Setting airport regulated charges: the choice between single-till and dual-till

Transcription

1 July 2014 Frontier Economics 1 Setting airport regulated charges: the choice A NOTE PREPARED FOR EASYJET Over the last twenty years, a recurring theme in the field of airport price regulation has been whether aeronautical charges, if regulated at all, should be set on a single-till or a dual-till basis. Under a single-till approach, profits from an airport s non-aeronautical activities (for example, from the provision of retail space or car parking) are deducted from the revenue requirement for aeronautical services before determining the level of aeronautical charges (typically covering aircraft take-off, landing, parking and use of air bridges, provision of terminal services to passengers and airlines, including processing and screening of passengers and their baggage). Under a dual-till approach, instead, the regulator excludes non-aeronautical activities from its calculation and focusses solely on aeronautical functions. The UK airports regulator most recently considered this issue in 2000/01 1, and ultimately determined to continue with single-till regulation. This decision was influenced by the UK Competition Commission, which at the time questioned whether dual-till arrangements would lead to improved efficiency and stated that a switch to dual-till would lead to a substantial transfer of income to airports from airlines and/or their passengers, potentially undermining regulatory credibility and creating regulatory uncertainty 2. Recently the Irish airports regulator has also consulted on the same topic and come to the same conclusion. In contrast, in recent years we have seen a move away from single-till regulation in both Spain and France 3. At the same time, the European Directive on Airport Charging is open-minded on the issue, leaving the matter to Member States 4. On the other hand, single-till approaches are generally supported by ICAO 5. Furthermore, the recently adopted new EU Guidelines on State Aid to airports 6 1 CAA (December 2000) The Single-till and the Dual-till Approach to the Price Regulation of Airports - Consultation Paper 2 Competition Commission (2002) report on price caps for BAA airports, Appendix Commission for Aviation Regulation CP4/2010 and CP1/ Directive 2009/12/EC of the European Parliament and of the Council of 11 March 2009 on Airport Charges. See preamble, para ICAO Doc 9082/7 paragraph 22(i) 6 Commission Communication C(2014) 963, Guidelines on State aid to airports and airlines.

2 2 Frontier Economics July 2014 implicitly recognise single-till principles by stating that [in] order to assess whether an arrangement concluded by an airport with an airline complies with the [state aid] test, expected non-aeronautical revenues stemming from the airline's activity should be taken into consideration. 7 It is clear that opinion remains divided on this topic. In this paper we try and bring some clarity to the economic issues surrounding single-till and dual-till regulation. We start by considering the matter from a theoretical perspective, asking: which approach is likely to most closely replicate the outcome of a competitive market, promoting efficient operating and investment decisions? We then go on to consider the practical issues of implementing price regulation. We provide guidance as to how the regulatory till should be defined and single- or dual-till regulation could be applied to result in efficient prices to airlines, and effective incentives and rewards for airport operators. We note that support for single-till or dual-till usually divides along clear lines: airlines and IATA in favour of single-till, airports in favour of dual-till. This says a lot about stakeholder perceptions of the issue and who are thought to be the winners and losers from the alternative approaches. Opponents tend to observe that single-till regulation reduces airport charges, discouraging investment in airport capacity. This may be perverse at a time when capacity at many major European airports is becoming increasingly stretched. Summary of conclusions The issue of regulating airport charges arises only when an airport has substantial market power (SMP). In these circumstances we conclude that regulation based on a single-till model is generally superior to solutions based on the dual-till approach. Airports that do not have SMP will respond to competitive pressures by offsetting non-aeronautical profits against aeronautical charges, when the nonaeronautical profits are a by-product of aeronautical activities. In a regulatory setting, the single-till approach mimics this dynamic more accurately than dual-till and therefore leads to more economically efficient outcomes. Specifically, even when single-till regulation reduces aeronautical charges, the airport should have efficient incentives to expand aeronautical capacity, provided regulation has been properly applied. 7 Ibid., para. 64.

