Wells Fargo Senior Analyst Anthony Gallo published an excellent document on March 28 highlighting critical problems with CSA 2010. You can contract me or contact Anthony directly at anthony.gallo@wellsfargo.com for the full document, but I want to highlight a couple of key points. The analysis shows little or no correlation between the CSA BASIC scores for unsafe driving and fatigued driving and accidents per million miles. Everyone wants to see continued reductions in truck-related accidents. It appears though, that the key CSA measures that would indicate a carrier should be a candidate for further scrutiny, bear no relationship to the actual safety ratings of those carriers. This is one of numerous reasons that shippers should not be using CSA data to create their own carrier credentialing methodologies. Here is an excerpt from the Wells Fargo publication.

FMCSA assigns a composite score in seven Behavior Analysis and Safety Improvement Categories (BASICs) along with a corresponding "limit" threshold. When a carrier breaches the threshold corrective actions must be taken to remedy the violations. Serious or persistent violations can result in enforcement actions against either the driver or carrier. Motor carriers initially receive a "warning letter" informing them of their violations. According to several reports as many as 50,000 letters were expected to be sent in Hi 2011.

When we dissected the underlying data in relation to the composite scores, we were somewhat surprised to see that there was little correlation between the scores and the actual number of accidents and injuries/fatalities. Specifically, we observed little correlation between poor scores in the Unsafe Drivers and Fatigued Drivers categories and the actual number of accidents or injuries/fatalities. Further, we could not find and are not aware of FMCSA data that statistically demonstrates a correlation behind these factors and accidents. Certainly it is intuitive that "unsafe drivers" and "fatigued drivers" should be more prone to accidents. However, this does not appear to be the case. We feel this may be due, in part, to the methodology by which the categories are scored. For example, unfavorable Fatigued Drivers scores typically result from Hours of Service violations, but also include Logbook violations and errors. So while a fatigued/tired driver (assumed to be tired if outside of HOS criteria) may indeed be more likely to be involved in an accident, one with Logbook errors would perhaps not be. Thus, if an unfavorable Fatigued Driver score is in fact being driven by Logbook errors, the composite score would be less likely to correlate with accidents, in our view.

We understand the Unsafe Drivers/Driving category is meant to capture speeding, reckless driving, improper lane changes, etc. Certainly we think these behaviors should be captured and likely have some correlation with accidents. However, there are literally dozens of subcategories within the Unsafe Driving category, and a severity weight is given to each violation subcategory. This may help explain some of the disconnect between the scores and the actual number of accidents. For example, "reckless driving" and "texting while driving" both carry 10 point weights. Conversely, "failure to use a seatbelt" has a 7 weight whereas "improper lane change" and "following too close" each carry a 5 weight, although intuitively the latter two appear to be less safe activities, in our view.