Impacts of Policy-Induced Freight Modal Shifts (2019)

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.

151 This chapter illustrates some of the possible applications of the mode choice models estimated as a part of this project. To give the numerical experiments a real-life flavor, the research team used some of the cases studies conducted as inspiration for the numerical experiments. Close examination of the case studies selected by the NCFRP Project 44 panel revealed that the two projects that were most suitable to illustrate the applications of the mode choice models were the Heartland Corridor and Crescent Corridor. These two cases were picked because they have sufficient basic information to build meaningful scenarios and exemplify the possible applica- tions of the models. It is worth noting that the applications discussed in this chapter are only intended to illustrate the use of the models; they do not purport to be an evaluation of the real- life impacts of the selected projects. Conducting real-life case studies requires access to basic data about demand patterns, as well as detailed data about network conditions before and after the implementation of network changes. Assembling such data requires a level of effort that is not within the scope of this project. The numerical experiments conducted in this section are nothing more than “real-life-inspired scenarios.” To construct these scenarios, basic details of the projects, publicly available data, and reason- able assumptions were used. The analyses demonstrate the application of both types of mode choice models: (1) market-share mode choice models, and (2) shipment-level mode choice models. The various applications of the models are shown in Table 55. The application of the market-share models to the scenario inspired by the Heartland Corridor estimated the impact of a shorter route on mode split for selected commodity types. The analysis for Crescent Corridor focused on the impact of travel distance and transit times on mode split for selected commodity types. The application of shipment-level models for the Heartland and Crescent Corridors focused on estimating the mode split along each corridor for selected commodity groups. The commodity types used in these analyses were based on the SCTG codes. In conducting the experiment, the effective hours of service for truck drivers based on current regulation were accounted for within total transit time for shipments (FMCSA 2011). An average truck capacity of 25 tons was employed; therefore, any shipment exceeding this threshold necessitates more than one truck. Two different mechanisms were used to estimate shipment sizes and the associated freight rates. In the first case, the research team used the shipment-size models estimated in this research (see Chapter 8) to estimate the freight rates. In the second case, the team used the PUMS dataset to obtain the average shipment sizes by commodity for the hypothetical flows. In terms of which set of results is the most reliable, it all depends on the quality of the estimates for the shipment sizes. The shipment-size models are based on the entire CFS microdata and, as a result, they rep- resent the national average pattern. However, since the shipment-size patterns may be dissimi- lar for different freight corridors, the national average pattern may not necessarily capture the C H A P T E R 1 0 Numerical Experiments

152 Impacts of Policy-Induced Freight Modal Shifts shipment patterns at specific corridors. If the shipment-size data for a given corridor are small or of questionable quality, the shipment-size models may provide the best option. Conversely, if the corridor-specific shipment-size data are large and of good quality, it would make sense to use them instead of the shipment-size models. The following sections succinctly describe the real-life corridors that inspired the numerical scenarios used to test the freight mode choice models and discuss the scenarios and the corre- sponding numerical results. Descriptions of the Real-Life Cases That Inspired the Numerical Scenarios The Heartland Corridor The Heartland Corridor, shown in Figure 44, is the most direct high-capacity intermodal route between the mid-Atlantic and the Midwest. The corridor connects the Port of Virginia to major destinations in the Midwest including Chicago, Detroit, Columbus and Cincinnati (FHWA 2014c). Prior to the Heartland Corridor, NS was forced to use the Norfolk and Western main lines across Virginia, Southern West Virginia, and Ohio to carry double-stack trains through Harrisburg or Knoxville. The Heartland Corridor project cut nearly 200 miles from each double-stack container moved to Chicago and reduced Norfolk-to-Chicago transit times from 3 days to 2 days. More information, including the major components of the project, is detailed in Chapter 9. Some of the key benefits realized by the completion of the corridor include the following: • Reduced transit times and costs for shippers via the corridor; • Improved mobility for truck freight and passenger cars attributed to diverted truck traffic; • Environmental benefits from reduced truck emissions; and • Improved access to global trade routes through Port of Hampton Roads for shippers and manufacturers in Virginia, West Virginia, Ohio, and Eastern Kentucky. The Crescent Corridor The Crescent Corridor, shown in Figure 45, consists of a 2,500-mile network of existing rail lines that extends from New Jersey to Memphis and on to New Orleans. The project includes straightening curves, adding signals, and building new track and rail terminals. In addition, a partnership between NS and the states of Tennessee, Pennsylvania, Virginia, Alabama, and Mississippi will improve the system and develop regional intermodal freight distribution centers (NS 2009a). More information on the project is presented in Chapter 9. Heartland Corridor Crescent Corridor - Using shipment-size models - Using PUMS data - Using shipment-size models - Using PUMS data - Using shipment-size models and PUMS data - Using shipment-size models and FAF data Analysis Impacts of shortening rail travel distance on rail market share by commodity type Estimation of truck market shares by commodity groups Impacts of travel distance on truck market share by commodity type Table 55. Numerical experiments utilizing the mode choice models.

