New estimates of the labor market effects of workers' compensation insurance.

• Author: Kaestner, Robert

1. Introduction

   The workers' compensation laws are one of the earliest examples of a government mandated employer provided benefit. Under a combination of federal and state laws, employers have a statutory obligation to provide medical and income benefits to injured workers. In most states, employers are required to purchase insurance to meet their financial responsibility. Recently, there has been growing legislative support for expanding the scope of mandated benefits to include items such as parental leave and health insurance. The political attractiveness of mandated benefits is that they do not require an increase in taxes, and that they represent a political compromise between liberals and conservatives over how best to carry out social policy [22].

   The labor market consequences of mandated benefits are potentially large and not well documented in the literature. The fundamental issue is to what extent an increase in benefits, which represents an increase in employer costs, is offset by lower wages. Previous studies that have attempted to measure such compensating differentials have reached dissimilar conclusions over the size and even the direction of the effects.(1) The findings from studies examining the effect of workers' compensation insurance on wages tend to be more consistent, often finding a negative effect, but are also characterized by considerable diversity. Thus, past research does not provide enough information to make an informed policy decision and continued research in this area is clearly warranted.

   The purpose of this paper is to provide new evidence on the effect of workers' compensation benefits on the labor market. In particular, the paper examines the tradeoff between wages and workers' compensation benefits and the relationship between workers' compensation benefits and job safety. The primary contribution of this paper is to identify two previously overlooked aspects of the problem, and to present evidence that these issues are empirically important. It is the first paper to measure benefits using both the income and medical components of workers' compensation payments. The majority of past studies have used only the income component, even though 44 percent of all workers' compensation payments in the U.S. in 1988 were medical benefits. In addition, in that same year the medical benefits share of all workers' compensation payments varied dramatically by state, ranging from a low of 20 percent in Rhode Island to 64 percent in Utah. Second, it is the first paper to examine the effect that regulation of the workers' compensation insurance market has on wages and job safety. Workers' compensation insurance is the most heavily regulated line of insurance, and in the past, most states required that rates be approved by the state insurance department before they can be put into effect. Several states, however, have recently implemented alternative regulatory policies, and there is evidence that some of these policies have had a significant impact on the price of insurance. Therefore, depending on the regulatory environment, firms may be paying a different price for workers' compensation insurance. It is of considerable public policy interest to examine how different regulatory practices affect the labor market.

   Several findings of the analysis are noteworthy. First, the evidence suggests that exclusion of the medical component of benefits is a serious omission. Estimates of the wage-benefit tradeoff obtained using solely the income component of benefits are 100 percent larger than those that use both the income and medical components of benefits. Second, the results show that changes in workers' compensation benefits have two countervailing effects on injuries. An increase in benefits is associated with both a moral hazard effect, which tends to increase injuries, and an employer safety effect, which tends to decrease injuries. This is the first study to separately identify these two effects. Previous studies in this area, which use only an income benefit measure, are unable to separate out the role of moral hazard from other benefit effects on injuries.(2) Third, the results indicate that the type of regulatory policy governing the workers' compensation insurance market has a significant effect on both wages and job safety. In states that have deregulated the workers' compensation insurance market, wages and injury rates tend to be higher than in states that require use of prior approved rates.

2. Analytical Background

   Wage-Benefit Tradeoff

   Competitive wage theory suggests that wages and workers' compensation benefits will be negatively related: an increase in benefits will reduce wages. The size of the wage reduction in response to a benefit increase depends on the employer's cost of providing the benefit and the employee's valuation of the benefit [22]. An increase in benefits increases employer costs and shifts the demand curve for labor to the left. From the employer's perspective, a benefit increase is similar to a tax on labor. Employees, however, value an increase in benefits because it shifts more of the liability for an injury from the employee to the employer. Thus, employees are willing to supply more labor at any given wage, resulting in an outward shift of the supply curve for labor. The magnitude of the wage decrease in response to an increase in benefits depends on the relative size of the supply and demand shifts.

