The economics of airport slots

Abstract

Constraints on airport profitability appear to preclude the use of market mechanisms to allocate capacity at congested airports such as London's Heathrow. But that is not quite the end of the story. Even if one accepts the political constraint on airport profitability, it is still possible to bring airport charges closer to levels that balance supply and demand for airport capacity. The prices of other goods and services supplied by airports can be reduced in compensation. Overall, there has to be a rebalancing of prices across the range of goods and services supplied by the airport operator and its agents. A rebalancing of tariffs would be greatly encouraged by 2 major complementary changes to public policy. First of the fiscal privileges enjoyed by airport operators and, 2nd, of the scope of economic regulation. There are real opportunities for policy-makers to improve the functioning of the market for airport services.

I. INTRODUCTION

Heathrow Airport Limited, a subsidiary of BAA plc, formerly known as the British Airports Authority, advertises a set of charges at which airlines can, in theory, buy its services. But as Richard Branson and many others have discovered, it is not enough to be willing to pay BAA's asking price to obtain slots at many times of day. The situation is set to continue for the foreseeable future, unless there is some government action. The 'Heathrow problem' has already spread to airports throughout Europe and North America, and will spread further over the next decade as the demand for air travel continues to outstIip the growth in airport capacity, particularly runway capacity. The emerging shortage of supply of airport capacity in western Europe and North America poses difficult problems for airline regulators.

II. AIRLINE DEREGULATION IN EUROPE

EC transport ministers have recently negotiated a major air service liberalisation package, described by a UK minister as a 'landmark agreement ... sweeping away at a stroke the restrictions which have governed services in the community for so long'. It applies from 1 January 1993. Key features of the package include:

* almost complete deregulation of air fares on intra-Community routes;

* abolition of route licensing agreements on international services (which have until now restricted competition on many intra-Community routes); and

* some freedom to offer domestic services in other Member States of the Community as extensions to international services from the home state. All remaining restrictions on entry to domestic services within Member States will be abolished on 1 January 1997.

It is widely recognised that for deregulation measures of this kind to improve airline efficiency, whether in the form of lower real fares for given levels of service, or improved service quality at given fares, or lower fares and improved service quality, competition has to be effective. Yet services on many of the major international routes in Europe are currently offered by what would be regarded in other industries as a small number of suppliers. Even on routes such as London-Amsterdam, which has been deregulated as a result of a private agreement between the UK and the Netherlands, the market is still dominated by the two respective national flag-carrier airlines.

Broadly competitive outcomes may be achieved even in markets with a small number of suppliers, as long as it is easy for new suppliers to enter and to leave the market. The strength of competition from new entrants has been identified as the main reason for the widespread improvements in US airline iIldustry performance which followed US airline deregulation (Levine, 1987; Kahn, 1988). Similarly the UKbus industry had little competition at the local market level before deregulation and has not had much more since. But following deregulation there is little evidence of excessively high prices or profits, and evidence that operators react to the fear of competition even where active competition has not taken hold.

However, for airlines to compete fairly, new airlines have to be able to get on reasonably even terms with existing airlines, and this means being able to offer services at popular times from popular airports, such as Heathrow and Gatwick. If younger airlines have severely restricted access to airports, the effectiveness of competition will be weakened, perhaps seriously. It is conceivable that far from leading to lower fares and improved service levels, deregulation could lead to even higher fares for Europe's long-suffering air travellers.

What is to be done? In the remainder of this article we first describe the processes by which landing rights are currently allocated at busy airports, including the activities of airport scheduling committees. We then consider some of the possible ways of addressing what is essentially a problem of market failure.

We find that there is no 'easy answer' and, indeed, that some proposed 'solutions' could actually make things worse. However, there are clear opportunities for improving the efficiency of the market for airport capacity which, if adopted, should encourage effective competition between airlines more strongly than current arrangements.

III. CURRENT ARRANGEMENTS FOR ALLOCATING SLOTS

Capacity at very busy airports such as Heathrow is currently allocated through a mixture of prices and administrative processes.

1. AIRPORT TARIFFS

Airports throughout the world publish a tariff for the use of their facilities. With a few exceptions, these tariffs around the world consist of a single scale of charges which is applied throughout the year. The level oflanding charge varies mainly according to factors such as the size or weight of the aircraft, plus a separate charge per passenger sometimes referred to as airport tax. The different charges paid by differently sized aircraft generally go far beyond any differences in cost which might be caused by the difference in size. The main explanation for the differences is ability to pay. Larger aircraft can be charged more mainly because they tend to make longer flights. On longer flights, airlines have a larger competitive advantage over other forms of transport and the landing charge is a smaller proportion of total travel costs. For these reasons, the larger charge has much less impact on the viability of the flight.

