Liability Insurance Pricing

The goal of this article is to test whether the threat of regulating (or of more stringent regulation of) automobile liability insurance as portrayed in the popular and industry press induces insurers to change the way they price their policies. More to the point, using quarterly state data from 1984 to 1993, the author attempts to determine whether insurance companies reduced premium increases to avoid regulation, a test the article will call the Regulatory Threat Hypothesis. The results suggest that automobile liability insurance premiums increased at a slower pace (or decrease) in the presence of a regulatory threat. INTRODUCTION AND MOTIVATION It is well known that firms react to outside pressure. Many companies have public relations departments to deal with pressure groups and other outside forces that may affect profits. Insurance companies faced such outside pressure in the midto late-1980s during the so-called insurance liability crisis. This crisis affected all types of liability insurance, including personal automobile liability insurance.1 The 1980s Martin Boyer is assistant professor, Department of Finance, École des Hautes Études Commerciales, Université de Montréal, and fellow, CIRANO, both in Montréal, Canada. The author would like to thank Sharon Tennyson for letting him use some of her data. He also thanks Diana Lee at the National Association of Independent Insurers; Robert Gagné, André Blais, Caroline Boivin, Daniel Parent, and seminar participants at the University of Minnesota; Université Laval; Insurance Federation of Minnesota; American Risk and Insurance Association; and Société Canadienne de Sciences Économiques for discussions and comments. The input of two anonymous referees was also appreciated. This work benefited from the financial and technical support of the S.S. Huebner Foundation at the University of Pennsylvania. 1 Berger, Cummins, and Tennyson (1994) examine the case of the general liability insurance crisis and its effect on the greater general liability reinsurance market. For a detailed exposition of the liability crisis in personal automobile insurance, see Cummins and Tennyson (1992). 38 THE JOURNAL OF RISK AND INSURANCE was also a period of great political pressure on state regulators. Consumer groups throughout the United States petitioned state regulators to mandate insurance firms to reduce premiums, or at least the rate of increase. One consumer group collected so many signatures that California held a referendum in November 1988 on automobile insurance premiums regulation. The referendum was known as Proposition 103.2 The referendum basically asked whether insurance companies should be mandated to reduce automobile liability premiums and whether any premium increase should be approved by an elected insurance commissioner.3 The popular vote was almost evenly divided, but ultimately Proposition 103 passed with 51 percent of the vote. A main driver of the vote was the behavior of city dwellers (especially in Orange County) who saw an opportunity to extract money from suburban residents. Higher premiums are paid in cities, and city dwellers voted in favor of Proposition 103 since it asked for rates to be based on experience rather than geographic location. This behavior would follow the argument initiated by Peltzman (1976), who argued that different groups use their political clout to influence regulation. The vote rocked the stock market, as the value of insurance companies publicly traded plummeted. Fields et al. (1990) found that insurance companies doing business in California had an average cumulative abnormal return of –6.9 percent, which means that insurers’ stock prices under-performed the market by 6.9 percent. In addition, the more business a company had in California, the greater the negative cumulative abnormal return. What is even more surprising is that a firm’s proportion of business in states neighboring California also had a negative effect on the cumulative abnormal return. Moreover, the stock price of some firms with no operation in California also fell.4 One possible explanation for this phenomenon is that investors in firms operating in states neighboring California were afraid that insurance rates were going to be controlled there as well. In fact, this concern may have been well founded; according to a survey, 90 percent of Americans would be in favor of passing a law similar to California’s Proposition 103.5 If investors perceived threats of regulation in states other than California, then one has to wonder whether the insurance companies themselves perceived such regulatory threats. If the insurance industry acknowledges the possibility of regulation, then it seems natural to conclude that it will do something to reduce the probability of such regulation. The question is, What should the industry do? The insurance industry can react to the threat of regulation in at least two ways. 2 The case of California is special, not only because it is the largest market in the United States, but also because California is one of the few states where popular referendums are binding. 3 More to the point, Proposition 103 asked for the following: premium rates to be cut by a minimum 20 percent; all rates would have to be approved by the insurance commissioner; the commissioner would be elected by the public rather than appointed by the governor; and the rates were to be based on driving history rather than geographic location. 