Source: (NS 2014b) Figure 45. Crescent Corridor. Source: (Appalachian Regional Commission 2009) Figure 44. The Heartland Corridor. It is important to mention the role played by freight rates. The freight rates used in the project closely resemble market rates. As a result, in corridors where, for instance, rail is very competitive, rail rates will be higher, in proportion to shipment distance, than in other corridors where rail is less competitive. These effects show up in some of the figures as small jumps, or drops, in the estimates of market shares (instead of a smooth and monotonic pattern of change in market shares). In the case of shipment-level models, which in some cases rely on the PUMS data, the results are influenced by the number of data points for a given interval of the variable used in the analyses. As in the previous case, using a small number of observations will lead to estimates of market shares that are not as smooth as expected. These observations ought to be taken into account when interpreting the results discussed in the next section.

154 Impacts of Policy-Induced Freight Modal Shifts 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% M ar ke t S ha re via Knoxville via Harrisburg via Heartland 35 - El ec tro nic 34 - M ac hin ery 30 - Te xti les 40 - M isc ell an eo us 24 - Pl as tic s a nd ru bb er 5 - M ea t 8 - A lco ho lic be ve rag es 4 - A nim al fee d 33 - Ar tic les of ba se m eta l 29 - Pr int ed pr od uc ts 26 - W oo d p ro du cts 43 - M ixe d f rei gh t Figure 46. Rail market share, P(R), Norfolk to Chicago. Market-Share Models The Heartland Corridor’s Inspired Scenarios Scenario #1: Impacts of Shortening Rail Travel Distance on Rail Market Share by Commodity Type The objective of this scenario is to illustrate the use of the freight mode choice market-share models—discussed in Chapter 8—to assess the impacts of a shortening of the travel distance by rail. To this effect, the research team assumed hypothetical flows of cargo that travel from Norfolk, Virginia, to Chicago, Illinois. The scenario assumes that the reduction in distance traveled is similar to the one produced by the Heartland Corridor, which shortened the dis- tance between these cities to 1,031 miles from 1,342 miles via the Knoxville route or 1,264 miles via the Harrisburg route. The analyses focused on the following commodities: electronic, machinery, textiles, miscellaneous, plastics and rubber, meat, alcoholic beverages, animal feed, articles of base metal, printed products, wood products, and mixed freight. These commod- ity types were selected, in part, because they are the commodities with the largest amounts of tonnage being transported in the corridor and according to the models that are available to estimate shipment size based on the GCD. The scenario considers two different cases that differ in the way the average shipment sizes were obtained, i.e., either from the PUMS data or from the shipment-size functions. Results: Using the Shipment-Size Models. In this scenario, the team used shipment-size models derived from the CFS microdata to compute the freight rates. The impacts on rail market share [P(R)] are shown in Figure 46. The figure shows that the reduction of the rail travel distance leads to increases in rail market share. In the case of wood products (SCTG 26), for instance, the rail market share assuming that all shipments use the Knoxville route is 6.0 percent and 6.3 percent if the Harrisburg route is used. Using the Heartland route, the estimated rail market share increases to 7.6 percent due to the shorter rail travel distances. In general, all com- modities exhibited similar behavior, although there are differences. Results: Using the Shipment Size from PUMS. In this scenario, the research team used the average shipment sizes from the PUMS data to compute the freight rates. As shown in Figure 47, the reduction in distance traveled leads to an increase in rail market share. However, the rail mar- ket shares for some commodities are very different from the market shares estimated using the shipment-size models. The commodities with the largest difference are wood products (SCTG 26) and mixed freight (SCTG 43). The difference in rail market shares for wood products can be