Heathrow and Gatwick are important exceptions to this general situation. Compared to the industry norm, their charges are distinctive because they change according to time of day and by season, and because they incorporate flat rate charges per aircraft irrespective of size. This means that there are wide differences between the effective charge per passenger between peak and off-peak periods for all types of aircraft, and there is more incentive to use larger aircraft, especially during the peak. For example, the total airport charge per passenger at Heathrow in 1991 for a Boeing 747 on a peak period international flight was L10.72, assuming 65 per cent of the seats are filled;(1) the equivalent off-peak rate was L2.49. For a small aircraft, such as a 30-seat Shorts 330, the peak period international charge was L20.17 per passenger, and the off-peak charge L4.01, based on the same assumption.

Despite this rather sophisticated peak pricing system, the demand for slots at Heathrow at busy times far outstrips the available supply. At Heathrow, as at other congested airports throughout the world, airport scheduling committees (ASCs) and scheduling conferences play a major role in allocating capacity.

2. SCHEDULING COMMITTEES AND CONFERENCES

AIRPORT SCHEDULING COMMITTEES

In the early days of aviation, commercial airlines devised their schedules without giving much thought to the availability of space to take off and land at airports, much as early motorists did not have to worry about traffic jams. It was taken for granted that some runway space would be available at a time reasonably close to the scheduled time.

However, as air travel became more popular, congestion built up at major airports, especially at certain times of the day and certain times of the year. If an aircraft is delayed because it has to wait to take off, it annoys the passengers and disrupts the schedule; stacking aircraft in the air for anything other than a short period before landing is unacceptable, for reasons of both safety and fuel cost. Airlines therefore had a common interest in the proper scheduling of the use of capacity at airports. Some accommodation could surely be reached in which unfortunate coincidences of demand are avoided. Airport scheduling committees were formed by airlines to schedule the capacity of major airports so as to avoid congestion caused by unnecessary coincidences in their timetables.

Today over 100 airports around the world, by no means all as congested as Heathrow, have scheduling committees, mostly operated under a set of rules made by the International Air Transport Association (IATA), which is the trade association for airlines. The committees are usually staffed by employees on secondment from the major airline at the airport, but there is little evidence of any favouritism in their decisions. If an airline wants to land or take off from one of these airports, it must apply to the relevant scheduling committee for a place in the schedule. In practice, aircraft are usually dealt with by air traffic control in pretty much the order they turn up, but unless the aircraft has a place in the schedule, in principle it will not be allowed to take off or land.

The airport authorities tell the scheduling committees the maximum rate of landings and take-offs that can be scheduled, although sometimes the number of take-offs or landings that can be accepted may be less than the declared capacity of the airways, because of limits on the number of passengers that can be handled by the terminal buildings. The committee shares out the capacity between airlines for periods of six months according to a set of rules, the main ones of which are listed below. Any spare space left over because of cancelled flights, irregular schedules or good weather is allocated temporarily to anyone putting in an application a day or a few days ahead.

SCHEDULING RULES

* Grandfathering. If you had the slot last year, you get it again this year.

* Use it or lose it. If you were allocated the slot last year but used it less than a certain proportion of the time, then it will be given to someone else next year.

* Priority for regular services. If there is competing demand for a slot which is available, then it will be allocated to the service which plans to use it most frequently.

* Directed discretion. If competing demands are still unresolved by the above, then the following are given precedence: rescheduling to cope with country differences in daylight saving time; rescheduling to accommodate a larger aircraft; devising a more realistic schedule; extending an existing schedule to give year-round operation. This is not sufficient to make the committee's choice mechanistic; rather, it gives a foundation upon which the committee can rest the exercise of its discretion.

SCHEDULING CONFERENCES AND SLOT TRADING

Because of capacity constraints, not all airlines receive all the slots they need at the times required to make up their schedules. A system of swapping or trading slots has therefore evolved, to try and achieve departure and arrival slots at more sensible times. Slot trading is at present organised through two international airport scheduling conferences each year, each applying to the six-month period ahead (a season). Trading can also take place on an ad hoc basis after the conferences.

Each airline submits its desired schedule to the ASCs about six months before the start of the season, and the ASCs then allocate slots according to the procedures listed above. The ASCs' decisions are formally announced at the start of the relevant international conference and this is also when airlines first see the planned schedules of their competitors. Trading then starts, with airlines who did not receive their desired slots seeking to improve their allocations and also ensuring that they have consistent sets of departure and arrival times. ASCs provide information showing slot mismatches and who 'owns' various slots. They might be described as taking on some of the functions of a broker. Airlines may trade slots at the same or different airports, alter the type of aircraft flown and the destination (origin) of the flight, subject to the approval of the ASCs. All trades must be authorised by the relevant ASC to ensure that there is sufficient terminal capacity and parking space to accommodate any changes. Trading can be complex, involving many parties in simultaneous swaps of slots.

