استراتژی های مدیریت از کار افتادگی کارفرما و سایر عوامل نرخ ها و هزینه های آسیب ادعایی: تجزیه و تحلیل بلند مدت ادعای بیمه از کار افتادگی مبتنی بر اشتغال

Problem: Recent research indicates that many injuries to workers are not job related and that more than half of these injuries are not compensable under Workers' Compensation. Prior research on predictors or risk factors for these injuries, and the determinants of their costs, is quite limited. Method: This study analyzes data on more than 1000 injury-related paid claims for long-term disability (LTD) benefits by workers in 271 U.S. firms. Multiple regression analyses are used to examine predictors of LTD injury claims rates and determinants of LTD benefit payments per claim. Results: Although less than 20% of these injuries are work related, employer characteristics (industry type, selected disability management practices, and recent exposure to layoffs) are significant predictors of injury claims rates. Availability and/or generosity of other disability income sources, such as Workers' Compensation and Social Security Disability Insurance (SSDI), also impact on observed claims rates. Other strong predictors are the demographic and occupational mixes of employees. Cost per claim is also related to employer characteristics, employee characteristics, and the availability of other fringe benefits offered by the employer. Surprisingly, the level of benefits available under the employer's LTD policy does not have a detectable effect on claims rates. Impact on Industry: The importance of employer characteristics and policies, as predictors of claims rates and costs, suggests that occupational and nonoccupational injury risks may be strongly correlated. Programs to reduce the risks and costs of occupational injuries may also have important payoffs in terms of lower risks and costs for injuries to employees outside of the workplace.

Nonfatal injuries to employed persons are a major economic burden in the United States. The landmark study by Rice, MacKenzie and Associates (1989) estimated that 37.3% of all injury costs were accounted for by nonfatal injuries to persons in the age range from 25 to 64 years. According to the March 1998 Current Population Survey data, more than 96% of all persons in this age range were employed and over 83% were employed full time. Thus, it seems reasonable to assume the bulk of nonfatal injury costs for this age group is the result of injuries to employed persons. There is substantial literature on the factors that influence the rates and costs of these injuries, and especially occupational (i.e., work related) injuries (e.g., Layne & Landen, 1997, Mitchell, 1988, Ore & Strout, 1997, Zwerling et al., 1996 and Zwerling et al., 1997). Much of this work focuses specifically on data from the Workers' Compensation program and/or the influence of variations across states and over time in the incentives and characteristics of Workers' Compensation programs (e.g., Conway & Svenson, 1998, Curington, 1994 and Ruser, 1998). Other studies of work-related injuries have emphasized the important role of employer-specific (or establishment-specific) factors in explaining variations among employers in injury rates (Intracorp, 1999). Hunt (1993) documented the persistence of injury rates over time for specific employers and interpreted this as the result of employer-specific risk factors or hazards. Several studies have emphasized the positive role that employers can play by implementing injury prevention programs and management strategies to reduce the costs of disabilities resulting from injuries Hansen, 1997, Intracorp, 1999, Kochaniec, 1999 and Rogers, 1995. Many of these latter works are case studies involving a small number of employers that document the effectiveness of particular programs. An important exception is the study by Hunt, Habeck, VanTol, and Scully (1993) that analyzed a survey database of approximately 170 employers in Michigan and incorporated summary measures of disability management activities as explanatory variables. Because of their focus on occupational injuries, previous studies paid little attention to the major rates or costs of nonoccupational injuries and to the major compensation sources for such injuries, such as private long-term disability (LTD) insurance, private short-term disability (STD) insurance, and Social Security Disability Insurance (SSDI). LTD insurance, which has been completely ignored in the prior literature, is an important source of financial protection for workers in the U.S. economy. The U.S. Bureau of Labor Statistics (BLS; Foster, 1998) estimates that 25% of all workers, and 31% of all full-time workers, are covered by LTD insurance. The total dollar amount of LTD benefits paid is only about 10–15% of the cash benefits paid to employees by Workers' Compensation insurance (U.S. Social Security Administration, 1996), but the number of claims under LTD may well exceed 15% of Workers' Compensation claims. One reason for this is that the dollar benefits from LTD claims are typically reduced when they are offset by benefit payments from Workers' Compensation and SSDI. More importantly, LTD claims are not limited to disabilities due to work-related injuries. Miller (1995) recently estimated that nonfatal injuries resulting in lost workdays were divided almost evenly between compensable and noncompensable injuries (45% vs. 55%). 1 This suggests that only half or less than half of all injury-related disability incidents for workers would be captured by Worker's Compensation data. The broader coverage of LTD insurance is a strength for the present study, though one might argue that impacts on employer policies on claims rates and costs are less clearly apparent because nonwork injuries are included in the LTD claims data. The present study contributes to the literature on injury and compensation programs by focusing specifically on experience under LTD insurance. We analyze data from a recent national survey of more than 270 employers that collected information about specific disability management practices. These survey data were linked to information on injury-related claims for (LTD) insurance benefits by employees of these employers, as well as to other data about the employers, the LTD policies covering their employees, and characteristics of the states and localities where these employers are located.