Numerical Experiments 155 explained, in part, by the small amount of data in PUMS for that commodity (only 35 observa- tions). Mixed freight, being a category with a heterogeneous composition of goods, is bound to exhibit different behaviors depending on the actual mix of commodities in a given shipment. Table 56 summarizes the expected changes in rail market shares between the original routes (i.e., Knoxville and Harrisburg) and the new Heartland Corridor. As shown, the market share increases by a range between 0.1 percent and 1.6 percent. This indicates that the shorter travel distance between the two points, due to the Heartland Corridor, resulted in an increased rail market share for the commodities in the analysis. As explained previously, the differences between the two sets of results are related to the number of observations in the PUMS data for each corridor. These results highlight the importance of gathering data about the shipment-size patterns for specific corridors. The Crescent Corridor’s Inspired Scenarios Scenario #2: Impacts of Travel Distance on Truck Market Share by Commodity Type The Crescent Corridor connects the Southeast and Northeast regions of the United States through a rail network. The objective of this scenario is to quantify the impacts of travel distance and transit time on mode split. Keeping this in mind, an origin and a set of destinations were selected 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% M ar ke t S ha re via Knoxville via Harrisburg via Heartland 35 - El ec tro nic 34 - M ac hin ery 30 - Te xti les 40 - M isc ell an eo us 24 - Pl as tic s a nd ru bb er 5 - M ea t 8 - A lco ho lic be ve rag es 4 - A nim al fee d 33 - Ar tic les of ba se m eta l 29 - Pr int ed pr od uc ts 26 - W oo d p ro du cts 43 - M ixe d f rei gh t Figure 47. Rail market share, P(R), Virginia to Illinois. via Knoxville via Harrisburg via Knoxville via Harrisburg 4 - Animal feed 0.5% 0.4% 0.1% 0.1% 5 - Meat 0.5% 0.4% 0.1% 0.1% 8 - Alcoholic beverages 0.5% 0.4% 0.1% 0.1% 24 - Plastics and rubber 0.4% 0.3% 0.4% 0.3% 26 - Wood products 1.6% 1.6% 0.2% 0.1% 29 - Printed products 0.4% 0.3% 0.4% 0.3% 30 - Textiles 0.3% 0.2% 0.3% 0.2% 33 - Articles of base metal 0.2% 0.1% 0.1% 0.0% 34 - Machinery 0.5% 0.4% 0.5% 0.4% 35 - Electronic 0.3% 0.2% 0.3% 0.2% 40 - Miscellaneous 0.4% 0.3% 0.4% 0.3% 43 - Mixed freight 0.8% 0.6% 0.3% 0.2% Shipment Size from PUMSShipment Size Models SCTG Table 56. Market share, P(R), difference compared to Heartland Corridor.

