7

Key Conclusions, Recommendations, and Ideas for Future Research

The goal of this study, as defined in the task statement, is to examine the effects of compensation methods and working conditions of long-distance drivers in the for-hire trucking and intercity passenger bus industries on the retention and safety performance of drivers in each industry. If there is good reason, based on a review of the empirical evidence and any identified limitations of that evidence, the report is expected to contain recommendations for data-gathering, analytic methods, and research designs that may be helpful for furthering understanding of the effects. This chapter, therefore, distills the report’s key conclusions from the report’s findings and then presents two recommendations about next steps for research and data-gathering.

Having focused on understanding the effects of driver compensation and working conditions but not on the many other factors that can have a bearing on safety, the study committee cannot know how the Federal Motor Carrier Safety Administration (FMCSA) should prioritize its data collection and safety research. Nevertheless, the study committee, informed by its findings, is expected to advise on whether further actions by the agency to strengthen this understanding would be desirable, potentially to include recommendations on appropriate data-gathering, analytic methods, and research designs for this purpose. The study’s central conclusions, distilled from the study findings, are therefore presented next, followed by recommendations for FMCSA research and data-gathering.

The recommendations are offered cognizant of this study’s narrow focus and limited view into FMCSA’s other data collection and research priorities. The emphasis, therefore, is on looking for potentially cost-effective

opportunities to leverage future research and data collection, including work initiated for other reasons, in ways that can inform studies of driver compensation and working conditions and their potential for safety effects. Consistent with this advice, some ideas for future data gathering and studies are provided following the recommendations, including thoughts on their practicality.

KEY CONCLUSIONS

The available data and empirical research are insufficient to determine whether driver compensation methods (and their implications for pay levels and regularity) and working conditions affect the driving behavior and safety performance of drivers in the long-distance truckload (TL) sector. While the proprietary nature of carrier compensation offerings complicates data collection for this research, a more fundamental challenge is that driver compensation methods (i.e., variants of piece-rate pay) and basic work requirements and conditions (e.g., irregular schedules, varied routings, extended time away from home) are largely uniform across the TL sector so as to severely limit the ability to compare the effects of alternative compensation methods (i.e., hourly and other non-piece-rate pay). Moreover, even if hourly pay and other non-piece-rate compensation became more prevalent in long-distance trucking, it is likely that comparisons of drivers paid by the hour with drivers paid by piece rate would be confounded by differences—many unobservable—in driver characteristics and work requirements.

Despite the nearly universal use of variants of piece-rate pay and considerable uniformity in working conditions among carriers in the long-distance TL sector, there is observed variability in carrier-level safety performance, including in rates of carrier compliance with traffic and motor carrier safety laws and regulations. Understanding the causes of this variability might be helpful for designing interventions to improve safety performance; however, even if firm causal relationships cannot be established, research to understand associations among carrier-level characteristics and safety performance may be helpful for informing FMCSA’s safety monitoring and enforcement priorities and reveal starting points for further research on causes.

Long-distance TL carriers experience universally and persistently high rates of driver turnover, which is consistent with the highly competitive nature of this trucking sector, as carriers are compelled to drive down their costs and search widely for paying loads to secure sufficient business. The sector’s observed uniformity of driver compensation methods

(largely variants of piece-rate pay) and working conditions (characterized by dispatching that leads to irregular work schedules) suggests that efforts by a TL carrier to reduce driver turnover by varying too far from this uniformity, such as by introducing alternative pay and driver dispatching methods, will reduce its ability to compete.

Drivers of intercity passenger buses have consistently strong safety records as a group and are usually paid by the hour, which creates analytic challenges for comparing the effects of different pay methods. However, the role of driver compensation and working conditions in explaining this performance has not been a matter of notable concern or inquiry and would require data that are not currently available to assess. Furthermore, intercity bus driver job requirements and working conditions differ far too much from the requirements and conditions of long-distance truck drivers to allow for meaningful comparisons between the two groups.

RECOMMENDATIONS

Based on these study findings and central conclusions, the committee offers the following two recommendations:

Recommendation 1: The Federal Motor Carriers Safety Administration should explore opportunities for leveraging research and data collection that may be planned and programmed for other purposes to help regulators, researchers, and industry examine the potential effects of driver compensation and working conditions on the safe driving behavior and performance of long-distance truck drivers. The focus should be on exploiting new data collection efforts, resourcefully and creatively tapping or enhancing existing sources of data, and supporting discrete, smaller-scale studies for identifying patterns and associations and possibly understanding causal relationships.

