Identification and Evaluation of Freight Demand Factors (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.

8 1. Introduction Background In the last 25 years, the U.S. freight industry has transformed for a number of reasons. Population growth and migration have been fundamental factors, occurring alongside changes in the economy from goods to services. Transportation industry factors include the deregulation of freight railroads and trucking, the deployment of hub-and-spoke-based air freight, growth in international and domestic containerized intermodal freight, global supply sourcing, computerized-warehouse-based logistics, and, of course, the internet. Changes in regulation led a new era of flexibility in the arrangements that providers could offer to their customers. R eal prices decreased and demand increased as shippers, suppliers and retailers began to remake their production and distribution systems to take advantage of lower transportation costs and improved services. While freight flows via truck and rail grew at rate that basically matched overall economic growth, essentially doubling their ton-miles carried between 1980 and 2007, waterborne freight decreased by 40%. But it would be simplistic to generalize and presume rail and truck transportation grows smoothly with GDP. The GDP index is made up of multiple components including service and government components that have grown faster than the non-durable and durable goods components. A million dollars of medical care or software does not generate the same demand for freight transportation as a million dollars of agricultural or automobile production does. This means that truck and rail transportation demands have actually grown faster than growth in the non-service sectors of the GDP. Second, a simple comparison of year-to-year GDP growth with truck and rail ton-miles yields very poor correlations of 0.331 a nd 0.656, respectively. Housing Starts are a g ood proxy for economic optimism and consumption, and the Industrial Production Index, which measures the relative output of manufacturing, mining and energy producers, might be better measures of the segments of the economy that influence freight transportation demand. There are other factors that affect freight transportation demand. Besides regulation, technological changes, such as intermodal containers and equipment tracking have made long- distance transportation less expensive and improved service quality. As customers continue to seek new products, best prices and high quality, suppliers and retailers have found it profitable to change their logistics strategies to utilize the advantages of the global marketplace. . Advanced logistics and increasingly complicated supply chains, along with computerized sales and inventory tracking, has allowed the freight delivery industry to respond to small demand variations faster than ever before. In such periods of change, freight models must be updated to use the newest available data, and the modeler must be continually aware of changes in industry structure that may create the need for older models to be re-estimated. In particular, where freight models are being used to predict future volumes, whether short-term or long-term, it is useful to follow the axiom “If you are forecasting the future, do it often” in order to capture the latest developments.

9 An important goal of this research into freight demand factors is, ultimately, to provide insights into the structure of the U.S. freight system which can be used effectively to produce volume forecasts, both short term and long term. Nevertheless, using freight models to forecast future freight flows based on estimates of the input variables should be treated cautiously. Some of the concerns that should be heeded by the modeler when forecasting are: • Historically estimated models may have become out-of-date. • The input or independent variables themselves may be more difficult to forecast than the freight movement variable. • The forecasts of the independent variables have, by definition, some amount of error. • And finally, forecasts from freight models that were developed for one purpose are often used by their practitioners to solve or explain other problems outside the original scope and intent of the models.1 This study focuses on the identification of independent variables that can be used to explain gross measures of freight demand over historical periods long enough to capture some of the structural shifts described above. Research Objectives The federal government, as well as state and local agencies that oversee our nation’s transportation system, seek to improve their understanding of factors that explain the magnitude and direction of changes in the demand for moving freight. Historically, projections of freight movement in the country have been imprecise. In particular, the more rapid growth experienced in freight compared to passenger transportation through the first part of this decade, the forecasts loosely described as a doubling of freight over the approaching years, and the depressed volumes that followed more recently, have together underscored the need to better understand the underlying factors affecting freight demand. The goal is to enable decision-makers to more effectively plan transportation infrastructure and develop policies that improve the country’s competitiveness. Consider that in 2008, 6.2% of the U.S, GDP was devoted to the cost of moving goods or services and 10.1% was spent overall on logistics. After almost two decades of efficiency gains following the deregulation of freight transportation in the early 1980s, a combination of higher fuel prices, increased demand, and realignment of supply meant that real costs of transportation were increasing and the 2008 figures were the highest percentage in eight years.2 The objective of this research is to describe and analyze various types of demand drivers that shape the volume and movement of freight in North America, principally within the U.S. Consumption of freight, the locations from which it is satisfied, and the modes, routes, and 1 “What’s Wrong with Freight Models?” Marcus Wigan, Frank Southworth, March 2009. “Confusion about what one can do with a given model is often founded on a lack of transparency as to the domain of application for which the model was originally designed and set up.” 2 Council of Supply Chain Management Professionals, State of the Logistics Union 2008

10 consolidations it u tilizes result from many factors that interact at different geographic levels – global, national, regional, and local – as well as at different temporal spans, producing changes in magnitude and mode of transport used. This research seeks to identify those drivers that have the largest influence in explaining the variability and geography of freight demand, with the ultimate goal of producing better forecasts of freight volumes. It is important to note that variables influencing demand cannot be analyzed in isolation, as it is often the specific combination and interaction of a collection of factors that influence the magnitude and direction of freight demand. Freight flows could be measured in several forms; the most common units are physical volumes like tonnage and vehicles, but units of work such as ton-miles and vehicle miles can be more expressive because they include the volume of goods combined with the distance moved. Analytical Process The following tasks comprised this research: Task 1 – Prepare, Investigate, and Evaluate a Candidate List of Factors Task 2 – Analyze the Candidate List of Factors Task 3 – Determine Cost-Effectiveness of Factors Task 4 – Define which Demand Factors are Good Early-Warning Indicators Task 5 – Hold a Peer Exchange at the Beckman Center, Irvine, California Task 6 – Address Outcome and Comments from Peer Exchange and Conduct Follow-Up Work Task 7 – Prepare Draft Final Report and Presentation The research team took the following specific steps: 1. Beginning with the bibliography from NCFRP-01 “Review and Analysis of Freight Transportation Markets and Relationships,” a review of literature and current models was conducted to investigate recent industry and academic thinking on “What affects the demand for freight transportation?” 2. The team then investigated which economic and demographic variables, as well as one-time events (for example, the North American Free Trade Agreement or NAFTA) seemed to affect national transportation demand by highway, railroad and waterway. The most likely “drivers” of transportation demand were identified and then tested for collinearity because many may be tied to the same underlying factors, such as population or overall macro- economic activity. 3. The team chose a sparse number of key measures of freight moving by truck, rail, and waterway -- all but pipeline and air cargo (which accounts for an insignificant portion of total tonnage) -- that makes up 95% of United States domestic freight movements. To reflect the underlying demand for freight movement, variables such as tons handled and ton-miles moved were considered for each mode.

11 4. The team tested a variety of statistical analyses to come up with a number of models that utilize one or more of the independent economic variables to explain each of the separate modal transportation demands. 5. The team conducted additional analyses using an assortment of time-lagged independent variables to determine if various economic and demographic variables could be used to predict subsequent demand for transportation. 6. Finally, the team verified several of the models using existing data through a “backcasting” process, comparing actual data against model predictions for past periods. 7. The team presented preliminary findings at a peer exchange involving 31 participants from various freight industry sectors, government agencies, ports, consultants, and academia in Irvine, California, in May 2010. A summary of the feedback from the peer exchange is provided in Appendix A. 8. The research team prepared this final report, incorporating the peer exchange and other feedback received on the preliminary work.