3 July 2014 Frontier Economics 3 We note that where an airport is subject to single-till regulation, the airport entity should expect to earn a competitive rate of return on its activities, provided it is operating efficiently. Against this benchmark, a switch to dual-till has the prospect of transferring profits into an unregulated entity while at the same time increasing aeronautical charges. The net result is dual- allows the airport to generate returns above competitive level. A move to dual-till is sometimes justified by arguing that a cross-subsidy is being removed. But we demonstrate that there is no cross subsidy in the first place. The commercial activities that properly belong within the single-till are byproducts of the main activity of the airport. And insofar as these by-products generate locational rents, these rents would be fully reflected in aeronautical charges set by an airport in a competitive market. We also consider the argument that has been advanced in some places that dualtill regulation is more appropriate for congested airports. We show that this is not correct, being based on an incorrect interpretation of theoretical models whose properties bear little relation to the reality of congested airports. However, we conclude that the application of single-till regulation does not always imply lower airport charges than dual-till. In many cases, especially at smaller airports, it is possible that charges should be the same under either approach. In the final section of this paper we also consider a range of practical issues that should be addressed to ensure that single-till regulation works effectively and as intended. This includes the need for arrangements that encourage and reward the airport for developing non-aeronautical profits, but also discourage regulatory gaming over the anticipated levels of those profits. Models of regulation Before we start our analysis, we need to describe in more details the two models of airport charges regulation which we consider in this paper. As noted above, in practice the regulation of an airport s aeronautical charges can be done in a variety of ways. For the purpose of this discussion, though, we only need a high-level definition of single-till and dual-till. This is because the key conclusions of our assessment do not depend on the specific implementation of the regulatory framework. The regulatory till represents the total of the airport revenues that need to be recovered via regulated aeronautical charges. Depending on the regulatory model chosen, a regulator might wish to focus only the provision of aeronautical services or may also take into account all or a share

4 4 Frontier Economics July 2014 of the profits from non-aeronautical services. A regulator is generally faced by a continuum of options along this spectrum. In practice, however, the regulatory models can be divided into two groups: single-till and dual-till. We consider each in turn. Figure 1 below shows in a stylised way how the regulatory till is defined under a single-till approach. Figure 1. Simple schematic for deriving the regulatory till under a single-till approach Total Airport Costs Less Non-Aeronautical Revenues Regulatory Till This contrasts with Figure 2, which shows how the till is derived under a dual-till approach. Figure 2. Simple schematic for deriving the regulatory till under a dual-till approach Total Airport Costs Less Non-Aeronautical Costs Regulatory Till On the face of it, it appears that the two approaches are very similar. In both cases non-aeronautical activities are excluded in some way to derive the regulated till. In the single-till approach, the regulatory till is determined by deducted non-

5 July 2014 Frontier Economics 5 aeronautical revenues from total airport costs, leaving the remainder to be recovered from aeronautical charges. In the dual-till approach it is nonaeronautical costs which are deducted. Considered at this conceptual level therefore, the difference between the two approaches is what happens to the profit from non-aeronautical services. In the case of single-till regulation this profit is deducted from the regulatory till and so reduces aeronautical charges, while in the case of dual-till regulation these profits (if there are any) are retained by the airport. We note that the above descriptions provide only a stylised view of the actual issues that a regulator would have when setting regulated charges. A key issue is that the definition of profit in the previous paragraph needs clarification, as the economist s definition of profit differs from the commonly-used accounting sense of the word. Furthermore there are many practical issues that a regulator would need to address. For example, with regards to single-till, there is an obvious issue associated with deciding which non-aeronautical services should be included in the regulatory till. On the other hand, dual-till solutions require a thorough costallocation to be carried out so that shared and common costs can be split between aeronautical and non-aeronautical activities. Failing to allocate costs properly could result in hidden cross subsidies between aeronautical and nonaeronautical activities. In turn this may lead to providing the regulated airport with misleading incentives. The regulator must address these practical issues when defining the regulatory framework. We believe that the high-level descriptions we have provided above are sufficient for the purpose of the discussion presented in this paper. We will, however, explore some practical issues with the application of airport charges regulation later on in the paper. Our approach The general aim of price regulation is to counter the effects of market power, preventing the generation of excess profits and promoting efficiency of operation and investment in sectors where the forces of competition are too weak to achieve these ends by themselves. While a purely competitive equilibrium is unattainable, it is a useful conceptual tool as it acts as a natural benchmark for assessing alternative regulatory options. Generally, regulatory frameworks that more closely induces a regulated company to behave as if they were subject to competitive pressures are superior to those that do not. Therefore, our approach involves defining two reference scenarios against which the stylised regulatory options listed above, single-till and dual-till, can be assessed. We first consider how an airport operating in a competitive market

