Auctioning Bus Routes : The London Experience

1 Introduction The London bus routes market provides an early example of the use of a combinatorial auction format in public procurement. This market covers about 800 routes serving an area of 1,630 square kilometers and more than 3.5 million passengers per day. It is valued at 600 million Pounds per year (roughly US $900 million). Prior to deregulation, bus services in the Greater London area were provided by the publicly-owned London Buses Limited. The London Regional Transport Act of 1984 reorganized the sector. The Act designated London Regional Transport (LRT) as the authority responsible for the provision and procurement of public transport services in the Greater London area, as well as for the development and operation of bus stations and for the operational maintenance. It also advocated a franchise system by empowering LRT to invite private operators to submit bids to carry out bus services. In order to enhance competition, LRT, which by virtue of the Transport Act acted as the holding company for the original public operator London Buses Limited, created a separate tendering division, independent from its operational division. The operational division, London Buses Limited, was split into 12 operational subsidiaries. These were privatized in 1994. In practice, the introduction of route tendering was gradual. There was an experimentation phase as LRT built up procurement expertise and identified the routes that represented the greatest potential for cost reduction and the least risk of disruption (Kennedy, Glaister and Travers 1995). Hence, though the first auction took place in 1985, it was not until 1995 2 that half of the network was tendered at least once. 1 Since then, tendering has reached its steady state regime with 15-20% of the network tendered every year. This chapter does several things. First, we describe the combinatorial auction format adopted by LRT and briefly discuss its properties (sections 2 and 3). Second, we describe the bidding patterns observed in the data that we collected for these auctions, with a special emphasis on package bidding and the effect of the auction size (section 4) Finally, as in any practical design problem, LRT was faced with a range of options when deciding on the auction format. A critical input into any such analysis is a better understanding of bidders' preferences, i.e., in our case, their cost structure. In section 5, we summarize a new method that we have developed to analyze bid data from …