There are four US airports where there is a formal market for slots for domestic flights. At these airports, slots are bought and sold for money, not swapped for other slots. Large prices are regularly achieved.(2) But for other airports, there is only anecdotal evidence of money occasionally changing hands when slots are traded at ASCs. There is no prohibition on buying and selling, but swapping appears to be the dominant method of trade. It is thought that provision of airline support services or other forms of 'co-operation' are sometimes included as part of the swap, essentially another form of barter.(3)

The present procedures enable airlines to take advantage of mutual gains from trade, although there are clearly associated costs, such as trying to devise the complicated multilateral swaps which might be acceptable. For this reason it seems likely that many potentially beneficial trades do not take place.

The potential gains from trade available under a pure barter system are almost certainly less than under a monetised trading regime. In particular, a barter system (in this context) requires that every buyer must also be a seller of the same 'commodity'. This restriction means that the volume of trading taking place would be less than under monetised trading in which every participant would, in principle, be able to expand or reduce slot holdings. An important issue, therefore, is why there is so little evidence of monetised trading, especially since under current IATA procedures airport slots are reasonably well defined, and something akin to a property right is established by virtue of the grandfathering principle.

OTHER ADMINISTRATIVE PROCEDURES

The task of the ASCs and Scheduling Conferences may be eased somewhat by additional rules imposed by governments in a well-meaning attempt to cope with excessive demand for particular airports. For example, in the UK the government makes 'Traffic Distribution Rules'. Under these rules freight-only flights and socalled general aviation flights are barred from using Heathrow, except on a highly restrictive 'prior permission' basis. This means that flights are allowed access only if there are spare slots available at short notice, because of cancelled flights or favourable operating conditions. Some other Traffic Distribution Rules have recently been rescinded, notably those barring Heathrow to all airlines which were not already flying there in 1976 and to charter flights.

The European Commission has two current proposals for reforming the existing administrative rules. The more radical involves established airlines giving up slots if 'new entrants' are unable to get slots within three hours of what they want.(4) The proposal with more chance of acceptance requires only that 'new entrant' airlines have priority in the allocation of half the available slots.(5) However, we would observe that when in the past political pressure has been brought to bear to admit emerging airlines to Heathrow, the newcomer has sometimes appeared to have made less effective use of the space than existing airlines (for example, by operating small aircraft at times when there is an excess demand for runway space).

IV. MARKET-BASED SOLUTIONS TO THE CONGESTED AIRPORT PROBLEM

The current prices for landing at Heathrow and other similarly congested airports result in a situation of excess demand, as for example shown in Figure 1. (Figure 1 omitted) Given the capacity available during peak periods, shown as 2 in Figure 1, and the demand from airlines and their passengers for the facilities (shown as line D), the current level of airport charges, shown as P sub 0 , is too low to balance demand and capacity. There is excess demand, shown as (D(P sub 0 ) - Q). In this situation, administrative processes are needed to manage demand.

The effectiveness of administrative procedures, however well intentioned, as a means of allocating scarce resources such as landing slots at congested airports is gravely weakened by inertia, imperfect information and politicised decisionmaking, and is likely to result in inefficiencies, in particular:

* There is no guarantee that scarce airport capacity is used by those airlines (and those passengers) who value it most highly.

* The system is administratively burdensome to airlines, especially as a result of the activities of scheduling committees and conferences.

* It inhibits airline competition, especially the development of new services by new or expanding airlines.

* The provision of airline services will be inefficient, because the true cost and value of a major input resource -- the landing slot -- is not clearly signalled.

There are no readily available estimates of the burden which these inefficiencies impose on air travellers and the wider economy. Experience in other areas of transport, and in other activities where market solutions have been substituted for administrative procedures, would suggest that the potential gains are certainly likely to be sufficiently large to justify some investment of effort to strengthen the role of market mechanisms.

Indeed simple common sense suggests that the answer to this problem would be to raise airport charges to a level (from P sub 0 to P in Figure 1) at which supply and demand would be in balance. But there are two very real obstacles to this neat and simple solution reflecting, first, uncertainty and imperfect information and, second, the institutional framework within which airport charges are set. We consider each issue in turn.

POSTED PRICES AND AUCTIONS

Whilst the physical capacity of airports, in terms of the maximum number of takeoffs and landings, and of passengers that can be processed through airport terminals, is known fairly accurately, information about the position of the demand curve is highly imperfect. As a result, there is also much uncertainty about the level of price that would clear the market.

Suppose that demand is incorrectly assumed to be more intense than it actually is (in Figure 2, shown as D sub 1 in comparison to true demand D), so that a high price, shown as P1 is mistakenly set. (Figure 2 omitted) This would lead to underused capacity. Conversely, if imagined demand were below true demand, prices might be set too low, leading to continuing excess demand, albeit less severe than currently. Administrative procedures to manage demand would still be required. In itself, the fact that mistakes were made might not be too serious. Prices would be reset periodically on the basis of observed outcomes, and there might be a quite rapid convergence towards the efficient market-clearing level, P. However, in practice, all sorts of external influence would intrude, resulting in constant shifts in the 'true' demand. Airlines would no doubt be quick to complain about the problems caused by instability and unpredictability of airport charges.

