Freight Forecasting Model
The principal tool in forecasting freight is FTR’s Input/Output model of freight transportation in the United States. This model was first constructed in the mid 1980’s and was completely re-estimated in the 1997-99 time period. This model was developed by estimating the movement of (virtually) all items with weight in the U.S. economy. There are several critical freight movement issues analyzed within the structure of FTR’s freight transportation model. These issues are:
• How far is each move?
• How many tons are moved?
• What products are moved?
• What are the weight changes between each move?
• How many times are the products moved?
• What type of move is made?
• Who does the moving?
Data on the tons of virtually every product originating in the U.S. economy were collected for each year from 1965 through 2004 (the latest year available as of this time). This data is then compiled into the following product categories:
• Coal, Petroleum and Natural Gas
• Aggregates
• Metallic Ores and Machinery
• Non-Metallic Ores and Chemicals
• Food and Feed
• Wood and Building Materials
• Household Consumables, Paper and Furnishings
• Post Office, Delivery, Moving and Other Services
• Waste, Trash, and Recycling
The next step in FTR’s freight transportation model is to move these tons of products between different levels of the country’s economic infrastructure. These levels are:
• Extraction
• Basic Manufacturing
• Finished Manufacturing
• Wholesale Distribution
• Retail Distribution
• Waste and Trash Removal and Recycling
These moves are done using four distinctly different types of moves, which are:
• To the Next Stage
• Intra (Movement Within the Stage)
• Imports
• Exports
These tons are then assigned to a 3-digit STCC (Standard Transportation Commodity Code) category using appropriate data from various sources, and when no appropriate source is found the U.S. Department of Commerce’s Commodity Flow Survey is used as an initial estimate on splitting the data into smaller categories before calibration. Using these estimates the 3-digit categories are calibrated back to the total tonnage moved.
The following modes of transportation are assigned to move the tonnage. They are:
• Rail
• Truck
• Pipeline
• Water (including domestic coastal)
Length-of-Hauls are set by mode at the 3-digit commodity level to get tonmiles.
Rail, Water and Oil Pipeline tons and tonmiles are calibrated to the numbers published in the ENO Transportation Foundation Report “Transportation in America.”
Rail average length-of-hauls are taken from the Rail Waybill database and calibrated, when necessary, to the total tonmiles published by ENO. These numbers are then split to the 3-digit level using the Public Use Waybill Database.
The Water Length-of-Haul data is taken from the U.S. Army Corps of Engineers’ Waterborne Commerce Statistics Center.
Natural Gas traffic is calibrated by assuming that their lengths of haul are the same as that of oil pipelines.
The next step is to convert the annual tons and tonmiles estimates to a quarterly series, which is done by using the most appropriate available economic sequence as a pattern. Forecasting of total freight tons and tonmiles is done by using the historical pattern of the relationship between the same economic series and the relevant tons series. Economic forecasts are furnished by the Center for Econometric Model Research of Indiana University.
The forecasts set the tonnage modal share constant from the last known value through the forecast period. The tonnage is then multiplied by the average length-of-haul to derive tonmiles. This is done at each 3 digit commodity by mode. The tonmiles for all modes are then added together to get the total tonmiles for that 3 digit commodity.
- Historical tons of Rail freight at the 3-digit STCC levels are obtained from AAR annual reports. Historical Rail modal shares are calculated and used to forecast Rail modal share of tons at the 3-digit STCC levels. These Rail modal share forecasts are used to calculate the forecast of Rail tons. Rail tonmiles at the 3-digit STCC levels are estimated in a similar fashion. Rail carloadings at the 3-digit STCC levels are forecast by using the historical method of average tons per car.
- This approach includes virtually all freight generated by the U.S. economy. Our estimation of freight modal shares are distinctly different from other published sources. There are two major reasons for these different modal share estimates. One is the inclusion of short haul trucking which is not included in many other sources. The second is the inclusion of natural gas, which also is missing in other reports. This has the effect of significantly raising truck modal share of tons and to a lesser degree that of tonmiles. Pipeline modal share is also higher, while Rail and Water modal shares are lower.
Truck tonnage is assigned by type to Classes 5-7 Truck, Class 8 Straight Truck, or Class 8 Tractor/Trailers at the 3-digit level. An average length-of-haul is set by type at the 3-digit level to get tonmiles. Using the average length-of-haul set at this point assigns it to a length-of-haul category. These categories are as follows:
• Short-Haul: 0-124 miles average length-of-haul
• Medium-Haul: 125-299 miles average length-of-haul
• Long-Haul: 300 or greater miles average length-of-haul
The 3-digit commodities are also assigned to a “product type” according to the intrinsic characteristics of the product from a transportation viewpoint. They are as follows:
• Raw
• Processed
• Manufactured
Using the above assignments we are able to show the commodities using a 3 x 3 matrix when the commodities are cumulated into 2 digit categories.
For more information regarding the Freight Model visit the following links:
For additional information on data sources please see the Sources page or contact Jon Starks at (888) 988-1699 ext. 42 or via e-mail at jstarks@ftrassociates.com.
