Freight Trip Generation and Land Use (2012)

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

37 This chapter discusses the connection between freight activity and land use, building on both empirical evidence and theory. Considering these interactions is important because both systems influence each other. On the one hand, land use patterns could impact FTG patterns as the different activities generate different amounts of freight-related trips; conversely, the freight system could also have a significant impact on land use, which is typically the case with large developments such as distribution centers, terminals, ports, and intermodal cen- ters, which not only influence the freight flows but also the geographic patterns of land use surrounding them. As clearly established in Chapter 2, the lack of consensus with respect to a definition for the term “land use” blurs the level of clarity needed to accurately describe the con- nections between freight and “land use.” Although there is some evidence of the application of the LBCS for freight, the comprehensiveness of the dimensions (e.g., activity, func- tion, structural characteristics, site development character, and ownership) would be very useful for understanding the relationships between the freight system and land use. For example, in studies that use the ITE Manual land use classi- fications (i.e., primarily structure-based or site descriptors), it should be possible to map these classifications to the LBCS Structure categories, while studies using employment codes (e.g., SIC or NAIC) could be mapped to the LBCS Function categories, and those using land use planning designations could be mapped to the LBCS Activity categories. Each of these dimensions could have a different impact on FG or FTG, making it essential to reclassify various study outcomes. In describing the connections between the freight system and land use, it is important to distinguish between two sepa- rate aspects: (1) how land use at the establishment level influ- ences FTG; and (2) how freight activity and land use interact with each other at the system level. These effects are shown in Figure 7. Although both aspects are important, since the main emphasis of this project is the impact of land use on FTG, the freight land use connections are not discussed here. Determining how land use impacts FTG requires resolv- ing and reconciling the difference of opinions between the economic/logistic and the transportation literature. NCHRP Synthesis 384: Forecasting Metropolitan Commercial and Freight Travel (2008a) identified as a modeling challenge how truck travel could be modeled without a direct connection to the economic activity that is generating the demand for the movement of the cargo. The economic/logistic literature sug- gests that FTG is determined by the FG (in itself the output of an economic process), along with a host of interactions con- cerning shipment size and total logistics costs. Interestingly, this body of literature barely mentions land use as a factor. The reason seems to be that, in most cases, land use is a con- straint to the production process, not an input. From the eco- nomic/logistic point of view, the input factors that determine FG and FTG include labor, capital, and other intermediate inputs to the process. In essence, the larger the employment or the capital, the larger the FG (while other factors, as discussed, determine the impact on FTG). The passenger transportation literature is inspired by a different paradigm, dating back to the influential work by Mitchell and Rapkin (1954), which established the impact that urban land use has on passenger traffic generation. From this perspective—which obviously does not explore the eco- nomic and logistic aspects of the underlying processes—land use variables such as built area are the ones that explain FTG. Critics point out, however, that such variables cannot mea- sure the magnitude of the use of space, and that other input factors, such as employment, are better explanatory variables. In contrast, the present research analyses conducted by the team stress the importance of studying FG as well as FTG. The analyses described in this report indicate that business establishments attract and produce cargo FG that translates into freight vehicle trips FTG. The amount and nature of the incoming and outgoing FG depends on the type of busi- ness, and its size. In contrast, the FTG depends on the cor- responding shipment sizes, and the ability of the carriers to C h a p t e r 4 Freight Trip Generation and Land Use