These difficulties have led some commentators to recommend the use of auction mechanisms as a means of allocating slots, either temporarily, in order to identify broadly the level of market-clearing price, or permanently. Such commentators claim auctions are an inherently more efficient approach to allocating capacity than posted prices, given the characteristics of the market. We now examine this proposition.

SLOT AUCTIONS

There are several different ways an auction might be used. An auction might be held every six months to allocate the slots for the coming season. But then airlines' schedules would need to have a complete upheaval every season, which would hardly be desirable for travellers. At the other extreme one could have one auction to sell the rights in perpetuity, after which slot trading (discussed below) would take over for future adjustments. In between, there is a proposal (Hulet et al., 1991) for a system in which seven-year leases would be auctioned, and a certain number would come up for re-auction every six months.

For the purposes of designing the auction, it does not matter too much which of these three models one has in mind. We will examine two forms of auction which have been proposed.

A SIMPLE AUCTION

One possible approach to auctioning slots, which has been recommended in a recent academic study (Grether et al., 1989), would be for airlines to send in sealed bids for each individual slot all at once. The airline making the highest bid for each slot would pay what it had bid.

The problem is that each airline would be most unlikely to get a set of slots that could be used to form a working timetable. Consider a small airline wanting to run one flight a day. If it bids for a landing and a take-off, it might be out-bid on one but not the other. Suppose to try and insure against that it bids for a few landings and take-offs. But it might end up with two take-offs and no landings, or an unmatched take-off and landing only a few minutes apart. The strategy of bidding a high price 'to make sure' would not work either, because whatever price is logical for one airline to bid is logical for others too. But suppose this airline gets its takeoff and landing slots at reasonable times. It is possible that it is proposing to bring in a 747 at the same time many other airlines are proposing to bring in 747s. There might not be enough space in the arrivals terminal to cope with the resulting flood of passengers.

In order to make the initial allocation from the auction into a workable schedule there would clearly have to be a lot of swapping. There would have to be far more swapping than currently takes place at scheduling conferences, because at least the current basis of grandfathering leaves airlines with an allocation that mostly works. The immediate outcome of a simple auction would be far less useful than the current working schedules, however imperfect. Many years' work in devising workable and useful schedules would be abandoned in one go.

A COMBINATORIAL AUCTION

In theory, airlines could bid not just for individual rights, but for combinations of rights (Rassenti et al., 1982). In bidding for a combination of rights, airlines would either get all or none of the combination, so would not find themselves holding pieces that could not be used. It is up to an airline to put what it wants together in a combination. A combination could be, for example:

* all the slots and rights to terminal space, gates and parking to make up one landing and one matched take-off;

* all the slots and other rights required by one aircraft as it makes its thrice daily round trip to a holiday destination; or

* all the slots and rights required to make up an hourly schedule on an important inter-city route.

Airlines should also be allowed alternative bids. In practice, this means that an airline bidding for a particular combination, could put in an alternative bid for the same combination a quarter of an hour later, stating that it wanted only one or the other. This would maximise its chances of getting something it could use.

To solve this auction, the bids would be processed by large computer which would attempt to choose the winning bids by maximising the revenue while respecting all the combinations and alternative bids. In fact it is extremely difficult to write a computer program to do this, and it is not yet possible, indeed may be impossible, to write one that does it exactly, but we can now get fairly close. Fairly close is probably good enough.

But a real problem is whether the bids could be written. There are many slight variations to the timings and the aircraft size and other factors which would still permit the service to be run. If all the alternatives are not considered and bid for, there is a risk of missing out. A complete set of bids could be so large that it could be impossibly time-consuming to write it out in full.

A combinatorial auction has never been operated outside the laboratory, even for something much simpler than an airport, so we do not know how damaging these problems are. We suspect that they could be serious, so the idea must be tested on something simpler first.

NO MARKET PRICES

The study of combinatorial auctions leads to a conclusion about airport rights which is at first sight rather strange. There is no market price for airport services at a busy airport.

We can think of an airport auction as a large commodity auction in which there are a lot of different commodities for sale. For example, one commodity is Heathrow take-offs at 08.45 on winter weekdays, of which there might be 9 or 10 for sale.(6) Normally a commodity auction will produce a single price, the market price, which separates 'winners' from 'losers'. We would expect there to be a single market price for the winter weekday 08.45s. But the reason that the computer finds the auction so difficult to solve is that it cannot find any such price. It can find an upper price above which all bids are accepted, and a lower price below which no bids are accepted. But in the middle is a 'core region' where some bids are accepted and some are rejected.

This corresponds to our own personal experience of trying to book something up at a busy time. If you pay a lot you will surely get it, and for a little you surely won't. But at intermediate prices it depends on whether you can be fitted in, depending upon your exact requirements. Similarly one airline might have bid a touch more than another for a particular slot, but because the other rights it needs to use the slot conflict with other airlines' needs, the lower bid will be accepted.

