Robust Combinatorial Auctions for Transportation Service Procurement with Uncertain Demand
In this project, we propose to study robust combinatorial auctions (CAs) for transportation services procurement. CAs are those auctions in which bidders can place bids on combinations of items, called “packages”, rather than just individual items. The advantage of CAs is that the bidders can more fully express their preferences. This is particularly important when items are complements, i.e., when a set of items has greater utility than the sum of the utilities for the individual items.Transportation services have distinct features that make CAs especially appealing. To carriers, a balanced network reduces the connection costs and can lower their overall costs. For example, the cost to haul some cargos from B to A can be much lower if the carrier also hauls loads from A to B, since this avoids vehicles move specially from other places to B with empty load. Hence, carriers can offer a lower price for such a round trip but can still maintain a high profit margin. Shippers, as a result, can also benefit from lower transportation prices thus caused. In fact, CAs have been widely used for truckload transportation in developed economies, such as US and Europe.CAs, however, do not always lead to the expected outcomes. The shippers usually procure transportation services for future demands based on demand forecasting, which is often not accurate enough. As a result, bids thus selected could be sensitive to demand variability. Bids are somewhat nonbinding contracts on both sides. It is well accepted by both shippers and carriers that by the time of implementation, shippers may not have the promised amount of loads, as stated in the contract, for carriers to move; and the carriers may not have the promised capacity always ready for the shippers because of operational issues. Hence, well-designed bids by carriers and well-selected bids by the shipper may perform worse at the time of implementation. Such practice somehow prevents carriers and shippers from benefiting from the theoretical advantages of CAs, unless various uncertainties have been incorporated when carriers design bids and when shippers select bids.In this project, therefore, we propose to combine CAs with the recent ideas from robust optimization. The essential motivation behind our work is that although individual forecasting, such as demand forecasting for a single lane, can not be expected accurate, aggregated forecasting, such as demand forecasting for a group of related lanes, can be expected more accurate. Taking advantage of this, we propose to build a general framework for carriers to design package bids and for shippers to select bids; the framework is aimed to reduce the risk of unnecessary higher costs incurred at the later stage due to earlier decisions, and to realize the theoretical benefits promised by CAs. The project consists of three parts. We first study bids selection problems for shippers, which serve as very good starting points because their counterparts when no uncertainty is involved have been widely studied. After knowing how proposed bids are selected by shippers, we then study for carriers bids formulation problems, which tend to be more complicated because, besides their own cost structures, carriers also need to consider their competitors in order to guarantee certain chances of winning bids. The third part, also our major objective, of the project is to promote robust CAs for transportation services procurement in mainland China and Hong Kong. We will collaborate closely with our industry partners to make sure our research practically applicable.
|Effective start/end date||1/04/10 -> 30/10/12|