156 Impacts of Policy-Induced Freight Modal Shifts along the Crescent Corridor. For the analysis, the origin selected was Metairie, Louisiana. The destination cities were selected based on the rail rate models that were available. The destination cities are organized in order of increasing distance. As with the analysis inspired by the Heartland Corridor, the commodity selection for the analysis was based on the top state-to-state commodity flows along the corridor and the available rail rate models. The selected commodities are plastics and rubber (SCTG 24), wood products (SCTG 26), textiles/leather (SCTG 30), base metal (SCTG 32), machinery (SCTG 34), electronic (SCTG 35), and mixed freight (STCG 43). The resulting truck market share, P(T), for the seven commodities is presented as a function of the following: 1. Distance (miles), using GCD; 2. Truck transit time (hours), which accounts for truck driver hours-of-service regulations; and 3. Truck rate (in U.S. dollars), which accounts for number of trucks required to fulfill the shipment. As previously stated, two methods for estimating shipment size—shipment-size models and PUMS data—were used for comparative purposes. The results are presented for both methods in the following sections. Results: Using the Average Shipment Size Obtained from Shipment Size Models. The fol- lowing graphs display the results for the analysis using the shipment-size models. Figure 48 illustrates the truck market share for all the commodities included in the analysis. In general terms and, as expected, the truck market share is relatively high for all commodities except mixed freight (SCTG 43). In all cases, there is a slight decrease in truck market shares as GCD, truck transit time, and truck rate increase. However, this drop in truck market share is more notice- able in the case of mixed freight products, which were found by the estimated models to be more sensitive than other commodities to changes in the values of these variables. To facilitate interpretation of results, the team decided to take out the results for mixed freight, as shown in Figure 49. Plastics and rubber (SCTG 24) is the most sensitive commodity among this group; it starts with a truck market share of 98 percent (Daphne, Alabama) and drops to 95 percent (Dorchester Center, Massachusetts). The graphs in Figure 50 also show that, although the truck market share decreases with these variables, there are segments where it does the opposite, as in the case of P(T) for plastics and rubber for GCD equal to 1,100 miles. The reason for this is that, at these locations, the rail rates (for SCTG 24) are higher relative to other locations. In summary, the analyses show that the truck market share decreases with GCD, truck transit time, and truck rate. The results for specific commodities are shown next. Figures 50 to 56 display each of the commodities individually in order to get a closer look at the market share patterns. Figure 50 presents the truck market share estimated for plas- tics and rubber (SCTG 24). The origin is Metairie, Louisiana, as it is for all the commodities. The destinations along the Crescent Corridor for this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; Atlanta, Georgia; McDonough, Georgia; Hinesville, Georgia; Greenville, South Carolina; Richburg, South Carolina; High Point, North Carolina; Barboursville, Virginia; Virginia Beach, Virginia; Baltimore, Maryland; Monkton, Maryland; Lincoln, Delaware; Philadelphia, Pennsylvania; and Dorchester Center, Massachusetts. Figure 51 shows the truck market share patterns for textiles (SCTG 30). The destinations along the Crescent Corridor used for this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Greenville, South Carolina; Richburg, South Carolina; Charleston, South Carolina; Charlotte, North Carolina; High Point, North Carolina; Durham, North Carolina; Roxobel, North Carolina; Richmond, Virginia; Virginia Beach, Virginia; Arlington, Virginia; College Park, Maryland; Baltimore, Maryland; Snow Shoe, Pennsylvania; Newark, New Jersey; Conway, Massachusetts; and Dorchester Center, Massachusetts.

Figure 48. Truck market shares, shipment-size models, all commodities. Figure 49. Truck market shares, shipment-size models, all commodities except mixed freight.

Figure 50. Truck market shares, shipment-size models, plastics and rubber (SCTG 24). Figure 51. Truck market shares, shipment-size models, textiles (SCTG 30).

