Impacts of Public Policy on the Freight Transportation System (2011)

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.

81 For a subset of the policies considered as part of this re- search, the research team performed a more in-depth explo- ration of freight system impacts. In some cases, this involved original analyses; in others, the research team merely syn- thesized and reported on impacts analyzed by others. This appendix presents this information for the following five pol- icy examples, all of which were introduced in Section 4. • Truck Speed Limits and Governors • Truck Size and Weight Rules • Inland Waterway Infrastructure Investment • Highway Tolls and Other User Charges • Lockage Fees for Inland Waterways Truck Speed Limits and Governor Rules Efforts to rein in the top speeds traveled by heavy trucks on U.S. highways have taken two approaches: differential speed limits and truck speed governors. The use of differential speed limits has been driven largely by concern for public safety, whereas the use of speed governors on trucks has been moti- vated by both public safety and an interest in achieving better fuel economy. The two approaches have similar impacts in that they result in trucks traveling more slowly and usually at speeds lower than those of the cars around them. Safety Impacts The safety impact of creating differential speed limits for cars and trucks (either through lower posted speed limits for trucks or the use of speed governors) has been the subject of much debate among researchers and policymakers. Research clearly finds that lower vehicle speeds reduce the severity of crashes and the incidence of fatalities. This is because impact force during a vehicle crash varies with the square of the ve- hicle speed.1 Lower speeds also improve truck braking dis- tances, which helps truck drivers avoid accidents. However, there is also a relatively strong consensus among researchers and practitioners that a higher variance in vehicle speeds (i.e., speed differential) increases the risk of accidents by increas- ing the number of vehicle interactions.2 There is no clear con- sensus as to whether the net impact of these factors is positive or negative. Analysis of crash data has provided mixed evidence on the safety impacts of lower speed limits for trucks. In a 1991 report, NHTSA found that more than 90 percent of combination-unit truck crashes and 95 percent of single-unit truck crashes oc- curred on roadways where the speed limit was less than 65 mph and where the incidence of truck speeding in excess of 65 mph was low.3 This analysis, although dated, suggests that speed governors could help prevent only a small portion of truck crashes. In the United Kingdom, all large combination trucks were speed limited after 1993. Between 1993 and 2005, the accident involvement rate for this vehicle class fell from 40 to 30 per hundred million vehicle-kilometers. During the same period, the accident involvement rate for all heavy goods vehicles increased slightly from 18.5 to 18.8 per hundred million vehicle-kilometers.4 Although this data does not isolate the effect of mandatory speed governors, it supports the hypoth- esis that they improve highway safety. TRB’s 2008 synthesis found a lack of relevant published research on how speed governors affect safety and instead A P P E N D I X B Details on Impacts of Selected Policies 1TRB. CTBSSP Synthesis 16: Safety Impacts of Speed Limiter Device Installations on Commercial Trucks and Buses. 2008. 2Johnson, Steven L. and Naveen Pawar. Cost-Benefit Evaluation of Large Truck- Automobile Speed Limit Differentials on Rural Interstate Highways. For the Research and Special Programs Administration, U.S. DOT. November 2005. 3NHTSA. Commercial Motor Vehicle Speed Control Safety. Report # DOT-HS- 807-725. May 10, 1991, p. ES-1. 4TRB. CTBSSP Synthesis 16: Safety Impacts of Speed Limiter Device Installations on Commercial Trucks and Buses. 2008.

82 conducted a small survey of fleet safety managers. For those fleets using speed governors, safety was selected as the primary consideration for selecting the maximum speed, followed by fuel economy. Roughly 56 percent of those surveyed said that the use of speed governors had reduced the frequency of crashes; 27 percent indicated that speed governors had had no impact, and 14 percent said that they could not deter- mine whether governors had had any impact. This response was not as conclusive as the responses fleet managers provided regarding the impact of speed governors on fuel economy and number of speeding violations. Operational Impacts By limiting the top speed at which trucks travel, speed gov- ernors can affect many aspects of a carrier’s operations. For example, a lower maximum speed improves fuel economy and likely reduces truck maintenance costs. At the same time, however, a lower maximum speed can result in trucks travel- ing fewer miles per day, which can affect revenues and labor costs. This section explores the impact of truck speeds on var- ious aspects of a carrier’s operations. Vehicle Modification Costs Obviously, trucks do not need any new equipment to com- ply with posted speed limits. However, to comply with a speed governor mandate, owners of late-model trucks (mid-1990s or later) would, at a minimum, need to access the engine’s elec- tronic control module and change its maximum speed setting. Fleet maintenance personnel would be able to do so with the correct electronic service tool. In a submission to a NHTSA rulemaking docket, the Truck Manufacturers Association (TMA) estimated the cost of this operation at $100 per truck. As shown in Table B-1, TMA also estimated that making it harder for the maximum speed setting to be changed by drivers or vehicle owners would increase the cost per truck. It would also require vehicle manufacturers to redesign and redeploy ECM software for approximately 40 different engine control systems. Hard wiring the ECM to make tampering even more difficult would entail the design of both new hard- ware as well as software. Trucks built from 1990 to around 1995 do not have the same type of programmable ECM as newer trucks. According to TMA, if these trucks were subject to a speed governor requirement, they would have to be out- fitted with a mechanical speed governor at a cost of $1,000 to $1,500 per truck.5 The USDOT has reported that in 2002 there were about 2.6 million Class 7 or 8 trucks in the U.S. fleet.6 If, as men- tioned previously, 75 percent of those trucks already have maximum speed settings in place, that would leave at least 0.6 million trucks without speed governors. In its submission to the NHTSA docket, TMA provided data on the maximum speeds set for a sample of truck purchasers in 2005. Of the vehicles sold with maximum speed limits that year, roughly 45 percent had maximum speeds set at 69 mph or higher.7 Assuming this proportion holds true for the entire fleet, an- other 0.9 million trucks would need to have their maximum speed limit adjusted downward if the Federal government were to set the maximum truck speed at 68 mph (as requested by ATA and the other petitioners). Therefore, at a minimum, a speed governor mandate could cost $150 million (1.5 mil- lion trucks at $100 per truck). This cost could be reduced by grandfathering older vehi- cles. For example, the nine large U.S. carriers that petitioned the Federal government for a mandatory speed governor rule requested that the rule apply to trucks of model year 1990 or newer. In its petition, ATA requested that the rule apply only to new trucks, which would eliminate the need to retrofit the existing fleet. Table B-1. Estimated cost of vehicle retrofits for mandatory speed governors. Cost Per Truck Non-Recurring Vehicle Manufacturer Cost Use existing electronic control m odule (ECM) $100 $0 Modify and deploy ECM software to ma ke maxim um road speed a factory password-protected feature $300 $100 mil lion Replace ECMs with ones that are “hard wired” to prevent tampering with maximu m road speed $2,000 $400 mil lion Install m echanical speed governors on older trucks $1,000 - $1,500 $0 Source: TMA. 5Robert Clarke, President, Vehicle Manufacturers Association, submission to Docket No. NHTSA-2007-26851, March 27, 2007. 6U.S. DOT, BTS, National Transportation Statistics, Table 1-21: Number of Trucks by Weight, http://www.bts.gov/publications/national_transportation_ statistics/html/table_01_21.html 7Robert Clarke, President, Vehicle Manufacturers Association, submission to Docket No. NHTSA-2007-26851, March 27, 2007.