6 6 Frontier Economics July 2014 would take non-aeronautical activities into account in setting aeronautical charges. In practice, of course, airport cost structures tend to involve high fixed costs, or rather costs that are fixed relative to the volume of traffic over the very long run (the runway and terminal capacity), and relatively low variable costs. With this cost structure it is inevitable that airport markets cannot correspond to a textbook model of perfect competition. As a result airports need to price above short run marginal cost to recover their capital costs. In order to achieve this, airports need to be differentiated in some way. The obvious differentiation is spatial: airports are spread out geographically to give each its own catchment. But other forms of differentiation exist, including serving different segments of the passenger market, such as low cost carriers or network carriers. Differentiation lends airports some degree of market power, but it does not follow that airports automatically need price regulation. Given the high fixed cost nature of the business, differentiation does not necessarily imply excess profits for the airport. For simplicity, however, we consider as our second reference scenario the case of an unregulated airport with a monopoly over passengers, and again ask the question, how would the monopoly airport take non-aeronautical activities into account in setting aeronautical charges? After defining how the competitive and monopoly airport would reflect nonaeronautical activities in its aeronautical charges, we consider the implication for airport regulation. Specifically, we compare the single-till and dual-till approaches with the competitive and monopoly scenarios and draw conclusions about which approach is likely to generate the most efficient outcome. The reference scenarios In this section we set out the reference scenarios, which we will use to assess the single-till and dual-till approaches to regulation. First, we set out a competitive scenario. We then consider how a monopoly airport will set aeronautical charges. Finally we move on to consider single-till and dual-till regulation and compare these approaches to the reference scenarios.

7 July 2014 Frontier Economics 7 Competitive market We start from considering the hypothetical case of an airport that only provides aeronautical services. If we assume that an airport is operating in a perfectly competitive market, and do not concern ourselves with the issues of cost recovery for fixed investment, competition will drive down airport charges to equate aeronautical charges with the marginal cost of providing aeronautical services c. At this level of price economic profit equals zero. In this scenario, all costs are variable with output/passenger numbers, including any capital costs; marginal costs are assumed to include a competitive return on any capital employed. We now consider whether this conclusion would change if airports started offering non-aeronautical services too. Clearly the airport would experience costs in providing non-aeronautical services and would likewise expect to generate revenues which at a minimum cover those costs. How then would this new factor affect aeronautical charges? There are a number of possibilities that must be considered. Is the non-aeronautical activity expected to cover its costs (i.e. are its economic profits greater than or equal to zero)? If this is not the case then the airport would be better off not performing the non-aeronautical activities in the first place. The airport cannot raise its aeronautical charges to recover non-aeronautical losses, because it is in a competitive market and could simply be undercut by rivals that did not attempt to provide non-aeronautical services. Provided economic profits from non-aeronautical services are positive, are these profits related to the throughput of passengers at the airport? If the airport expects to generate increasing non-aeronautical profits as a result of increasing passenger numbers (for instance because the fixed assets costs embodied in in-terminal retain space are used more intensively) then for any additional cost per passenger the airport expects to incur, it will perceive an offsetting non-aeronautical profit per passenger, reducing its perceived marginal costs In more formal terms, if the airport has a marginal cost c of accommodating additional passengers and generates non-aeronautical profits of per passenger, then its perceived marginal cost, of providing aeronautical services is reduced from to. Hence the airport would equate its airport charges at equal to its perceived marginal costs:

8 8 Frontier Economics July 2014 So we see in this very simple case that the completive airport would pass on all economic profits from non-aeronautical services in full to passengers in lower aeronautical charges. This situation is illustrated in Figure 3 below. Without non-aeronautical profits output is set at, where the downward-sloping (red) demand curve cuts the horizontal supply curve at price equal to. With non-aeronautical profits, output is set higher, at, where the demand curve cuts the horizontal supply curve at the lower price, reflecting non-aeronautical profits. Figure 3. Impact of non-aeronautical profits on aeronautical charges competitive case 60 Price Quantity120 There are a number of important observations to make about this result: Competitive aeronautical charges can only be reduced by non-aeronautical; activities. Non aeronautical losses cannot be recovered because of undercutting from competitors. Competitive aeronautical charges will only be adjusted if the non-aeronautical activity is a complement to the aeronautical that is if non-aeronautical revenues increase as passenger numbers increase. Commercial activities undertaken by the airport but unrelated to passenger numbers should not be expected to impact on aeronautical charges. Aeronautical charges will only be reduced if the airport generates economic profits from, non-aeronautical activities. If the non-aeronautical activities