Numerical Experiments 159 Figure 52 shows the truck market share patterns for mixed freight (SCTG 43). The destina- tions along the Crescent Corridor used for the analysis of this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; Memphis, Tennessee; Antioch, Tennessee; Knoxville, Tennessee; Charleston, South Carolina; Roxobel, North Carolina; Barboursville, Virginia; and Monkton, Maryland. Figure 53 shows the truck market share patterns for electronic and other electrical equipment and components (SCTG 35). The destinations along the Crescent Corridor used for the analysis of this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Atlanta, Georgia; Greenville, South Carolina; Richburg, South Carolina; Charlotte, North Carolina; Durham, North Carolina; Roxobel, North Carolina; Richmond, Virginia; Pittsburgh, Pennsylvania; Snow Shoe, Pennsylvania; Newark, New Jersey; and Dorchester Center, Massachusetts. Figure 54 shows the truck market share patterns for machinery (SCTG 34). The destinations along the Crescent Corridor used for the analysis of this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Durham, North Carolina; Virginia Beach, Virginia; Camden, New Jersey; Newark, New Jersey; and Dorchester Center, Massachusetts. Figure 55 shows the truck market share patterns for base metal (in primary or semi-finished forms and in finished basic shapes) products (SCTG 32). The destinations along the Crescent Corridor used for the analysis of this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Hinesville, Georgia; High Point, North Carolina; Virginia Beach, Virginia; Baltimore, Maryland; Monkton, Maryland; Snow Shoe, Pennsylvania; and Newark, New Jersey. Figure 56 shows the truck market share patterns for wood products (SCTG 26). The desti- nations along the Crescent Corridor used for the analysis of this commodity, indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Richburg, South Carolina; Roxobel, North Carolina; Richmond, Virginia; Virginia Beach, Virginia; Baltimore, Maryland; and Philadelphia, Pennsylvania. Results: Using the Average Shipment Size Obtained from the PUMS. The results presented in this section were estimated using a process similar to the one presented in the previous section, with the difference that the shipment sizes are obtained using the PUMS data instead of ship- ment size models. As before, the truck market share is presented as a function of GCD, transit time, and truck rate for seven commodities. The results are shown in Figure 57. In this version of the analysis, the truck market share for plastics and rubber (SCTG 24) is the most sensitive of the commodities to distance, transit time, and truck rate. In order to facilitate interpretation of the results, plastics and rubber (SCTG 24) was removed to get a closer look at the other commodity types, as shown in Figure 58. As expected, mixed freight (SCTG 43) is one of the commodity types that is more sensitive to changes in the pre- sented variables. The results for specific commodities are shown next. Figures 59 to 65, display each of the commodities individually in order to get a better look at the market share patterns. Figure 59 presents the truck market share estimated for plastics and rubber (SCTG 24). The destinations along the Crescent Corridor for this commodity, as indi- cated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; Atlanta, Georgia; McDonough, Georgia; Hinesville, Georgia; Greenville, South Carolina; Richburg, South Carolina; High Point, North Carolina; Barboursville, Virginia; Virginia Beach, Virginia; Baltimore, Maryland; Monkton, Maryland; Lincoln, Delaware; Philadelphia, Pennsylvania; and Dorchester Center, Massachusetts.

Figure 52. Truck market shares, shipment-size models, mixed freight (SCTG 43). Figure 53. Truck market shares, shipment-size models, electronic (SCTG 35).

Figure 54. Truck market shares, shipment-size models, machinery (SCTG 34). Figure 55. Truck market shares, shipment-size models, base metal (SCTG 32).

Figure 56. Truck market shares, shipment-size models, wood products (SCTG 26). Figure 57. Truck market shares, PUMS shipment size, all commodities.

Figure 58. Truck market shares, PUMS shipment size, all commodities except plastics (SCTG 24). Figure 59. Truck market shares, PUMS shipment size, plastics and rubber (SCTG 24).

164 Impacts of Policy-Induced Freight Modal Shifts Figure 60 presents the truck market share estimated for textiles (SCTG 30). The destinations along the Crescent Corridor for this commodity, as indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Greenville, South Carolina; Richburg, South Carolina; Charleston, South Carolina; Charlotte, North Carolina; High Point, North Carolina; Durham, North Carolina; Roxobel, North Carolina; Richmond, Virginia; Virginia Beach, Virginia; Arlington, Virginia; College Park, Maryland; Baltimore, Maryland; Snow Shoe, Pennsylvania; Newark, New Jersey; Conway, Massachusetts; and Dorchester Center, Massachusetts. Figure 61 presents the truck market share estimated for mixed freight (SCTG 43). The destinations along the Crescent Corridor for this commodity, as indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; Memphis, Tennessee; Antioch, Tennessee; Knoxville, Tennessee; Charleston, South Carolina; Roxobel, North Carolina; Barboursville, Virginia; and Monkton, Maryland. Figure 62 presents the truck market share estimated for electronic (SCTG 35). The destina- tions along the Crescent Corridor for this commodity, as indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Atlanta, Georgia; Greenville, South Carolina; Richburg, South Carolina; Charlotte, North Carolina; Durham, North Carolina; Roxobel, North Carolina; Richmond, Virginia; Pittsburgh, Pennsylvania; Snow Shoe, Pennsylvania; Newark, New Jersey; and Dorchester Center, Massachusetts. Figure 63 presents the truck market share estimated for machinery (SCTG 34). The destina- tions along the Crescent Corridor for this commodity, as indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Durham, North Carolina; Virginia Beach, Virginia; Camden, New Jersey; Newark, New Jersey; and Dorchester Center, Massachusetts. Figure 64 presents the truck market share estimated for base metal (SCTG 32). The destina- tions along the Crescent Corridor for this commodity, as indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Hinesville, Georgia; High Point, North Carolina; Virginia Beach, Virginia; Baltimore, Maryland; Monkton, Maryland; Snow Shoe, Pennsylvania; and Newark, New Jersey. Figure 65 presents the truck market share estimated for wood products (SCTG 26). The des- tinations along the Crescent Corridor for this commodity, as indicated by the markers on the graphs, are Daphne, Alabama; Wetumpka, Alabama; Birmingham, Alabama; McDonough, Georgia; Richburg, South Carolina; Roxobel, North Carolina; Richmond, Virginia; Virginia Beach, Virginia; Baltimore, Maryland; and Philadelphia, Pennsylvania. Shipment-Level Models The main goal of this analysis, for both the Heartland and the Crescent Corridors, is to estimate the market shares using the choice probabilities for all the shipment records included in PUMS along the two corridors. The analysis employs the unweighted mode choice models, which use generalized cost (with 5 percent as opportunity cost) as the input variable. Commodity-specific models were used where they were available; otherwise, the pooled models were used. To obtain the shipment-level data required by these models, the PUMS dataset was post-processed to only include the relevant records, i.e., those that may be impacted by the corresponding network improvements. The way in which these subsamples were selected varies for each corridor. For each corridor, the three origin zip codes with the highest sample size were selected to apply the models, and to evaluate the impact of the commodity type and the transit time by truck in the aggregated probabilities along each corridor for all the records starting at the selected origin.

