THE THEORY AND PRACTICE OF INSURANCE: WHY THE DISJUNCTION?
Economists and insurance experts have studied the industry for many years and have developed a series of theoretical concepts to explain how insurance markets function. The prevailing view of the demand for insurance was summed up by one conference participant who noted that an economically rational consumer would understand that apart from certain tax benefits, when you buy insurance, you’re making a bet with an insurance company, which the insurance company wins on average because it must cover administrative costs and earn a competitive return for its shareholders. From this perspective it makes sense to insure only against potential losses so large as to affect your standard of living. But that is not always how consumers behave.
For example, one of the fastest-growing insurance markets is repair or replacement insurance for relatively inexpensive electronic products that seldom fail. Usually the insurance is priced as if the risk of failure were much higher than it is. Nonetheless, for some products, nearly 80 percent of consumers opt for insurance even though the loss would have no appreciable effect on their standard of living.
Similarly, the greatest risk facing a 70-year-old widow with three children, all wealthier than she, is loss of income. Thus she should buy annuities rather than life insurance. But if the widow behaves like her peers, she is seven times more likely to buy life insurance than annuities.
Just as the demand for insurance seems to diverge from the theoretical, rational paradigm, the supply of insurance also seems to depart from the standard model. For example, in most markets, an increase in demand raises the price and increases the supply. But when the perceived risk of a terrorist incident went from negligible to plausible in the wake of the September 11 attacks, demand for terrorist coverage increased, but the voluntary supply of such insurance virtually disappeared.
Professors David Cutler and Richard Zeckhauser of Harvard University use these conundrums to argue, in a comprehensive and thought-provoking paper, that there is a systematic tendency for practice to diverge from theory in insurance markets. One consequence is that insurance is often purchased when theory suggests it should not be (life insurance among the elderly), while many substantial risks that should be insured (terrorism) are not. A second consequence is that there are significant mismatches between entities that should bear risk (financial markets) and those who do (governments and private and mutual insurance companies).
The authors explore several reasons for the discrepancy between theory and practice. On the demand side, they suggest that the standard model of utility maximization needs to be extended to take account of some findings of behavioral economic research. For example, individuals appear to be averse to loss of any kind and will pay far more than the expected loss to protect against even small losses. Similarly, individuals often greatly overestimate the negative experience from a loss.
On the supply side, the authors argue that insurers are often risk averse and feel unable to diversify their risks, as in the case of terrorism (where events may be immensely costly and highly correlated) and long-term health care insurance (where it is not possible to diversify the risks cross-sectionally). The solution to this supply problem, they argue, is to encourage risk-spreading beyond the “narrow confines of primary insurance and reinsurance” to include financial instruments that tap the enormous pool of insurance dollars in global financial markets.
THE CRISIS IN MEDICAL MALPRACTICE
In several states around the nation, medical malpractice insurance has become either prohibitively expensive or totally unavailable. This experience follows a relatively quiescent period in the market, when such insurance was broadly available, but seems to echo an earlier crisis in malpractice insurance in the 1970s. At that time many states changed their tort laws, making it more difficult to sue physicians or capping awards for noneconomic damages suffered by patients, or both. Insurers also took steps to limit their own risks, most prominently by replacing the “occurrence” policy (which covered malpractice as long as it occurred during the policy period) with a “claims-made” policy (covering only those claims filed during the policy period). Moreover, doctors began to self-insure by forming their own mutual companies and “risk retention groups” to cover their malpractice liabilities. And some states adopted measures to assure the availability of insurance and reduce its cost to physicians by introducing both joint underwriting associations for physicians unable to buy coverage in the voluntary market and patient compensation funds that limit the defendant physician’s liability at some threshold, but provide additional compensation to the patient up to a higher threshold.
In view of these reforms, the reemergence of the malpractice insurance crisis is somewhat surprising. But the fact that the crisis is hitting some states far harder than others—general surgeons’ malpractice insurance rates increased 75 percent in Dade County, Florida, but only 2 percent in Minnesota—suggests that it may be useful to evaluate the earlier reforms, which also differed widely from one state to another. In the second paper presented at the conference, three experts in medical malpractice insurance from the Wharton School—Patricia Danzon, Andrew Epstein, and Scott Johnson— wrestle with these issues, focusing primarily on the reasons for the cross-state variations.
The authors reach several conclusions, which they take pains to characterize as “tentative” given the paucity of appropriate data. They report that while “shocks to insurer capital”—or high payouts—have not contributed to premium increases, they have induced some insurers to exit the market. The authors also find evidence that malpractice insurers did not set aside sufficient reserves against losses in the late 1990s and that subsequent loss forecast revisions have been positively associated with premium increases. Consistent with general evidence of underwriting cycles (discussed by Scott Harrington in the paper that follows), it appears that “excessive competition” during the “soft market” of the 1990s contributed to the “hard market” of recent years, with smaller insurers, especially recent entrants, leaving the market.