9 July 2014 Frontier Economics 9 only generate a normal (accounting) profit, hence economic profit is zero, aeronautical charges will be unaffected by the non-aeronautical activity. This last point raises an important question. Does this conclusion imply that an airport will only set lower airport charges if it has market power over its nonaeronautical activities? The answer to this is no. First, if the market for the non-aeronautical activities is corresponds to textbook perfect competition, then the retail prices for the non-aeronautical services will be determined by the market outside the airport and will equal the marginal cost of the service. In this limiting case there will be no economic profits from nonaeronautical services and the issue does not apply. However, in reality some services could generate economic profits without market power, because of the presence of fixed costs. Take retail as an example. Assuming that the retail prices of goods in the airport shops are competitively determined i.e. determined by the cost of those goods in shops outside the airport. It is still possible that economic profits could be generated because the density of customers created in the terminal building means that sales per m 2 of floor space exceed anything that could reasonably be expected from a high street shop. In this case the retail space generates locational profits as a result of the concentration of potential customers created by the terminal building. Hence we might expect to see larger airports, which are more likely to achieve these locational profits, setting lower aeronautical charges, other things being equal. One final observation is required on this scenario. It is sometimes said that the outcome described here is inefficient because the price of aeronautical services is set below their marginal cost. But this outcome is not inefficient. The only non-aeronautical services captured in this system are those that are a by-product of the aeronautical activity a complement in economic terms. As such the airport is behaving as an efficient profit-maximiser in discounting aeronautical charges. Overall the airport is generating normal profits consistent with the economic risks of the activity as a whole, and the total marginal revenue from expanding the airport is equal to the total marginal cost of doing so. Market power In this section we consider the case of an airport with unregulated monopoly power. This is the most extreme case of market power, but the conclusions we reach apply also in situations where the market is not perfectly competitive, such as oligopolistic markets. Again, first we consider the case of an airport that only offers aeronautical services. Having monopoly power, the airport will set aeronautical charges at the level at which its marginal revenue for aeronautical activities (that is, the increment in revenues from serving an additional passenger) equals its marginal

10 10 Frontier Economics July 2014 cost (that is, the increment in cost from serving an additional passenger). This is the condition for the monopolistic equilibrium. If the marginal revenue were higher than the marginal cost, the airport would have an incentive to generate more demand by reducing its price. If the marginal revenue were lower than the marginal cost, the airport would increase the unit price to cover its cost. We now think about the introduction non-aeronautical services as in the previous case. The same two questions apply, but the conclusions are somewhat different. Is the non-aeronautical activity expected to cover its costs (i.e. are its economic profits greater than or equal to zero)? In this case, because the airport has no competitors, it is able to pass on losses from non-aeronautical services in higher aeronautical charges. As we shall see later, this has important implications for the application of single-till regulation. Are these profits related to the throughput of passengers at the airport? If the airport expects to generate increasing non-aeronautical profits as a result of increasing passenger numbers, then for any additional cost per passenger the airport expects to incur, it will perceive an offsetting nonaeronautical profit per passenger, reducing its perceived marginal costs In more formal terms, if the airport expects to generate non-aeronautical profits per passenger, then in effect its perceived marginal cost,, of providing aeronautical services is altered from to, where could be positive or negative, and so could be higher or lower than. The profit maximising monopolist without non-aeronautical services will set aeronautical charges at a level, which is higher than the equivalent competitive level. How much higher depends on how sensitive demand is to price the less sensitive the higher will be the monopoly price. If profits from non-aeronautical services are introduced, it is straightforward to show that the profit maximising monopoly airport will set aeronautical charges at, where: That is, the monopolist will reduce (assuming is positive, increase if negative) by half the economic profits from non-aeronautical activities. Moreover, this level of unregulated charges will always be higher than the charge that would be set by a competitive airport taking non-aeronautical activities into account.

11 July 2014 Frontier Economics 11 This is illustrated in Figure 4 below. A monopolist will maximise profit by choosing a level of output that makes marginal revenue equal to marginal cost. Marginal revenue is the red-dotted line on the figure below. Because marginal revenue has a steeper slope than the demand curve, output is set at a lower level and prices are set higher. The monopolist with no non-aeronautical profits sets aeronautical prices at, which is higher the equivalent competitive level, and output at, which is lower than in the competitive scenario. Introducing nonaeronautical profits reduces the monopolist s aeronautical charges to and output rises to. Figure 4. Impact of non-aeronautical profits on aeronautical charges monopoly case 60 Price 50 Marginal Revenue Quantity120 Overall we see that the case of unregulated monopoly has some similarities to the competitive outcome. In particular the profit maximising airport will offset nonaeronautical economic profits against aeronautical charges. However, market power allows the airport to retain some of that profit, which would not be possible in a competitive market. Furthermore, market power allows the airport to recover losses from non-aeronautical services, at least partially, which is not possible in the competitive scenario. It is sometimes observed 8 that the presence of non-aeronautical profits may remove the need for the regulation of aeronautical charges, even if the airport has 8 For instance, see Starkie & Yarrow (2001)