consolidate their shipments (e.g., with the shipments of other establishments). Other factors, such as storage capacity con- straints, inventory and transportation costs, etc., play a key role in determining shipment sizes, delivery frequencies, and the amount of inventory. This suggests that the establishment’s land use is, at best, a proxy for the underlying economic activity being conducted. However, in the absence of detailed information about an establishment’s economic characteristics, assuming that FTG depends on general characteristics of land use may just be a pragmatic solution. The weakness of this decision is that vari- ous land use classes group together economic sectors with fun- damentally different FTG patterns. In essence, the adequacy of land use attributes as explanatory variables depends on how well the land use class matches the FTG patterns of the indus- try segments that have been included. In cases where there is a good match, land use is likely to be a good predictor. In con- trast, if a land use class groups together disparate economic activities, it is unlikely to be a good explanatory variable. To illustrate the divergent ways used to define a land use class, and the implications in terms of FTG analysis and modeling, a subsample of the available disaggregate data (collected from about 800 businesses in the New York City area) was analyzed, and the best FTG models for each two- digit SIC groupings were estimated. Then the team mapped some of the land use definitions reported in the literature (Fisher et. al. 2001). See Table 24. Although the analysis is based on a subsample, it provides some interesting conclu- sions. The table shows, for each SIC code, the parameters of the FTG models estimated with the New York City data. (Only statistically significant parameters are shown.) Two parameters are displayed: a constant (the number of deliv- eries per establishment) and the number of deliveries per employee. In some cases, if only the constant is shown, the FTG for that industry sector does not depend on employ- ment level. If only an FTG rate per employee is shown, it means that the FTG increases proportionally to employment level. If both parameters are listed, the implication is that the FTG for that particular industry sector has a minimum value that increases with employment level. As shown in the table, in the case of New York City, the majority of industry sectors have constant FTG. The second largest group has both a constant and a term that increases with employment. The minority of the industry sectors exhibit FTG that increases proportionally to employment. Key implications from the analyses are as follows: • There is a lack of uniformity in the definition of land use classes. • The land use classes typically group together a number of highly heterogeneous industry sectors, with different FTG patterns. See for instance the industry sectors listed under “Retail.” • It is difficult to borrow FTG rates from one location to another. • In computing an FTG rate as a function of a land use vari- able such as square footage, the analyst assumes that FTG depends on business size, when in fact this is not the case for a sizable number of industry sectors. (Since for the same line of businesses, employment is likely to be corre- lated with business area, it is also likely that the widely used square footage only plays a role in industries that exhibit such correlation between FTG and business size.) • FTG rates based on land use classes that group industry sectors that do not share similar FG and FTG patterns are not likely to do a good job of explaining FTG. 38 Inputs (cargo) Freight trip generation Notation Output (Freight generation) Establishment: * Economic activity * Size: - Employment, - Area, etc. * Land use class * Other attributes Carrier: Based on shipments from the establish- ment and others, decides on: * Vehicle type * Delivery frequency (Freight generation) (Freight Trip Generation) Main focus of the project Surrounding Land Use System: * Spatial distribution of activities * Externalities (positives and negatives) Figure 7. Schematic of connections between freight and land use.

39 .r G rotce S CI S noit pircse D CI S ts E/le D p m E/le D Z A ,xi neo h P a de mal A A C ,yt n uo C A G , at nalt A E M ,rog na B 1 I I I I 2 I I I I 7 I I I I 8 I I I I 9 I I I I 10 I I I I 12 I I I I 13 I I I I 14 I I I I 15 0.132 OB BS I S 16 2.467 OB BS I S 17 2.508 OB BS I S 21 3.377 I I I I 22 3.377 I I I I 23 3.778 I I I I 24 0.066 I I I I 25 1.434 0.027 I I I I 26 3.377 I I I I 27 3.377 I I I I 28 3.377 I I I I 29 3.377 I I I I 30 3.377 I I I I 31 3.377 I I I I 32 3.377 I I I I 33 3.377 I I I I 34 2.875 I I I I 35 3.377 I I I I 36 3.377 I I I I 37 3.377 I I I I 38 3.377 I I I I 39 Agricultural production-crops Agricultural production-livestock and animal specialties Agricultural services Forestry Fishing, hunting, and trapping Metal mining Coal mining Oil and gas extraction Mining / quarrying of nonmetallic minerals, except fuels Building constr-general contractors and operative builders Heavy construction other than building construction-contractors Construction-special trade contractors Tobacco products Textile mill products Apparel and other finished products made from fabrics and similar material Lumber and wood products, except furniture Furniture and fixtures Paper and allied products Printing, publishing, and allied industries Chemicals and allied products Petroleum refining and related industries Rubber and miscellaneous plastics products Leather and leather products Stone, clay, glass, and concrete products Primary metal industries Fabricated metal products, except machinery and transportation equipment Industrial and commercial machinery and computer equipment Electronic and other electrical equipment and components, except computer Transportation equipment Instruments and related products Miscellaneous manufacturing industries 3.377 I I I I 4 g nir utcaf u na M 1 ,er utl ucir g A d na , yrtser of seire hsif 2 lare ni M seirts u d nI 3 -c urts n o C n oit seirts u d nI 40 Railroad transportation OB BS I S 41 Local/suburban transit/interurban highway passenger transportation OB BS I S 42 Motor freight transportation and warehousing OB BS I S 43 United states postal service OB BS I S 44 Water transportation OB BS I S 45 Transportation by air OB BS I S 46 Pipelines, except natural gas OB BS I S 47 Transportation services OB BS I S 48 Communications OB BS I S 49 Electric, gas, and sanitary services OB BS I S 5 ,n oitaci n u m m o C , n oitatr o ps nar T seitilit U d na Table 24. Mapping of SIC and land use definitions found in the literature. (continued on next page)

