Guidebook for Developing Subnational Commodity Flow Data (2013)

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.

1 Overview of the Guidebook and Key Issues 1.1 Introduction Information about commodity flows is one of the most critical data needs for freight planning. Commodity flows describe the quantity of commodities that are shipped between origins and destinations and the transportation modes that are used. In other words, com­ modity flow data describe what moves, where it moves from and to, how much of it moves, and the manner by which it moves. This information is critical to freight planning for the following reasons: • Commodity flow data provide a direct link between the inputs and outputs of an economy and the freight flows that the economy gives rise to. Commodity flow data indicate what is produced and consumed in an economy and can be used to generate estimates of demand for freight transportation. Commodity flow information also can be used to understand which industries generate demand for freight and therefore benefit from freight investments. This is a critical input for decisions that prioritize freight investments. • Commodity flow information—the types of commodities, the amount being shipped, and the distances commodities are being shipped—is a critical determinant of mode choice. There­ fore, commodity flow data are critical in modal diversion studies. In order to optimize multi­ modal transportation systems, state departments of transportation (DOTs), metropolitan planning organizations (MPOs), and federal transportation agencies are often interested in examining the costs and benefits of competing modal investments. Commodity flow informa­ tion is critical to determining how much freight is divertible from one mode to another based on transportation agency investment decisions. • Commodity flow data are often used as inputs to state and multistate models that forecast transportation demand. Commodity flow data are critical in multistate corridor studies because they provide the best representation of freight transportation demand at this level of geography. Commodity flow data also are increasingly being used to estimate air quality impacts of the transportation inputs to various industrial activities. Purpose of NCFRP Report 26: Guidebook for Developing Subnational Commodity Flow Data The purpose of NCFRP Report 26: Guidebook for Developing Subnational Commodity Flow Data (Guidebook) is to provide state DOT and MPO freight transportation planners with an easy­to­use manual on how to develop subnational commodity flow databases to meet freight transportation planning needs in their region. The Guidebook describes the limitations of using existing national commodity flow databases for local planning. The Guidebook also describes methods to develop primary commodity flow data using local data collection along with how to augment local data collection efforts with information from published data sets and commodity C h a P T E r 1 . 0

2 Guidebook for Developing Subnational Commodity Flow Data flow disaggregation techniques. The Guidebook provides a detailed, step­by­step description of each method. It also describes the limitations of each method and how these limitations can be mitigated. Uses of the Guidebook There are several ways that the Guidebook can be used. A Guidebook user may want to learn how to apply one of the methods of developing subnational commodity flow data described in the Guidebook. The Guidebook provides detailed instructions on how to use the method, what types of problems it might best be applied to, and its limitations. Chapters 2.0, 3.0, 4.0, and 5.0 describe methods of developing subnational commodity flow data. A Guidebook user may want a comprehensive approach that uses the different methods in combination to produce a subnational commodity flow database, taking into consideration the relative strengths and weaknesses of each of the methods. The Guidebook provides instructions for a procedure that combines the methods. Also, users may not have sufficient resources to build a comprehensive commodity flow database solely from primary research. In this case, combining primary data collection methods with methods that use existing public sources may be necessary. A Guidebook user may want information on the strengths and weaknesses of specific freight data sources. This can be found in a subtask report associated with the development of the Guidebook titled “Review of Subnational Commodity Flow Development Efforts and National Freight­Related Data Sets,” which includes descriptions of existing databases and references on what is included in these databases and is available at www.trb.org/Main/Blurbs/169330.aspx. For primary data collection, the Guidebook provides information about sampling frames and their cost and availability as well as illustrative survey instruments that have been demonstrated during the course of this project. Structure of the Guidebook The Guidebook is structured to emphasize collecting new freight flow data for developing subnational commodity flow data. It is complemented by data analysis efforts such as using locally available freight data and disaggregating national commodity flow data. This structure is designed to allow transportation agencies to choose the approach that best fits with their freight planning needs and available resources. The specific Guidebook sections are as follows: • Chapter 1.0—Overview of the Guidebook and Key Issues • Chapter 2.0—Collecting Subnational Commodity Flow Data Using Establishment Surveys • Chapter 3.0—Collecting Subnational Commodity Flow Data Using Roadside Intercept Surveys • Chapter 4.0—Developing Subnational Commodity Flow Data Using Supplemental Sources of Local Economic Activity • Chapter 5.0—Developing Subnational Commodity Flow Data Using Disaggregation • Chapter 6.0—Playbook Following this introduction, the Guidebook is divided into two main sections. Chapters 2.0 through 5.0 comprise the “Methods” section. They are designed to introduce the user to specific methods that can be used to develop subnational commodity flow data. Each chapter describes a particu­ lar method and is structured to provide step­by­step procedures for implementing that particular method. Examples are provided from actual practice, and there are worksheets that can be used by practitioners to design an application of the method for their own planning studies. Chapter 6.0 is the “Playbook” section. Borrowing an analogy from football, the Playbook section is designed to help practitioners design a “Game Plan” consisting of mixing and matching different