   A number of previous studies have attempted to measure the tradeoff between wages and workers' compensation benefits.(3) Generally, these studies estimate a cross-sectional wage regression that includes a variable measuring the risk of injury or death on the job, and a workers' compensation benefit measure. The typical model may be written as follows:

   W = [[Alpha].sub.0] + [[Alpha].sub.1]R + [[Alpha].sub.2]B + [Epsilon]. (1)

   In equation (1), W is a measure of wages, R is a measure of risk on the job, B is a measure of the income component of workers' compensation benefits, the [a.sub.i]'s are parameters, [Epsilon] is an error term, and observational subscripts have been suppressed. In general, the results from these studies indicate that higher risk is associated with higher wages, and that increases in workers' compensation income benefits lead to lower wages.

   The first issue which this paper addresses relates to the use of only the income component of workers' compensation benefits. The relevant variable to use in equation (1) is total benefits, which is the sum of both income and medical payments. The decrease in the demand for labor depends on the change in the employer's total cost of insurance, and the increase in supply of labor depends on the change in the total benefit package offered the employee.(4) Therefore, the more appropriate model is the following;

   W = [[Delta].sub.0] + [[Delta].sub.1]R + [[Delta].sub.2]WC + [Eta], (2)

   where WC is the total workers' compensation benefit, the [[Delta].sub.i]'s are parameters, and [Eta] is an error term. Total workers' compensation payments include both the income (B) and medical (M) components of benefits. Thus, equation (2) may be rewritten as

   W = [[Delta].sub.0] + [[Delta].sub.1]R + [[Delta].sub.2]B + [[Delta].sub.2]M + [Eta]. (2a)

   In general, [[Alpha].sub.2] of equation (1) is a biased estimate [[Delta].sub.2] of equation (2), because of the exclusion of (M) from the model. The direction of the bias depends on the sign of the partial correlation between B and M, holding R and any other variables that are included in equation (2) constant.(5)

   All but two previous studies investigating the wage-benefit tradeoff have used a model that includes only the income component of workers' compensation benefits. The estimates obtained from these studies may be biased and the conclusions drawn from them misleading. For example, Viscusi and Moore [26] use estimates of the wage-benefit tradeoff from their study to evaluate the adequacy of workers' compensation benefits. Based on a simple model of consumer choice, and their empirical estimates, Viscusi and Moore conclude that workers' compensation benefit levels are inadequate: workers are underinsured. If the estimates obtained by Viscusi and Moore are seriously biased, this conclusion may be erroneous.(6) Two papers that do not use solely the income component of benefits are Dorsey and Walzer [10] and Gruber and Krueger [12]. In these two papers, a measure of total workers' compensation costs are used, but these measures are a combination of total benefits (price) and the quantity of injuries. In both studies, the authors frequently find that the effect of workers' compensation costs on wages is either negative and insignificant, or positive and significant. These anomalous results may be due to the fact that the measure of workers' compensation costs used in these studies depends on the quantity of injuries, which is expected to be positively related to wages.(7)

   This paper estimates a model similar to equation (2) using separate measures of the income (B) and medical (M) components of workers' compensation benefits. Inclusion of the medical component of benefits in the model will result in better estimates of the tradeoff between wages and workers' compensation benefits than those found in prior studies. Estimates of the parameters of equation (2) will not be subject to the omitted variable bias due to the absence of the medical component of benefits. Furthermore, the two separate measures of workers' compensation benefits do not depend on the number of injuries, and estimates of their effects should more accurately reflect the influence of benefits on wages, compared to the estimates obtained by Dorsey and Walzer [10] or Gruber and Krueger [12].

   In virtually all of the previous studies concerned with the effect of workers' compensation benefits on wages, it has been assumed that all employers are paying the same price for insurance. In Carroll and...