Recommendation 2: The Federal Motor Carrier Safety Administration should support further research to obtain a better understanding of why safety performance varies among truckload carriers despite uniformity in compensation methods and working conditions, ideally to make progress in identifying and understanding the underlying causes of these differences but also to detect patterns and associations that may inform safety monitoring and enforcement strategies.

IDEAS FOR FUTURE RESEARCH

This study’s charge specifies that its report “will contain recommendations to the sponsor and potentially to Congress on a research agenda that outlines the kinds of analytic methods, data gathering, and study designs that would be helpful for expanding and strengthening” understanding about “the effects of compensation methods and other relevant factors on driver retention and safety performance.”

For a variety of reasons this part of the study charge proved difficult to satisfy. As discussed in this report, there are theoretical reasons why long-distance truck drivers who are paid by the hour would have less incentive to engage in riskier driving behaviors than long-distance truck drivers who are paid by the mile or by other forms of piece-rate pay, but the theories cannot be tested adequately with available data for reasons explained in Chapter 5.

Looking forward to future possible research to better understand how compensation methods may affect safety, the sections that follow summarize the problems encountered and lessons learned from past research and suggest possible ways forward. The suggestions are for “possible” ways forward because the chain of cause and effect is complex with multiple confounding influences. The causes of any given crash involving a heavy truck are usually numerous and interacting. They include the many proximate causes of a crash, including the vehicle’s condition (brakes, tires), road and traffic conditions, and the actions of drivers of other vehicles. They may also include less obvious, and often unrecorded, factors such as the truck driver’s skill level and experience, the vehicle’s maintenance schedule, and possibly the effects of driver compensation on driving behavior. Many challenges can arise in trying to account for all significant influences, and efforts to measure them using available data—or without costly efforts to obtain the needed data—may not be possible.

Recognizing these challenges, the study committee offers several suggestions for research and data gathering that FMCSA might pursue incrementally to obtain a better understanding of how compensation methods can influence truck driver safety performance. They are characterized as ideas for further exploration rather than recommendations for specific actions considering this study’s inability to identify compelling reasons to suspect that driver compensation is sufficiently powerful as a safety-related explanatory variable such that it warrants specialized and prioritized data gathering. As a general matter, the suggestions imply that FMCSA could start with small efforts to test ideas and methods and, when promising, to follow up with more rigorously designed studies. Ideas for driver-level studies are offered first, followed by ideas for carrier-level research.

Driver-Level Studies

Grouped below are three fundamental challenges associated with conducting driver-level (as well as carrier-level) studies that have been encountered in the past with ideas on possible ways to address them.

Causal Mechanism

Chapter 5 (see especially Figure 5-3) postulates that if compensation method affects safety, it must do so through some influence on driver behavior, while recognizing that there are multiple influences on truck driver behavior that are unrelated to compensation and multiple influences on safety unrelated to truck driver behavior. Holding those complexities aside for the moment, we must ask what the mechanism is through which compensation methods, level, or regularity would affect driver behavior. Chapter 5 organizes its review of observational studies with the expectation that compensation might affect (a) driver fatigue (and thereby affect safety) and (b) driver behavior independent of fatigue through incentives to speed or engage in other risky driving behaviors. The studies reviewed turned out to be unpersuasive for all the reasons described in the chapter, but they may also suggest possible paths forward.

The review of several naturalistic driving studies using highly instrumented vehicles suggests that driver errors are compounded by long daily work hours, implying that fatigue may be affecting behavior and safety, but these studies did not attempt to account for how compensation might contribute to fatigue. Regarding future research, the technology already exists to measure driver behavior in real time, and several large fleets are using such technology to measure and monitor driver behavior. Use of this technology might be able to shed greater light on the proximate determinants of safety critical events and crashes and, therefore, some of the causal mechanisms affecting safety, including mechanisms that might involves influences of compensation.

Data

A second fundamental problem in the compensation-safety literature begins with lack of data at the individual level about the income, benefits, and bonuses drivers receive, as described in Chapter 3. The list of missing variables extends to accounting for driver skill and experience, physical condition at the time of the crash as affected by fatigue or medical conditions, carrier-influenced motivations affecting driving behavior, contributing roles of local conditions at the time of the crash, and so forth. Efforts to obtain more complete data through driver surveys or from company records of starting

per-mile rates for drivers from a sample of large firms have still not obtained sufficiently rich data to disentangle the many confounding influences on safety performance and provide credible evidence, as discussed in Chapter 5.

Method

The ideal method for approaching the questions posed in the study charge would be to conduct a randomized controlled trial (RCT). This method is the best one for shedding light on cause and effect, but no such experiment has not been. Randomly assigning drivers that are otherwise equivalent into groups paid by the mile or the hour under the same set of working conditions has not been feasible.

