in the negative.
Part IV considers an alternative model of selection in insurance markets, in which it is the good risks who buy more insurance. The standard adverse selection models assume that insureds are homogenous except for differences in the probability of loss. In particular, everyone is assumed to be equally risk-averse, and there is therefore no relationship between an insured's risk aversion and her riskiness. Once the assumption of homogenous risk aversion is relaxed, however, alternative selection mechanisms become possible. I therefore discuss the theoretical and empirical support for a model of "propitious selection," in which low-risk individuals are willing to buy insurance even at "unfair" rates. I conclude that propitious selection is at least as plausible as the standard adverse selection story in many cases.