The Pricing of Workers Compensation Insurance: Effects on Safety and Claims

Transcription

1 The Pricing of Workers Compensation Insurance: Effects on Safety and Claims Michael M. Barth Georgia Southern University Robert W. Klein Georgia State University Gregory Krohm International Association of Accidents Boards and Commissions Presented at the American Risk and Insurance Association Annual Meeting Washington, DC August, 2006 Contact Author: Tel: ; Fax: ;

2 The Pricing of Workers Compensation Insurance: Effects on Safety and Claims Abstract The effects of risk-based pricing of workers compensation insurance and the use of experience rating affect employers safety incentives and workplace accidents, injuries and claims. Previous research related to this topic tends to suggest that pricing incentives do improve safety and reduce injuries but there are number of limitations to this research and most workers compensation experts do not find it to be definitive. In this paper we present the results of some initial analysis of data on individual employer claims experience the first time that data at this more disaggregated, micro level has been examined to address this research question. Our initial analysis of individual firms workers compensation experience suggests that experience rating does have the intended effect of encouraging employers to improve safety. More specifically, our analysis yields supporting evidence of an ex post pricing effect higher experience modifications decreases the number of claims in subsequent years. We also found that employer size, as measured by payroll, tends to reduce the number of claims, consistent with economies of scale associated with employer investments in safety. Introduction The importance of risk-based pricing of insurance is widely accepted by many scholars and practitioners. Risk-based pricing is believed to have a number of desirable effects, including the mitigation of adverse selection, protection against insolvency, and, under some circumstances, enhancing incentives for insureds to reduce their risk of loss. One of several devices used to construct risk-based prices in commercial insurance, especially in workers compensation, is experience rating. While experience rating is an established practice in workers compensation insurance in the US and certain other countries, its merits and effects continue to be a matter of debate. One of the prominent questions in this debate is whether experience rating enhances employers incentives to reduce accidents and worker injuries and actually results in greater safety and fewer (and perhaps less severe) worker injuries. The literature and previous research on this question tends to support the safety effects of experience rating, but many consider the evidence underwhelming and not definitive. 1

3 The purpose of this paper is to contribute to the theoretical and empirical research on experience rating in workers compensation by further exploring its dynamics and testing its effects with new data and methods that have not been used previously. More specifically, in the empirical component of this paper, we assess the effects of experience modification rating factors on employers subsequent loss experience. In our initial research, we analyze a large sample of experience-rated Wisconsin employers to assess the interaction between changes in workers compensation experience rating modification factors and employers lost-time workers compensation claim counts. While a number of econometric studies of the effects of experience rating have been published, this study is unique in several respects. First, it estimates the experience rating-claims relationship using a large sample of individual employer data, while previous studies have relied on data aggregated at a state or class level. Second, this study includes several variables to control for the risk characteristics of individual employers. Finally, it explores the interaction of employer size and employer response to experience rating. Our results are consistent with the hypothesis that increases in an employer s experience modification factor leads to a decrease in its workers compensation claims. It is important to note that our research is at an early stage and subject to a number of qualifications. Our results are specific to Wisconsin although it would be reasonable to hypothesize that the underlying links between experience rating and employers subsequent experience would be more universal. We will test this proposition in further research using similar data acquired from other states. As data from other states are incorporated into our research, we will also be able to begin to examine the effects of state differences in experiencerating systems and other factors. This extended research, supplemented by new data and models 2