This problem arises from demand externality. It is the reason why simple market solutions at busy airports are never quite going to work. An administrator will always be needed to sort out the fine details and do the 'fitting in', even if the airport charges were more realistic.

POSTED PRICES AND THE DEVELOPMENT OF SECONDARY MARKETS

Whilst any kind of slot auction process appears to be unattractive in terms of the ransaction cost burden, it is also the case that the rigidity of a posted price regime would be reduced if there was an active secondary market in slots. If slots were traded, then arguably they could be sold by those who make less valuable use of them to those who could make more valuable use of them. If slots could be leased out for a short period of time, then innovative airlines would be able to try out their new ideas. Airlines in financial difficulties would be able to realise their assets as a more flexible alternative to a take-over bid.

The need for administrative intervention would be reduced. Yet we see little slot trading or leasing. In this section we examine why it does not happen, find ways of helping it to happen, and finally examine whether it is worth it.

WHY DON'T AIRLINES TRADE SLOTS FOR MONEY ALREADY?

Apart from the four US airports where there is a formal market for slots for domestic flights, there is little evidence of slots changing hands for money. Airlines do not currently own landing slots in any legal sense, nor do the scheduling committees have any legal status. But the system creates a kind of ownership through the grandfathering rule.

There are several other claimants to the ownership of slots. In Britain, the government has statutory power to dissolve scheduling committees and take over slot allocation via the Civil Aviation Authority (CAA). The European Commission also has powers to dissolve the ASCs. The owners of airports would probably claim that ultimately they should control access to their land. Even under the current system, if a slot is left vacant, another airline will be moved in without any agreement by the departing airline. We could almost say that airlines have only 'squatters' rights' to their slots. They might be evicted any time, and have possession only as long as they stay.

This uncertain control over the slot reduces trade andprevents leasing. If you are an airline occupying a slot and someone wants to buy it from you, you would want a lot of money because it is the basis of your business and you might be able to hold onto the slot for free for many years yet. But if you want to buy a slot for money, it would be a gamble, because you might be evicted next year. It is not an asset of sufficient quality to put in your corporate balance sheet. The likelihood of finding a willing buyer and seller for money in these conditions is small.

HOW DO WE MAKE SLOT TRADING POSSIBLE?

This leads naturally to the suggestion that these 'squatters' rights' should be converted into formal ownership rights. Legal ownership would give an airline secure knowledge of what its rights are, and would give not just the right but also the confidence to trade slots or to lease slots out on whatever terms are agreed.

The first step in setting up such a system is to define the nature of the right. The airline must have rights to all the critical elements which make up the slot -- runway space, terminal space, parking, gate and whether it is a take-off or landing. All of these vary with the type of aircraft, since different aircraft can take up different lengths of time on the runway, and need different amounts and types of the various facilities. But the individual parts of the right must be individually tradable, because otherwise you would not be able to change terminal or change the aircraft.

They should not be defined as absolute rights to use the facilities at an exact time, because this would be too inflexible both for the airport and for the airline, either of which can suffer disruption and delay. So it would be defined as a right to 'schedule' for a particular time, which would be backed up by an official scheduled capacity to the airport.

WILL SLOT TRADING WORK?

We do not think that trade in slots will be particularly brisk. It heips to understand why by looking at the current trade through swapping. Currently airlines often get incompatible slot allocations from scheduling committees, and try to swap them amongst themselves to make up something more workable. However few swaps are made in comparison to the total size of the market. The problem is that there are not many willing traders.

Once an airline has a set of slots which make up a workable schedule, it is probably not in a position to make many trades with another airline. Every landing slot is matched by a take-off slot from the same airport, and a take-off slot from another airport. Each landing slot is matched by space in a specific arrivals terminal, and so forth. When one starts making a few alterations, there are many consequential changes which all have to take account ofwhether there is capacity. It is the same problem of demand externality that we met earlier.

When trades are made they are often more complicated than a simple swap. Commonly three or four airlines all shift around a bit in some deal brokered with the aid of highly skilled schedulers and computers. The swaps are often accompanied by other trades in kind, such as deals over ground-handling services, airline catering and the like.

Even when absolute rights exist which appear to make a slot a safe trade for money, a one-sided trade of some slot element for money will remain risky. In buying a slot you need all the matching components for your slot. An airline may frequently have to rely on getting all the components and matching parts off several vendors, each of whom will themselves be looking for oddments. For this reason, it is possible that the current swapping system will not be instantly replaced or superseded by monetised trades. Deals may still need to be made by diligently seeking for the clever multilateral swap.

We favour the possibility of selling slots for money for the following reasons:

* It makes trades more likely than currently.

* New airlines can enter the market without needing to have existing slots to swap.

* It facilitates the leasing of slots.

* It should allow airlines to count slots as assets in their balance sheets and mortgage them.

* It allows airlines in difficulties to sell out i/n whole or part.

* Airlines can lease out or sell a slot rather than deliberately run loss-making flights so as to avoid application of the 'use it or lose it' rule.

