Review of U.S. Department of Transportation Truck Size and Weight Study - Second Report: Review of USDOT Technical Reports (2015)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

37 5. MODAL SHIFT MAP-21 requires the USDOT study to estimate the following [Section 32801(a)(6)]: (A) the extent to which freight would likely be diverted from other surface transportation modes to principal arterial routes and National Highway System intermodal connectors if alternative truck configuration is allowed to operate and the effect that any such diversion would have on other modes of transportation; (B) the effect that any such diversion would have on public safety, infrastructure, cost responsibilities, fuel efficiency, freight transportation costs, and the environment; (C) the effect on the transportation network of the United States that allowing alternative truck configuration to operate would have; and (D) whether allowing alternative truck configuration to operate would result in an increase or decrease in the total number of trucks operating on principal arterial routes and National Highway System intermodal connectors. The modal shift technical report describes the estimates, for each alternative configuration scenario, of diversion of freight from railroads to trucks and the change in annual truck VMT by configuration, highway system, and operating weight that would occur as a consequence of the alternative trucks’ greater capacities. The estimates are of VMT in a single year, with respect to a base case defined as 2011 actual traffic volumes, and assume that freight patterns have reached equilibrium at some time after introduction of the alternative configurations. The report also presents single-year estimates of changes in shippers’ transportation and nontransportation logistics costs, railroad revenue, traffic congestion costs, pollutant emissions, and energy consumption (Modal Shift, ES-9–ES-11, 43).

38 Responsiveness to the Questions Identified by Congress The USDOT report presents estimates of the quantities required in the congressional charge: diversion of freight from other modes and effects on the other modes (i.e., revenue loss) caused by allowing the alternative trucks, and effects of allowing the alternative trucks on freight transportation costs, truck traffic volume, fuel efficiency, and the environment. However, the method used to estimate the diversion of freight from rail to truck and from preexisting to alternative configuration types is of questionable validity. Shortcomings in the mode choice analysis have important implications, because the estimates of changes in truck travel by configuration and weight determine the magnitudes of the infrastructure, safety, and enforcement cost estimates in the USDOT study. The estimates of effects of changes in truck travel on fuel efficiency and the environment are produced by simple but practical methods that are appropriate for the study, although the results will be inaccurate if the mode choice estimates are inaccurate. Methods and Data Application of the Intermodal Transportation and Inventory Cost Model Freight mode choice models can be classified into two families: disaggregate (including discrete choice models and supply chain–based mode choice models) and aggregate (including market share models, elasticity-based models, and economic cost models). Disaggregate models estimate the preferred mode for each shipment of each establishment, while aggregate models directly estimate the market share for each mode. Disaggregate models can realistically represent the underlying decision-making behavior of shippers. If aggregate estimates of market shares are needed, as in the USDOT truck size and weight study, the estimates produced by these models must be aggregated into market shares. This aggregation requires data on shipment characteristics from a suitable sample of establishments. The quality of the

39 market shares estimates depends on both the ability of the model to represent each shipper’s mode choice decisions and the quality of the disaggregate shipment characteristics data available. The USDOT study used the Intermodal Transportation and Inventory Cost (ITIC) model (Modal Shift, ES-6–ES-8), a typical example of supply chain–based mode choice models. ITIC is a disaggregate single-shipper model that computes the optimal (lowest total logistics cost) combination of shipment size and freight mode (choosing between rail and truck and among truck configurations). The model replicates the decisions made by shippers and receivers. ITIC needs data at the establishment level relevant to logistics costs, including inventory costs, amount of cargo to be transported from the shipper to each receiving location, and transportation cost. These shipper and shipment characteristics vary greatly with industry sector, geographic location, company size, and other factors. As a result, it is not possible to select “typical” values as input to ITIC that will provide meaningful market share projections. Among the required inputs to ITIC are data on shipper-to-receiver flows of freight. The only readily available database containing records of shipments by origin, destination, and shipment size is the Carload Waybill Sample, submitted by the railroad industry to USDOT, which contains data only on rail shipments. The Commodity Flow Survey, conducted by the Bureau of the Census, collects shipment size data for all modes, but this information is confidential. The survey does not record truck configuration and body type used for carrying specific commodities, information needed to project market penetration of alternative configurations. In the absence of a comprehensive database of establishment-level freight shipments by origin, destination, and shipment size, application of ITIC requires estimating the flows by means of a model or by assumption. The USDOT study generated the required freight flows by first estimating annual county- to-county flows by commodity and mode. These were disaggregated into shipments by assuming that shipment size equals the maximum payload of the vehicle carrying the freight and that all shipment origins and destinations are county centroids (Modal Shift, 9, 16). This procedure is equivalent to assuming that all shippers of a commodity in a county cooperate in consolidating their shipments and that all have identical preferences for level of service. The effect of these strong and arbitrary assumptions on