4 at a cross-state level, also should help us to discern the ex-ante and ex-post effects of experience rating (discussed further below). Our paper begins with a brief discussion of how experience-based adjustments of an employer s workers compensation premiums might induce it to increase its expenditures on improving safety and reducing accidents and injuries. This is followed by a review of previous research related to this topic, its indications and its limitations. The next section explains the unprecedented nature of the data that we utilize for our analysis and the methods we employ. We then present our initial empirical results and their qualifications. We conclude with a summary of our findings and our future research plans. Experience Rating, Safety and Claims Experience The primary question addressed by this analysis is whether the experience-based adjustment of employers workers' compensation insurance premiums has any noticeable effect on their subsequent lost-time claims. It is important to provide at least a brief explanation of how experience rating is implemented in workers compensation insurance in the US. Class rates of employers meeting minimum premium requirements are modified based on their relative claims experience over the previous three years. Employers with better than average experience than their class are given a modification factor of less than 1.0 that decreases the net or adjusted premium they are charged and vice versa. The rationale underlying experience modification is that it passes the costs of credible injury experience back to the employer, thus more equitably allocating costs according to risk and providing incentives for employers to promote workplace safety and reduce their risk of loss. If there were no experience modification or other employerspecific premium adjustment for risk, then each employer in a particular class would pay the 3

5 average loss rate for that class, even if they consistently had more or fewer claims than the average for the class. In the US, experience modification factors are calculated using a somewhat complex formula that contains a number of actuarial adjustments to an employer s historical claims experience. These adjustments are intended to produce a more accurate measure of relative risk among members of a class. Because the severity of a given claim is believed to be more subject to random chance than underlying differences in the risk of loss, the experience rating formula caps all losses entering the historic loss base of the employer at $5,500 i.e., the experience rating formula gives more emphasis to claim frequency than claim severity. Moreover, because the Law of Large Numbers predicts that the loss experience of small employers will be more volatile than the loss experience of large employers, the experience modification formula allocates more and more weight to the employer s actual experience as its payroll base increases. Further, employers must meet certain minimum premium thresholds in order to be subject to any experience rating. There are a number of other adjustments not mentioned here that affect the conversion of an employer s raw experience to a factor used in the determination of the premium that it will pay. 1 What effect should experience rating have on an firm s behavior and ultimately the safety of its workplace and the incidence of worker accidents and injuries? From an economic perspective, in a world of perfect information, firms would be expected to optimize the tradeoff between expenditures on workplace safety and the cost of worker injuries. The economic costs of worker injuries would include those that are incurred through a workers compensation system insurance premiums and compensation losses retained by the firm as well as other costs, such 1 Readers interested in obtaining a better understanding of how experience rating works in workers compensation insurance are referred to the NCCI s The ABC s of Experience Rating. 4

6 as lost productivity, lost investments in the training of an injured worker, etc. We would expect a firm to invest in a particular safety measure if the marginal cost of the investment was less than the expected associated reduction in worker injury costs. 2 If safety investments are subject to diminishing returns, firms would continue to invest in safety until the marginal cost of an additional investment would exceed the corresponding reduction in the cost of worker injuries. Of course, firms do not operate in a world of perfect information and there are other real world factors that could affect their safety efforts and worker injuries. First, firms face information constraints and uncertainty concerning the benefits of a particular investment in or expenditure on safety. This may cause them to under-invest in safety because they cannot determine the full benefits of safety investments at the time they make them. Second, employers may consider non-economic costs of worker injuries or may be motivated by considerations other than the pure economic tradeoff between the cost of safety expenditures and the costs of worker injuries. Third, firm managers may be influenced by views and biases that either contribute to or diminish their efforts to increase safety and reduce injuries. Fourth, firms may not be able to fully control the level of safety and the incidence of worker injuries. There may factors outside their control or influence, e.g., worker attitudes and safety behavior, external events affecting safety, etc. Fifth, mechanisms designed to assess and price higher risk, such as experience rating, are themselves imperfect. All of these complexities may interfere with the intended effects of risk-based pricing of workers compensation on firms safety efforts and worker injuries and claims. Experience rating should have a positive effect on workplace safety but it has been difficult to quantify the significance and magnitude of this effect. Further, it is reasonable to 2 Harrington and Niehaus (2003) contains a chapter on loss control and discuss how firms might optimize their expenditures on safety and preventing/mitigating losses. 5