12 12 Frontier Economics July 2014 market power. The logic of this argument is that the profit maximising aeronautical price is lower than the full monopoly aeronautical price for in the absence of non-aeronautical activities. Indeed one can go further and observe that in certain circumstances it is possible that the unregulated monopolist sets an aeronautical charge that is lower than, the marginal cost of aeronautical services. These observations are true but the conclusion about the need for regulation of an airport monopolist is incorrect. First, note that the efficient competitive price for aeronautical services in this situation is, which is always lower than. So an unregulated monopolist will price above the efficient competitive level, which suggests price regulation would continue to improve efficiency in this case. Secondly, it cannot be taken for granted that for an unregulated monopoly the economic profit from non-aeronautical activities,, is always positive. If it is negative and the monopoly is unregulated, the airport will set aeronautical charges even higher than the stand-alone monopoly price,. It would be sound to argue that no airport would plan on being negative, as this reduces overall profits. But there is an important difference between what is planned and what occurs in practice, and this is an important distinction when considering regulation. A competitive airport will invest in non-aeronautical activities if it sees the opportunity to gain competitive advantage 9. However, if that investment turns out badly, the airport will not be able to recover losses from its aeronautical customers. This is exactly the same as if it were to invest unwisely in aeronautical facilities. An unregulated monopolist, on the other hand can recover some of the costs of its inefficiency or misjudgements from its customers. This reduces the incentive to act efficiently in the first place. The fact that the unregulated monopolist can recover some part of non-aeronautical losses from passengers reduces its incentives to be efficient in its commercial developments as well as its aeronautical activities. Economic regulation can protect passengers from this risk and promote efficiency on the non-aeronautical as well as the aeronautical side. 9 In this simplified example that advantage comes entirely through the ability to undercut rivals on aeronautical charges, but in practice this advantage could also result from improved quality of passenger experience.

13 July 2014 Frontier Economics 13 Implications for regulation Dual-till regulation leads to prices set above the competitive level We have shown that an airport operating in a competitive market will fully offset expected profits from non-aeronautical activities against aeronautical charges when those profits are a by-product of aeronautical activities, i.e. when those profits increase in line with increasing passenger numbers. We have shown that a competitive airport would set aeronautical prices equal to its perceived marginal cost, net of those non-aeronautical profits). This outcome (a price of ) is equivalent to the expected outcome of single-till regulation. This also holds when the airport has market power as shown in Figure 4. By contrast, dual-till regulation would be equivalent to capping aeronautical charges at the marginal cost of aeronautical services, allowing the airport to retain the profits of non-aeronautical activities. In this case the regulated airport may charge below the regulated cap, because it is still in its interest to share some of the non-aeronautical profits with customers, so it will in practice charge the lower of or. Profit maximising prices will only be lower than the dual-till regulated price if demand is very elastic and/or if economic profits from nonaeronautical activities are large relative to aeronautical marginal costs. In any case, the aeronautical prices that the airport sets will definitely be higher than the competitive level. This also means that the profit generated by the airport as a whole will be higher than the competitive level, implying excess profits to the airport operator. It is unclear why the retention of these excess profits could be justified. If they are, as described here, created by locational rents on activities that are by-products of the main airport activity then it is clearly appropriate for a regulatory authority to consider (and where necessary limit) the profits of the whole aeronautical activity, which by definition includes the profits of any by-products. Single till regulation corresponds more closely to the competitive outcome From these observations we conclude that single-till regulation corresponds, in principle, to the competitive paradigm. It will lead to lower aeronautical charges than dual-till regulation and charges that correspond more closely to the efficient level. We note that, in the presence of SMP, single-till regulation can always be expected to improve welfare and pricing efficiency relative to the unregulated situation. In most cases it is likely that dual-till regulation would also improve efficiency, but to a lesser extent than single-till. In the specific circumstance that the monopoly price after allowing for non-aeronautical profits (that is, as explained above, with the monopolist returning half of any non-aeronautical