Figure 61. Truck market shares, PUMS shipment size, mixed freight (SCTG 43). Figure 60. Truck market shares, PUMS shipment size, textiles (SCTG 30).

98.0% 98.2% 98.4% 98.6% 98.8% 99.0% 99.2% 99.4% 99.6% 0 200 400 600 800 1000 1200 1400 T ru ck M ar ke t S ha re GCD (miles) a. Truck Market Share P(T) vs Great Circle Distance (GCD) 98.0% 98.2% 98.4% 98.6% 98.8% 99.0% 99.2% 99.4% 99.6% 0 10 20 30 40 T ru ck M ar ke t S ha re Truck Travel Time (hours) b. Truck Market Share P(T) vs Truck Travel Time 98.0% 98.2% 98.4% 98.6% 98.8% 99.0% 99.2% 99.4% 99.6% 0 500 1000 1500 2000 T ru ck M ar ke t S ha re Truck Rate (USD) c. Truck Market Share P(T) vs Truck Rate Figure 63. Truck market shares, PUMS shipment size, machinery (SCTG 34). Figure 62. Truck market shares, PUMS shipment size, electronic (SCTG 35).

Figure 64. Truck market shares, PUMS shipment size, base metal (SCTG 32). Figure 65. Truck market shares, PUMS shipment size, wood products (SCTG 26).