Overview of the Guidebook and Key Issues 3 methods from Chapters 1.0 through 5.0 ( the Methods section) into their own “Playbook” for attack­ ing a particular freight planning problem. The Playbook section also includes examples of actual planning problems and guides the user through the process of selecting and combining methods. Many users will start in the Playbook section after reading this introduction. They will figure out what they need to know and start designing their game plan; the Playbook will guide them back to the Methods section to do the detailed planning of the surveys, data mining, or data disaggregation that they may need to solve their problems. The Playbook section is designed to work together with the Methods section and guide the reader back and forth between the two sections. Other users, those who want to get a complete picture of the whole topic, can start at the beginning of the Guidebook and just read from cover to cover. 1.2 Background on Commodity Flow Data The most commonly used commodity flow database is FHWA’s Freight Analysis Framework (FAF) database. In 2011, FHWA released the third version of this database. FAF pivots off of the Bureau of Transportation Statistics (BTS) Commodity Flow Survey (CFS), and then integrates data from a variety of sources to create estimates for tonnage and value by origin, destination, commodity, and mode for a base year (currently 2007 to be consistent with the CFS) and fore­ casts, currently through 2040. The FAF database is national in scope and the sampling and data collection embodied within it (such as the CFS) are designed to produce statistically valid data at the national level and for very limited subnational geographies. FAF provides data for each state and for 89 regions around the country. Figure 1.1 shows the FAF regions. Table 1.1 shows a snippet of FAF data for a hand­ ful of agricultural commodities originating in Alabama and destined for any location. Table 1.2 shows truck flows from the Norfolk, Virginia, metropolitan region to the Chicago, Illinois, metropolitan region. While there is a wealth of information in the FAF database, states and MPOs need subnational databases that provide more disaggregate geographies for the origins and destinations of flows. Subnational commodity flow databases might involve county­level detail, city­level detail, or even traffic analysis zone (TAZ)­level detail. These databases can be used in several types of applications, including the following: • Trade flow analysis • Development of truck trip tables for travel demand model • Feasibility analysis for new modal services • Estimating the economic impact of freight activity • Regional or statewide freight plans • Corridor studies of freight­intensive roadways • Modal diversion studies Trade flow analyses include studies of the trading partners for states or substate regions (such as counties or MPOs). These analyses are useful for understanding the relationship between geo­ graphic entities such as the relationship between two different states or between an MPO region and an adjacent, external region. Trade flow analyses also provide information on the relative size of internal, internal­external, and through freight trips for a region. Trade flow analyses can be done for total freight flows, specific modes, and specific commodities or commodity groups. They are useful for understanding how a region’s economies relate to the economies in other regions. These analyses also provide a sense of where freight is trying to move to and from for a region. Trade flow data can be depicted graphically as desire lines of freight moving between

4 Guidebook for Developing Subnational Commodity Flow Data regions. Figure 1.2 shows a map depicting trade flow between MPO regions in Georgia using a preliminary output from the state’s travel demand model. The development of a truck trip table is one of the initial steps in creating a truck component for a travel demand model. These tables provide estimates of the number of trucks moving between TAZs in the model area, including external regions. In recent years, a number of states and MPOs have developed truck trip tables from commodity flow data by gathering data on the average payload (or loaded weight) for trucks carrying different commodities and adjusted for empty movements. If the total tonnage of a commodity moving by truck between an origin and destination is known, this can be divided by an average payload for trucks carrying that com- modity to determine the number of truck trips. Generally, the last step in the travel demand model process is route choice, in which the truck trip table is assigned to the road network. While routed commodity flow data can be useful to calibrate and validate travel demand models or to determine which industries use different cor- ridors, routed flows are not required since the model assigns trucks based on predetermined criteria. It should be noted that it is theoretically possible to develop vehicle trip tables for other modes using commodity flow data, but the routing logic for these nontrucking modes may be more complex and is usually not incorporated into state or MPO models. Feasibility analyses for new modal services include studies on the demand for new or altered ser- vice for a specific mode, for example, a study to determine the need for a new rail intermodal yard. Source: FHWA FAF. Figure 1.1. FHWA FAF regions.