All these things have happened in the US. But we should not forget the difficulties. It will not be a perfect market, because the problem of matching slots to make a usable timetable is a barrier to trade. Trading will allow new entrants into the market more easily, but it will also allow existing strong airlines further to strengthen their positions. We note that the US Federal Trade Commission has blocked at least one large trade in slots (the sale of Eastern's Washington-New York service to US-Air). We would expect the European competition authorities to be equally vigilant in policing trades which significantly increased the level of concentration at major hub airports.

CONCLUSION ON MARKET MECHANISMS

The principal conclusion to be drawn from this discussion is as follows. If market mechanisms are to be given a more important role in allocating airport capacity, a combination of 'reformed' posted prices, set closer to a market-clearing level, and more extensive slot trading based on the establishment of formal property rights in slots, is to be preferred on grounds of economy and efficiency to an approach based on the use of auctions.

This conclusion is reinforced by wider considerations. Despite the moves towards airline deregulation, a large proportion of scheduled international air services between the UK and non-EC countries will be the subject of restrictive agreements with overseas governments, which carry with them an obligation on the part of the UK government to ensure that airport capacity is available to operate the service. Also as we have seen, IATA procedures regarding the use of congested airports are firmly based on the grandfathering principle.

Both slot trading and posted price regimes are compatible with international obligations of this kind, since both preserve incumbent airlines' rights to retain the existing slots. Auction mechanisms present far more of a difficulty, because they are fundamentally incompatible with the principle of grandfathering. 'Solutions' to the problem, for example, which differentiated between regulated and deregulated routes, making only slots used for deregulated services subject to the auction, are not only unattIactive on allocative efficiency grounds, but would also fall foul of the IATA principles.

It is notable in this context that US experiments with trading mechanisms have been restricted to domestic flights.' By analogy, the application of market allocation mechanisms could be restricted, at least in the first instance, to UK domestic traffic and possibly to international charter services. This would involve setting aside a proportion of slots at each airport for these classes of traffic, and would permit some limited re-allocation of traffic. However this solution would be rather less attractive than in the US, because the size of the US means the domestic market is very large there. For our small islands, a restriction making it difficult for international scheduled services to displace charter or domestic services would be serious, particulaIly for Gatwick.

V. THE CONSEQUENCE OF HIGHER AIRPORT CHARGES

In general, the potential efficiency gains would be smaller the more restricted the scope of the market. There could also be complaints of discrimination between UK airlines with different mixes of domestic and international services. Overseas carriers seeking to establish new scheduled services might feel aggrieved at not being allowed to buy slots.

Increasing BAA's airport tariffs to market-clearing levels would, in turn, greatly increase BAA's total revenues and profitability. Just how large an increase in charges might be required in order to balance supply and demand is highly conjectural; indeed, we have already noted that the very uncertainty about the size of increase required is an argument in favour of allowing the level to emerge from an auction process.

However, it is possible to make an educated guess at the order of magnitude, and a heroic attempt to do so is reported in CAA (1989). The results of this exercise, which was based on the use of an airport traffic forecasting model developed by the CAA, suggested that airport charges per passenger might need to increase by approximately 150 per cent at both Heathrow and Gatwick to equate supply and demand at the two airports, given the expected level of demand for access and the available airport capacity in the mid-1990s. Our initial assessment of the CAA model, and the assumptions made, lead us to believe that this is probably an overestimate, especially in the case of Gatwick. For illustrative purposes, we assume that market-clearing prices would be 100 per cent higher than at present at Heathrow, and 50 per cent higher at Gatwick.

In order to assess the possible effects on BAA' s profitability of setting marketclearing prices, we have used data on revenues and returns at BAA's South-East airports published in a recent report on BAA by the Monopolies and Mergers Commission (MMC). Table 1 shows that, in 1990-91, approximately 40 per cent of BAA's revenue at the three South-East airports was obtained from airport charges. (Table 1 omitted) The remaining 60 per cent of revenue came from other commercial activities, such as retailing and car park operation, carried on by BAA and its agents. The data on operating margins show clearly that these other commercial activities were very much more profitable to BAA than the activities financed through airport charges.

If airport charges had been set at market-clearing levels, and if revenues and costs from other commercial activities had remained unchanged,B then, as shown in Table 2, total revenue acIoss Heathrow, Gatwick and Stansted would have increased by 34 per cent.(9) (Table 2 omitted) Operating profits would have approximately doubled and the current cost return on capital employed in this segment of BAA's business would have risen from nearly 9 per cent to approximately 18 per cent.

In the report on BAA referred to earlier, the MMC concluded that the profitability of BAA's South-East airports was already sufficiently high given the risks faced by the business. The MMC recommended a new price control formula which would lead to a significant reduction in airport charges in real terms over the next several years. These recommendations were largely accepted by the CAA (CAA, 1991).