83 Effect on Fuel Efficiency There is consensus that reducing the top speed of trucks im- proves fuel economy, but the estimated amount of savings varies. The research team used EPA’s Physical Emission Rate Estimator (PERE) model to estimate fuel consumption rates of motor vehicles under different driving conditions. Modeling of a “typical” tractor-trailer produced a fuel-efficiency penalty of about 0.08 mpg for every mph increase above 55 mph. Johnson and Pawar estimated that each mile-per-hour increase in speed beyond 55 mph decreases fuel efficiency by 0.03 to 0.08 mpg, depending on the type of road and the speed vari- ance of traffic flow.8 ATA’s 1996 estimate of the fuel efficiency penalty of higher speeds was slightly higher. ATA estimated a penalty of 0.10 to 0.14 mpg for each mile-per-hour increase in speed beyond 55 mph.9 Another study estimated that reducing the freeway driving speed of a typical long-haul combination truck from 70 mph to 65 mph would reduce fuel use per truck by 972 gallons per year, a 6 percent savings.10 Effect on Equipment Maintenance Costs A 1996 ATA publication indicated that operating equip- ment above 55 mph generally decreases component service life and shortens preventive maintenance intervals.11 How- ever, in 2005, Johnson and Pawar found no support for more frequent maintenance intervals. Regarding tires, Johnson and Pawar found no objective research data related to the effect of speed on tire wear at the speeds commonly traveled on rural interstates. However, the majority of maintenance managers surveyed by the authors said that tire wear increases beyond a 65 mph operating speed. Several studies also attributed lower brake maintenance costs to lower maximum speeds. Insurance Costs Studies found that when setting premiums, insurers do not offer “front-end” premium discounts to carriers using speed governors. Rather, insurers look only at a company’s experi- ence ratings.12 Instead, insurers will reduce premiums for car- riers with the best demonstrated safety records. Therefore, it is possible that speed governors eventually generate insurance savings for carriers that use them, but there is no data avail- able to estimate the amount of savings. Driver Retention The common perception is that truck drivers are strongly opposed to the mandatory use of speed governors, especially when they are paid by the mile or by the trip. A 2007 OOIDA survey of 3,400 members who are company drivers found that 82 percent would prefer to work for a company that does not use speed governors, all other things being equal. Only 4 percent said they would choose a company using speed governors.13 In contrast, in a 2008 survey of fleet safety managers, 64 per- cent said that driver attitudes toward speed limiters were largely neutral, and 23 percent said driver attitudes were pos- itive. Seventy-seven percent of the managers said that the impact of speed limiters on driver hiring and retention was neutral.14 It may be that driver opposition is softening as the voluntary use of speed governors becomes more widespread among carriers. Driver Compensation Because some drivers are paid by the number of miles they drive, reducing the maximum speed of trucks could result in loss of income for those drivers. The amount of reduction would depend on what percentages of miles were driven at speeds exceeding the new limit. Drivers who log many miles in western states where speed limits are higher would stand to lose more income than other drivers. In its rulemaking petition to NHTSA and FMCSA, ATA suggested that, because of the chronic shortage of long-haul drivers, carriers would need to compensate drivers for any loss of income.15 However, there is very little in the literature about how the voluntary adoption of speed governors has affected driver compensation thus far. Profitability Johnson and Pawar found that the profitability of oper- ating a fleet at higher truck speeds (specifically, 70 mph vs. 65 mph) was highly dependent on the characteristics of the fleet and various external variables such as the price of fuel. The authors were able to construct a plausible scenario in which operating trucks at the higher speed of 70 mph increased profits, but fuel was assumed to cost $2.00 per gallon.16 At higher prices, it would presumably be more difficult to con- struct scenarios where operating at higher speeds actually increased profits.8Johnson and Pawar, pp. 128–129. 9ATA (The Maintenance Council). 55 vs. 65+: An Equipment Operating Costs Comparison. 1996, p. 13. 10Jeffrey Ang-Olson and Will Schroeer, “Energy Efficiency Strategies for Freight Trucking: Potential Impact on Fuel Use and Greenhouse Gas Emissions,” Trans- portation Research Record 1815, 2003. 11ATA, 1996. 12Johnson and Pawar, p.127; TRB, 2008, p. 32. 13TRB, 2008, p. 14. 14TRB, 2008, p. 33. 15ATA, rulemaking petition to NHTSA and FMCSA, October 20, 2006. 16Johnson and Pawar, p. 119.

84 Given the highly competitive nature of the trucking indus- try, one could reasonably conclude from the widespread vol- untary adoption of speed governors that they probably increase profits and, at worst, have no impact at all on profits. Other Types of Impacts Two other potential impacts of lower speed limits for trucks are worth mentioning. First, lower speed limits could affect the amount of congestion experienced on the nation’s highway system. Second, how the truck speed limits are applied and enforced could affect competitiveness. Traffic Flow Lower speed limits for trucks have a mixed effect on con- gestion, and it is not yet clear whether the net impact is posi- tive or negative. On the one hand, some studies suggest that lower speed limits for trucks can increase congestion, because trucks end up clustering together and impeding the flow of traffic.17 On the other hand, if differential speed limits or speed governors reduce the frequency and severity of crashes involving trucks, then they also reduce the hours of delay asso- ciated with such crashes. A 2002 study for NHTSA estimated the hours of delay caused by heavy vehicle crashes in the year 2000. The results are shown in Table B-2.18 The study valued hours of delay at $13.86 in urban areas and $16.49 in rural areas; the difference is due to the differ- ences in average vehicle occupancy in the two settings. Using these estimates of dollar values and hours, a fatal crash on an urban interstate causes more than $300,000 in time delays, while an accident with property damage only on a rural major arterial causes $4,200 in time delays. Truck Size and Weight Rules The regulations of truck size and weight have a multitude of impacts on the truck industry as well other modes. Truck size and weight regulations can also affect overall highway safety, traffic operations, fuel consumption, and emissions. When the first Federal limits were imposed in 1956, a grandfather clause allowed states to retain any truck size and weight lim- its exceeding the Federal limits as long as these limits were in place at that time. As a result, the current size and weight lim- its reflect a patchwork of Federal and state limits, with many situations in which equipment acceptable in one state cannot be used in neighboring states. In 2000, the USDOT completed a 6-year, comprehensive study of truck size and weight policy options. This study in- cluded modeling of a “uniformity scenario” (later referred to as the “Federal uniformity scenario”) in which the grandfather provisions in Federal law would be revoked and states would be required to adopt the Federal weight limit of 80,000 pounds on all Interstates and National Network highways.19 In a follow-on study published in 2004, the USDOT analyzed a “western uniformity scenario” in which the maximum gross vehicle weight limits of the grandfathered western states would be harmonized at 129,000 pounds. (This limit is near the high end of the range among the grandfathered states.) Thus, the USDOT has looked at the likely impacts of harmonization at the high end and low end of the range of possible choices. The impacts modeled by these two studies are compared below. In comparing the impacts of the two studies, one should note that the two studies did not use the same time periods for their analyses. For its 2000 comprehensive study, the USDOT used 1994 as the base year and compared policy impacts in the year 2000. For the later study of the western uniformity scenario, the USDOT used the year 2000 as the base year and compared policy impacts in 2010. Despite this disparity, it is worth comparing the direction of policy impacts (increases vs. decreases) and the percentage changes projected. Changes in Freight Distribution by Type of Truck and Mode As shown in Table B-3, in the Federal uniformity scenario, the imposition of the Federal size and weight rules on the grandfathered states would result in a projected increase in total truck VMT of 3.5 million miles. This increase in over- all VMT is caused by shifting freight traffic from longer and heavier vehicles to 5-axle tractor semitrailers. More of the Table B-2. Hours of delay per heavy vehicle crash, 2000. Road Class Property Damage Only Injury Fatality Urban Interstate 2,260 7,344 21,749 Other Freeway 1,766 5,737 16,990 Major Arterial 949 1,929 9,127 Rural Interstate 814 2,646 7,835 Other Freeway 416 1,350 3,999 Major Arterial 255 829 2,454 Source: NHTSA, 2002. 17Johnson and Pawar, p. 93. 18NHTSA, The Economic Impact of Motor Vehicle Crashes 2000. May 2002. 19A few states have weight limits below Federal limits on non-Interstate portions of the National Network. Under the uniformity scenario, those states would be required to bring weight limits up to Federal limits on those highways.