7 postulate that experience rating may have both ex ante and ex post effects on employer safety efforts. Theory would suggest that there is an ex ante effect in that firms anticipating higher premiums due to adverse claims experience might be motivated to increase safety and reduce injuries proactively to avoid higher premiums. Arguably, firms would reap the maximize savings from safety investments by implementing them sooner rather than later, understanding that that anticipated savings or positive cash flows would need to be discounted back to the time when safety expenditures would be made. However, there may be reasons that would lead firms to make certain safety investments only after receiving an adverse experience modification. One set of reasons has to do with information. Firms may not be aware they have a safety problem until they have actually experienced higher-than-expected worker injuries and compensation claims. Indeed, a firm may use their claims experience to inform them about their safety level and risk of worker injuries. Firms may also use their claims experience to evaluate the effects of certain safety measures before they may further safety investments. Other potential reasons for ex post safety effects may include management myopia or managers failure to pay attention to safety and the risk of work injury until they encounter workers compensation claims and higher experience modification. When the premiums of these firms rise because of adverse experience, this may draw attention managers attention to safety problems and compel them to do something about the problems. If one examines data on the experience rating and claims experience of a set of firms for a particular period, there may be a mix of ex ante and ex post effects confounded in the data. There may be some firms that have implemented safety measures ex ante to avoid adverse experience modifications, causing their claims experience and experience modifications to be 6

8 lower than they would be otherwise. Finally, there may be some firms that implement initial or additional safety measures and achieve reduced claims following higher experience modifications received during the period of time measured by the data. Therefore, our ability to discern ex ante safety effects or ex post effects that occurred prior to our sample data period is constrained by the nature of our data. With the data we have for this initial stage of our research, we can test the manifestation of the ex post claims effects resulting from higher experience modifications received during our sample data period. Our data does not enable us to test the ex ante effects of an experience rating system or the ex post effects of experience rating modifications prior to our sample period. In further research, we hope to be able to test for both ex ante and ex post effects using data from multiple states. Previous Research There have been some published studies on the effects of workers compensation insurance pricing on the incidence of workplace accidents but this literature leaves many workers compensation researchers and practitioners unsatisfied. One constraint on this research is that the kind of information that would be highly desirable for research on this topic is not publicly available. These would be what we term micro-data on individual employers (who could remain anonymous) that would include information on how their premiums are adjusted for risk (e.g., experience rating modifications) and their actual claims experience. Insurers are not comfortable in sharing these data with researchers and national statistical agents that are the sole repositories of much of these data are also reluctant to make these data available to researchers. Therefore, the lack of public or researcher access to micro-data has limited the kind of research that has been performed to date. 7

9 Consequently, the previous studies that shed any light on this topic (intentionally or not) have used proxies for employers exposure to experience rating and class-level injury rates or have examined the effect of regulatory systems on safety-related outcomes. These indirect methods have produced results that are interesting they tend to be consistent with the expectation that more sensitive risk-based pricing improves safety - but they are not considered definitive by most workers compensation experts. There also have been event studies of injury rates in systems prior to and after they have strengthened their use of experience rating and surveys of employers regarding their opinions on experience rating systems. However, the event studies cannot control for the effects of other factors or changes coincident with experience rating changes and the employer surveys are subject to selection bias and most employers inherent distaste for experience rating regardless of how it affects their actual behavior. Among the first category of studies, Ruser (1985) looked at the relationship between increases in workers compensation indemnity benefits and injury rates under experience-rating and suggested that larger firms would have a different reaction to experience rating than smaller firms. Ruser found that higher benefits had less of a positive effect on the frequency of injuries as firm size increased. In a similar line of research, Worrall and Butler (1988) produced findings consistent with the concept that experience rating works to reduce injuries, but their data did not lend itself to direct measurement of the effect because they used firm size as a proxy for experience rating, but could not distinguish between large firms that self-insured and large firms that purchased commercial insurance. In the second vein of research, Danzon and Harrington (2001) examined the relationship between regulatory suppression of worker s compensation rates and loss costs and found that rate suppression, measured by the lagged residual-market share of payroll, increased loss growth. 8