Overview of the Guidebook and Key Issues 5 In the case of intermodal rail, the key factors for determination of the feasibility of a new service include the types of commodities that have a trip end near the yard location and the distance that the goods are being carried. For this type of study, the specific route being used by current traffic is not relevant, therefore routing information is not needed. However, a commodity flow database could be used to estimate the commodities that are demanded and supplied at a particular location, in addition to the distances that those goods travel to reach their corresponding trip ends. Estimating the economic impact of freight activity also is something that can be accomplished without using routing information. This type of study is generally used to determine the size of the freight­related components of a region’s economy relative to the regional economy as a whole. In this case, commodity flow databases can be used to estimate the size of the freight­related components of the economy. These databases also can be used to estimate how freight­related economic activity is distributed among various commodities and industries within a region. Regional or statewide freight plans can be developed in several different fashions. However, in order to understand freight activity in a comprehensive regional fashion, it is critical to under­ stand how many freight vehicles there are at different locations on the freight network, the travel patterns of these vehicles, and the freight network that is used. Typically, routed flow information is not provided with commodity specificity, but information on truck routing on the highway network is a critical component of developing a statewide or regional freight plan. Corridor studies on freight­intensive roadways are the most direct example of freight plan­ ning efforts in which national­ or state­level commodity flow information by itself is insufficient. 2007 Origin Commodity Mode Total Tons (Thousands) Total Dollars (Millions) Alabama Live animals/fish Truck 1,569.38 1,691.28 Alabama Live animals/fish Rail 0.01 0.05 Alabama Live animals/fish Air (include truck-air) 0.38 14.51 Alabama Live animals/fish Multiple modes and mail 0.01 0.03 Alabama Cereal grains Truck 3,070.71 642.71 Alabama Cereal grains Rail 0.36 0.08 Alabama Cereal grains Air (include truck-air) 0.02 0.04 Alabama Cereal grains Multiple modes and mail 8.65 1.57 Alabama Other agricultural products Truck 3,678.31 1,457.25 Alabama Other agricultural products Rail 583.76 293.73 Alabama Other agricultural products Air (include truck-air) 0.63 6.15 Alabama Other agricultural products Multiple modes and mail 216.07 149.31 Alabama Other agricultural products Other and unknown 187.27 27.83 Alabama Animal feed Truck 3,997.60 1,169.09 Alabama Animal feed Rail 2.54 9.60 Alabama Animal feed Air (include truck-air) 0.06 1.83 Alabama Animal feed Multiple modes and mail 70.13 40.76 Alabama Animal feed Other and unknown 334.94 20.38 Alabama Meat/seafood Truck 2,069.60 4,705.65 Alabama Meat/seafood Rail 0.02 0.02 Alabama Meat/seafood Air (include truck-air) 3.58 4.01 Alabama Meat/seafood Multiple modes and mail 4.49 12.48 Alabama Meat/seafood Other and unknown 68.77 128.17 Source: FHWA FAF3 Database, 2007. Table 1.1. Example of FHWA FAF data.

6 Guidebook for Developing Subnational Commodity Flow Data 2007 DMSa_ORIG DMS_DEST SCTG2 DMS_ MODE Total Tons (Thousands) Total Dollars (Millions) Norfolk VA-NC MSAb (VA Part) Chicago IL-IN-WI CSAc (IL Part) Live animals/fish Truck 0.04 0.13 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Cereal grains Truck 0.01 0.01 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Other agricultural products Truck 7.89 17.33 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Animal feed Truck 0.44 0.20 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Meat/seafood Truck 8.17 20.96 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Milled grain products Truck 2.06 3.39 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Other foodstuffs Truck 16.62 88.40 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Alcoholic beverages Truck 11.46 26.70 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Tobacco products Truck 0.02 0.19 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Building stone Truck 2.55 0.06 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Natural sands Truck 24.70 1.18 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Gravel Truck 0.01 0.00 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Nonmetallic minerals Truck 15.05 0.07 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Coal Truck 0.00 0.00 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Gasoline Truck 0.04 0.02 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Fuel oils Truck 0.28 0.50 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Coal-n.e.c. Truck 0.07 0.01 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Basic chemicals Truck 6.25 21.05 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Pharmaceuticals Truck 0.00 0.00 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Fertilizers Truck 29.65 4.69 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Chemical products Truck 1.98 7.93 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Plastics/rubber Truck 23.05 36.75 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Logs Truck 0.03 0.01 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Wood products Truck 9.90 10.12 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Newsprint/paper Truck 9.16 5.12 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Paper articles Truck 0.50 1.46 Table 1.2. Sample truck flow data from FHWA FAF database.