85 smaller vehicle combinations are needed to transport the same amount of freight. For this scenario, the USDOT did not attempt to estimate the diversion of freight from truck to rail. In the western uniformity scenario, total truck VMT in the region is estimated to decrease by 4.8 million (25 percent). Currently, LCVs are not often used for shipments for which one or both trip ends are outside the 13-state region. About half the VMT within the region for such shipments is pro- jected to shift to LCVs. This shift would require carriers to as- semble and disassemble the twin and triple trailers for travel outside the region. Despite the extra cost this would impose on carriers, the USDOT concluded that the net cost savings would still be attractive to carriers. For shipments entirely within the region, the percent of VMT in LCVs was projected to increase from about 9 to 78 percent. Less than one-tenth of 1 percent of rail traffic in the region was estimated to divert to LCVs. Safety Impacts The issue of safety is probably the most studied and most controversial aspect of truck size and weight policy. Truck size and weight rules affect safety in several ways. First, these rules affect the total number of miles traveled by trucks, which in turn affects the exposure of the overall truck fleet to crashes. Second, these rules affect vehicle performance, such as mini- mum braking distance and the propensity to roll over. Many other aspects of truck trips also affect safety, including driver performance, roadway design, vehicle maintenance, traffic conditions, and weather. Because of the many factors, isolating the effect of truck size and weight rules has proven difficult. In its 2000 comprehensive study, the USDOT did not present quantitative assessments of the safety impacts of the various scenarios that it analyzed. Instead, the agency pre- sented data on the crash rate history of different vehicle types and findings from engineering studies of vehicle safety performance. Regarding crash rates, the agency compared the fatal crash rates of single-trailer combination trucks and multi-trailer combination trucks during the period 1995 to 1999. As shown in Figure B-1, for most roadway classes, the fatal crash rates for single-trailer and multi-trailer combination trucks did not differ greatly. The one exception was the road- way class of “other rural roads,” on which multi-combination trucks had a much higher fatal crash rate.20 The 2000 study did not draw clear conclusions regarding the safety impacts of each scenario under scrutiny. However, from the information presented, one can conclude that, all other things being equal, the increase in heavy truck VMT resulting from the Federal uniformity scenario would result in more fatal truck accidents. At the same time, the shift of freight traffic from multi-combination trucks to single-trailer trucks might reduce the number of fatal truck accidents. The net impact is unclear. Like its predecessor study, the USDOT’s 2004 analysis did not offer a quantitative assessment of the net safety impacts Table B-3. VMT by vehicle configuration: Federal uniformity scenario versus policy baseline, 2000 (national level). Federal Uniformity Scenario Vehicle Configuration Base Case VMT (millions) VMT (millions) Percen t Change 5-axle Tractor Semitrailer 83,895 91,205 +9% 6- or 7-axle Tractor Sem itrailer 6,605 3,660 -45% 5- or 6-axle Double 5,994 5,986 -- 5- or 6-axle Truck Trailer 2,358 2,455 +4% 7-axle Double 632 290 -54% 8- or mo re axle Double 759 198 -74% Triple 126 54 -57% Total 100,369 103,848 +3.5% Source: USDOT Comprehensive Truck Size and Weight Study Final Report, Vol. 3, Ch. 4. 20It is worth noting that only 5 percent of the VMT by multi-combination trucks was accumulated on that type of road. U.S. DOT Comprehensive Truck Size & Weight Study, Vol. 3, Ch. 8, p. VIII-4. Table B-4. VMT by vehicle configuration—western uniformity scenario versus policy baseline, 2010 (13-state level). Source: USDOT, Western Uniformity Scenario Analysis, Table ES-2.

86 of the western uniformity scenario. Instead, the agency con- cluded that the fatal crash and travel data did not allow a detailed examination of LCVs separately from the 28-foot double trailers currently allowed on the National Network under Federal rules. According to the USDOT, the measure- ment problem was threefold: (1) fatal accidents are rare occur- rences, (2) there are few LCVs currently operating, and (3) there is only limited travel data collected on them. Regard- ing this last point, the agency noted that there is no Federal requirement to collect data for specific types of multi-trailer combination vehicles and, at the time of publication, only 2 of the 13 states actively collected separate VMT for different types of multi-trailers.21 In the end, the agency concluded that, even though the reduction in VMT by heavy trucks would lower crash exposure, there were too many other uncertain- ties regarding other safety impacts of LCVs to reach a firm conclusion on the net safety impact of the western uniformity scenario.22 Fuel Consumption and Air Emissions Under the Federal uniformity scenario, truck VMT was es- timated to increase by 4 million miles, because more truck trips would be required to move the same amount of freight. This increase in VMT translated into increased fuel consump- tion of 635 million gallons. For the western uniformity scenario, truck VMT was pro- jected to decrease by 4.8 million miles (25 percent) because the use of longer, heavier trucks would translate into fewer truck trips. The 25 percent reduction in truck VMT was estimated to result in a reduction in fuel consumption of 613 million gal- lons (12 percent). Fuel savings were not directly proportional to the reduction in VMT because fuel economy decreases as vehicle weight increases.23 For both studies under consideration, the U.S. DOT assumed that total truck emissions would vary directly with changes in fuel consumption. DOT did not attempt to quantify how changes in emissions would translate to changes in air quality. The research team calculated the change in greenhouse gas (GHG) emissions, shown in Table B-5. Total U.S. GHG emis- sions from heavy trucks are approximately 386 million metric tons of CO2-equivalent. So the Federal uniformity scenario would increase this total by 1.6 percent, while the western uniformity scenario would decrease U.S. heavy-truck GHG emissions by 1.5 percent. Traffic Operations Because of the shift of freight from heavier and longer vehi- cles to 5-axle semitrailer combinations at 80,000 pounds, the Federal uniformity scenario was projected to increase traffic congestion and associated costs in the year 2000 by 100 mil- lion vehicle-hours (0.4 percent). Figure B-1. Fatal crash rates by vehicle type and road type, 1995–1999. 21Western Uniformity Scenario Analysis, p. VII-20. 22Western Uniformity Scenario Analysis, p. ES-6. 23USDOT, WUSA, p. ES-8.