10 Although not directly related to the effect of experience rating, Danzon and Harrington s study does suggest that there may be a strong empirical relationship between injury rates and workers compensation pricing. However, because they focused on loss costs rather than injury rates per se, it is not possible to infer whether the effect they found acts through safety or other elements of loss control or both. Barkume and Ruser (2001) assessed the effects of deregulation of workers compensation insurance on prices and injury rates in the U.S. They concluded that the relaxation of regulations on the pricing of workers compensation led to reductions in both premiums and injury rates. Thomason, Schmidle and Burton (2001) performed an extensive study of how alternative insurance arrangements for workers compensation in the US affected various system outcomes, including injury rates. They found that injury rates were higher in exclusive-state-fund jurisdictions than in states that permit private insurers to underwrite workers' compensation insurance policies. Their findings also indicated that injury rates in jurisdictions with competitive fund states were lower than injury rates in states that only had private workers' compensation insurers, which introduced additional questions as to the true underlying effects. These studies tended to focus on the regulatory regime rather than specifically on the effect of experience rating, but they tend to support the idea that more appropriate pricing systems lead to reductions in injury rates and insurance costs. Kralj (2000) and Wright and Marsden (2002) review these and other studies of the effectiveness of experience rating in the US, Canada and other countries. Both of these literature reviews conclude that the balance of the evidence is consistent with the proposition that experience rating has a beneficial effect on reducing the frequency and severity of claims but point out that the literature is not conclusive and that the studies to date have suffered from 9

11 methodological weaknesses. Taken together, these studies and their reviews provide some indication that employers respond to price incentives but leave considerable room for further research that would examine the link between risk-based pricing and workplace safety more directly. The analysis discussed in this paper seeks to take a step in that direction and future development of this stream of work will hopefully make further progress. Data and Methodology While it is difficult to obtain micro-data from national statistical agents the authors have sought the cooperation of several state rating bureaus or advisory organizations in providing data they collect or posses. The Wisconsin Compensation Rating Bureau (WCRB) provided nonpublic data from its database for this study. Firm-level data for a randomly selected pool of 3,000 employers that were nominally experience rated in policy year 2003 were provided. The data included the number of lost-time claims for each policy year from 1999 through 2003, aggregate payroll, the governing class code of the firm, and the experience rating modification factor at the beginning of the policy year. Data anomalies and/or low payroll volume prompted us to discard some of the observations, leaving 13,148 observations out of the potential 15,000 firm-years. This information was supplemented by pricing information from the National Council on Compensation Insurance (NCCI) Experience Rating Plan Manual for Wisconsin, which includes the Expected Loss Rate (ELR) factors for each of the 321 different governing class codes reflected in the employer data sample. The governing class code is an indication of the primary type of work activity in a particular firm, but does not directly measure the risk. Ideally, the number of workers in each class code, plus the associated payroll, would give a more precise 10