Even if the UK regulatory authorities were prepared to accept a large increase in BAA's profitability, which is unlikely given the MMC and CAA's conclusions, its compatibility with the UK's international obligations would certainly be strongly challenged by the US government. In particular, Article 10.3 of the Bermuda II air services agreement with the US government specifies that:

User [i.e. airport] charges may reflect, but shall not exceed, the full cost to the competent charging authorities of providing appropriate airport and air navigation facilities, and may provide for a reasonable rate of return on assets, after depreciation. In this provision of facilities and services, the competent authorities shall have regard to factors such as efficiency.... User charges shall be based on sound economic principles and on generally accepted accounting practices within the territory of the appropriate contracting Party.

BAA could also find itself subject to investigation by the EC under Article 86 of the Treaty of Rome, which outlaws the 'abuse of a dominant position' which affects trade between Member States.

VI. WHERE DO WE GO FROM HERE?

We seem to have reached an impasse. Constraints on airport profitability appear to preclude the use of market mechanisms to allocate capacity at congested airports such as Heathrow. But this is not quite the end of the story.

Even if we accept the political constraint on airport profitability, it is still possible to bring airport charges closer to levels which balance supply and demand for airport capacity. The prices of other goods and services supplied by airports can be reduced in compensation. Overall, there has to be a rebalancing of prices across the range of goods and services supplied by the airport operator and its agents.

A rebalancing of tariffs would be greatly encouraged by two major complementary changes to public policy, in respect, first, of the fiscal privileges enjoyed by airport operators such as BAA and, second, of the scope of economic regulation.

FISCAL PRIVILEGES

The profitability of commercial activities at major airports largely reflects the hugely profitable activity of selling duty-free goods. Such a lucrative fiscal privilege to airport operators may have made some sense during the 'infant industry' stage of civil aviation, when heavy investment in infrastructure facilities had to be made far in advance of demand.

But in the current circumstances of the civil aviation industry in the UK, the existence of the duty-free privilege introduces a major distortion into the market for airport services.

A significant step towards reforming this situation will arise if the European Commission is successful in its attempt to abolish duty-free allowances in respect of travel within the EC. If this were implemented, BAA would be obliged to increase airport charges, in order to maintain current levels of profitability. As noted in the recent MMC report on BAA:

The current cross subsidy to airport charges, from the profits of duty- and tax-free goods, will clearly be much reduced when the concession on sales of intra-EC duty and tax-free goods is abolished. We therefore accept the need for BAA to be compensated [through an increase in airport charges] for this loss of income. (CAA, 1991, paras 13.95-98)

THE SCOPE OF ECONOMIC REGULATION

Currently BAA's South-East airports are subject to economic regulation by price control. The limit is expressed as the maximum permitted revenue per passenger from airport charges. Charges at Heathrow and Gatwick are each subject to a separate price cap, and there is an overall price cap for the South-East system as a whole. The separate price caps at Heathrow and Gatwick were introduced in order to limit the scope for tariff rebalancing between the three airports. Additionally, Section 41(3) of the Airports Act, 1986, requires prices at any airport to be set to cover the costs of supply, which may be said to establish a floor to prices at Stansted.

Although the price cap mechanism applies only to airport charges, its level is determined after taking account of the expected profitability of BAA's commercial activities. If, as now, the profitability of these activities is expected to increase, then this puts automatic downward pressure on the level of airport charges, an utterly perverse outcome given current market conditions.

The rationale for applying direct economic regulation only to airport charges is itself highly dubious, because BAA and its agents enjoy a strong market position in the supply of facilities such as on-airport car parking and the on-airport services to airlines. Extending the scope of economic regulation to cover a wider range of services supplied by BAA and its agents would not only recognise the need, as perceived by the Office of Fair Trading (OFT),(1O) to protect consumers of other types of service supplied at airports; it would also give the company freedom to rebalance charges across a wider set of outputs than just those covered by airport charges. In circumstances where there was excess demand for slots at the prevailing published tariffs, this would encourage BAA to increase the price of slots relative to the price of all other airport outputs where there was not excess demand at prevailing prices.

WINNERS AND LOSERS

An important political issue is who wins and loses from the type of changes we are examining. We believe that the current beneficiaries of the underpricing of airport slots are mainly the established airlines and their owners. In particular, we doubt that inter-airline competition on many routes is sufficiently strong under normal, i.e. non-recessionary, market conditions, to ensure that the benefits of these low input costs are passed on to airline passengers to a significant degree, because of the barriers to entry created by the shortage of airport capacity and the small number of airlines operating on most routes. Support for this view comes from the high values attributed to packages of airport slots and routes when airlines change hands.

The extent to which any increase in airport charges was passed on to airline passengers in higher fares on services to and from congested airports would depend upon the precise nature of the price changes introduced by the airport operator, and upon competitive conditions in airline markets. We believe that considerably less than the full increase in airport charges would be passed through into fares, because currently the benefit of low airport charges is substantially enjoyed by airlines, following from the same argument above. Some airline passengers who use complementary services at airports, such as airport car parking, would benefit from the compensatory reductions in charges in these services which resulted from a rebalancing of tariffs.