87 Unfortunately, the congestion model used in the compre- hensive study published in 2000 was not applicable to the west- ern uniformity scenario because the model does not allow for analysis at less than a national level. In its analysis of the west- ern uniformity scenario, the USDOT made only qualitative assessments of the likely impacts on traffic flow. The agency concluded that because of the shift of the reduction in total truck VMT, one could expect a slight decrease in delay in the 13 western states. Shipper Costs and Railroad Revenues Changes in truck size and weight regulations affect the trans- portation costs incurred by freight shippers. If the regulations become more restrictive, then amount of payload per truck will decrease and the cost per ton-mile of transportation will increase. Conversely, if the regulations become more permis- sive, then the amount of payload per truck will increase and the transportation cost per ton-mile will decrease. Changes in truck size and weight regulations affect rail shipper trans- portation costs as well, because some shippers will divert their freight to trucking or will obtain reduced rates from the rail- roads as they compete with lower truck rates. As shown in Table B-7, in the Federal uniformity scenario, the USDOT estimated that the transportation cost for shippers using trucks would increase by $6.4 billion per year, or about 3 percent. For the western uniformity scenario, the USDOT estimated savings to shippers of about $2 billion annually, or about 4 percent of total shipper costs for moves by truck in and through the region. These additional costs estimated for the Federal uniformity scenario are higher than the projected savings in the western uniformity scenario, because the removal of the grandfather provisions in Federal law would affect more than the 13 west- ern states analyzed in the USDOT’s 2004 study. For the Federal uniformity scenario, the USDOT did not es- timate the impact on rail shippers, but the agency surmised that the impact would be small because most of the potentially affected freight trips were of relatively short distances.24 For the western uniformity scenario, the USDOT was able to esti- mate savings for shippers using rail. The agency estimated that the increased competition of the longer, heavier trucks would generate minor savings of about $30 million per year for ship- pers using rail. Of this amount, $3 million in savings would accrue to shippers who actually switch from rail to trucking; the rest would accrue to rail shippers through lower rates.25 Level of Investment in Inland Waterway Infrastructure Lack of investment in inland waterway infrastructure in- creases the probability of a lock or dam failure. There is no information to reliably estimate how this policy decision affects the probability of failure. But the research team can estimate the cost of a failure. Complete quantification of the cost of a lock or dam failure would require data on the actual delay and the value of the particular goods being moved. In the case of a total failure and a forced mode shift, one would need to know the actual reduc- tion in transit time and the difference between barge and rail rates for the specific cargo involved, as well as the value of that cargo. Given that the research team is examining a hypotheti- cal case, we rely on the average value of goods moving by barge 24USDOT, Comprehensive TS&W Study, p. XII-3. 25USDOT, Western Uniformity Scenario Analysis, p. ES-9. Table B-5. Impacts of scenarios on fuel consumption and GHG emissions. Federal Uniformity Scenario Western Uniformity Scenario Fuel Consumption +635 million gallons -613 million gallons (-12%) GHG Em issions 6.2 million metric tons CO2-eq 6.0 million metric tons CO2-eq Table B-6. Impacts of scenarios on traffic operations. Federal Uniformity Scenario Western Uniformity Scenario Traffic Delay +100 million vehicle-hours (+0.4%) Small decrease Congestion Costs +$1.9 billion Small decrease Table B-7. Impact of scenarios on shipper costs. Federal Uniformity Scenario Western Uniformity Scenario ($2000) Shippers Using Trucks +6.4 billion (+3%) -$2 billion (-4%) Shippers Using Rail Not estimated -$30 million (<1%)

88 and the difference between average barge rates and average rail rates. We also know tonnage moving on the Upper Mississippi and the Ohio Rivers. This allows us to say something about the potential magnitude of the costs of a structure failure. Delay Costs Average value per ton for shallow-draft, domestic water carriage is $250.26 Delays, as opposed to complete stoppages, are far more likely to occur on the Ohio River than on the Upper Mississippi, because all the Ohio River locks are doubles. In 2006, 241.5 million tons moved on the Ohio.27 Thus, the total value of this traffic was $60.4 billion. For sim- plicity in developing an approximation, we assume that the traffic was evenly distributed over the 20 locks on the Ohio. Thus, the value of the annual traffic moving through any one lock was also $60.4 billion. All but three of the locks have 1,200-foot main chambers; of these, all but one has 600-foot auxiliaries. A 15-barge tow can pass through a 1,200-foot lock in about 30 minutes.28 With a 600-foot auxiliary, the tow has to be broken up, moved in two passes, and put back together. We assume this proce- dure adds 1 hour to the time for locking through. There are often queues at locks, so waiting time would also increase. We assume an average of 3 hours of extra waiting time, so 4 hours is added to the transit time for each tow. We now assume that it takes 2 months to repair the failed lock (1/6th of a year). Assuming no seasonal variations, $10.1 billion worth of goods will experience a 4-hour delay (60.4 billion ÷ 6 = 10.1 billion). We assume the owners of the cargo are paying annual inter- est at 6.5 percent.29 The result is a delay cost of $300,000 ((0.065 ÷ 365 ÷ 6) × $10.1 billion). This value is offered strictly to give a rough sense of order of magnitude. If, as is likely, we have underestimated the effect of queuing, the value would be greater but still not large. An increase by a factor of 5 would bring the amount to $1.5 million. This reflects the low value per ton of traffic moving on the rivers. Forced Mode Shift A total blockage, forcing a shift of cargo to rail or possibly truck, could occur from a lock failure on the Upper Missis- sippi where only 3 of 29 locks are doubles. In 2006, 71 mil- lion tons moved on the Mississippi above the mouth of the Missouri.30 We assume a lock fails at or near the mid-point between the mouth of the Missouri and Minneapolis and that it affects half the total tonnage, or 35.5 million tons. Assum- ing, again, 2 months for lock repairs, affected tonnage would be approximately 6 million tons (35.5 ÷ 6 = 5.9). There could be various responses to the blockage. Blocked traffic could move between a point downstream from the failed lock and origins and destinations to the north by rail or truck. Some traffic could be transferred to rail or truck for portage around the blockage and put back on the river. To offer some rough sense of the magnitude of impact, we assume a scenario in which traffic moves 100 miles by rail when it would otherwise have been on the river, and we assume no change in distance, only change in mode. In 2007, average rail rates were $0.03 per ton-mile and barge rates were $0.014 per ton-mile, a difference of $0.016.31 For our scenario, we assume a greater difference for several reasons. A rail carrier might be in a strong bargaining posi- tion because of the blockage. Also, extra costs for transload- ing would be spread over a relatively short move. It is reason- able to use a difference of $0.02 per ton-mile. Thus, shipping cost for each ton would increase by $2.00. Given that 6.0 mil- lion tons are affected, the total cost is $12 million ($2.00 × 6.0 million = 12 million). Cost Summary These cost estimates are clearly rough and are intended only to give a sense of the order of magnitude of impact. If any- thing, they are probably low. For example, the assumption of 2 months to fix a failed lock could be optimistic. It may be use- ful to think of a range in which the above estimates are the low end and the high estimate is greater by a factor of five. This is shown in Table B-8. Highway Tolls and Other User Charges Tolls affect which roads truckers use, because tolls change the relative costs of the roads available for a given move. The most direct impacts on the freight system are the costs to Table B-8. Cost of a lock failure. Nature of cost Possible range of cost Cost of delay $300,000 to $1.5 million Cost of mode shift $12 million to $60 million 26Calculated from 2007 Commodity Flow Survey, Preliminary, December 2008, Table 1. 27Waterborne Commerce of the United States, USACE, 2006, Part 2, p. 61. 28Michael Bronzini, “Inland Waterways: Still or Turbulent Waters Ahead?” Annals of the American Academy of Political and Social Science, Vol. 553, p. 70, Septem- ber, 1997. 29Federal Reserve Board, data on business lending in November 2007. http:// www.federalreserve.gov/releases/e2/200712/default.htm 30Waterborne Commerce of the United States, USACE, 2006, Part 2, p. 203. 31Rail rate from AAR, Railroad Facts, 2008, p. 30. Barge rates calculated using a rev- enue amount extrapolated from the 2002 Economic Census to 2007 and ton-miles from the CFS, Preliminary, December 2008, Table 1.