168 Impacts of Policy-Induced Freight Modal Shifts For presentation purposes, the results were aggregated by taking the mean value of all the probabilities corresponding to records that lie within a specific interval of driving time by truck. The intervals were constructed starting at 0 and with a length of 4.5 hours. Driving times exceeding 13.5 hours were grouped in the same bin based on the current hours-of-service regulations for truck drivers, previously discussed (FMCSA 2011). The results were aggregated based on the nine super groups of the SCTG codes that the CFS presents in the data user guide for the PUMS. A general overview of the composition of the groups is presented in Table 57 (U.S. Census Bureau 2018). Since the aggregation process is based on several sampling stages and does not account for all the possible records under each one of the conditions analyzed, there is a component of error in the results presented. To overcome the fact that some aggregate estimates might potentially exhibit counterintuitive results because the sampling error is high, a sample size threshold of 50 observations was defined. For each one of the bins corresponding to a driving time interval and a SCTG super group, the average probability estimated is considered reliable if the number of observations in that bin is greater than or equal to 50. The chosen threshold is based on the fact that the data showed relatively small 95-percent- confidence intervals (in terms of length) when the subsample size is more than 50 observations. It should also be noted that the sample size within each one of the bins in the analysis and the reliability of the estimate obtained for each bin are not in the research team’s control because the sample sizes are determined by the PUMS data. The Heartland Corridor’s Inspired Scenarios Scenario #3: Estimation of Truck Market Shares by Commodity Groups For the purpose of this analysis, observations from the states and jurisdictions interacting with the Heartland Corridor were the only ones considered. These were the states of Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, Ohio, Pennsylvania, and Virginia and the District of Columbia. Other observations that were eliminated to arrive at an effective sample size of 249,279 (93.123 percent of the original sample of interest) include the following: (1) intrastate shipments; (2) observations lacking specific information on commodity type; and (3) observations with coal (SCTG 15) as the commodity type, as there was no shipment- size model available for this commodity. Essentially, the results presented in this section repre- sent the estimates of the mode shares from the ZIP codes shown in Table 58 to destinations in the hinterland of the Heartland Corridor (See Figures 66 through 71). In interpreting the results, the reader ought to keep in mind that the market shares are bound to exhibit variability on account of sampling error because the market shares are estimated as the average of the shipment-level probabilities, and the number of observations is relatively low. That is the most likely reason for apparently counterintuitive results, as well as the fact that this numerical experiment does not control for the variability of all the factors involved in the mode SCTG Super Group SCTG General Description 01 - 05 Animals and agriculture 06 - 09 Milled grains, oils, alcohol, and tobacco 10 - 14 Building stone, natural sands, non-metallic minerals, metallic ores 15 - 19 Coal, petroleum, gasoline, fuel oils, other coal and petroleum products 20 - 24 Chemicals, pharmaceutical products, fertilizers, plastics and rubber 25 - 30 Wood, paper and paperboard, printed products, textiles, leather 31 - 34 Non-metallic and base metal materials, machinery 35 - 38 Electronics, motorized vehicles, transportation equipment, precision instruments 1 2 3 4 5 6 7 8 9 39 - 41, 43, 99 Furniture, mattresses, lamps, miscellaneous, waste and scrap, mixed freight Table 57. Super groups considered in the analysis.

Numerical Experiments 169 State of Origin Zip Code Sample Size Illinois 60629 18,314 Ohio 45856 16,628 Philadelphia 16874 15,656 Table 58. Sample sizes for the top three origins chosen in the Heartland Corridor. choice process. However, the reader must be aware that some aggregate estimates were cataloged as “highly unreliable” because of having a low sample size (fewer than 50 observations) and are not displayed in the graphs. Despite the removal of all the observations for coal, the SCTG super group # 4 that contains coal, petroleum, gasoline, fuel oils, other coal and petroleum products (SCTG codes 15 to 19) is displayed in the graphs because there were other observations associ- ated with the remaining commodities in this group that were included in the analysis. Origin ZIP Code: 16874 (Philadelphia) 75% 80% 85% 90% 95% 100% 0 5 10 15 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 66. Truck market shares, ZIP code 16874 (Philadelphia), PUMS shipment size. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 67. Truck market shares, ZIP code 16874 (Philadelphia), Shipment-size models.

170 Impacts of Policy-Induced Freight Modal Shifts 75% 80% 85% 90% 95% 100% 0 5 10 15 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 68. Truck market shares, ZIP code 45856 (Ohio), PUMS shipment size. Origin ZIP Code: 45856 (Ohio) 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 69. Truck market shares, ZIP code 45856 (Ohio), shipment-size models.

Numerical Experiments 171 Origin ZIP Code: 60629 (Illinois) 70% 75% 80% 85% 90% 95% 100% 0 5 10 15 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 70. Truck market shares, ZIP code 60629 (Illinois), PUMS shipment size. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 71. Truck market shares, ZIP code 60629 (Illinois), shipment-size models.