12 measure of firm-specific risk, but for the majority of firms, and especially for small to medium size firms, the governing class code is a reasonable proxy. As would be expected, most of the employers in the data set did not have a lost time claim in a particular year, which creates a preponderance of zeros in the measured claim counts. Of the approximately 120,000 covered employers in the state, the majority have fewer than 5 employees. For the smallest employers, especially in non-hazardous industries, a lost time injury is a very unusual event. While we have used ordinary least squares (OLS) regression for this initial analysis, alternative methods such as Poisson regression or categorical modeling processes may provide more accurate measures in future development of our research. We tested several different modeling techniques with our current data set, but their results were consistent with those that are reported for the OLS model. As more data become available, one of these alternative models may yet prove to provide a better fit. Beyond just random chance, there are many potential risk-related factors affecting a given employer s claim counts (or the employer s propensity to experience one or more claims) from one period to the next. Because of the limitations to our data set, we could not model all potentially significant causal factors. 3 Payroll volume, for example, provides an estimate of the number of workers exposed to injury and is included in the model. Additionally, we include a measure of growth in payroll from one year to the next to evaluate the effect of changes in the worker mix. However, we must point out that payroll is a function of salary or wage rates as well as the number of employees so it is an imperfect measure of an employer s exposure to accidents and injuries. 3 In seeking the cooperation of the WCRB and other state rating bureaus, we have balanced the risk of asking for too little with the risk of asking for too much. As such organizations become more accustomed to sharing data with researchers, it may be possible to obtain more employer-specific variables these organizations possess short of the actual identities of employers. To obtain additional variables, it will be necessary to link employer identities to information in other databases that would expand the employer characteristics and factors that we might model. 11

13 The size of the firm would also be expected to have an impact on the number of claims beyond just the number of workers exposed to loss. Firms may experience economies of scale with respect to the gains from investments in or expenditures on certain worker safety measures because of fixed costs associated with those measures, e.g., the fixed cost of developing a safety education and training program or the fixed cost of an ergonomic study of the causes of worker injuries. If economies of scale are present, it would suggest that larger firms should have relatively lower claims per worker, all other things equal, because fixed-cost expenditures on worker safety are spread more broadly across a larger number of workers. There are potentially other factors related to firm size that are harder to measure. Hiring costs would be expected to increase for a large employer that consistently had worse-thanaverage loss experience (it might develop a bad reputation among prospective workers), so there may be a natural tendency for larger employers to focus more efforts on safety and loss prevention than smaller employers. Smaller employers may have diminished incentives to increase safety and prevent accidents because of informational asymmetries that make it harder for prospective employees to compare the safety record of one small employer to the next. A very large employer, such as Wal-Mart, on the other hand, is subject to greater public attention and that attention may provide a greater incentive to invest in workplace safety efforts to reduce hiring costs, negative publicity or regulatory scrutiny. If the experience rating pricing system has its desired safety effect, there should be a negative relationship between claim counts and experience rating. The higher price of insurance should provide incentives to increase workplace safety. However, as discussed above, the design of the pricing system may be send imperfect signals to employers or employers may be price inelastic in their response to higher workers' compensation premiums. Further, our data can 12

14 only reveal the ex post effects of experience rating on claim frequency we are not in a position to discern any ex ante effects of experience rating that may already be reflected in an employer s claims experience. Indeed, actuaries claim that credible samples of individual employer experience over three years is a fairly good predictor of their losses in the upcoming policy year. Hence, actuaries tend to view the modification factor as a means of producing ex post equity among all employers in a class at the end of the policy year rather than a device that will tend to cause an employer to experience a decrease in claims. We also include lagged values in our model to evaluate whether there are delays in the impact of any increased safety incentives due to higher experience mods. If the experience modification factor increases, the cost of workers compensation insurance increases as well. The base cost of workers compensation insurance as a percentage of payroll can be a relatively small number (e.g., less than 0.5 percent for bank workers) or it can be a relatively large number (e.g., 7 percent for factory workers). The effect of a change in the experience modification factor, which increases or decreases the base rate in response to historical loss experience, can have a material effect on the cost of production and can also interact with other factors. If a firm s labor costs increase materially because of worker injuries, the firm should have an incentive to implement additional worker safety measures to reduce that cost, but those changes may take time to implement and have a demonstrable effect. Comparative Statistics by Size of Employer If the average firm s workers compensation costs are materially affected by the change in its workers compensation experience modification factor that, in turn, leads to a significant ex post safety effect, then we would expect to see an inverse relationship between changes in the 13