89 trucks diverting to alternate routes to avoid tolls. These are the costs and operating problems of switching to different roads or different times from those otherwise preferred. But these are not the only impacts. Other users of the roads to which trucks divert may be affected by the increase in truck traffic on those roads. Also, reduction in truck traffic on the tolled road may affect, positively, other users of that road. Highway tolls can also affect railroads. All charges to truck- ers affect the total cost of highway freight carriage and, there- fore, the relative costs of highway and rail carriage. The relative costs of these modes determine, in part, their relative shares of freight traffic. This affects the revenues and earnings of truck and rail carriers. Beyond that, the efficiency of the freight system is affected if freight movement on highways is mispriced. If truckers are paying less (or more than) marginal cost, the freight system will not function at maximum efficiency, and the effects will be felt throughout the economy. Accordingly, the following are the four principal areas of impact from highway pricing: • Costs to trucks that divert from a tolled road • Impacts on trucks that stay on a tolled road • Impacts on mode share between highway and rail • Effects on the whole economy and society from an ineffi- cient freight system Potential for Quantification of Impacts Diversion Effects Regarding costs to trucks that divert to alternate roads, there is enough information to permit estimates of changes in crash rates, fuel consumption, and speed. This allows us to estimate crash costs, fuel and other operating costs, and delay costs per diverted truck VMT. Two available estimates of diversion rates, one based on an actual tolled road, give us a basis for making a plausible approximation of total costs to diverted trucks from similar roads. Details on these cost estimates are presented below. We cannot make com- parable estimates of the effects on other traffic on the roads to which trucks divert or the roads from which they divert. We can, however, offer some speculation as to whether such impacts would be noticeable. No data support a national aggregate estimate of the costs (or benefits) of diversion. Effects on Truck/Rail Mode Share We are not aware of any useful data or analyses that would permit a quantitative estimate of revenue impact on rail car- riers related to a given change in the overall price of high- way carriage. The prevailing view in both the trucking and rail industries is that higher costs for truckers would shift some traffic from highway to rail but not a large amount. ATA supports a fuel-tax increase to improve highways and is not concerned about possible loss of traffic. Regarding tolls, our discussions with rail executives suggest they do not expect much of an impact on mode share simply because tolls only apply to a small portion of traffic. Higher fuel taxes or a general VMT-based tax would have a stronger effect in this view. There is a widespread view among rail managers, rail- industry analysts, and shippers that the quality, especially reliability, of rail service is a far more important factor than price in determining shipper choice of mode in markets where there is significant rail-truck competition. In particular, these markets are rail-intermodal service and carload service. (Rail carload service is movement in shipments of one or a few cars at a time, as opposed to shipments that require a full train.) It is also worth noting here that some large truckload carriers that offer rail intermodal service are making a deliberate effort to shift more of their long-haul traffic (over 1 day’s drive) to rail intermodal. Wider Impacts on the Freight System and the Economy As a matter of economics, there is no question that the freight system would be more efficient if the inputs used for providing freight service were correctly priced. Where inputs are purchased in open markets, and there is no monopoly power, we can assume relative prices of inputs accurately re- flect their relative costs, and there are no significant distortions. Highway use is not priced in an open market, and the providers of highways, the Federal and state governments, have monop- oly power. There is no alternative to buying fuel and paying the Federal tax, and the same is true for many state taxes. Although trucks and other motor vehicles can switch to alternate routes, toll authorities have a high degree of market power. There is consensus among economists and others who study transportation that use of highways is not optimally priced. To our knowledge, there are no usable data or analyses that would provide a basis for estimating the value of the efficiency gain that would ensue if highways were correctly priced. We know there would be gains, possibly significant gains, but we have no way for making a plausible estimate of their magnitude. The same observation applies to any benefits from congestion pricing, whether to trucking or to the wider economy. Estimate of Costs to Diverted Trucks Although we cannot establish a national estimate for costs to diverted trucks, we can estimate a range of costs for what might be a typical tolled highway. In order to do this, we need to estimate diversion rates and the changes in crash, fuel,

