Post written by Professor Richard Steinberg, Chair in Operations Research and member of the Managerial Economics and Strategy Group within the Department of Management at the London School of Economics and Political Science. Professor Steinberg specialises in auctions and mechanism design, and internet economics.
International air transportation demand is currently growing at a formidable rate, with total revenue passenger-kilometres (RPK) predicted to double between 2014 and 2029. While this growth is led by developing countries, notably China, even developed countries with historically high demand continue to see growth. Airport runway capacity remains arguably the most significant constraining factor limiting growth in flight operations. Because of this, decision-makers with oversight for airport access regularly employ procedures to restrict airport access to limit congestion.
The most predominant of these procedures is the International Air Transport Association’s Slot Allocation System, which involves defining a set of airport “slots” and allocating them to flight operators. This system is used at virtually all large airports within Europe, and in the three largest US airports: Kennedy International and LaGuardia airports in New York, and Reagan National airport in Washington, DC.
All large airports in Europe currently deal with congestion by allocating runway slots to airlines twice a year. Unfortunately, they do this in a very inefficient way. Under current regulations, if a runway slot is used by an airline 80% of the time over a season, then it has use of the slot for the following season, and in this way the airline can retain the slot in perpetuity.
This rule has led to some shockingly wasteful behaviour. Airlines have been known to organise “ghost flights,” i.e., flights with empty planes, so as not lose valuable runway slots and to prevent entry by potential competitors. This ruse became especially common during the early stages of the Covid-19 pandemic. As Covid restrictions lifted, the demand for air travel returned to pre-pandemic levels in around April this year, but ghost flights remain. Britain, Europe, and China have been reviewing their slot allocation procedures. In particular, the issue is now being studied by the UK’s Department for Transport, who have pointed to auctions as a possible solution.
Indeed, a market-based approach, in which airlines bid for slots — rather than having them be assigned by committee — would be a far more efficient procedure. Economic theory and historical data strongly indicate that slot auctions would result in increased airport capacity and fewer flight delays, as well as decreased flight costs to both airlines and passengers. Further, slot auctions would eliminate the incentives for airlines to schedule empty aircraft. In fact, by making more efficient use of airport slots, including eliminating the incentive for ghost flights, slot auctions would reduce each airline’s overall carbon emissions, moving them closer to environmental targets set by the ESG rating agency Vigeo Eiris.
How then, would slot auctions work?
When British Airways requires a slot for a flight departing from London and going non-stop to Los Angeles, it clearly also requires a slot at the Los Angeles airport for the flight’s arrival. Likewise, for a Lufthansa flight departing from London and stopping in Frankfurt before arriving in Los Angeles, three slots are required, and for an Air Canada flight headed from London to Toronto, then to Chicago, and finally arriving at Los Angeles, four slots are needed. In each case, the airline will be reluctant to place separate bids for its two, three, or four required slots because winning some, but not all, would commit it to paying for one or more slots that it does not need. In each case, the airline will instead prefer to bid for a package of slots, that is, an “all or nothing” bid.
Auctions that allow package bids are called combinatorial auctions. Over the past 20 years, combinatorial auctions have been developed for many industries, most notably for the allocation of leases for radio bands for use by telecoms companies for the provision of mobile phone service, which has transformed the industry and raised billions of pounds in revenue.
How would a combinatorial auction for airport slots operate? The standard approach is to have bidders submit all their bids — including package bids — to the auctioneer, who then attempts to piece them together and accept those bids that together maximize auction revenue. This complex jigsaw puzzle, called the winner determination problem, is unfortunately in a class of mathematical problems called NP-hard. What this means is that as the size of the problem—i.e., the number of slots—increases, the time required to determine the allocation ramps up so exceedingly quickly that even the fastest computer cannot solve it.
The PAUSE auction
However, there is an alternative. Cambridge professor Frank Kelly and I have developed a combinatorial auction procedure called PAUSE (Progressive Adaptive User Selection Environment), where the auctioneer never faces the winner determination problem. This procedure would be especially suitable for the auction of airport slots.
PAUSE proceeds in stages. In stage 1, airlines bid on individual slots only. In stage 2, airlines can submit package bids that contain two slots. However, each airline must submit its package bid as part of a composite bid, which is a set of non-overlapping bids that cover all the slots in the auction. Of course, an airline is unlikely to be interested in all the slots in the auction; however, it can fill out its composite bid with previously-submitted bids by any of the bidders. In stage 3, airlines can submit package bids for up to three slots; in stage 4, up to four slots, and so forth, as long as the bids are submitted as part of a composite bid. Each stage can have several rounds of bidding. The point here is that, when the bidding stops, there is no need to solve the winner determination problem. The slot allocation is displayed to all by the final composite bid.
By avoiding the winner determination problem, the PAUSE auction procedure is applicable for slot auctions of any size. PAUSE also possesses several other advantages over other combinatorial auction procedures, including the feature that virtually any additional rules or regulations required by the Airports Council International can be easily accommodated by PAUSE
Insufficient slot capacity and levels of carbon emissions have both reached a critical stage, so now is the right time to introduce a combinatorial auction procedure that can handle slot allocation problems for airports of any size.