Methodology for the Freight Analysis Framework-2:
Forecasts of Inter-regional Commodity Flows

2. Forecast Methodology

The foundation of the approach to the freight forecasting is the consistency in outlook provided by using Global Insight's macroeconomic, regional, inter-industry, and intrastate forecast modeling capabilities. These economic forecasting models are built and maintained with a common framework and perspective that provides comprehensiveness, consistency, and detail unique for freight transportation forecasting. Most importantly, this means that the resulting detailed freight flow forecasts are derived in a manner consistent with the path of the economy at a national, regional, and sub-state level.

This section provides a general overview of the forecasting methodology, while the following two sections provide a more detailed examination of the steps taken in producing the domestic and international forecasts.

The initial calibration in the forecasting process involves two distinct steps. The first is constructing the desired level of geography in the Business Demographics Model and the Business Transactions Matrix relative to the 2002 FAF2 base year data. The creation of the FAF2 region geography in these two models is an aggregation process, grouping the county-level data into the FAF2 regional market definitions, and summing the values.

The second step during this initial stage entails the development of the crosswalk between the North American Industry Classification System (NAICS) industry sector classifications and the two-digit level of the Standard Classification of Transported Goods (SCTG) commodity classification. This was done through a review of existing commodity classification concordance files, which detail the relationships between various combinations of NAICS, SCTG, the Standard Transportation Commodity Classification (STCC) and Standard Industry Classification (SIC) codes at various levels of detail. The crosswalk between industry and commodity classifications is important because it provides the bridge between the value and weight of the physical commodities and products shipped through the transportation system and the industry activity measured by economists on an industry establishment level, typically using the value of output or purchases and the associated employment.

The development of the baseline commodity tonnage forecasts uses a multi-step approach:

1. Establish national control totals by commodity;
2. Apply specific shipment growth by market and commodity;
3. Apply specific purchasing and consumption growth by market and commodity;
4. Summarize & compare the results from steps 2 & 3 with the national controls;
5. Adjust the resulting freight flows so that the volumes correspond with the national control levels as follows:

For each market and commodity, adjust so shipments match purchases.

For each commodity, adjust so that national control totals are satisfied.

2A. Domestic Forecast Methodology

The first step in creating the forecast of the FAF II data set is to extract the county-level employment and the U.S. dollar value of output information, by 6-digit NAICS code, from Global Insight's Business Demographics Model (BDM). This data covers each of the forecast years from 2002 to 2030. A separate routine is used to extend the forecast data from 2030 to the 2035 forecast horizon. Section 3 of this report contains a detailed description of the BDM.

The employment data from the BDM is then converted from NAICS into SCTG categories, and aggregated to the 2-digit SCTG level to conform to the FAF II 2002 baseline data. The concordance table identifying the relationships between NAICS and SCTG coding systems is used in this processing. The programming code with the NAICS to SCTG logic is presented in Section 4, in a tabular representation.

The NAICS to SCTG relationships were refined using input from a variety of sources including the Harmonized System commodity classification (HS) to SCTG mapping created by MacroSys, Inc. for the FAF2 project. Additional commodity coding relationship tables were consulted and reviewed including mappings from the U.S. Census Bureau between NAICS, SCTG, HS, SIC and STCC. Extensive cross-referencing was required to insure that all detailed NAICS industry categories in the BDM were assigned to a SCTG commodity code, and also that all SCTG commodity category codes in the FAF2 data were assigned to a NAICS industry sector classification.

Concurrent with the extension of the BDM output and employment forecasts to 2035, the county-level data is summarized to match the geographic market region definitions used in the 2002 FAF2 base year data set. The counties were mapped to the FAF2 geographic regions using the definitional assignments provided by FHWA. The output and employment data is then converted to growth rates and the results are cross-checked and verified against the growth rates for the individual constituent counties.

The independent forecast variables include data from Global Insight's Business Transaction Matrix (BTM), described in Section 5 of this report. The BTM Input/Output (I/O) tables require a similar methodology for translation of the NAICS industry classification codes to SCTG commodity category codes, and the county-level geography to the FAF2 geographic market regions. Minor adjustments to the NAICS to SCTG relationships were made to insure that all SCTG categories had assignments from NAICS industry categories, as follows:

NAICS 212322 25% to SCTG11 & 75% to SCTG12

NAICS 211111 45% to SCTG16 & 55% to SCTG19

NAICS 324110 26% to SCTG17, 25%t to SCTG18 & 50% to SCTG19

The procedure next takes the original 2002 baseline FAF2 domestic freight flow data set, and creates two versions for subsequent processing. The first version maintains all of the original modal detail (truck, rail, air, water, truck & rail, pipeline, and other) and their associated tonnage and dollar volume values. This version of the file is then used later in the process to apply the appropriate modal share distribution across the forecast lane volumes. The second version of the file aggregates each of the modal volumes, again maintaining both the tonnage and dollar values, into region-to-region traffic lane totals.