15 modification factor and future claims. That is, as the modification factor increases, we would expect to see the number of claims in subsequent years to decrease if the associated increase in a firm s premium prompts it to further improve safety with a beneficial effect on its accidents, injuries and ultimately its workers compensation claims. Alternatively, if the full cost of worker injuries was not reflected or captured adequately by the experience modification system, or if employers were not sensitive to the price signals created by higher experience modifications, or if employers have already optimized safety expenditures (and their results), then we would expect to see no statistical association between changes in the experience rating modification factor and the number of claims. However, it is difficult to hold all things equal across firm sizes. The riskier types of firms (e.g., contractors) tend to be smaller, which suggests that there will be an inverse relationship between firm size and claim count. Using the WCRB data, we segregated the observations into ten groups of roughly 1,300 observations each based on payroll size. For each of the groups, we then took simple statistical measures of the number of claims, the number of claims per $100,000 of payroll, and the average experience modification factor. Table 1 below shows the summary data for each of the ten size groups. Table 1 shows that the average number of claims per firm increases with payroll size, which is consistent with our expectations. Since total payrolls should increase with the number of workers, one would expect that the average number of claims per firm would be higher for the firms with larger payrolls. However, when we use the average claim rate (claims per $100,000 of payroll) we see that the claims rate actually declines with payroll size. Part of that decline may be due to economies of scale in safety expenditures or to the other factors previously discussed, but some of it may also be attributable to differences in the inherent risk of job sites or working 14

16 environments in large versus small firms. The average ELR, a measure of expected claims, declines with firm size. Table 1 Comparative Data Across Size Groups Average # of Claims Prob (Claims =0) `Prob (Claims >0) Claims Per 100K of Payroll Avg Mod Factor Avg ELR Factor Group Payroll Range No. of Obs. Claims 1 10K - 74K 1, % 4% K-130K 1, % 12% K-200K 1, % 14% K-290K 1, % 18% K-416K 1, % 24% K-605K 1, % 29% K-945K 1, % 37% K-1,557K 1,315 1, % 52% ,557K-3,146K 1,315 1, % 58% ,146K - 578,840K 1,313 7, % 79% ALL 13,148 12, % 33% Interestingly, the average experience rating modification factor is remarkably stable across firm sizes. That is, given this data, we detect no size-related anomalies in the assignment of the modification factors. However, because the ELR factor, which determines the base rate that the modification factor is applied to, is higher for the smaller firms, the relative impact of the modification factor might be greater for the smaller firms. We should also note that the approach and method used to calculate modification factors would be expected to produce an average modification factor that is close to 1.0 which is what we observe in Table 1. OLS Regression Model Results The regression model we used to measure the effect of the experience modification factors on claims experience is: 15

17 Claim Count it = b 0 + b 1 *MOD it + b 2 *MOD it + b 3 *MOD it-1 + b 4 *log(payroll it ) + b 5 *log(payroll it *MOD it ) + b 6 *ELR j + b 7 *log(payroll it /Payroll it-1) + error term. Where: Claim Count it =the number of claims for Employer i in Year t. MOD t = the experience modification factor for Employer i in Year t. Payroll it = the total covered payroll (in $) for Employer i in year t. ELR j = the Expected Loss Rate for Employer i s primary classification. We used ordinary least squares (OLS) regression to estimate parameters for each of the explanatory variables in the model. Our regression results are presented in Table 2. We evaluated one lag period (Model 2) and two lag periods (Model 3) for the experience modification factor (MOD) and found that the lagged values had no predictive power. Other than the lagged variables, our other explanatory variables were all significant at the level. Table 2 OLS Regression Results Using Claim Count As Dependent Variable Variable Variable Description Model 1 Model 2 Model 3 B0 Intercept b1 MOD(t) Current Experience Factor b2 MOD(t-1) Last Year Experience Factor b3 MOD(t-2) Prior Year Experience Factor 0.05 b4 log(payroll) Size b5 log(payroll*mod(t)) Interaction of volume and price b6 ELR Relative Risk b7 log(payroll(t)/payroll(t-1)) Growth Number of Observations 10,012 10,012 7,189 Adjusted R-square The residuals from this model indicated non-constant variance and potential nonlinear relationships between the dependent variable and some of the explanatory variables. A second set of models was run, using the square root of claims as the dependent variable, and those results are shown in Table 3. The parameter estimates and signs were similar to those produced 16