The authors also explore the effect of various tort reforms on the malpractice insurance market. Based on their statistical analysis, they conclude that states that capped noneconomic damages and limited joint and several liability had much lower premium increases than states that did not enact these reforms. In contrast, joint underwriting associations and patient compensation funds have reduced neither premium increases nor the probabilities that insurers will leave the market. In fact, they may have increased costs. Two states with joint underwriting associations—Pennsylvania and South Carolina—had among the largest cumulative premium increases over the period.
In a dinner speech to conference participants, Jay Fishman, the chairman and chief executive officer of The St. Paul Companies, the largest company to withdraw from the medical malpractice insurance market, provided an important perspective on the authors’ somewhat puzzling finding that shocks to insurer capital (higher payouts) led to exits from the market rather than to higher premiums. He explained that St. Paul decided to exit the malpractice market after concluding it could break even only if it raised premiums 30 percent annually for three years consecutively. In St. Paul’s view, such large premium increases would lead to a “death spiral” of adverse selection, in which only the highest-risk physicians would purchase the insurance, thus raising future losses and requiring still higher future premiums. Fishman also cautioned that caps on pain and suffering awards were not the appropriate solution to the malpractice problem. In his view, what is needed is a more realistic definition of malpractice, which acknowledges that even when physicians take all reasonable precautions, the innovative and experimental treatments that modern medical science makes possible and that patients demand do not always produce the desired result.
TORT LIABILITY, INSURANCE RATES, AND THE INSURANCE CYCLE
Medical malpractice is only a subset of larger concerns about the tort system generally and its relation to insurance rates and availability. In the next paper, a leading analyst of liability insurance markets, Scott Harrington of the University of South Carolina, reexamines these controversial issues.
It is common for analysts of insurance and participants in the market to characterize the business as going through a regular “underwriting cycle”—fluctuations between “soft” periods, when premiums are stable or falling and coverage is readily available, and “hard” periods, when premiums soar and coverage is less available. Harrington notes that the most infamous example of a “hard” market was the liability insurance crisis of the mid-1980s. More recently, the market for commercial property and casualty insurance has hardened in the wake of 9/11, while markets for general liability and medical malpractice have also grown harder.
Harrington explores various theories advanced by analysts to explain the insurance cycle. Clearly, much of the variation in rates and availability reflects changes in the discounted value of expected claims (the so-called “perfect markets” hypothesis). But capacity constraints on the supply of capital available to insurers may also play a role. Mixed evidence supports this explanation in the case of hard markets. But evidence on the role of capacity constraints in soft markets is much weaker.
Nonetheless, taking all factors into account, Harrington notes that there is good reason for believing that as the tort liability system expands—both in types of situations in which damages are awarded and in the amounts of damage themselves—insurance rates will increase in the long run. Yet he concludes that the case for tort reform does not rest solely on the impact of the tort system on insurance markets. The fundamental issue is whether the current system efficiently deters undesirable behavior.
INSURING AGAINST TERRORISM: THE POLICY CHALLENGE
No single event during the past several years has focused public attention on the importance of insurance as intensely as the September 11 terrorist attacks. That disaster led virtually all insurers to withdraw or seek to withdraw coverage for terrorist-related events, actions which most states approved. At the same time, insurers and real estate developers, among others, urged the federal government to provide reinsurance to the primary insurers so that they could again offer terrorism coverage without the fear that another large-scale event could wipe them out. Eventually, Congress responded and President Bush signed into law the Terrorism Reinsurance Act of 2002 (TRIA), under which insurers would pay premiums to the government “ex post,” or after another such event. Under TRIA, the federal government agreed to pay for 90 percent of an insurer’s losses after the insurer first absorbed losses up to 7 percent of its earned premiums (the so-called “retention” level). With TRIA set to expire next year amid calls to extend it, this seemed an appropriate time to review the act. Kent Smetters, who worked on the legislation at the Treasury Department while on leave from The Wharton School, is particularly well-qualified to undertake this review.
Smetters is highly skeptical of the need for government-provided reinsurance for terrorism. Indeed, he argues that other government policies have contributed to the unavailability of terrorism insurance after 9/11 and to the inadequacy of insurance coverage for other sorts of catastrophes.
Government tax policies, for example, prevent insurers from setting aside tax-deductible reserves for catastrophes. And government regulatory and accounting policies make it harder for insurance companies to securitize large risks. Smetters believes that in the absence of these restrictions, investment banks and insurers would develop capital market instruments that would diversify the risk of terrorism and other catastrophic losses. He notes that losses ten times larger than the $40 billion loss on September 11 are not uncommon in world capital markets. Indeed, U.S. capital markets alone routinely gain or lose $100 billion a day and often several trillion dollars a month.