90 and other operating costs, as well as speed and delay costs that trucks would incur from taking sub-optimal routes. We focus the analysis on combination trucks, which account for almost all of non-local highway carriage. Five-axle truck- trailer combinations with 18 wheels are, by far, the preponder- ant configuration. We are looking at inter-city traffic, so we concentrate on rural roads. Diversion rates will be much lower for short, urban trips where switching to alternate routes may cause a disproportionate increase in distance and where alter- nate roads in a feasible distance are unlikely to be Interstates or high-quality freeways. Diversion Rates There have been two recent systematic attempts to esti- mate the degree of diversion of truck traffic from a road after a toll is imposed: a study of potential diversion from a toll on I-81 in Virginia and an empirical study of the diversion im- pact of tolls on the Ohio Turnpike.32, 33 The I-81 study was based on estimating costs to truckers of diverting from I-81 with estimates for various classes of traffic, including varying lengths of haul. For this study, the authors assumed truckers would compare costs of staying on I-81 with costs of diverting and choose the least-cost alternative. The authors of the Ohio Turnpike study used data on Class-8 truck VMT nationally, for Ohio, and for the Ohio Turnpike to estimate a demand curve as a function of the toll rate and speed. These two efforts led to somewhat different results, but we can use them together to establish a plausible range for diver- sion effects. The I-81 study yielded toll division impacts shown in Table B-9.34 The results of the Ohio Turnpike study are shown in Table B-10. We have already noted the difference in method between these studies. The I-81 diversion estimate is based on estimates of comparative costs between a tolled I-81 and alternate routes for loads with various origins and destinations. The Ohio study is based on an empirical demand curve applied to rates and speeds. The Reebie estimate of diversion is much higher than the one offered by Swan and Belzer. One reason is that the trips are probably longer on I-81. The Reebie study states that average length of haul for trucks on the Virginia segment of I-81 is 1,000 miles.35 This is through freight moving between the Southeast and the Northeast. We do not know average trip length for the Ohio Turnpike, but, given that Ohio is a major manufacturing state with several large metropolitan areas, it is likely that higher proportions of moves have either origins or destinations in Ohio or are entirely intrastate. Longer moves are more likely to divert than shorter ones, because more alter- nate routes are feasible. There is no reason to expect that these two studies would yield closely similar results; they used different methods applied to quite different traffic. It is reasonable to suppose, however, that the very high diversion rates for I-81 at $0.20 per mile and higher would not often be found on short tolled segments with higher percentages of local trips. To get a sense of actual toll rates in addition to the Ohio Turnpike, we looked at rates for 5-axle trucks imposed in Indi- ana and Illinois. For the Indiana Toll Road (157 miles), the rate is $0.17 per mile.36 Rates are higher on the Illinois toll roads.37 The low end of the range is $0.21 for night rates and $0.28 for day rates on the longer segments—I-90 (76 miles) and I-88 (96 miles). The high end is day rates of $0.42 to $0.53 on the shorter segments—I-94 and I-355 (both 30 miles). In summary, this gives us the following per mile truck toll rates in these states: • Ohio: $0.13 • Indiana: $0.17 • Illinois: $0.28–$0.53 (day rates) Toll authorities, whether public or private, do no set prices at levels that lead to high diversion rates; they lose revenue if they do that. The Reebie study suggests that the maximum revenue rate for I-81 would be in the range of $0.12 to $0.15.38 Reference to the above table with the Reebie results suggests maximum diversion rates, at these prices, of 15 percent to 25 percent. The Ohio case shows that toll authorities will not Table B-9. Estimate of truck diversion on I-81 in response to tolls. Toll (dollars per mile) Percentage of truck VMT diverting 0.05 09.0 0.10 14.0 0.12 16.0 0.15 23.0 0.18 31.0 0.20 36.0 0.30 67.0 0.40 81.0 Source: Bryan, J., et al., Reebie Associates and Atherton, Mease & Company. “The Impact of Tolls on Freight Movement for I-81 in Virginia”; prepared for Virginia Department of Transportation, 2004. 32Bryan, J., et al., Reebie Associates and Atherton, Mease & Company. “The Impact of Tolls on Freight Movement for I-81 in Virginia”; prepared for Virginia Department of Transportation, 2004. 33Swan, Peter, Pennsylvania State University, and Michael Belzer, Wayne State University, “Empirical Evidence of Toll Road Traffic Diversion and Implications for Highway Infrastructure Privatization,” 2007. 34Values in table are from Bryan, et al., p. 9. 35Memorandum from Reebie to VDOT, February 11, 2004. 36Indiana Toll Road website, https://www.getizoom.com/index.jsp 37Illinois Tollway toll calculator, http://www.illinoistollway.com/portal/page?_ pageid=133,1397406&_dad=portal&_schema=PORTAL 38Bryan et al., p. 10.

91 always price to maximize revenue. This suggests that we can think of $0.15 to $0.25 as a likely range for tolls on longer roads; and we can think of 5 percent to 25 percent as a plau- sible range for diversion rates. Crash Costs Estimating crash costs requires values for • Cost per crash for 5-axle trucks and • Increment in crashes per VMT for shift from Interstate to lower quality roads. We assume the cost per crash for combination truck (trac- tor and one trailer) to be $164,000.39 This is the average over all crash types: fatality, injury, and property-damage only (PDO). Estimating change in crash rate due to diversion poses some difficulty, because of the nature of the data. FMCSA reports fatal crash rates for large trucks by FHWA road class but not rates for other crashes. (This is because available data on fatal crashes are better than data on other crashes.) One way to deal with this problem is to find a way to scale up from fatal crashes to all crashes. FMCSA does provide data on all large-truck crashes broken out by crashes on divided highways and on highways not divided.40 These data show that fatal large-truck crashes, as a percentage of all large- truck crashes, are virtually the same on these two road types: 1.3 percent on undivided highways and 1.4 percent on divided highways. These data do not separate rural and urban crashes. This diminishes their accuracy for our purposes but does not elim- inate their usefulness. For one thing, fatal crash rates on rural and urban Interstates are almost the same: 1.3 and 1.2, respec- tively, per 100 million VMT. Fatal crash rates on rural arteri- als and urban “Other” facilities are also very close: 2.1 and 1.8, respectively.41 Crash rates on rural “Other” are much higher, 4.4, but only 17.0 percent of rural VMT for combination trucks is on “Other” roads.42 For these reasons, we believe that these data on divided and undivided highways give us a plau- sible approach to an estimate. FMCSA’s definition of large trucks includes all trucks over 10,000 pounds. Thus, the FMCSA data on large-truck crashes include many vehicles in addition to the combination vehi- cles doing most of the hauling of highway freight. (FMCSA puts out data on combination trucks’ fatal crashes but does not relate them to type of road.) This is a potential distorting factor, but we note that trucks over 26,000 pounds account for a disproportionate share of large-truck crashes, espe- cially the more severe ones. Crashes involving trucks over 26,000 pounds are 89 percent and 58 percent, respectively, of large-truck fatal and injury crashes.43 (This particular data set does not provide full information on PDO crashes.) We con- clude that using data for large-truck crashes will not unduly distort our estimates. The divided highways should be roughly comparable to Interstates and Non-Interstate Principal Arterials in FHWA’s classification of rural highways, and the undivided highways should be comparable to rural “Other” roads. Therefore, we can use fatal-crash percentages of all crashes on undivided and divided highways, 1.3 percent and 1.4 percent, respec- tively, as the basis for scaling up to all crashes. For this pur- pose, we take the reciprocals of 0.13 and 0.14 as scaling fac- tors to obtain rates for all crashes from the reported rates for fatal crashes. These factors are, respectively, 77 for undivided roads and 73 for divided highways. We apply these factors to FMCSA’s reported fatal crash rates per 100 million VMT for three classes of rural roads: Interstate, Non-Interstate Principal Arterials, and Other.44 Table B-11 shows the results. Table B-10. Estimate of truck diversion on Ohio turnpike in response to tolls. Year Actual toll (dollars per mile) Percentage of truck VMT diverting 2001 0.18 13.62 2002 0.18 13.68 2003 0.18 13.56 2004 0.18 13.00 2005 0.13 05.01 Source: Swan, Peter, and Michael Belzer, “Empirical Evidence of Toll Road Traffic Diversion and Implications for Highway Infrastructure Privatization,” 2007. Note: The speed limit on the turnpike was raised from 55 mph to 65 mph in early September of 2004. 39Eduard Zaloshnja and Ted Miller, “Unit Costs of Medium and Heavy Truck Crashes,” prepared for F MCSA, December 2006, p. 8, Tables 3 and 5. The value in Table 3 for quality of life (QALY) was adjusted using the number in Table 5 for QALY when value of a statistical life (VSL) is $5.75 million. DOT’s estimate of VSL was increased to $5.8 million in a directive from OST of February 5, 2008. 40FMCSA, “Large Truck Crash Facts 2005,” (LTCF), 2007, Table 30. 41Fatal crash rates are from FMCSA LTCF, Table 19. 42VMT data are from FHWA, “Highway Statistics,” Table VM-1. 43FMCSA LTCF, Table 39. 44FMCSA LTCF, Table 19.

