STANSTED AIRLINES CONSULTATIVE COMMITTEE

York Aviation STANSTED AIRLINES CONSULTATIVE COMMITTEE STANSTED AIRPORT FORECASTS REPORT May 2006

Originated by: Louise Congdon Dated: 22nd May 2006 Reviewed by: James Brass/Richard Kaberry York Aviation Dated: 25th May 2006

STANSTED AIRLINES CONSULTATIVE COMMITTEE STANSTED AIRPORT FORECASTS Contents Page 1 INTRODUCTION...1 2 BAA S TRACK RECORD...6 3 BAA S FORECASTING METHODOLOGY...12 4 COMPARISON TO OTHER FORECASTS...34 5 KEY ISSUES...43 6 REALISTIC MARKET PROJECTIONS...61

1 INTRODUCTION 1.1 York Aviation LLP (YAL) was appointed in autumn 2005 to advise the Stansted Airlines Consultative Committee (ACC) regarding the air traffic forecasts for Stansted Airport presented by BAA in connection with the proposed development of the Airport. This Report covers the air traffic forecasts prepared by BAA in connection with the Interim Master Plan and G1 development, those prepared in support of the G2 Project and those supporting the current quinquennial review process, including those included with the most recent Stansted Capital Investment Programme 2006 (CIP). 1.2 Stansted Airport is facing a critical period, with substantial decisions due to be taken in the next few years which will affect the course of its development for the foreseeable future. The forecasts for growth, the facilities required to meet demand and the cost of providing such facilities are the fundamental building blocks to understanding the appropriateness of decisions to be taken regarding the future of the Airport. This Report concentrates on the air traffic forecasting aspect and the relationship between growth and any pass through of development costs into airport charges. History 1.3 A number of presentations and documents relating to the G1 and G2 forecasts had been given by BAA to the ACC prior to York Aviation being asked to advise the ACC. The first of these was a presentation made by Stan Maiden to the ACC on 14 th June 2005. This presentation was put forward by BAA as a first step in the process of constructive engagement. We discuss our understanding of BAA s methodology as presented in Section 2. However, the presentation, and subsequent working papers issued by BAA, focussed on the principal demand drivers for traffic growth in the London area as a whole and some particular issues relating to Heathrow and Gatwick, rather than addressing specific issues related to forecasts for Stansted. A further presentation was made to the ACC in September 2005 on these system demand drivers. York Aviation LLP 1

1.4 The ACC indicated that, at the time, it was not able to engage in discussions regarding the forecasts underpinning development at Stansted without clearer information regarding, in particular, the business case for G2 and specifically the implications for user costs. The ACC also made clear that it wished to understand how BAA had taken account of the specific needs of the low fare airlines, making up the majority of Stansted users both now and in the future, in considering the relationship between the market for Stansted and the cost of development to be passed on to users through airport charges. 1.5 In July 2005, BAA published its draft Interim (G1) Master Plan. The ACC continued to press for further details supporting the passenger, movement and cargo forecasts contained therein. 1.6 In October 2005, the ACC wrote to BAA with a list of requests for further information needed by the ACC to understand the basis of the proposals contained in the draft Interim (G1) Master Plan. It was made clear that the airlines were not in a position to comment on the G1 plan without adequate information regarding the traffic forecasts and the justification for specific development projects necessary to support the growth of Stansted to 35 mppa and/or the capacity of its existing runway. Included within this was a request for further clarification regarding the traffic mix underpinning the overall passenger and movement forecasts for Stansted and how these related to busy hour flows giving rise to the need for specific facilities. 1.7 On 2 nd December 2005, BAA responded with partial responses to the questions posed by the ACC. This included a basic breakdown of the 35mppa passenger forecast by segment. Correspondence continued regarding the information still being sought by the ACC in response to the earlier questions about the G1 development and the inability of the airlines to make meaningful comment in the absence of the relevant information. The issues are now being pursued in the context of the G1 planning application. 1.8 In December 2005, BAA published its G2 Consultation Document, based on the demand forecasts previously issued. The ACC repeated its requests for more detailed explanation of the derivation of these forecasts and the assumptions underlying them. At the ACC s request, a further presentation on forecasting was made on 17 th January 2006. This was largely a repeat of the earlier presentation of June 2005 and failed to address the questions raised by the ACC over the previous 7 months. 2 York Aviation LLP