40 These considerations suggest that ensuring a good match between land use classes and the underlying FG and FTG pat- terns could be accomplished by either one of the following: • Redesigning FG and FTG modeling so that it could be properly linked to the land use classification system being used at a particular jurisdiction, and/or • Fostering the use of land use classification systems that are consistent with the underlying patterns of FG and FTG. An attractive way to redesign FG and FTG modeling to deal with the challenge associated with the potentially signifi- cant number of different definitions of land use is to: • Estimate FG and FTG models for the various industry sec- tors captured by the CFS data. • Develop tools that enable Metropolitan Planning Organi- zations (MPOs) and State Departments of Transportation (DOTs) to mix the industry sector FTG models in the proper proportions, according to the local mix of indus- try types. The latter could be readily obtained from the ZIP code Business Patterns data (U.S. Census Bureau 2011). This modeling strategy is illustrated in Figure 8. As shown, if FG and FTG models are estimated for the various indus- try sectors, they could be mapped into any local definition of land use classes using properly defined mixing functions that reflect the proportions in which the different industry sectors—e.g., the number of wholesale trade and eating and drinking places—are found in that particular jurisdiction. Among other things, this modeling strategy enables one to take full advantage of the CFS micro-data, and to match the resulting models to the land use classes currently in use by the relevant transportation agencies. Recent studies in urban areas such as Seattle, Washington, Sacramento, California, and San Francisco show clear descrip- tions of the economic benefits of freight activities at the regional level. They also clearly describe the negative impacts of freight activities at the local level, such as conflicts with neighboring community developments, noise, etc. (see Cambridge System- 50 Wholesale trade - durable goods 3.071 0.054 R R R R 51 Wholesale trade - nondurable goods 1.813 0.074 R R R R 52 Building materials, hardware/garden/mobile home dealers 0.353 R R R R 53 General merchandise stores 2.899 R R R R 55 Automotive dealers and gasoline service stations 0.353 R R R R 56 Apparel and accessory stores 1.314 0.032 R R R R 57 Home furniture, furnishings, and equipment stores 3.714 R R R R 59 Miscellaneous retail 2.902 R R R R 20 Food and kindred products 1.609 0.010 I I I I 54 Food stores 2.764 0.011 R R R R 58 Eating and drinking places 2.017 0.034 R R R R -e v o G ,)I( lairts u d nI ,) R( liate R ,eciff O sesse nis u b re ht O ,s dl o hes u o H ,t ne m nr ) B O( seci vres sse nis u B ,)I( g nir utcaf u na M ) O( re ht O ,) R( liate R ,) S B( ,eciff O ,) R( liate R ,)I( lairts u d nI n oital u p o P ,)I( laicre m m oc w ol/lairts u d nI ,) R( liate R ,) S( la n oit utits ni /eciff o/seci vre S dl o hes u o H 8 d o o F LAND USE CLASSES USED IN THE MODELING EFFORT: 6 -el o h W elas e dar T 7 e dar T liate R Note: For some industry segments where not enough data were available, models were estimated for industry group. .r G rotce S CI S noit pircse D CI S ts E/le D p m E/le D Z A ,xi neo h P a de mal A A C ,yt n uo C A G , at nalt A E M ,rog na B Table 24. (Continued).