172 Impacts of Policy-Induced Freight Modal Shifts General Findings. For the scenario of Philadelphia, the results indicate that most of the commodities exhibit a strong preference for trucks. Only the groups of coal, petroleum, etc., and chemicals/pharmaceuticals have different patterns. The former shows high sensitivity to driving time while the latter shows the lowest propensity to use trucks. When the shipment-size models are used, the most sensitive groups are grains, oils, etc., and coal, petroleum, etc., with the former reaching probabilities close to zero. In the case of Ohio, the constant patterns are exhibited by all commodity groups with chemicals showing the lowest probabilities, using PUMS information. For the analysis using the shipment-size models, the coal, petroleum, etc., and grains, oils, etc., groups are the ones that appear to be more sensitive. The building stone group also shows a drop in the probabilities under this scenario. The results for Illinois are very similar to those for Philadelphia and Ohio. Building stone, chemicals, and coal groups are the most sensitive to variations in the driving time when the information about shipment size is taken from PUMS. In the context of the shipment-size models, animals and agriculture, and grains, oils, etc., show very sensitive patterns by decreas- ing very rapidly as long as the driving time increases. The Crescent Corridor’s Inspired Scenarios Scenario #4: Estimation of Truck Market Shares by Commodity Type For the Crescent Corridor, the states and jurisdictions under consideration were Alabama, Delaware, District of Columbia, Georgia, Kentucky, Louisiana, Maryland, Massachusetts, New Jersey, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, and West Virginia (17 in total). Similar to the analysis of the Heartland Corridor, the following were eliminated to determine the effective sample of 366,649 observations (90.256 percent of the original sample of interest): (1) intrastate shipments; (2) observations lacking specific information on commodity type; and (3) observations with coal (SCTG 15) as the commodity type, as there was no shipment-size model available for this commodity. Table 59 shows the three origins chosen for the analysis in this corridor. Results are shown in Figures 72 through 77. State of Origin Zip Code Sample Size Georgia 30318 21,589 Philadelphia 16874 20,317 New Jersey 07104 17,226 Table 59. Sample sizes for the top three origins chosen in the Crescent Corridor.

Numerical Experiments 173 Origin ZIP Code: 07104 (New Jersey) 94% 95% 96% 97% 98% 99% 100% 0 5 10 15 20 25 P ro ba bi lit y of c ho os in g T ru ck Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 72. Truck market shares, ZIP code 07104 (New Jersey), PUMS shipment size. 94.00% 95.00% 96.00% 97.00% 98.00% 99.00% 100.00% 0 5 10 15 20 25 P ro ba bi lit y of c ho os in g T ru ck Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 73. Truck market shares, ZIP code 07104 (New Jersey), shipment-size models.

174 Impacts of Policy-Induced Freight Modal Shifts Origin ZIP Code: 30318 (Georgia) 94% 95% 96% 97% 98% 99% 100% 0 5 10 15 20 25 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 74. Truck market shares, ZIP code 30318 (Georgia), PUMS shipment size. 94% 95% 96% 97% 98% 99% 100% 0 5 10 15 20 25 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 75. Truck market shares, ZIP code 30318 (Georgia), shipment-size models.

Numerical Experiments 175 Origin ZIP Code: 16874 (Philadelphia) 95% 96% 97% 98% 99% 100% 0 5 10 15 20 25 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 77. Truck market shares, ZIP code 16874 (Philadelphia), shipment-size models. 95% 96% 97% 98% 99% 100% 0 5 10 15 20 25 Pr ob ab ili ty o f c ho os in g Tr uc k Driving time by truck (h) All Animals and Agriculture Grains, Oils,… Chemicals/Pharmaceutical… Wood, Paper, Textiles… Non-metallic/Metallic Materials Electronic/Electric Equipment Furniture and Miscellaneous… Building stone… Coal, Petroleum,... Figure 76. Truck market shares, ZIP code 16874 (Philadelphia), PUMS shipment size.

176 Impacts of Policy-Induced Freight Modal Shifts General Findings. For the case of New Jersey, the group of wood, paper and textiles is the most sensitive group no matter what the source for the shipment data are. Additionally, the probabilities exhibited by this group are the lowest among the nine groups. The groups for building stone and coal are the ones that present the biggest changes with respect to the source of the shipment information. In the case of Georgia, use of the shipment information from PUMS or use of the shipment- size models produce results that do not differ significantly. Once again, the wood, paper, textiles group shows the lowest probabilities and the highest sensitivity to variation in the driving time by truck. Philadelphia exhibits the same conclusions already mentioned with regard to the wood, paper, textiles group. The commodity groups that are more sensitive to the source of the information about the shipment are the coal group and the building stone group, with the latter showing high sensitivity in both cases.