14 14 Frontier Economics July 2014 profits to customers) is lower than the marginal cost of aeronautical services (i.e. > ) then dual-till regulation would actually be redundant as the airport can be expected to set charges below the cap set by the regulator anyway. But the outcome would still be worse for passengers than the single-till outcome as the price set by the airport would still be higher than the single-till (or the competitive) outcome. In the following section we address some of the practical issues of setting singletill or dual-till regulation. Before moving on, however, we cover two additional complications. These are: What happens if the airport has market power in non-aeronautical services; and The case of congested airports. Market power in non-aeronautical services In the preceding analysis we have assumed that the airport does not have market power over non-aeronautical services. This means that passengers are paying an efficient competitive price for retail services, car parking, etc.. Nevertheless, it is possible that the airport generates economic profits from these activities, because of fixed costs and the high density of use that these services achieve. If, however the airport has unregulated market power over non-aeronautical services then passengers will be being over-charged for these services (meaning there is a welfare loss) and economic profits from these services will be increased. In these circumstances, considering Figure 3 and Figure 4, the behaviour of the airport will remain the same, but aeronautical charges will be set lower, reflecting the market power in non-aeronautical services. This outcome is not ideal. In this case aeronautical are in efficiently low the airport is leveraging its market power in non-aeronautical into the aeronautical market. As passengers will not all consume non-aeronautical equally, this also involves a cross subsidy from the users on the non-aeronautical services to all other passengers. Given that in these circumstances the single-till regulated charge will be too low, while the dual-till charge will be too high it is not possible to say definitively which leads to the greater welfare loss. The correct solution from a regulatory point of view, however, should be to address the issue of market power in the non-aeronautical activities as a separate issue, and having done so, set airport charges on single-till principles. This approach offers the prospect of achieving the best possible solution for all passengers.

15 July 2014 Frontier Economics 15 Congested airports There is a limited literature which purports to address the relative merits of single-till and dual-till regulation by solving theoretical mathematical models. These papers tend to quote a result that single-till regulation is more efficient for uncongested airports while dual-till regulation may be more efficient at congested airports. We have reviewed this literature and find that these results depend on a specific stylised formulation of congestion that bears little relation to how congestion actually impacts on airports. Importantly in this literature congested should not be confused with full. In this literature airports always have spare capacity. For this reason alone little or nothing can be inferred from this literature. In reality congestion delays at major airports tend to be synonymous with the airport being almost full and slots being in short supply. Under these conditions ticket prices rise to choke off demand which is reflected in slot values if aeronautical charges are prevented from rising by regulation. In these circumstances passenger welfare is largely unaffected by the level of aeronautical charges or the method by which they are set. In this case the level of aeronautical charges primarily determines the share of congestion rents between airlines and the airport. It is sometimes argued that higher airport charges would act as a signal to expand the airport when the airport is full. We note that higher capacity prices when an asset is congested can act as a signal for market entry, but it is by no means clear that higher prices would give an incumbent airport the incentive to expand that it is facing no prospect of competitive entry. In our view this literature cannot be used to draw any useful conclusion about the merits of different forms of regulation in reality. It does not contradict the fact that single-till principles remain the benchmark for price-setting in a competitive market, while decisions over airport expansion, in reality are complex and balanced policy decisions in which congestion costs form one part of a complex agenda. A further explanation of the significance of these issues is given in an annexe.

16 16 Frontier Economics July 2014 Practical issues associated with single-till and dual-till regulation We recognise, of course, that there is a big difference between a theoretical model of pricing and a real regulatory mechanism. The method by which a rule is applied in practice can have a significant bearing on the effectiveness of the solution. We have argued that an airport in a competitive market will apply single-till principles. But this is not the same as saying any form of single-till regulation must be superior in practice to dual-till regulation. The application of regulation raises a number of important practical issues, which include: The scope of the regulatory till and the definition of included and excluded non-aeronautical services. Cost allocation issues. Incentive mechanisms. Non-aeronautical revenue yardsticks. Timing issues. The scope of the regulatory till As previously explained, regulation of airport charges, whether single-till or dualtill should not cover all activities simply because they belong to the airport or its owners. There is a clear rationale for deciding what should or should not be included in the regulatory till. All services that are ring-fenced and offered to airlines under competitive conditions (whether by the airport or third parties) should be excluded from any calculation of regulated charges. This primarily relates to ground handling services. All activities relating to the handling of aircraft on the runway and apron, and the processing of passengers through the terminal, which typically are covered by aeronautical charges should be included in the till. All non-regulated activities which are clearly a by-product of the aeronautical function of the airport should be included in the till. A by-product is a service which can be expected to increase in value as the number of passengers passing through the airport increases.

