Workers’ Compensation Insurance Pricing

Workers’ compensation insurance provides medical payments and cash benefits to over seven million injured workers per year. As such, it is probably the largest social insurance program in the nation after Social Security. In the private market alone, employers paid approximately $20 billion in 1986 to insure their obligations under the various state workers’ compensation statutes. Add to that the coverage provided by the federal government, state funds, and self-insurance, and the total expenditures for compensation of injury arising “out of, or in the course of employment” must exceed $30 billion. Given its size and scope, workers’ compensation insurance must be the most under-researched social insurance program in the nation.

In recent years, a growing number of research papers have applied modern neoclassical labor economics to address incentive issues in the workers’ compensation system. How these incentives can be detected in distributions of workers’ compensation indemnity claims, and what they imply for those distributions, are the subjects of this research. We develop a simple statistical model that breaks down the observed distribution of indemnity payments into choice, chance and heterogeneity components, with particular attention given to how changes in the benefit structure (given market wage rates) and in experience rating affect claims. There is growing evidence (see our review in Worrall and Butler, 1985a) that the structure of benefits and experience rating may significantly change both the frequency with which claims are filed as well as the duration of a claim once it is filed. Given the dynamic changes taking place in this enormous social insurance program, even small response effects have large distributional and efficiency impacts.

This chapter describes a prospective payment system for workers’ compensation benefits, modeled on the DRG system which is used for paying hospitals.

Workers afflicted with occupational injury or disease are compensated for their losses in a variety of ways. The mechanism that is most often considered is workers’ compensation insurance which, under suitable conditions, provides compensation for such losses without regard to fault on the part of the employer. Workers may also be compensated through the existence of wage differentials related to the anticipated losses or to the portion of those losses not properly compensated from other sources. In addition, victims of occupational injury or disease may obtain compensation through the tort system if the cause of injury or disease is a product whose producer has failed to provide it in a safe form or with suitable warnings regarding the dangers involved in its use. Other sources of compensation include private disability insurance, private health and accident insurance, and payments under public programs such as Social Security. This paper discusses the problems associated with structuring a reasonable system of compensation for occupational disease in a context in which various sources of compensation exist.¹

Does the experience rating of workers’ compensation insurance affect employers’ provision of safety? It is possible that an experience rating system that provides some firms with other than the actuarially fair premium may result in a reduction of safety provision by some employers and a concomitant increase in job injuries. Despite the fact that one of the goals of the workers’ compensation system is to provide safety incentives, there has been very little empirical evidence provided to help us answer the public policy questions about the efficacy of safety provisions. By contrast, there has been a growing body of evidence that claim frequency (Bartel and Thomas, 1982; Butler, 1983; Butler and Worrall, 1983; Chelius, 1983, 1982, 1977; Ruser, 1985; Worrall and Appel, 1982) and severity (Butler and Worrall, 1985; Worrall and Appel, 1982; Worrall and Butler, 1985) are directly related to workers’ compensation benefits. Most of the claim frequency studies have focused on the risk-bearing or claim-filing behavior of employees. Using different specifications and time periods, these studies have provided evidence that benefit increases may raise workers’ compensation costs in at least three ways: an increase in the daily cost of each claim; an increase in claim cost due to increasing claim duration; and an increase in claim frequency. With the exception of Ruser (1985), the frequency studies cited above have actually provided crude measures of the net effect of benefit changes on claim filing or injury rates. This net effect arises from countervailing forces: the employee’s increased incentive for risk bearing and claims filing, and the employer’s increased incentive to economize on claims. Employee effects seem to be dominating. In this chapter, we shall follow Ruser (1985) and explicitly test for both employee and employer incentives. Before we consider our empirical scheme, however, we shall briefly describe the underlying choice-theoretic model that generates the claim-filing (employee) and safety-providing (employer) behaviors.

The National Council on Compensation Insurance (NCCI) prospective experience rating plan is used in workers’ compensation pricing in a large majority of states. This plan commonly is applied to about 15 percent of employers, and approximately 90 percent of the labor force may be employed by these firms. The NCCI plan adjusts an individual employer’s manual premium based on its own loss experience during a three-year experience period to produce the standard premium for the firm. In principle, this process is designed to produce expected standard premium loss ratios for each class of employment that will equal the NCCI target for a state, which is called the standard permissible loss ratio. The standard premiums then are subject to premium discounts and the addition of an expense constant to incorporate differences in insurer expenses for firms of different sizes.¹ Hence, loss ratios based on net premiums, i.e., standard premiums adjusted by premium discounts and the expense constant, are expected to decline with increases in firm size.

Private insurance carriers sell workers’ compensation insurance in all but six states and now account nationally for about 60 percent of all benefit payments.¹ Traditionally, the procedure used to determine workers’ compensation insurance rates limited the amount of price competition among carriers. Recently, however, several states have changed their laws or regulations to permit more competition in rates. This study examines in detail the impact on the employers’ costs of workers’ compensation insurance as a result of the January 1, 1983 introduction of open competition in Michigan.

The marketplace in which property—liability insurers operate has become increasingly risky. Recent years have witnessed price competition far more demanding than any experienced since the formation of the insurance cartels earlier in this century. Historically, high interest rates have led to intense competition among insurers for investable funds coupled with demands by policyholders to be given credit for interest on funds held in reserves. The insurance sector has experienced higher inflation than the economy as a whole due to liberalization of jury verdicts, broadened interpretations of contractual terms, and other factors (see Cummins and Nye, 1981). Losses on long-tail lines have become less and less predictable. These forces have increased the volatility of underwriting profits, made insurers more vulnerable to potential financial difficulties, and increased the risk of insolvency.

The derivation of an appropriate price for a reinsurance contract is a formidable problem. This is due in large part to the difficulties encountered in estimating the probability distribution for underlying loss payment. For example, facultative reinsurance contracts on individual direct policies usually apply to the large or catastrophic losses described in the right-hand tail of the loss distribution. But the infrequency of these large losses implies that the tail of the loss distribution is extremely difficult to estimate.¹ Similar estimation problems arise with treaty reinsurance, especially when it is arranged on a nonproportional basis. Despite these problems, some estimate of the loss distribution (or at least of summary characteristics of the distribution) is essential to the derivation of the reinsurance premium. This is true no matter what technique is used to derive an appropriate premium for the reinsurance contract.