4 National Underwriter, Property/Casualty edition, October 7, 1989. 5 National Underwriter, Property/Casualty edition, June 26, 1989. MEDIA ATTENTION, INSURANCE REGULATION, AND LIABILITY INSURANCE PRICING 39 The first is to influence the regulator so that it becomes more conciliatory toward insurers.6 The second is to persuade the population through voluntary price reductions not to support state insurance commissioners’ threat of more stringent regulation. The former tactic is in line with the capture theory idea introduced by Stigler (1971) and Posner (1974), while the latter is the main subject of this article. Obviously, the industry can use both tactics at the same time. Think of a game in which insurance firms first set premiums (perhaps voluntarily reduce them in certain instances) and then compete with so-called consumer groups to capture the regulator. If premiums are set low, consumer groups would be less inclined to devote money and energy to capturing the regulator since they do not have much to gain. As a consequence, it becomes easier for the industry to capture the regulator since they are the only ones applying pressure. Once the regulator has been captured, insurers can then increase premiums to normal levels. The use of voluntary price restraints by producers has been shown by Erfle and McMillan (1990; see also Erfle, McMillan, and Grofman, 1989, and Glazer and McMillan, 1992). They showed that during the oil crisis of the Seventies, firms voluntarily reduced price increases of the most visible sort of oil to convince the federal government that no price ceilings were needed. In this article, the author applies a technique similar to that of Erfle and McMillan (1990) to test the regulatory threat hypothesis for the property and liability insurance industry. According to that hypothesis, insurance companies in markets where price regulation (or more stringent price regulation) is possible should voluntarily reduce premiums or premium increases to signal to the population and the regulator that premiums are not too high and that further regulation is not necessary. This article uses the automobile liability insurance industry because it is already a heavily regulated sector in many states. Furthermore, a large movement toward insurance rate suppressions occurred during the insurance liability crisis of the mid-1980s (see Harrington, 1992, and Kramer, 1992) and before that (see Harrington, 1987). Therefore, the threat of regulation or the threat of more stringent regulation may have been more credible than in any other given economic sector. To test the model, the author used the Fast Track data tapes available through the National Association of Independent Insurers (NAII). These tapes provide basic, quarterly insurance market data for every state, plus the District of Columbia. Such data include total premiums paid, number of exposure units, and total losses incurred. The tapes span 1984 to 1993 inclusively. This period is significant because the liability crisis occurred in 1987-89. Therefore, the tapes offer a sufficient number of observation quarters before and after the crisis to test the article’s regulatory threat hypothesis. 6 Throughout, the article supposes that the only possible intervention from the insurance commissioner’s office is to reduce price ceilings for automobile liability insurance premiums. Insurers can capture the regulator by having it not reduce the ceilings or increase them. The article shall assume that insurance commissioners cannot set price floors. 40 THE JOURNAL OF RISK AND INSURANCE This article’s contributions are two-fold. The first is the empirical analysis of the pricing behavior of insurers threatened by regulation.7 The second concerns the construction of the regulatory threat variable. The author researched close to 130 newspapers and business journals from 1984 to 1993 and highlighted all the threats of regulation that arose in every state for any quarter of any year. The regulatory threat variable took the value one if there was discussion of liability insurance premium regulation. The main result the author obtained is that the threat of regulation had a significant effect on the pricing behavior of insurance companies in the personal automobile liability insurance industry. It was also found that price increases were significantly smaller after the passage of Proposition 103 in California. The author concludes from these two results that the insurance industry reduced premium inflation as a result of regulatory threats reported by the news media. The author also concludes that the passage of Proposition 103 sent a signal to the industry that regulation or more stringent regulation was a serious possibility after 1988. The article is structured as follows. In the section titled “Model,” the author develops a simple model of how firms price their product. The primary results are presented in the section of the same name. The ordinary least-square analysis is first presented to determine what affects the average premium increase8 in a state in a quarter. It was found that price increases are smaller whenever a threat of regulation exists. The next section presents a two-step estimator procedure to control for the simultaneity between premium increase and the presence of a regulatory threat. After controlling for many outside factors, evidence is found to support the regulatory threat hypothesis that is quite robust to model variations (“Robustness” section). Finally, the article concludes.