Overview of the Guidebook and Key Issues 7 Commodity flow information needs to be complemented or supplemented by specific routed flow information. This can come in many forms. To estimate truck flows using a standard travel demand model, commodity flow information must be disaggregated to the TAZ level and assigned to the network taking into account the corresponding routing of passenger cars and other vehicles. Truck roadside intercept surveys also are a useful tool for understanding corridor­ level truck flows. They can be used to augment or validate disaggregated commodity flow data. A comprehensive roadside survey program can be used as a stand­alone tool to develop commod­ ity flow information for a corridor. Similarly, establishment survey data can be used to better understand commodity flow information at the corridor level. Modal diversion studies also require information about routed flows to fully understand the impact of freight moving on one mode relative to another. Typically, these studies are designed to determine the effect of diverting freight from highway to rail, and the studies attempt to quan­ tify the impact on congestion, safety, road maintenance, and the environment. This requires an Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Printed products Truck 2.28 5.61 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Textiles/leather Truck 0.00 0.00 2007 DMS_ORIG DMS_DEST SCTG2 DMS_ MODE Total Tons (Thousands) Total Dollars (Millions) Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Nonmetal min. products Truck 36.40 16.28 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Base metals Truck 13.23 24.93 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Articles-base metal Truck 14.31 25.46 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Machinery Truck 43.87 361.79 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Electronics Truck 10.29 76.05 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Motorized vehicles Truck 10.60 67.33 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Transport equipment Truck 0.08 5.36 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Precision instruments Truck 0.51 12.21 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Furniture Truck 0.50 0.54 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Misc. mfg. products Truck 7.67 46.93 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Waste/scrap Truck 0.68 0.09 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Mixed freight Truck 2.48 1.64 Norfolk VA-NC MSA (VA Part) Chicago IL-IN-WI CSA (IL Part) Unknown Truck 0.00 0.00 aDMS = Domestic bMSA = Metropolitan Statistical Area cCSA = Consolidated Statistical Area Source: FHWA FAF, 2007. Table 1.2. (Continued).

8 Guidebook for Developing Subnational Commodity Flow Data understanding of the current freight flow routing that is likely to divert from highway to rail. Similar to corridor studies, travel demand models are one useful source of these types of esti- mates. These models typically rely upon disaggregated commodity flow databases, and they can be augmented by roadside surveys or establishment surveys. 1.3 Geographic Issues Related to Subnational Commodity Flow Data As stated earlier, one of the primary issues related to subnational commodity flow data is that available freight flow data are not provided with the geographic detail needed to conduct several different types of common state and MPO freight planning activities. Figure 1.3 shows the differ- Source: GDOT Statewide Travel Demand Model as of October 20, 2010. Figure 1.2. Georgia MPO trade flow map.

Overview of the Guidebook and Key Issues 9 ence between FAF zones and MPO regions in the Pacific Northwest. For the Seattle and Portland metropolitan regions, the MPO boundaries and the urban boundaries that are defined in FAF do not match. Also, county-level detail within these two metropolitan regions is too disaggregated to be covered by FAF. The Spokane MPO falls into an FAF zone characterized as “remainder of Washington.” It includes over 75 percent of the state and by itself would not be able to provide meaningful infor- mation on Spokane’s regional freight flows. As noted previously, national commodity flow databases are developed from samples that do not provide statistically valid data at the subnational geographies that are needed for most state and MPO freight analysis. One approach that has been used extensively to overcome this obstacle is to disaggregate national databases to the subnational level by using economic data as an indicator of the level of commodity production or consumption in the subna- tional zones. This approach is described in more detail in Chapter 4.0 of the Guidebook. Many researchers have noted that in some cases there are weak relationships between commodity production/consumption and the economic indicator data that are often used in the dis- aggregation procedures. For example, industry-specific employment is often used as an indica- tor variable for consumption or production within an industry. However, it has been found that for certain commodities there is not a strong correlation between the quantity of goods produced in an industry and the number of people employed in the industry. Chapter 4.0 of the Guidebook describes a number of disaggregation techniques, identifies some of the weak- nesses in the relationships between commodity production/consumption and specific eco- nomic indicator data, and discusses how best to use disaggregation techniques in combination with other approaches. Figure 1.3. FAF zones and MPO regions in the Pacific Northwest.

