Practice Summaries: American Airlines Uses Should-Cost Modeling to Assess the Uncertainty of Bids for Its Full-Truckload Shipment Routes

American Airlines, Inc. (AA) is the world’s largest airline in passenger miles transported; its annual revenue is over $21 billion. Although its goal is passenger transportation, AA must minimize the costs of ancillary functions that are not related to its core business. One of these functions includes the full-truckload (FTL) freight shipment of maintenance equipment and in-flight service items for passengers. At any given time, the system inventory for these goods can be over $1 billion, creating a web of complexity as goods move point-to-point worldwide. The purchasing department in AA’s transportation and logistics group presented the Supply Chain Resource Cooperative at North Carolina State University with an opportunity to explore the costs associated with its FTL shipments in the United States. Specifically, AA desired a should-cost model to assess bid quotes for its FTL point-to-point freight shipment routes to ensure that AA does not overpay or underpay its FTL suppliers. We developed a probabilistic model using a decision tree that estimates reasonable costs for FTL freight shipments. AA begins the bid process for an FTL point-to-point shipment route by sending out a request for quote (RFQ) to prospective suppliers. Suppliers respond to the RFQ, and AA does a side-by-side comparison of the bids. In any given year, AA’s purchasing department has approximately 500 RFQs in its bid process. Bid quotes in the long-haul trucking industry can differ significantly. AA is no exception; its singlecontract bids can deviate by as much as 200 percent. To account for this variability and to identify potential overbids and (or) underbids, we (1) collected base-case and range data from primary and secondary sources on the cost variables that can affect an FTL bid, (2) ensured that the cost variables were mutually exclusive and collectively exhaustive, and (3) modeled the uncertainty using the DPL decision analysis software. Primary research involved interviewing a number of trucking firms and truck manufacturers. Secondary research involved collecting data from various trucking websites, government websites, and trucking forums, and reviewing the literature. AA piloted our should-cost FTL model against an RFQ consisting of bid quotes from six FTL carriers. We used DPL to conduct one-way sensitivity analyses of the cost variables (see Figure 1). The basecase deterministic value (shown by the vertical line