18 in Table 2 for the untransformed dependent variable, and there were fewer violations of the basic OLS assumptions. Table 3 OLS Regression Results Using Square Root of Claim Count As Dependent Variable Variable Variable Description Model 4 Model 5 Model 6 B0 Intercept b1 MOD(t) Current Experience Factor b2 MOD(t-1) Last Year Experience Factor b3 MOD(t-2) Prior Year Experience Factor 0.12 b4 log(payroll) Size b5 log(payroll*mod(t)) Interaction of volume and price b6 ELR Relative Risk b7 log(payroll(t)/payroll(t-1)) Growth Number of Observations 10,012 10,012 7,189 Adjusted R-square We again note the inverse relationship between the claim counts and the experience modification factors. In Models 5 and 6, all variables were statistically significant at the level, including the single lagged value for the experience modification factor in Model 5. However, in Model 6, the lagged variables both drop out again, and we conclude at this point that the historical experience modification factors are not significant predictors of future claim experience. As additional data become available, we may revisit that conclusion. These results indicate that the most current modification factor (MOD t ) has a negative and statistically significant effect on the number of claims reported in subsequent years. In other words, the higher the experience modification factor, the fewer claims that an employer tends to incur in a subsequent year. The magnitude of the estimated coefficient for MOD t suggests that an increase in an employer s mod from 1.0 to 2.0 will be a followed by 3 to 4 fewer claims, all other things equal. However, we should note that most of the employers in our sample had experience mods ranging from 0.8 to 1.2 so that the actual consequences of experience rating are small 17

19 because the range of experience mods tends to be small. Still, our result is consistent with the hypothesis that experience rating has a significant ex post effect on employers claims experience. Interpreting the parameter estimates of the variables involving an employer s payroll is somewhat more complicated because of statistical issues associated with the correlation of these variables with each other and other variables in our model. Essentially, our review of the data and our statistical results leads us to believe that larger employers have more claims simply by reason of their greater size (i.e., they have more employees who can be injured), but the increase of claims with size diminishes with size. In other words, large employers experience more claims but at a diminishing rate. This inference is consistent with our hypothesis that there are economies of scale in improving safety and reducing accidents, injuries and claims. Further refinement of our data, model specification and regression techniques will be needed to better support the inferences we are drawing on the relationship between experience modifications, employer size and employers claims experience. Alternative models may provide yet a better fit to the data. The dependent variable (number of claims) does not follow a normal distribution and is not a continuous variable, which is a violation of one of the assumptions of OLS regression. In a very large sample, the effect of this violation is minimal, but the estimators may not be optimal. Insurance claims counts are believed to more closely follow the Poisson distribution and this commends the use of a more complex modeling technique such as poisson regression or zero-inflated poisson regression. While using a more complex regression estimator may produce more accurate estimates of the parameters, the real test of the efficiency of this model will be its ability to produce stable results with larger samples of data and/or with data from other states. 18