17 July 2014 Frontier Economics 17 Any service that does not fall within these categories should have its costs and revenues excluded from any regulatory calculation, whether the approach taken is one of single- or dual-till. However, the separation of the costs of different services creates specific issues. Cost allocation In the discussion above we have talked about economic profits from nonaeronautical activities being offset against aeronautical costs. In practice single-till regulation does not work in this way. Instead non-aeronautical revenues are netted off against total airport cost (after excluding ground handling and the unrelated services identified above.) to arrive at a residual that makes up the aeronautical charge. This represents a significant simplification of the regulatory process, compared to dual-till regulation. In the latter case it is necessary to exclude not only the revenues of the nonaeronautical activity from the regulatory till, but also its costs, but also its costs. This latter step if fraught with difficult and provides huge opportunities for regulatory gaming on the part of the airport. This is because a significant proportion of the costs associated with nonaeronautical services will in fact be incurred as costs in common with aeronautical services. Most obviously, separating the costs of retail space in the terminal building from space used for aeronautical services involve judgement and calculation. Does the regulator allocate costs pro-rata on the basis of floor space? If so, on currently used floor space, or future projected space, or the maximum space that could be allocated to retail? How are airport central overheads dealt with, including, but not restricted to the negotiation and management of contracts with franchise operators? Under a dual-till approach there is a clear incentive for the airport to attempt to allocate as large a proportion of costs as possible to the regulatory till, while excluding the revenues from these activities. Incentive mechanisms While we believe single-till principles most closely match the appropriate competitive outcome, we recognise that the incentive properties of any specific regulatory scheme will not be identical to those of a competitive market. As already described a profit maximising airport will reflect non-aeronautical economic profits in its aeronautical charges to some extent. But in the real world it will not simply pass through its actual non-aeronautical economic profits. Rather it will seek to pass through a competitive level, based on the equivalent profits that its rival airports can generate from similar activities. So, if an airport is exceptionally good at generating no-aeronautical profits, it will get to retain a

18 18 Frontier Economics July 2014 proportion of that profit, reflecting the extent to which it is able to out-perform its rivals. This is a healthy process as it provides the airport with an incentive to improve efficiency. To the extent that this gives the airport a commercial, advantage over its rivals, they will be induced to try harder and, in a dynamic market, may be expected to catch up their rival eventually. As the airport s lead over its rivals in the non-aeronautical arena is eroded, that airport will be forced to pass on the benefits of its performance in lower aeronautical charges. This means that in a competitive market we would expect to see a lag between the generation of any specific economic profits on non-aeronautical activities and the passing-back of these profits in in lower aeronautical charges. By contrast the process by which non-commercial profits are passed back in single-till regulation tends to be somewhat different. Typically a forecast of actual airport-specific revenues is made for the next regulatory period and that is deducted from total costs to arrive at the regulatory till. This process has the benefit that once the regulatory till is fixed the airport keeps the benefit of any out-performance until the next review of charges. However, the airport also has an incentive to understate its forecast of revenues in the first place, to minimise the reduction in the till and maximise the scope for outperformance. Furthermore, as the next regulatory review approaches, the airport may have a reduced incentive to further improve non-aeronautical revenues, because it will rapidly see any benefits taken away in lower aeronautical charges. These are problems which have similarities with the promotion of cost-efficiency under price cap regulation. There are potential solutions to the problem that could be considered, including benchmarking of non-aeronautical revenues against other airports to avoid gaming, and rolling adjustments that allow the airport to retain the benefit of any out performance for a fixed period regardless of when the outperformance is achieved. Yardsticks for non-aeronautical activities The previous section mentioned the possible use of benchmarking as a method of countering the potential gaming by an airport when assessing non-aeronautical revenues. The use of yardsticks, as opposed to benchmarking, is when a regulator actively substitutes estimates calculated from other firms (in this case airports) in place of the airport s own estimates of costs or revenues. Such an approach has its uses in the application of either dual-till or single-till regulation. Starting with single-till, we have discussed the fact that competitive airport charges reflect single-till principles, and that the effect of non-aeronautical

19 July 2014 Frontier Economics 19 activities should be one-way, that is they can reduce airport charges but not increase them. In contrast there are scenarios in single-till regulation where the charges could actually be increased by taking a single-till approach. In other words, aeronautical charges calculated on a single-till basis could exceed those calculated using a dualtill. By way of an example, consider an airport with a new, and largely empty terminal. This fact by itself may not suggest inefficiency: merely that the airport has built to accommodate expected future growth. Now, consider that, for simplicity, half the new terminal space was built to accommodate retail outlets, but at present these are largely empty. Consider then one possible outcome under a single-till regulatory approach: the regulator takes all retail costs into account (including half the empty terminal) and deducts forecast retail revenues, which even if they are normal on a perpassenger basis, are low absolutely because there are few passengers at present. Contrast this with a dual-till approach where the regulator identifies that half the terminal building is for the purpose of retail and excludes all these costs, along with forecast revenues. As the airport almost certainly making a loss on its retail space, aeronautical charges set on a single-till basis will almost certainly be higher than those set on a dual-till. Of course this result has occurred because in the dual-till case the regulator has made a sensible judgement on cost allocation, while in the single-till case the regulator has failed to benchmark adequately terminal costs. If it had done so it would probably have concluded that the terminal was excessively large and prevented the airport from recovering these terminal costs, at least until passenger numbers have risen sufficiently to justify the new space. It is equally possible however for a regulator using dual-till to make a similar mistake. In this case suppose half the terminal is intended for retail space but only 10% of the space is actually in use. If the dual-till regulator deducts 10% of the terminal costs (as well as the retail revenues) from the regulatory till then passengers will equally end up subsidising the excessive retail space. For the single-till regulator the use of yardsticks presents one potential solution to this problem. The key is not to allow a loss incurred by the airport on nonaeronautical activities to be cross-subsidised from aeronautical charges. In the above case the regulator could either disallow a proportion of terminal costs from the regulatory till (on grounds of inefficiency) or it could substitute actual retail revenues with a yardstick of revenues, based on what would be reasonable to expect from a terminal building g of the relevant size.

