Shortly after the Federal Motor Carrier Safety Administration (FMCSA) put into effect changes to the CSA’s Safety Measurement System last week, the American Trucking Associations (ATA) released a white paper explaining that “the CSA system lacks sufficient data on the majority of the industry to render meaningful scores for most motor carriers.” 1
According to the ATA, the FMCSA “has sufficient violation data to assess 40% of active carriers in at least one category but only enough to assign a percentile rank or score in at least one category to 12% of active carriers,” which the FMCSA claims is non-problematic since ‘those carriers are involved in 83% of the crashes.’” 1
But the ATA emphasizes that many accidents are not even reported to the FMCSA (less than 75% according to the University of Michigan Transportation Research Institute), leading to inaccurate CSA ratings.
The FMCSA believes that their statistics are accurate due to utilization of their State Safety Data Quality (SSDQ) system, which estimates the number of non-fatal accidents by looking into the number of fatal crashes, comparing these estimates with the number of actual reported non-fatal crashes, giving those states a rating of “good,” “fair,” or “poor.” 2
Although the FMCSA’s findings note that the majority of states are classified as being “good” when indicating their crash numbers, the ATA points out three main flaws that lead these results to be considered inaccurate.
*The FMCSA model of determining whether or not a state is doing an appropriate job in reporting non-fatal accidents is too broad, giving the state a wide threshold to report numbers in. For instance, “Texas needed to transmit between 12,551 and 25,163 accidents, a range of 12,612 accidents,” a number that far exceeds accuracy. 3
*Although the number of reported non-fatal accidents may only slightly vary in small states year after year, the FMCSA’s threshold varies depending on the number of fatal accidents in that state. As the ATA noted, “Alaska experienced an annualized fatal accident count of between 1 and 3 for each month in 2012. When the annualized count was 3, the model expected between 58 and 136 non-fatal, reportable accidents, a range of 78. When the annualized count was 1, the model predicted between 17 and 47 accidents, a range of 30.” 2 The ATA concluded that the number of fatal accidents is not a reliable indicator in determining the number of non-fatal accidents in a state.
*Finally, although the FMCSA claims that most states do a “good” job in reporting crash statistics, which the agency considers as evidence that their ratings are accurate; the previous explanations explain these numbers as unreliable.
The ATA concluded by stating that “Thorough examination of state police accidents reports (PAR) remains the only way to accurately measure State accident reporting to MCMIS.” 2
Read the ATA’s White Paper: Underreporting of Commercial Motor Vehicle Crashes: Assessment of FMCSA’s Crash Reporting Measurement Tool December 2102 at
Do you agree with the ATA that the FMCSA’s CSA system contains insufficient data? What do you find to be the consequences of having this “inaccurate” in the system?