An alternative approach has been to rely on a case control method, although this method has fallen short of allowing for credible causal analysis. FMCSA has funded studies in the past that have applied the case control method to large numbers of drivers at individual large firms that agreed to participate. For example, the Dunn et al. (2020) study cited in Chapter 5 on commercial driver safety risk factors (CDSRFs) focused on health and demographic indicators of crash risk for 20,000 drivers at a single large truckload firm. Moreover, the CDSRF study included a modest, supplemental effort to gather information about compensation on a small subsample of crash-involved drivers and matched drivers not involved in crashes who had consented to participate in a follow-up study. The researchers surveyed these drivers about their methods of compensation and then, for each crash-involved driver, found five matches of drivers who were active drivers at the time of the crash but not involved in crashes. Drivers were matched solely on years of experience and being active at the time of the crashes, only two of undoubtedly many relevant characteristics.

Because the CDSRF study did not analyze these data, Guo and Xiang (2024), in response to a request from this study, analyzed the data. They found associations between compensation method and crash risk, with differences in odds ratios of sufficient magnitude to suggest that a future study with a stronger design might be fruitful. In this preliminary exploration of whether the effects of compensation methods could be evaluated as part of a study designed for another purpose, the results are compromised by flaws in the design of the compensation questions used in the follow-up survey in the original study, the inability to control for the type of driving that drivers were engaged in, inadequate matching of crash-involved drivers with drivers not involved in crashes, and low response rates and other sources of sample attrition among both crash-involved and matched drivers. Nonetheless, if these shortcomings could be overcome in a carefully designed and well-resourced and executed study, it might be possible to add compensation questions to future large case-control studies that

FMCSA intends to pursue for other reasons. Designing an “add on” to a study FMCSA wants to pursue for other reasons would keep the cost of learning about the compensation-safety relationship reasonable, especially when considering that the complexity of causal mechanisms involved may preclude achieving a clear result.

Broad Considerations for Future Driver-Level Studies

The suggested driver-level approach that follows is premised on the willingness of carriers and drivers to cooperate and share data that would be strictly protected. The suggestions could yield more robust results if causal mechanisms could be observed with instrumented vehicles.

Although an RCT is the gold standard for supporting causal analysis, the committee cannot imagine a feasible and ethical application of such a method for examining the effects of compensation on safety. As noted, however, FMCSA has funded large-scale case-control studies in the past. The effort to evaluate the compensation-safety relationship as part of the CDSRF study described above suggests that it might be feasible as an add-on to a study that FMCSA would undertake for other reasons, and it could yield some insight if better designed and executed. Whether a case control or alternative observational approach is taken, what is critical is the rigorous design and execution of research that adopts the kinds of methods described by an earlier study of driver fatigue and health (NASEM 2016). In a case control study building on the CDRSF supplemental analysis, it would be important for a future study to not only obtain better data on compensation and driving conditions but also limit sample attrition (via nonresponse or other mechanisms) and collect sufficiently rich data to allow for high-quality matching of cases (using, for example, propensity score methods). Otherwise, the validity of the causal analysis would not be convincing, and the findings would not be credible.

Carrier-Level Studies

As explained in Chapter 5, several studies have reported that larger carriers on average have better safety records than smaller ones on average despite the prevalence of piece-rate compensation across all carriers and higher turnover among larger carriers. The reasons why are not understood. Accordingly, this report’s suggestion is to conduct case studies with companies to explore whether there are replicable policies, programs, and practices, including compensation methods, that carriers have employed to improve safety.

An initial set of case studies has been done by Camden et al. (2022) of 9 carriers with 50 or more power units that achieved notable improvements

in safety. Among the policies, programs, and practices that the carriers employed in attempting to improve safety are recruitment, hiring, and compensation practices to attract and retain safe drivers; more rigorous driver training than the norm; company-wide emphasis on safety culture; driver-supportive dispatching; application of advanced vehicle safety technologies; and systematic vehicle maintenance strategies. Conduct of additional case studies of companies with exemplary safety records, including companies of all sizes, would possibly reveal common efforts that seem to have been successful. It would also be helpful to conduct case studies of companies that have not experienced improved safety to identify how their policies and practices differ from those of the carriers with better safety records. Although case studies are qualitative by nature, they could suggest additional studies to gain more insights about what works. If case studies identify specific, replicable practices that seem promising, it would be important to design and conduct further studies to explore whether such practices really do improve safety and could be broadly adopted.

REFERENCES