Our analysis has identified a large number of predictors that are associated with LTD injury claims rates. Occupation and industry types are very important predictors even though less than 20% of the cases studied were reported to be work related. Presumably, this reflects the influence of variations in job requirements across occupations and industries, and the interactions of these requirements with functional deficits caused by injuries that result in disability. Among the LTD policy characteristics, the elimination period and rules for defining disability were significant predictors, but the direction of effect for the latter factor was unexpected. Moreover, variables relating to benefit levels and benefit duration limits, which capture economic incentives to seek LTD benefits, did not impact claims rates. The relationships between claims rates and other benefit or compensation programs were strong but varied. Results for measures of generosity of Workers' Compensation benefits suggest that they are serving as substitutes for LTD benefits, while the availability of other employer-provided fringe benefits and SSDI benefits appear to strengthen employees' incentives to go on LTD benefits. Disability management results were varied and not always significant. The integration of disability management responsibilities within the firm was important, but we did not find evidence that when outside contracts integrated disability management with other human resources services, claims rates were reduced. In our analysis of benefit cost per claim, employee gender and industry type were again significant in the same directions as in the claims rate analysis. Age and employer-provided benefits (STD, RET, INFSTD) were also strongly positive, reinforcing the idea that such benefits are complementary to the use of LTD benefits. On the other hand, the results for other benefit programs (SSDI and Workers' Compensation) and for the salary and benefit level variables show no substantial evidence that employee behavioral responses to economic incentives impact on the cost of the LTD claim. Indeed, the negative impact on cost of continuing group medical coverage for employees on LTD is opposite to what a simple economic incentive hypothesis would predict, and may instead reflect the influence of continuity of health care in returning employees to the job. Finally, our measures of disability management activities were not consistently related to cost per claim, though some evidence of cost reductions was observed. The prudent reader will treat the results we have reported with caution for several reasons. First, in our analysis of claims cost, the information describing each injury is limited to the average cost for claims in the same ICD9 category (ICDCOST). This is obviously a relatively crude and incomplete measure of the severity of each injury. As a result, an important component of severity is omitted from our model, and this omitted component may be correlated with other explanatory variables included in our model, resulting in a bias in the estimated coefficients. Second, in our analysis of claims rates, the essentially cross-sectional nature of our data set is a potential limitation. Many of our explanatory variables, such as the disability management efforts and the fringe benefit offerings, are chosen by the employers and thus are potentially correlated with omitted employer or firm-level characteristics. Fixed effects estimation, which mitigates this problem, would require analysis of a larger pooled data set covering a larger number of years for each employer and variations over time in the explanatory variables of interest. A further limitation is the relatively small size of our sample of employer respondents in relation to the number of potentially relevant explanatory variables. This limitation is a particular concern in that the outcomes we are studying (i.e., LTD injury claims) are relatively infrequent and, as a result, a number of employers do not experience claims in at least one of our study years. The small size of our study sample is also a result of a low survey response rate, which is typical of employer mail surveys Dillman, 1978 and Jensen & Gabel, 1992. While we do not find evidence of response bias (see Appendix B), it remains a potential concern and future studies need to address this concern by increasing the response rate through additional telephone and in-person follow-up contacts. Finally, the analysis reported here is the first to examine claims rates and benefit costs for LTD injury claims. Thus, we cannot compare our results with previous studies or use evidence from previous studies to reinforce our major findings. Confirmation through further studies is clearly needed before the findings reported here could be regarded as definitive.