10 Guidebook for Developing Subnational Commodity Flow Data Primary data collection is the most straightforward approach for developing subnational commodity flow data. Chapter 2.0 of the Guidebook examines data collection methods for estab­ lishment surveys. Chapter 3.0 of the Guidebook examines data collection methods for roadside truck intercept surveys. The use of global positioning system (GPS) data is an emerging field of tracking freight flows. The following is an overview of these data sources: • Establishment Surveys—The establishment survey procedures described in this Guidebook include sampling and recruiting strategies, survey instrument design, sample sizes for dif­ ferent levels of geography, and sampling frames for identifying potential recruits. One of the issues associated with using establishment surveys at the subnational level is the need to sur­ vey for both outbound and inbound shipments. National establishment surveys, such as the CFS, only need to survey for outbound shipments. Establishment surveys also must weigh the tradeoffs associated with trying to get high levels of geographic detail on the origins and destinations of shipments relative to achieving high survey response rates. • Roadside Intercept Surveys—These data are extremely useful for identifying commodity flows through a specific geographic location. The primary disadvantage is that the locations at which data can be collected are severely limited and likely to be insufficient to establish comprehensive geographic coverage. Roadside surveys also are not useful for collecting data on commodity flows within a metro area. Additionally, these surveys can only be applied to highway flows, so they are not truly multimodal. Finally, roadside intercept surveys provide information on truck origins and destinations, but not the true trip pattern of the cargo being carried. • GPS Databases—It is now possible to purchase GPS data directly from vendors of truck fleet tracking systems. These databases are very useful for understanding truck operational char­ acteristics. The problem in using these for commodity flow studies is that there is no infor­ mation about the commodities carried. Commodity information can sometimes be inferred from the land uses at trip ends, but detailed land use data are not available in most locations. Additionally, it is difficult to control the sampling procedures used for GPS databases, so the statistical validity of the data is questionable. Also, similar to roadside intercept surveys, the data are only available for the truck mode. There are a number of specialized databases, many in the public domain, that provide data on commodity production that can be used to enhance or develop subnational commodity flow data. These are described in the Guidebook. Typically, these databases do not provide geographic detail that is more disaggregated than county­level detail. There also are establishment­level databases that provide employment and value of shipment data at the level of the individual firm that can be used to enhance subnational commodity flow databases with much more geographic detail. These establishment­level databases also are described in the Guidebook. Economic input­output model databases can be used in conjunction with specialized databases to develop information about consumption of commodities. These economic input­output databases are typically used in conjunction with data on the industries that consume these commodities. Chapter 5.0 of the Guidebook describes this approach to developing subnational commodity flow data. The biggest limitation of these specialized databases is that few include information about the origins or destinations of flows. Therefore, the greatest use of specialized databases tends to be in establishing control totals for inbound and outbound flows at subnational geographies of interest. 1.4 Commodity Issues Related to Subnational Commodity Flow Data There are several commodity classification issues that arise in developing subnational com­ modity flow databases. The largest establishment survey in the United States is the BTS CFS.