Smetters argues that reforming these policies would eliminate the need for government-provided terrorism reinsurance. He believes that unfettered insurance and capital markets could absorb large, even catastrophic, losses without undue stress.
Some of the most common arguments in favor of direct government intervention into the terrorism insurance market include the difficulty of forecasting future losses, the magnitude of potential terrorist losses, asymmetric information between the government and private sector, and the fact that many people will rationally forgo insurance because they believe the government will bail them out after a major loss. Smetters reviews these arguments and finds most deficient because they fail to explain why the private market solution is inefficient. He does concede, however, that mandatory coverage might be justifiable for certain risks that are difficult to diversify in capital markets and that are borne by groups such as homeowners and farmers whom the government is likely to bail out after a big loss.
BROKERS AND THE INSURANCE OF “NON-VERIFIABLE LOSSES”
Although insurance contracts are tightly worded documents, losses are often followed by disputes over whether coverage applies and, if so, to what extent. Indeed, according to Professors Neil Doherty and Alexander Muermann of the Wharton School, insurers today seem more prone than ever to contesting large claims. Doherty and Muermann call such contested situations “non-verifiable losses” and seek to explain how insurance brokers play a role in resolving—or preventing—the disputes.
Insurers often try to prevent disputes. Because they make an investment in acquiring information about their larger customers, and do not want to lose them, insurers have incentives to make reasonable offers to pay non-verifiable losses and thus avoid disagreements. Some insurers might also assist their customers in controlling losses by providing engineering or risk-reduction services.
Doherty and Muermann point out, however, that because the insurance brokerage market is highly concentrated—three brokers dominate the market—insurers who derive business from them have strong incentives to avoid the reputation of being unwilling to pay claims. Otherwise, the insurers may lose not only future referrals but some of their existing book of business as well. In addition, insurers known as unwilling payers may have to accede to broker demands for more compensation to steer business in their direction.
In short, the highly concentrated nature of the brokerage industry tends to work in favor of insureds if and when they experience non-verifiable losses. This may be the rare exception where a highly concentrated market structure works in favor of customers rather than producers.
CONSOLIDATION IN THE EUROPEAN INSURANCE INDUSTRY
The development of a single European market has led to much consolidation among firms, both within European countries and across them, over the past decade. The insurance business has been no exception to this pattern. In the final paper presented at the conference, David Cummins of the Wharton School and Mary Weiss of Temple University examine whether these European mergers create value for shareholders.
In similar studies of mergers in the U.S. market, principally among banks, analysts have found a predictable pattern: while the stock prices of acquired companies typically go up after a merger proposal is made, the prices of stocks of acquiring companies generally remain unchanged or even drop. Does this pattern hold up for European insurance mergers?
Cummins and Weiss’s analysis of stock prices before and after the European mergers generally confirms this pattern. Although insurance mergers across borders did not change shareholder value for acquirers one way or another, mergers within the same country produced significant shareholder shareholder losses on average. In contrast, target companies saw their shareholder value increase in both cross-border and within-country mergers, but more so for transactions where the two companies did business in the same country. In sum, cross-border mergers clearly appear to generate net gains to shareholders (acquirers and targets combined), but within-country mergers seem more likely to destroy than create shareholder value.
These results are relevant to policymakers, especially those concerned with market structure. For example, where shareholder gains arise from entrenchment of market power rather than from efficiencies, antitrust intervention may be warranted. Because insurers in cross-border deals probably are not direct competitors to begin with, their combinations should pose fewer antitrust concerns than within-country mergers (especially where insurance markets may already be concentrated). Unlike the banking sector, where national bank regulators often discourage cross-border mergers, national insurance regulators seldom intervene to protect national champions. These findings are also useful to investors and managers of insurance companies as they consider which kinds of mergers are most likely to produce efficiencies and enhanced market opportunities and thus shareholder gains.
If any common lesson emerges from this conference, it is that insurance is too important to the fabric of the American economy—and indeed to our society generally—to be ignored. If individuals and businesses could not obtain insurance against various causes of financial misfortune, the American economy would be much less productive and consumption choices much more constrained. These papers represent the most sophisticated and best-informed economic analysis of the key policy issues concerning the insurance industry now available. Yet the authors have been careful to underline the limits of their analysis. Many aspects of insurance markets remain difficult to explain in terms of conventional economic analysis.
Economists must thus develop richer models to explain the behavior of insurance markets. But even in the absence of new breakthroughs in economic analysis, current research provides a strong rationale for policymakers to reexamine the tax and regulatory policies that impede the performance of insurance markets and the allocation between public and private responsibility for the efficient provision of insurance.