40 the accuracy of the ITIC mode share projections is unknown, and consequently the USDOT study is not a correct application of the model. 7, 8 The USDOT study assumes that total ton-miles of freight traffic are unchanged from the base case in the alternative configuration scenarios. In reality, a reduction in the cost of freight is likely to induce additional traffic. (For example, a producer may take advantage of lower transportation costs by consolidating production at fewer locations or by choosing a more distant source for a raw material, or a firm trading internationally may change the ports it uses for shipping its goods.) The desk scan cited research that concluded that the effect is difficult to estimate (Modal Shift, 91–94). However, because the effect of changing size and weight limits on truck traffic volume drives all other impacts, the USDOT study should have included an estimate, presented as a range of possible induced freight volumes, reflecting the range of estimates in the literature. As the USDOT report explains (Modal Shift, 171–172), in predicting shippers’ choices between two modes, the ITIC model assigns all freight to the mode that the model predicts to have the lower cost, regardless of the magnitude of the cost difference, whereas models based on observations of shipper behavior have found that, when costs appear close, both modes may retain a substantial share of traffic (presumably on account of cost-related factors not included in the models). A sensitivity analysis might reveal whether this simplifying assumption is likely to greatly affect the aggregate mode share projections of ITIC. Logistics Cost Estimates The estimated cost savings per vehicle mile of truck traffic reduction (Modal Shift, ES-10 and Table ES- 3) are strikingly large in Scenarios 1, 2, and 3 ($6.68, $4.71, and $4.58 per vehicle mile in Scenarios 1, 2, 7 For an example of correct use of a logistics cost model, see Leachman et al. (2005) and Leachman (2008). 8 The modal shift desk scan (Modal Shift, 94) cites a Government Accountability Office report using ITIC that expressed a similar concern about the uncertain effect of missing input data on the reliability of the estimates produced by the model (GAO 2011, 60).

41 and 3, respectively). The ratios for Scenarios 4, 5, and 6 ($0.79, $1.00, and $1.01, respectively) are consistent with past studies’ results. In the USDOT study analysis, shifts from rail to truck are small and no new freight is generated when limits are relaxed, so cost savings on the order of the reduction in truck VMT times the average operating cost per truck mile (roughly $2.00 for truckload general freight carriers) would be expected. ITIC estimates inventory and other logistics costs in addition to vehicle operating costs. As an example of a nontransportation cost savings, allowing heavier tractor-semitrailers might enable a shipper to avoid transferring cargo into trailers from arriving marine containers that are too heavy to carry on the highway under present weight limits. However, savings in nontransportation costs of the magnitude that the study’s total logistics cost estimates imply in the alternative configuration scenarios would be remarkable and should be explained in the report. Environmental Impacts The emissions impact analysis assumes that changes in emissions of carbon dioxide and oxides of nitrogen are proportional to changes in gallons of fuel consumed (Modal Shift, 52). As the USDOT report notes, this is a simplification for oxides of nitrogen emissions, which vary with the vehicle drive cycle. Apparently, no estimate of changes in emissions from trains was made. The lack of an estimate of changes in particulate matter emissions is a significant omission. The report projects reductions in traffic congestion in all the alternative configuration scenarios but does not estimate the effect of this reduction on fuel consumption and emissions of all motor vehicles. Treatment of Uncertainty The USDOT report presents point estimates for the modal shift analysis results. The concerns raised above about the accuracy of the ITIC projections in the study could have been alleviated through sensitivity analysis. In a sensitivity analysis, the model would be run repeatedly, with assumptions about

42 input values varied over plausible ranges (e.g., with alternative assumed shipment size distributions). The range of the resulting model estimates would provide an indication of the uncertainty of the estimates and reveal which input assumptions were most critical in view of the intended policy application of the estimates. Recommendations USDOT should undertake a significant effort to improve understanding of the behavioral factors that influence freight demand and freight mode choice, the analytical techniques used to depict freight markets as they are affected by public policy, and the data required for analysis and modeling of freight markets. Research should be performed to develop and test the three methods of predicting mode choice described in the desk scan (Modal Shift, 96–99): disaggregate models, aggregate econometric models, and expert opinion. The Commodity Flow Survey, in its current form, will not be able to provide the data needed to estimate mode choice in a future truck size and weight study because it does not collect data about vehicle choice. USDOT, together with the Bureau of the Census, should resume the conduct of the Vehicle Inventory and Use Survey, which was discontinued in 2002. The survey was the only source of systematic data on truck shipment sizes, truck cargo capacities, and the truck configurations and body types suitable for carrying specific commodities. Data on these truck transportation characteristics are needed for credible projections of truck travel. In addition, qualitative surveys that capture shipper, carrier, and receiver behaviors and decision-making processes are much needed to improve projections. References Abbreviation GAO Government Accountability Office

43 GAO. 2011. Intercity Passenger and Freight Rail: Better Data and Communication of Uncertainties Can Help Decision Makers Understand Benefits and Trade-Offs of Programs and Policies. GAO-11-290, Feb. Leachman, R. C. 2008. Port and Modal Allocation of Waterborne Containerized Imports from Asia to the United States. Transportation Research Part E: Logistics and Transportation Review, Vol. 44, No. 2, pp. 313–331. 10.1016/j.tre.2007.07.008. Leachman, R. C., T. Prince, T. R. Brown, and G. R. Fetty. 2005. Port and Modal Elasticity Study. http://narc.org/uploads/File/Transportation/Library/MemberDocs/SCAG_Elasticity.pdf.