The total domestic shipment volumes are projected out to the forecast horizon using the forecast information from the BDM, converted to annual growth rates. The result is a table that for each region-to-region SCTG commodity flow, a forecast tonnage and dollar value for each of the forecast years is produced.

The BTM Input/Output data are integrated with the 2002 base year FAF2 data, so that for each region - SCTG commodity combination there is a complete set of associated SCTG commodity volumes that are purchased or consumed. The base year purchases or consumption volumes are then forecast to each year of the forecast period using the BDM output and employment data growth rates.

At this point the national level freight forecast, based on the most recent Global Insight U.S. quarterly-frequency economic data, is used to establish aggregate level benchmark freight volumes for each SCTG commodity category. The total 2002 base year FAF2 freight flows, by SCTG commodity, are forecast using the national level forecasts of output and consumption.

Once these benchmark values have been established the next step is a rebalancing of the original BDM-based region-to-region forecast. This is an iterative process whereby the detailed regional geographic market-to-regional geographic market commodity flow volumes are adjusted to and constrained by the national benchmarks. Upon completion of this iterative scaling, a series of tables are created that analyze the forecast changes in annual growth rates. These results were extensively reviewed before the final reallocation of the total forecast lane volumes by mode. This is where the modal distributions from the 2002 base year FAF2 data by commodity, by origin and destination are used to distribute the total forecast lane volume.

2B. International Forecast Methodology

The procedure for forecasting the international components of the FAF2 data are similar in nature to those used for the domestic traffic, but some adjustments are needed due to the different underlying growth drivers for international business transactions and the additional gateway or port market definitional dimension that are incorporated. The process of producing the international forecasts treated the import and export portions of the international data separately, as the treatment of suppliers and consumers is asymmetrical with respect to the level of detail available on each end of the transaction, with much more detail available on the U.S. end of the shipment.

The base year FAF2 data is maintained throughout the processing in separate files of imports and exports, and one set of files maintains the original mode distinctions (truck, rail, air, water, truck & rail, pipeline, and other), while a second set of files is created that provides the lane total volumes for the sum of all of the modes. Tonnage and dollar values are maintained in each.

Unlike the domestic data, the international records also contain the gateway or port market which identifies where flows enter or exit the U.S. The originating foreign geographic market for imports and the foreign destination geographic market for exports is identified by the foreign region in the base year 2002 FAF2 data and in the forecasts.

The growth rates of U.S. imports and U.S. exports, by commodity, from Global Insight's World Trade Service World Trade Model (WTM) are applied to the FAF2 base year international data to obtain forecasts flows by the gateway/foreign geographic regional market/SCTG commodity combination. Additional information about the WTM is found in Section 3C.

In order to apply the WTM to the FAF2 2002 base year data, the commodity classifications of the WTM commodity trade models had to be translated to SCTG commodity categories. The relationship between the WTM commodity groups and SCTG commodity codes is shown in Section 4B. Additionally, the geographic regional market areas used in the WTM were translated to match those used in FAF2, as shown is Section 4C.

With the needed commodity and geographic regional market mappings complete, export volume growth was established by regional market and commodity from the BDM output data, and the WTM foreign import purchases data. For U.S. international import volume growth, also by geographic regional market and SCTG commodity category, the shipment level import freight flow forecast was a function of the WTM import forecast, and the purchases from the BTM.

As in the domestic forecasts, national-level constraints by SCTG commodity category where applied in an iterative process. Import shipment level forecasts were controlled by purchases, and export purchases are controlled by shipments. Once the national level constraint was applied, a similar process was completed for each port and SCTG commodity combination. The resulting file at this stage contains total tonnage and dollar values for each forecast year for each regional market-gateway region-SCTG commodity combination.

A quality control step similar to that used for the domestic forecasts produced output formatted with annual growth changes for each SCTG commodity, gateway and SCTG, and foreign geographic region market and SCTG commodity category. After review of this summary data, the modal share splits were then applied to create the final detailed import and export international traffic forecast files.

 

previous | next