92 trucks is the average of the crash rates of the three road types weighted according to the distribution of the diverted traffic. The diverted crash rate is 155, so the change in crash rates is 60 crashes per 100 million VMT. (Diverted crash rate is cal- culated as: 0.6 × 95 + 0.2 × 153 + 0.2 × 339 = 155.) With a cost per crash of $164,000, this yields a crash cost of approximately $10 million per diverted 100 million VMT (164,000 × 60 = 9,840,000). Fuel and Other Operating Costs Fuel, maintenance, and tire costs will vary directly with the change in road types. To estimate fuel costs of diversion, it is necessary to make assumptions about average speed of com- bination trucks by road type and to obtain data on variation in fuel consumption with truck speed. Discussions with people in the trucking industry suggest that large trucking firms tend to set governors in the 60 to 65 mph range, so their trucks average less than 60 mph on Interstates. Owner-operators and small firms that do not use governors would have somewhat higher speeds. We assume speeds by road type as shown in Table B-15. Because of the lower speeds and our focus on rural roads, fuel consumption per mile drops for the diverted trucks. Given the assumed distribution of diverted traffic over road types, fuel consumed per mile is 0.157 gallons (0.6 × 0.172 + 0.2 × 0.146 + 0.2 × 0.126 = 0.157). We assume $2.50 per gal- Table B-11. Estimated crash rates by rural road type per 100 million VMT. Road Type Fatal Crashes Scaling Factor All Crashes Interstate 1.3 73 95 Non-I-S Principal Arterials 2.1 73 153 Other Roads 4.4 77 339 Table B-13. Road types used by diverted trucks. Road Type I-81 Percentage of VMT Ohio Turnpike Percentage of Trucks Interstate 69.0 55.9 All Non-Interstate 4-lane 2-lane Non-Interstate 31.0 44.1 22.8 21.3 Source: I-81 values calculated from Bryan et al., p. 14; data in Figure 6. Ohio Turnpike values calculated from Swan and Belzer, p. 18, Table 10. Table B-14. Assumed distribution of diverted truck VMT. Road Class Percentage of VMT Interstate 60.0 Non-I-S Principal Arterials 20.0 Other Roads 20.0 Table B-12. Combination-truck VMT by rural road type. Road Type Percent of VMT Interstate 52.2 Non-I-S Principal Arterials 31.0 Other Roads 16.8 To estimate change in crash rates due to diversion, we have to know the distribution of diverted truck traffic over road classes. The current distribution of combination-truck VMT over rural road types is shown in Table B-12.45 In the I-81 study, Reebie estimated the percentage of VMT diverted to Interstates and to all other roads and found 69 per- cent diverted to Interstates. Swan and Belzer estimated num- ber of trucks diverted by road type. The findings are shown in Table B-13. The high percentage of diversion to Interstates from I-81 surely reflects the long average length of haul on I-81 in Virginia. The estimated truck diversions from the Ohio Turn- pike show percentages by road type not dissimilar to the actual VMT percentages from FHWA data. (If average length of haul for combination trucks on the Ohio Turnpike is close to the national average length of haul, the percentage of trucks divert- ing is an acceptable proxy for percentage of VMT.) It is reason- able to assume that the 4-lane non-Interstate is roughly comparable to non-Interstate arterials and the 2-lane non- Interstate is roughly comparable to “Other” roads. Both of the estimates suggest that truckers choosing alternate routes do not use 2-lane roads if they can possibly avoid it. From these estimates we may infer that diverted combination-truck VMT will be distributed over road types, with some adjust- ment, in essentially the same way as all combination-truck VMT. For purposes of this analysis, we assume the following distribution of diverted-truck VMT (see Table B-14). We assume, for all cost-estimation purposes, that the tolled road is an Interstate or a freeway of comparable quality. There- fore, the pre-diversion crash rate is 95 per 100 million VMT as shown in the above table of crash rates. The rate for diverted 45“Highway Statistics,” Table VM-1.

93 lon for the price of diesel fuel.48 The result is a reduction in fuel cost of $.035 per mile. Costs of maintenance and tire wear are currently a little less than 20 percent of fuel cost per mile.49 We assume that ratio to hold for this analysis; therefore, we can increase the change in fuel cost by 20 percent to obtain an estimate of $0.042 as the reduction in operating cost per mile. This is equal to $4.2 million per 100 million VMT. Delay Costs We estimate delay costs with the same average speeds as- sumed above and the same distribution of diverted trucks over road types. The result is an average speed of 55.2 mph for diverted trucks (0.6 × 62 + 0.2 × 50 + 0.2 × 40 = 55.2). This leads to an increase in trip time of 0.002 hours per mile (pre- diversion mile).50 This yields approximately 200,000 addi- tional hours per 100 million VMT without allowing for any additional miles. For cost per hour of delay, we look at the revenue that tractor and driver generate in an average hour. If it takes a load an extra hour to reach its destination, the firm has lost an hour’s use of the tractor and, hence, the revenue it would generate in an average working hour. Approximately a year ago, industry executives in private conversations with us indi- cated an average revenue per day for a tractor in truckload service of about $700 to $725. In today’s market, that could be somewhere from $600 to $650. Current market condi- tions are not typical, however, so we assume $700. Available data tells us that 12 hours is an acceptable estimate of the av- erage working day of a long-haul driver.51 On this basis, lost revenue is $58.33 per hour. The cost of 200,000 additional hours is $11,590,307, which we may round to $12 million for our estimate. Cost Summary Table B-16 summarizes the annual cost changes per 100 mil- lion diverted VMT. Some of the alternate routes chosen will re- sult in longer distances. The Reebie paper estimates the increase at 1.1 percent.52 This yields the adjusted total of $18 million per 100 million VMT. Table B-17 shows the amount of VMT diversion esti- mated for I-81 and the Ohio Turnpike under different pric- ing scenarios. For I-81, tolls of $0.12 and $0.15 were cho- sen because the Reebie paper suggested that these are the maximum-revenue toll rates. For the Ohio Turnpike, diver- sion amounts for 2004 and 2005 were chosen because of the change in toll rates: $0.18 in 2004 and $0.13 in 2005 and the change in the speed limit in September of 2005 (55 mph to 65 mph). This cost estimate does not include increased mainte- nance costs on roads to which trucks divert. It is based on some reliable data and some plausible assumptions based on reliable data. It could be adjusted up or down to some degree. Nonetheless, it gives a rough sense of the magnitude of the direct costs from diverted trucks in some typical toll scenarios. The estimate does not include impacts on other traffic on roads to which trucks divert and from which trucks divert. We do not have sufficient data to estimate quantities for such impacts. It appears, however, that impacts on other traffic could be significant, at least for some segments. The Reebie estimate for I-81 suggests 15 percent to 25 percent of truck Table B-16. Summary of annual cost changes per 100 million diverted VMT. Cost Component Cost (million) Crashes $10.0 Fuel and operating costs $-04.2 Delay $12.0 Total $17.8 Total adjusted for added distance $18.0 Table B-15. Assumption for speed and fuel economy by road type. Road Class Average Speed 46 Miles per Gallon 47 Gallons per Mile Interstate 62 5.83 0.172 Non-Interstate Principal Arterials 50 6.83 0.146 Other 40 7.95 0.126 46Speeds are based on ICF team’s own expertise and e-mails from two trucking executives. One executive stated that his company expected an average speed of 45 mph off the Interstate. Thus, we assigned 50 mph to higher-quality non- Interstate roads and 40 mph to other non-Interstate roads. 47ATRI data. 48Transport Topics, March 23, 2009, gives current price just over $2.00, but these are abnormal conditions. One year ago, the price was $3.97, and that may have been an extreme. We assume $2.50, for this estimate. 49ATRI, “An Analysis of the Operational Costs of Trucking,” December 2008, p. 21, Table 4. 50Change in time per mile is the difference between time required to travel one mile at new speed and one mile at old speed. (1÷55.2) − (1÷60). 51FMCSA Field Survey. 52Bryan et al., p. 13.