20 20 Frontier Economics July 2014 One approach or the other is needed to avoid aeronautical charges subsidising a loss on non-aeronautical activities. Timing issues Continuing the theme from the previous section, there are further complications that should be considered. The example given may represent an example of inefficiently scaled terminal expansion or simply a timing issue. As is often the case with commercial investment, costs may be front-end loaded and revenues may be back-end loaded. In this case the relevant benchmark for the single-till adjustment we would expect to see in a competitive market is not the actual in year profit (we have already noted that it is really the profit that rivals can make, rather than the airport s own profit that determines the adjustment), but rather the average rate of profit that a typically efficient operator can achieve. Given this observation, regulators faced with the problem described above might think about a method of averaging of costs and revenues over time, possibly on a per-passenger basis, to smooth out the variations caused by the inevitable fact that capacity and demand are unlikely to rise smoothly in line with one another at an airport. This applies to the issue of adjustments for non-aeronautical activities, but also more generally to the evolution of aeronautical costs and demand.

21 July 2014 Frontier Economics 21 Annexe the treatment of congestion costs There is a limited literature which purports to address the relative merits of single-till and dual-till regulation by solving theoretical mathematical models. A number of these articles are listed in the references to this paper. These articles tend to quote a result that single-till regulation is more efficient for uncongested airports while dual-till regulation may be more efficient at congested airports. We have reviewed this literature and find that this result depends on a specific stylised formulation of congestion that bears little relation to how congestion actually impacts on airports. Importantly in this literature congested should not be confused with full. In this literature airports always have spare capacity. For this reason alone little or nothing can be inferred from this literature. Interpretation of congestion It is worth starting with the observation that in these models the level of aeronautical charges that maximises welfare in the absence of congestion clearly reflects single-till principles, so surplus (or profit, depending on the exact objective function) from non-aeronautical activities is passed back in full in lower aeronautical charges. So the literature confirms our observations in this paper, that other things being equal aeronautical charges should follow single-till principles. However, the literature adapts this picture by adding the effects of congestion. The economic definition of congestion being applied is passenger delays, measure in terms of time and multiplied by an assumed value of time. It is assumed that delays increase as a linear function of the number of passengers. One should start by noting that this has nothing to do with an airport being full, as in having no spare runway slots. Indeed in these models there are no capacity constraints at all. Any level of demand can be accommodated, just with delays increasing in a linear fashion. This has huge implications for the results. Any increase in demand increases delays, which is a negative externality experienced by all passengers. Demand for air travel is assumed to depend on the total cost of travel which is the sum of the ticket price and the value of time lost through delays. Having set up this system, the authors make the observation that, in the presence of an externality of this kind, profit maximising behaviour by firms will not maximise social welfare. This is a well-known general result and not special to this literature. Typically profit-maximising prices will be lower than the social welfare maximising level and congestion will be too high.

22 Average Delay Per Flight 22 Frontier Economics July 2014 This would be true under all forms of price regulation, but in this case, observing that dual-till prices are higher than single-till prices and social welfare maximising tariffs are also higher than single-till prices, then if the costs of congestion are high enough (although high enough cannot be quantified) re is a window it is possible that dual-till prices are closer to the social welfare maximum than singletill prices. It should be emphasised that this fact, while true in this model, is by coincidence, not by design. There are a number of problems with drawing policy conclusions from this result. The first is that as a matter of fact a linear function linking passengers to delays bears no relation to the reality of delays at a busy airport. In reality delays at an airport tend to be low until the airport has nearly run out of runway slots, when delays rise sharply. But even then delays do not rise exponentially: even when an airport is full it cannot gridlock like an over-used road, because the number of ATMs is rationed to the available slots. Delays rise to a higher level because the fullness of the airport reduces the ability of the system to recover from shocks, but that effect is always likely to be limited. This is illustrated in Figure 5 below. Figure 5. Typical airport delay curve stylised example Demand/Capacity Ratio