|Table of Contents|
Both fatal and non-fatal motor vehicle crash injuries were estimated by state and industry. The state fatality estimation used two data sets: NHTSA’s 19982000 FARS and the U.S. Bureau of Labor Statistics’ 2000 Census of Fatal Occupational Injuries (CFOI).
Occupational fatalities by state generally are the CFOI motor vehicle fatality count. To reduce the effects of random variation with fewer small sample sizes, if the CFOI state count was fewer than nine, the 19982000 average of occupational motor vehicle traffic fatalities from FARS was used instead. This resulted in an estimated 2114 U.S. on-the-job motor vehicle fatalities. Off-the-job motor vehicle-related fatalities were computed by subtracting the state occupational highway fatality estimates from the 19982000 FARS state totals (all ages). The calculations assumed that all people under age 65 are workers or dependents.
The number of non-fatal occupational motor vehicle injuries was computed using four data sources: the 19982000 FARS, the 2000 police-reported state non-fatal injury counts (Blincoe et al, 2002), the 2000 CFOI, and the National Health Interview Survey (NHIS) 19871992. (The NHIS is a nationwide sample of civilian households. It includes information on injuries, whether they were on-the-job, and where they occurred.)
The number of injured on-the-job motor vehicle crash survivors by state was computed in four stages.
To distribute the injured survivors of on-the-job crashes by industry, the SOII distribution of survivors of lost-workday occupational injuries by two-digit Standard Industrial Classification Code was used. The SOII excludes medically treated survivors without workdays lost, whom the NHIS estimates are 41.5 percent of the total. It also does not cover all workers. Notably, it excludes government workers and self-employed truck and taxi drivers. (To estimate injury survivors in the government sector, the number of survivors per motor vehicle fatality for the service sector was multiplied by the CFOI motor vehicle fatality count for government employees.) Beyond its under-coverage problem, the SOII appears to under-count motor vehicle crash injuries. It records only 50,336 of the estimated 195,000 injured survivors of on-the-job crashes.
This update relies on CFOI data to calculate on-the-job motor vehicle fatalities. In previous estimates, the CFOI was used in conjunction with the FARS to estimate on-the-job fatalities. The CFOI has increasingly been accepted as a reliable source of fatality counts. Relying on the CFOI data results in a decrease in the estimated number of occupational fatalities (and the associated costs) from prior reports. The decrease does not indicate a trend in the number of fatalities.
Incidence of commercial vehicle crashes used to calculate cost per crash was estimated from aggregated data purchased from the Insurance Services Office (ISO). Com-mercial vehicle miles traveled were estimated from two sources. Total vehicle miles traveled in 1994 (2.75 billion) is from Traffic Safety Facts, 2000. The percentage of vehicle miles driven by commercial vehicles (6.7 percent) was calculated from the Nationwide Personal Transportation Survey, 1995. This percentage was multiplied by the 2000 total number of vehicle miles traveled for an estimate of 184 billion commercial vehicle miles traveled in 2000.
Medical, productivity, emergency services, property damage, legal, and non-liability insurance claims processing costs were estimated with SOII occupational injury survivor counts by vehicle type occupied or pedestrian status and costs per crash victim by vehicle type occupied or pedestrian status from Zaloshnja et al. (2003). The costs then were distributed into more detailed categories with the distribution in Miller (1992). Other costs per case and costs in Table 4 are from Miller (1992). These costs were inflated to 2000 dollars using inflators (medical spending per capita, employment cost index, and consumer price index all items) calculated from the 2002 Economic Report of the President. Employer crash costs were adjusted to specific states using ratios of state to national costs. The medical and composite state price adjusters were calculated from the ACCRA Cost of Living Index. The wage adjuster was calculated from estimates of personal income per capita by state in the 2001 Statistical Abstract of the United States. Costs per employee in Table 8 were calculated using the number of employees by state from the 2001 Statistical Abstract of the United States.
Total employer health fringe benefit costs (Figure 1) were computed following the methods in Miller (1992). Sources of data were as follows: sick leave, health and other insurance, and Workers’ Compensa-tion per hour worked from Table 701, 2000 Statistical Abstract of the United States; number of workers and average weekly hours worked from Table 580, 2001 Statistical Abstract of the United States.
Costs in Table 10 were assigned to non-fatal injuries by vehicle type. These costs were determined in an analysis of highway crash costs (Miller et al., 1996). Fatal crash costs were assigned on a per case basis. The number of employees by industry came from the 2000 Survey of Occupational Illnesses and Injuries (SOII). The wage-risk premium was calculated as an average per worker across all industries. Wage-risk premiums are implicit costs to employers to compensate workers for increased risk levels. Since the premiums are implicit and not a direct monetary cost, they were separated from the other costs. Due to lack of information, costs related to crash-caused traffic delays of commercial vehicles or employees not involved in crashes were not included in the estimate of the total cost of highway crashes to employers.
Employer costs of safety restraint non-use and alcohol-involved motor vehicle crashes were estimated from a combined file from NHTSA’s 1999 FARS, NHTSA’s 1999 Crashworthiness Data System (CDS), and NHTSA’s 1999 General Estimates System (GES). Costs per victim were merged into the file following Zaloshnja et al. (2003). The GES file was adjusted for under-reporting of alcohol involvement by police through the method described in Blincoe et al. (2002). The costed file was used to estimate employer costs of alcohol-involved injuries and the portion of employer costs per unrestrained victim that can be attributed to restraint non-use. The latter was estimated as a difference between the actual total cost and the hypothetical cost of crashes in the case that all vehicle occupants were restrained. Property damage was kept constant because it is not affected by restraint use. Given that the information on the commercial nature of the vehicle is not available for all vehicles in CDS and GES, a probability model constructed from 199199 FARS was used to estimate the probability of a non-fatal injury being work-related in cases when it could not be directly determined. Employer costs of alcohol-involved injuries by state (Table 9) were estimated by assuming that the ratio between employer and societal costs remains the same across the states. Societal costs of alcohol-involved injuries by state were found in Taylor et al. (2002).