94 VMT diverting at revenue-maximizing tolls. Depending on the amount of other traffic on a segment, this could have some effect on level of service. The Swan and Belzer estimate for the Ohio Turnpike shows 13 percent of VMT diverting at the toll of $0.18 per mile. This could be noticeable on some segments but would likely have little impact on most segments. Regarding roads to which trucks are diverted, Swan and Belzer present data showing that some segments could see increases of several thousand trucks per day.53 That is certainly enough to degrade the level of service on some segments. Lockage Fees for Inland Waterways There have been recent proposals to phase out the fuel tax to towboats and replace it with a lockage fee. The most signif- icant impact of such a policy would be the increase in the tax burden on inland towing, which might cause an increase in barge rates and have some effect on mode shift. We estimate the change in the tax burden on the inland towing industry that would result from the proposed lockage fee. For this pur- pose, we compare projected revenues from the lockage fee with current and projected payments on the fuel tax. We can then compare the new tax burden and the change in tax bur- den with projected towing-industry revenue. The latest published estimate of towboat revenues is from the 2002 Economic Census: $2,557 million from inland towing.54 Published data from the 2007 Economic Census do not yet include revenues from inland towing. Available data from the Commodity Flow Survey (CFS) do, however, include ton-miles of shallow-draft water carriage in 2007. By compar- ing this figure with ton-miles carried in 2002, we obtain the change in the amount of freight carriage sold. Over this period, ton-miles decreased from 212 billion to 164 billion, by a factor of 0.77.55 If we also adjust for the change in the price of inland carriage, we can obtain a reasonable estimate of 2007 revenue from inland towing. Price indices for water transportation show a price increase from 2002 to 2007 by a factor of 1.19.56 This leads us to an estimate of 2007 rev- enue of $2,356 million.57 It is desirable to estimate industry revenue for three refer- ence years: 2004 and 2013 in addition to 2007. Industry tax burden in 2004 is relevant because that was the last year in which the full deficit-reduction tax of $0.043 per gallon was levied on towboats. The Office of Management and Budget (OMB) has projected revenue from the proposed lockage fee out to 2013. We can therefore compare total tax payments to industry revenue in 2004, 2007, and 2013. Industry revenue declined from 2002 to 2007 at an annual rate of 1.63 percent. Assuming a constant rate of decline over that period, revenue in 2004 would have been $2,474 million.58 In projecting fuel- tax revenue out to 2013 (as a baseline), OMB assumes that the decline in barge traffic in recent years stops and fuel tax revenue grows at a very slight rate from 2007 to 2013: 0.36 percent.59 We use this growth rate to project from 2007 revenue and obtain 2013 revenue of $2,407 million.60 We obtained tax payments data for 2004 and 2007 from the Congressional Budget Office, Tax Analysis Division. Estimated tax payments for 2013 are available from the FY 2009 Presi- dent’s Budget, cited above. Table B-18 shows the relative bur- den of tax payments in the three reference years, assuming the implementation of the lockage fee in FY 2009. We see that the highest burden is for the lockage fee in 2013, 5.2 percent of industry revenue. (The burden is actually higher in 2012, but the fee does not stay at the 2012 level; it drops back because of the balance in the trust fund.) The second highest burden was in 2004 when the industry was still paying the full deficit-reduction tax of $0.043 per gallon in addition to the user tax of $0.20 per gallon. The burden was lightest in 2007 when the deficit-reduction tax was zero. The most meaningful measure of the increase in burden is the tax increase as a percentage of revenue. This tells us how much the industry would have to increase its prices in order to recover fully the tax increase. And this, in turn, tells us Table B-17. Truck VMT diversion and associated annual cost. Toll Rates VMT Diverted (millions) Annual Cost (millions) $14.8 I-81 $10.0 $23.9 Ohio Turnpike $0.15 $0.12 $0.18 $0.13 82 56 133 54 $09.7 53Swan and Belzer, p. 18, Table 10. 542002 Economic Census, Product Lines, Transportation and Warehousing, Table 1, p. 7 (NAICS 483211 [coastal and inland freight, product codes 44010, 44030]). 552002 CFS, Table 1a, p. 1; 2007 CFS, Preliminary Release. 56BEA, Gross Domestic Product by Industry, indices for water transportation http://www.bea.gov/industry/gpotables/gpo_action.cfm?anon=94425&table_ id=23984&format_type=0 57Calculation: 0.77 × 1.19 × 2,557 = 2,356 58Calculation: 2,557 × (1−0.0163)2 = 2,474 59Calculated from OMB projections in FY 2009 President’s Budget, Analytical Perspectives volume, Table 17-3 (p. 266) and Table 17-4 (p.269). 60Calculation: 2,356 × (1.0036)6 = 2,407

95 the degree to which the tax increase would have an impact on mode share and on total movement of freight. It cannot be assumed, of course, that the towing industry would, or could, raise rates by enough to fully offset the tax increase. But the amount of a 100-percent passthrough shows us the maximum possible effect on mode share. As shown in Table B-18, the 2013 tax increase ($33 mil- lion) is 1.4 percent of total revenue. For the traffic using locks, the tax increase would be more than 1.4 percent, because the traffic not using locks would not pay the fee at all. The Economic Census data on revenue from inland towing show a very small share, less than 10 percent, coming from coastal carriage.61 Some inland river traffic also does not use locks, although it is not likely a large share of total movement. Let us assume as an upper bound that one-third of the total rev- enue would come from non-lock traffic. If one-third of the traffic pays no tax, the percentage of revenue coming from lock-using traffic would be 1.5 times the percentage coming from all traffic. So the extreme case would be that the lockage fee would mean a tax increase of 2.1 percent of the revenue of lock-using traffic. A 2.1 percent increase in average rates of lock-using barge traffic is unlikely to have a significant impact on mode share between barge and rail. It is not likely that all of the increase could be passed through, so the real effect would be a slight increase in barge rates and a slight decrease in earn- ings from carriage using locks. The aggregate impact would likely be negligible. Table B-18. Inland waterway towing revenue and tax payments. Fiscal year Industry revenue (million) Tax payment (million) Tax % of revenue Change in tax as % of revenu e 2004 fuel tax with $0.043 $2,474 $110.0 4.4% NA 2007 fuel tax without $0.043 $2,356 $87.3 3.7% NA 2013 fuel tax projected $2,407 $93.0 3.9% NA 2013 lockage fee projected $2,407 $126.0 5.2% 1.4% Change in tax as a percentage of revenue is only relevant for 2013; we do not have an estimate of what the revenue from the lockage fee would have been in the earlier reference years. 612002 Economic Census, cited above.