Airline customers, in general, would be expected to benefit in the longer term from the increased effectiveness of airline competition stemming from the reduction in entry barriers, and from changes in the competitive position of services operating from different airports. To the extent that airline fares on services using congested airports such as Heathrow increased relative to services using less congested airports such as Stansted, then services into Stansted would compete more effectively with services into Heathrow. Finally, we would expect over time that there would be changes in the mix of airline services into Heathrow, with services where customers' willingness to pay was relatively low being displaced by services where customers' willingness to pay was high.

A WAY FORWARD

We conclude that there are real opportunities for policy-makers to improve the functioning of the market for airport services, by a combination of policies which both allow the establishIllent of property rights to airport slots and reduce the present cross-subsidy from other commercial activities to airport charges. These policies, in turn, would improve the plospects for effective airline competition in an increasingly deregulated environment. The impact of such initiatives would, however, be quite limited if they were pursued in isolation by the UK government. As we said at the outset, the market failure in the market for airport services is a worldwide phenomenon, deserving the urgent attention of the relevant authorities in Europe, North America and the Far East.

1 As reported in Civil Aviation Authority (1992). Sixty-five per cent load factor is about average for an international scheduled flight.

2 The four airports are Chicago O'Hare, New York J. F. Kennedy, New York La Guardia, and Washington National. The last two are domestic only airports and show relatively brisk trade. Kennedy is predominantly international, so there is little trade. At O'Hare, trade is fairly slow because there are two airlines which control about 80-85 per cent of the domestic slots. There is no disclosure of trade, but NERA's perception is that the price per slot has varied in the ranpe $300,000-$3,000,000. Obviously slots fetch more at certain times of day. But a slot will fetch more if it forms part of useful package, or if it is of special value to the buyer.

3 The IATA Procedure Guide talks often of trades taking place in an environment of mutual co-operation and goodwill.

4 Proposal for a Council Regulation (EEC) on common rules for the allocation of slots at Community airports', OJ C43, 19 February 1991. At Article 9: 'When requests for slots by entrants ... cannot be accommodated ... the Member State shall ... reclaim slots used by carriers to the extent that these carriers operate more than six slots on that route on the day in question.'

5 'Draft Commission Regulation (EEC) on the application of Article 85(3) of the Treaty [of Rome] to certain catepories of agreements, decisions and concerted practices concerning joint planning and coordination of schedules, joint operations, consultations on passenper and cargo tariffs on scheduled air services and slot allocationat airports', OJC253, 30 September 1992. At Article 5: 'New entrants have priority in the allocation of 50 per cent of newly created or unused slots and slots which have been given up by a carrier'.

6 Take-offs and landings are grouped into quarter hour blocks.

7 The US slot allocation rules allow for confiscating slots from domestic carriers if international agreements require more slots to be given to foreign carriers. In practice this has effect only at Chicapo O'Hare, and only the two major carriers are required to give up slots. It is, nonetheless, an unattractive aspect of the system.

8 This is a simplifying assumption. In practice, increases in airport charpes would almost certainly affect the mix of aircraft using the three airports, with a shift towards larper aircraft carrying more passengers. Any such increase in passenger throughput would affect other commercial revenues and margins.

9 Recalling that less than 50 per cent of BAA's income comes from airport user charges.

10 See OFT Press Release 7/90 for a summary of the OFT's concerns.

REFERENCES

BAA plc (1991) A Report on the Economic Regulation of the South-East Airports Companies, Report No. MMC 2, July, London: CAA.

Civil Aviation Authority (1989) 'Traffic distribution policy for the London area and strategic options for the long term. A consultation document', Report CAP 548, App. 9, London: CAA.

Civil Aviation Authority (1991) Economic Regulation of BAA South East Airports 1992-1997, Report CAP 599, London: CAA.

Civil Aviation Authority (1992) ManchesterAirportplc, Report No. MMC 3, July, London: CAA.

Grether, D.M., Isaac, R.M. and Plott, C.R. (1989) The Allocation of Scarce Resources, Experimental Economics and the Problem of Allocating Airport Slots, Boulder, Col.: Westview.

Hulet, J., Lake, H. and Perry, G. (1991) Study on Airport Slot Allocation, September, London: Department of Transport.

Kahn, A.E. (1988) 'Airline deregulation -- a mixed bag but nevertheless a success', Transportation Law Journal, Vol. 16, No. 2, pp. 229-51.

Levine, M.E. (1987) 'U.S. deregulation: evidence from ten years of experience', Revue international de droit economique, March.

Rassenti, S.J., Smith, V.L. and Bulfin, R.L. (1982) 'A combinatorial auction mechanism for airport time slot allocation', Bell Journal of Economics, Vol. 13, No.2.

* Ian Jones, Ivan Viehoff and Phillipa Marks are at NERA (National Economic Research Associates). London.