41 • It is expected to have a solid connection to the kind of eco- nomic variables used in transportation planning forecasts. This alternative, however, does require a complementary step, involving a simple model to estimate freight traffic from the estimates of tonnage. This could take the form of a lookup table that estimates FTG as a function of the cargo produced/ received. These estimates must also take into account the gen- eration of empty trips, which typically represent 20% of truck traffic in urban areas, and 30–40% of interstate truck traffic. Although these approaches require commodity flow data, this should not be problematic because the data can be obtained from the CFS. The CFS micro-data are available for use at the Regional Data Centers sponsored by the Census Bureau, though securing access requires a lengthy process. Since the CFS micro-data is collected from about 100,000 establishments and atics, Inc. 2010b; The Tioga Group et al. 2006; Hausrath Eco- nomics Group and Cambridge Systematics, Inc. 2004). In these regions, there is a growing maturity with respect to understand- ing freight and land use, which provides an opportunity to uti- lize local knowledge to advance FG and FTG models. It is important to understand that explicitly modeling FG and FTG separately is very convenient. Table 25 shows a sum- mary of the key pros and cons associated with the use of the different metrics and approaches. The most obvious feature of the table is that there are many tradeoffs to consider. How- ever, from the conceptual point of view, using tonnage to measure the FG has obvious advantages, such as the following: • It enables one to treat FG and FTG as separate concepts. • It enables one to explicitly consider the shipment size and its impacts on vehicle/mode choice. FG model 7 FTG model 7 FG model 8 FTG model 8 FG model 1 FTG model 1 FG model 2 FTG model 2 FG model 3 FTG model 3 FG model 4 FTG model 4 FG model 5 FTG model 5 FG model 6 FTG model 6 FG and FTG models estimated for industry sectors (NAICS) using the CFS microdata Land use classes defined and used by a particular planning jurisdiction FG model 9 FTG model 9 FG model 10 FTG model 10 Retail Light Industry Office W1,R W2,R W4,R W3,LI W5,LI W7,O W8,O W9,O W6,LI W10,O Note: In this example, the weighting factors (Ws) correspond to the mix of industry sectors (i.e., number of establishments per industry sector) for a given land use. Figure 8. Schematic of proposed approach. Approach / Metric Advantages Disadvantages Solid connection to economic variables Able to consider the role of shipment size Could use the CFS micro-data Requires commodity data Easy to measure Weak/No connection to economic variables Consider loaded and empty trips In some cases, not related to business size Easy to measure Weak/No connection to economic variables In some cases, not related to business size Only reflects the loaded trips Requires the use another model to estimate vehicle-trips (loaded and empty) generated Freight Generation Models using Commodity tonnage Freight Trip Generation Models Using Vehicle-trips Freight Trip Generation Models Using Deliveries Table 25. Advantages and disadvantages of different freight demand metrics.

42 • Develop simple computational tools to convert the FG and FTG models by industry sectors into models that match the land use classes used by the transportation agencies in charge of the analyses as well as estimate freight traffic from the FG models developed from the CFS micro-data. Summary The analyses indicate that the ability of land use variables to explain FG and FTG depend on how well the different land use classes are able to represent the economic/logistic processes that impact FG and FTG. In this context, if a land use class encom- passes a set of disparate industry sectors with very different FG and FTG patterns, the corresponding land use variables can- not be expected to be good explanatory variables. On the other hand, if the industry sectors under a given land use class exhibit similar FG and FTG patterns, land use variables are likely to do a better job. As a result of these considerations, ensuring a good match between the land use class and the industry sectors within it is a must. Achieving this would require: • Developing FG and FTG models by industry sector that could be mapped into the land use classes used by a given planning agency, using an appropriate mixing function. • Fostering the adoption of land use classification systems that provide a good match between the land use classes and the underlying economic sectors. contains about 4.9 million shipments nationwide, it should provide a solid foundation for FG/FTG modeling. For refer- ence purposes, the authoritative and important ITE Trip Generation Manual contains data collected from about 4,800 trip generation studies (Institute of Transportation Engineers, 2008). This means that the CFS micro-data could provide, every 5 years, an amount of data equiva- lent to 20 ITE Trip Generation Manuals. Furthermore, by using the CFS micro-data, the freight modeling commu- nity would need to do the following: • Use the best approach from the conceptual point of view that decouples the generation of demand from the genera- tion of freight traffic. • Take advantage of a massive data set that is collected every 5 years and covers almost all relevant economic sectors in the nation. • Produce FG/FTG models for all freight-related industry sectors across different regions. • Map these industry sector models into the various defini- tions of land use adopted by the different MPOs and state DOTs using mixing distributions that reflect the local employment distributions. • Still require to estimate freight traffic from the FG mod- els developed from the CFS micro-data. However, in the opinion of the team, this is a small price to pay for exploit- ing the potential of the CFS data.