Overview of the Guidebook and Key Issues 11 However, it only covers certain commodities. The industries that are covered by the CFS fall into the following North American Industry Classification System (NAICS) sectors: • Mining • Manufacturing • Wholesale trade • Select retail trade industries—specifically electronic shopping, mail­order houses, fuel dealers, and publishers • Auxiliary establishments (i.e., warehouses and managing offices) of in­scope multi­establishment companies. An advance survey of approximately 40,000 auxiliary establishments was con­ ducted to identify auxiliary establishments with shipping activity The CFS does not include establishments classified in the following sectors: • Farms • Forestry • Fishing • Utilities • Construction • Government­owned entities (except government­owned liquor stores) • Transportation • Most retail and services industries • Foreign­based business importing to the United States. However, in theory, domestic portions of imported shipments can be captured in the CFS once arriving at a U.S.­based establishment (assuming it is an eligible shipping establishment included in the CFS) Level of Commodity Detail and Variation in Commodity Characteristics The level of commodity detail that is desired in a primary data collection effort will also have a significant impact on the size of the sample required and the complexity of the survey proce­ dures. The tradeoff between level of detail in the final database and cost/complexity of the data collection effort is described in Chapter 2.0 of the Guidebook. If disaggregation of national com­ modity flow databases is utilized, then the strength of the relationship between the economic indicator variable and the estimated tonnage generated will be directly impacted by the level of commodity aggregation that is used. For commodities that are included in the CFS, data accuracy varies across commodity, mode, and origin­destination patterns. This is in part due to the varying sample sizes used for different industries. FAF is derived from CFS and therefore inherits its accuracy issues for in­scope com­ modities. For out­of­scope commodities, FAF weaves together a variety of sources; this generates new accuracy issues that must also be considered. Note that for the last three CFSs, the Stan­ dard Classification of Transported Goods (SCTG) coding scheme was utilized. SCTG codes were developed for the 1997 CFS and are based on the Harmonized System (HS) codes. The level of commodity detail is important for freight planning applications that require analysis of the value of goods moving in particular corridors or amongst regions. The level of commodity detail can also have a significant impact on the ability to accurately estimate vehicle flows from commodity flow data because these estimates are typically calculated on an average ton per vehicle basis. However, commodity detail can vary significantly even within two­digit commodities. Table 1.3 shows the ton­value ratio of the textile industry (SCTG 30) and com­ pares it to the ratios for its three­digit subcomponents. At the two­digit level, the ton­value ratio is 99. However, for the textile industry’s three­digit subcomponents, the ton­value ratio ranges from 46 for footwear to 209 for textiles not elsewhere classified. Therefore, converting vehicles

12 Guidebook for Developing Subnational Commodity Flow Data to tons at the two­digit level for textiles can lead to inaccurate results if the actual commodities being moved are actually concentrated into one of the three­digit categories. Variation of commodity characteristics at more disaggregated levels of commodity detail as compared to a more aggregate classification scheme also is important for expanding survey data collected from new sources at the local level. If similar expansion factors are used for commodi­ ties that are the same at the two­digit level, but vastly different at the three­digit commodity level (for example, with respect to commodity shipment weights, ton­value ratios, or commodity production/consumption per employee), then inaccurate expanded survey data may be gener­ ated. Additionally, accurate commodity classification data are important when disaggregating commodity flow data to the local level since often the local data are provided in units of dollar value and these must be translated into tons. Commodity Classification Systems Commodity flow databases need to use a classification system, and there are a variety of choices, including the following: • Standard Classification of Transported Goods (SCTG). This system was developed by the U.S. Department of Transportation, U.S. Bureau of the Census, Statistics Canada, and Trans­ port Canada to replace the Standard Transportation Commodity Code (STCC) for the CFS and to integrate the U.S. coding system with the Canadian coding system. SCTG tends to be directly tied to industries that create and ship goods. SCTG identifies major commodities car­ ried by each mode of transportation and each significant intermodal combination, and it can be easily linked to the classifications used for international trade. This system has been used in the last two versions of CFS and FAF. • Standard Transportation Commodity Code (STCC). This system was developed initially in the 1960s by the Association of American Railroads for analyses involving the railroad indus­ try. Railroad waybill data, a comprehensive rail commodity flow database available to states, is still published using the STCC coding system. The first two versions of FAF and the CFS also used this system. • Harmonized System. This system was developed by the World Customs Organization as a customs tariff and statistical tool. It also is used by governments, international organizations, and the private sector for setting trade policies, monitoring price and quota controls, and compiling national accounts. SCTG Code Commodity Description Value (Millions Dollars) Tons Ton/ Value Ratio 30 Textiles, leather, and articles of textiles and leather 473,610 46,728,000 99 301 Textile fibers, yarns, and broadwoven or knitted fabrics, except coated or treated 56,837 10,954,000 193 302 Textile clothing and accessories, and headgear (except safety) 262,867 12,452,000 47 303 Textiles and textile article, not elsewhere classified 98,757 20,627,000 209 304 Footwear 37,234 1,706,000 46 305 Leather and articles, luggage of related materials, and dressed furskins and articles 17,916 990,000 55 Source: 2007 BTS Commodity Flow Survey. Table 1.3. Ton-value ratios for textiles and textile subcomponents, 2007.