20 The results here are limited to Wisconsin and we cannot assume a priori that the results for other states would be the same. While there are modest differences among the states with respect to how experience modification factors are calculated, there may be a number of other factors that vary more substantially and also influence the relationship between experience modification factors and subsequent claims experience. As data from other states are added, we can test the robustness of our model results as well as assess the effects of state-specific factors. Conclusions and Future Research Our initial analysis of individual firms workers compensation experience suggests that experience rating does have the intended effect of encouraging employers to improve safety. More specifically, our analysis yields supporting evidence of an ex post pricing effect higher experience modifications decreases the number of claims in subsequent years. We also found that employer size, as measured by payroll, tends to reduce the number of claims, consistent with economies of scale associated with employer investments in safety. While our results are promising and consistent with economic predictions regarding riskbased pricing and experience rating, it is important to raise several caveats. First, our data allows us to observe employers experience modification factors and subsequent reported claims. From these data we infer a relationship between experience rating, employer safety efforts, worker injuries and workers compensation claims. However, we cannot observe actual safety efforts nor the incidence of worker injuries hence, we can only infer that the complete desired chain of behaviors and outcomes exist. Higher experience mods could also motivate some employers to discourage workers from filing claims. We cannot rule out the possibility that this phenomenon, if it exists, could be contributing to the reduction in claims that we observe. 19

21 Our data also did not allow us to discern an ex ante effects of experience rating. On the one hand, this means that risk-based pricing and experience rating could have additional desirable effects on safety, injuries and claims beyond that indicated by our results. On the other hand, if we raise the possibility that experience rating has undesirable ex post effects, we must acknowledge that it might also have such undesirable effects ex ante. As we have noted, other model specifications and regression techniques may yield results that come closer to the reality that we are trying to assess. Our initial data is limited to Wisconsin and additional data sets from other states may provide more insights on the effects of experience rating across jurisdictions. Ultimately we hope to expand our data sets at the employer level and at the state level, as well as refine our methods and models, to better understand how the pricing or cost of workers compensation and other factors affect the incidence and severity of workplace injuries. References Barkume, Anthony J. and John W. Ruser, 2001, Deregulating Property-Casualty Insurance: The Case of Workers Compensation, Journal of Law and Economics 44: 37. Danzon, Patricia M. and Scott E. Harrington, 2001, Workers Compensation Rate Regulation: How Price Controls Increase Costs, Journal of Law and Economics 44: 1. Harrington, Scott E. and Gregory R. Niehaus, 2003, Risk Management and Insurance, 2 nd Ed. (Boston, Mass.: McGraw-Hill/Irwin). Klein, Robert W. and Gregory Krohm, 2006, Pricing in Workers Compensation Insurance: Incentives for Safety, unpublished paper, Georgia State University, Atlanta. Kralj, Boris, 2000, Occupational health and safety: effectiveness of economic and regulatory mechanisms, in Workers Compensation: Foundations for Reform (Toronto: University of Toronto Press): National Council on Compensation Insurance, 2002, Experience Rating Plan Manual for Workers Compensation and Employers Liability Insurance (Boca Raton, Fla.). 20

22 National Council on Compensation Insurance, 2000, Basic Manual for Workers Compensation and Employers Liability Insurance (Boca Raton, Fla.). National Council on Compensation Insurance, 2004, The ABC s of Experience Rating (Boca Raton, Fla.). National Council on Compensation Insurance, 2005, Classification Codes and Statistical Codes for Workers Compensation and Employers Liability Insurance (Boca Raton, Fla.). National Council on Compensation Insurance, 2005, Experience Rating Update, (Boca Raton, Fla.). Ruser, John W., 1985, Workers Compensation Insurance, Experience-Rating and Occupational Injuries, Rand Journal of Economics 16: Thomason, Terry, Timothy P. Schmidle, and John F. Burton, Jr., 2001, Workers Compensation: Benefits, Costs and Safety Under Alternative Insurance Arrangements (Kalamazoo, MI: W.E. Upjohn Institute for Employment Research). Worrall, John D. and Richard J. Butler, 1988, Experience Rating Matters, in Philip S. Borba and David Appel, eds., Workers' Compensation Insurance Pricing: Current Programs and Proposed Reforms (Boston, Massachusetts: Kluwer Academic): Wright, Michael and Sara Marsden, 2002, Changing business would bearing the true cost of poor health and safety performance make a difference?, Norwich, England: Health and Safety Executive, Contract Research Report 436/