Overview of the Guidebook and Key Issues 13 • The U.S. Census Vehicle Inventory and Use Survey (VIUS) System—VIUS was part of the U.S. Economic Census, but it has been discontinued. It was used extensively by freight analysts to develop payload factors for converting commodity tonnage into vehicle trips (see previous discussion of creating truck trip tables from commodity flow data). Some analysts continue to use this older data for this purpose. This product developed its own classification system roughly equivalent to SCTG, but aggregated to better match the collected data. • Waterborne Commerce Statistical Center (WCSC) Codes. These codes were developed by the U.S. Army Corps of Engineers. These codes have been standardized to reflect the hierarchical structure of the Standard International Trade Classification Codes, but are focused on com­ modities that are most likely to utilize water transport. These codes are closely tied to HS codes. • U.S. Department of Agriculture (USDA) Crop Report Codes. These are detailed codes with classification schemes that differentiate specific crops (e.g., winter wheat versus spring wheat versus durum wheat). USDA Crop Report codes can be easily rolled into other commodity classification codes. • North American Industry Classification System (NAICS)—NAICS is a collaborative effort by Mexico’s Instituto Nacional de Estadística, Geografía Informática (INEGI), Statistics Can­ ada, and the U.S. Office of Management and Budget. The system is designed to be compatible with the United Nations Statistical Office’s International Standard Industrial Classification System (ISIC), and it was designed to replace the Standard Industry Code (SIC) codes which were previously utilized. Updated NAICS versions are released every 5 years. As noted in its definition, NAICS is not actually a commodity classification system but rather is an industrial classification system. However, some analysts have used it to classify commodities based on the idea that the industry code can be used to represent the primary product produced by each industry. This makes the linkage between industry data and commodity data more transpar­ ent for some users. The research team for the Guidebook discourages use of the NAICS system for the classification of commodities. Other freight flow data sources such as the Port Import­Export Reporting Series (PIERS) and the Census Foreign Trade Data Bureau also have their own classification systems. These classification systems are generally based off the more commonly used codes, but are tailored to specific uses. New commodity classifications are encountered (for example, specialized classifications developed by trade associations to represent their products) when incorporating local data sources. Similarly, when conducting surveys of the private sector, researchers may find that each industry (and some­ times each company) has a unique method of classifying the commodities they move. When conducting local surveys to collect subnational commodity flow data, researchers may find it most useful to collect data and present the final results in SCTG format so that the data can be compared with CFS and FAF data. CFS and FAF data may also be used in state and regional planning to provide control totals; thus, having consistency in the commodity classification scheme is important. However, there is the potential to use other commodity classification codes for specific data collection efforts. These efforts can include surveys focused on the railroads or surveys in which the survey sample of companies already has been predefined into a different system that does not easily translate to SCTG. If different sources of data are going to be used to develop a subnational commodity flow database, they may incorporate various different classification systems. In this case, a methodol­ ogy will have to be developed to bridge from one classification system to the other. Developing these bridge tables must be tailored to the commodity detail that is desired in the final database and the commodity detail that is available from existing sources. Inevitably, there is some loss of accuracy in utilizing a bridge table. For example, in bridging from STCC 24 (Lumber or Wood Products) to SCTG, there are two potential commodity codes that can be utilized: SCTG 25 (Logs and other Wood in the Rough) and SCTG 26 (Wood Products). While a bridge table can be improved by using commodity detail with more specificity (e.g., three­ and four­digit commodity

14 Guidebook for Developing Subnational Commodity Flow Data codes), the usefulness of more specificity will be limited by the amount of detail available in the existing data sources. Relationship Between Commodities and Industries Conducting local establishment surveys typically includes collecting information on both outbound and inbound freight flows for individual companies. To relate outbound and inbound flows, the relationship between industries and commodities needs to be well understood. Gen­ erally, industries produce commodities that are generally consistent with their industrial classification—textile mill companies produce textile products and food product companies produce food products. Table 1.4 shows that essentially all commodities (99 percent) produced by the food manufacturing industry are food manufacturing products, and Table 1.5 shows that 80 percent of commodities produced by the textile mill industry are textile mill products. However, it is common for industries to consume products within the industry as well as products from outside the industry. For example, the food manufacturing industry consumes not only food products (34 percent), but also animal products (31 percent) and crop products Industry $ Millions (2002) Percent Total Food Products 447,676 99% Basic Chemicals 2,233 0.5% Beverage Products 1,183 0.3% Pharmaceuticals and Medicines 555 0.1% Soaps, Cleaning Compounds, and Toiletries 191 0.04% Other Chemical Products 38 0.01% Plastics and Rubber Products 27 0.01% Yarn, Fabrics, and Other Textile Mill Products 22 0.005% Converted Paper Products 11 0.002% Source: BEA Use Table after Redefinitions, 2002. Table 1.4. Commodities produced by the food manufacturing industry. Table 1.5. Commodities produced by the textile mill industry. Industry $ Millions (2002) Percent Total Yarn, Fabrics, and Other Textile Mill Products 35,724 80% Resins, Rubber, and Artificial Fibers 4,774 11% Nonapparel Textile Products 2,441 5% Basic Chemicals 526 1% Plastics and Rubber Products 492 1% Apparel 318 1% Medical Equipment and Supplies 89 0.2% Printed Products 63 0.1% Paints, Coatings, and Adhesives 61 0.1% Soaps, Cleaning Compounds, and Toiletries 59 0.1% Other Miscellaneous Manufactured Products 48 0.1% Converted Paper Products 39 0.1% Furniture and Related Products 35 0.1% Scrap, Used, and Secondhand Goods 14 0.03% Source: BEA Use Table after Redefinitions, 2002.

Overview of the Guidebook and Key Issues 15 (15 percent). It also consumes products that are in completely different industry sectors, such as converted paper products, plastics, and rubber products as well as many others. This contrasts significantly with the commodities it produces, which essentially all fall under the food prod­ ucts category, as seen above. In addition, the textile mill industry consumes 38 percent of textile products, 35 percent of resins and fibers, as well as more than 20 percent of products from other industries, including crop products, basic chemicals and so on. Tables 1.6 and 1.7 show the distri­ bution of goods consumed by the food manufacturing and textile mill industries. A comparison of Tables 1.4 and 1.5 with Tables 1.6 and 1.7 shows how the goods consumed by the food manu­ facturing and textile mill industries differ from the industries themselves according to make and use tables produced by the U.S. Department of Commerce Bureau of Economic Analysis (BEA). Therefore, in collecting subnational commodity flow data for specific industries, it is impor­ tant to understand who the producers and consumers of a specific commodity are. For example, to develop a subnational commodity flow database of the food manufacturing industry, it would not only be necessary to survey companies in food manufacturing, it would also be necessary to survey companies in industries that produce animal products, crop products, and paper prod­ ucts on the inbound side. At the metropolitan and state level, it is likely that several of the producers of a key commodity will be located outside of the jurisdiction of the transportation planning agency. Therefore, it would likely be necessary to survey companies that consume the commodity and are located within the agency’s jurisdiction. Table 1.6. Commodities consumed by the food manufacturing industry. Commodities $ Millions (2002) Percent Total Food Products 83,159 34% Animal Products 74,883 31% Crop Products 37,559 15% Converted Paper Products 16,445 7% Plastics and Rubber Products 8,445 3% Boilers, Tanks, and Shipping Containers 3,974 2% Fish and Other Nonfarm Animals 3,835 2% Basic Chemicals 1,276 1% Other Commodities 7,775 6% Source: BEA Use Table after Redefinitions, 2002 (excluding service sectors). Table 1.7. Commodities consumed by the textile mill industry. Industry $ Millions (2002) Percent Total Yarn, Fabrics, and Other Textile Mill Products 8,020 38% Resins, Rubber, and Artificial Fibers 7,416 35% Crop Products 1,725 8% Basic Chemicals 914 4% Soaps, Cleaning Compounds, and Toiletries 616 3% Semiconductors and Electronic Components 508 2% Other Fabricated Metal Products 360 2% Apparel 317 2% Converted Paper Products 259 1% Animal Products 246 1% Nonmetallic Mineral Products 226 1% Plastics and Rubber Products 178 1% Paints, Coatings, and Adhesives 164 1% Source: BEA Use Table after Redefinitions, 2002 (excluding service sectors).