1.9 Following that presentation and in order to facilitate engagement on the question of air traffic forecasts for Stansted, the ACC formed a Forecasting Working Group, which York Aviation was asked to advise, in order to continue a dialogue with BAA. The aim was to gain a greater understanding of the forecasts being presented for Stansted and with a view to seeing the forecasts amended to reflect the views of the airport s users, as BAA had accepted at Heathrow and Gatwick airports. It had been hoped that BAA would set out more realistic growth forecasts for Stansted as a result of the process. However, BAA has appeared unwilling to engage fully. 1.10 A letter was sent by the ACC to BAA on 23 rd January 2006, setting out the information and explanations it was seeking, relating specifically to how BAA s forecasts for Stansted had been derived. This is attached in Appendix A. It was intended that the responses to this letter would form the basis of a meeting between the ACC s Working Group and BAA in order to move towards an agreement on passenger and movement forecasts underpinning the case for both the G1 and G2 developments. 1.11 To assist the process, the ACC requested York Aviation to examine some of the critical issues underlying future growth projections for Stansted, particularly in terms of the drivers for growth at Stansted, the sensitivity of growth to increases in airport charges, and the extent to which there would be additional spill of traffic from Heathrow and, to a lesser extent, Gatwick to Stansted. The early findings from this work were shared with BAA at a meeting on 17 th February 2006. Little further clarification was obtained from BAA at this meeting, although the ACC was promised that a full explanation of BAA s forecasts would be given, as requested in January, along with some sensitivity tests following submission by BAA of the updated forecasts for all the London airports to the Civil Aviation Authority (CAA) at the end of February. 1.12 Our preliminary findings were also presented to the full ACC meeting on 24 th February and have subsequently been summarised in responses to the CAA s Regulatory Issues Paper and in response to the G2 Consultation. We give a fuller explanation of our projections of future demand at Stansted in Section 5. 1.13 Following these meetings, a letter dated 1 st March 2006 was sent to the BAA setting out the ACC s understanding of what BAA had agreed to provide by way of additional information and sensitivity tests. This letter is also contained in Appendix A. York Aviation LLP 3

1.14 In the meantime, BAA responded to some of the questions posed by the ACC but without answering the fundamental questions regarding the basis upon which the forecasts had been made and the sensitivity of these forecasts to varying assumptions about capacity at the other airports or the price sensitivity of Stansted demand. On 30 th March, the ACC was sent a summary spreadsheet showing how BAA s projections for Stansted would be affected by varying the assumptions about the underlying price of oil and assuming a 100% pass through of the proposed increase in airport charges, taking into account actual Stansted air fares. This was construed by BAA as the ACC s forecast but it is important to recognise that the results still derive from BAA s model and are not accepted by the ACC. We discuss these issues further in Sections 2 and 4. 1.15 Although BAA has supplied some information on the results obtained from its forecasting model, the ACC has been unable to obtain any explanation for the basis of the assumptions about Stansted s market share upon which the forecasts are based. At the present time, as we go on to discuss, we are unable to recommend to the ACC and other key stakeholders that the BAA s model represents a robust basis upon which to consider the market for Stansted over the period to 2030. We discuss this further in the next section. 1.16 On 3 rd May 2006, BAA formally launched its 10 year CIP for its airports. The ACC will be responding formally to the Stansted CIP in due course. A separate background paper explaining the forecasts was issued to the ACC. This paper misconstrues the ACC s position regarding the forecasts for Stansted as will be clear from a reading of this Report. 1.17 In this Report, we explore BAA s passenger forecasts for Stansted, their robustness, some of the key issues raised and present an alternative view of the growth prospects for Stansted. The Report is structured as follows to set out: BAA s track record in Section 2; our understanding of BAA s methodology in Section 3; a comparison between BAA s forecasts and earlier forecasts for Stansted in Section 4; the key issues raised in Section 5; our projections for growth at Stansted in Section 6. 4 York Aviation LLP

1.18 We do not deal with air freight forecasts in this paper as generally the magnitude of growth projected is not queried by the ACC. York Aviation LLP 5

2 BAA S TRACK RECORD 2.1 In this section and the next, we set out the ACC s understanding of the methodology adopted by BAA and the resulting forecasts. We also comment on some of the issues raised, which are discussed further later in this Report. Our comments here are limited by the lack of information supplied by BAA regarding its modelling, despite repeated requests for information by the ACC. In particular, whilst some information has been supplied on the numerical values of assumptions used, there is no information or evidence as to the validation of these assumptions. In our mind, this places severe doubt on the overall robustness of the process and the resultant forecasts. 2.2 BAA adopts a top down methodology for producing its airport passenger and movement forecasts. Under this approach aggregate forecasts are produced for the UK as a whole (Department for Transport) or, in BAA s case, for the London area and then allocated to individual airports. This approach assumes that the drivers of demand remain broadly constant over time and that the factors influencing which airports passengers will choose to use remain broadly constant, i.e. that passengers will respond in a consistent manner trading off access to an airport, with the frequencies available and the air fares offered, along with any other relevant factors. BAA s approach differs from classic top down approaches in being far more deterministic and reliant on the judgement of the forecaster as to how demand and capacity at airports will evolve over time than the more usual approach in modelling changes in behaviour in relation to changes in air fares, service frequency and other airport specific factors, such as surface access using logit or other probability models. 2.3 The top down approach assumes, in essence, that there is a fixed quantity of demand, in this case, to use the London airports, which has to be allocated between airports, albeit that at the margin, when all airports are full, some demand is assumed to be lost, i.e. not able to be accommodated. This traditional approach tends to be robust when the relationships between airports are stable but less so when there are complex competitive changes in the market. This is evident in BAA s poor track record at Stansted where the dynamics of the effect of ongoing capacity constraints at Heathrow and of the generative effect of low fare services have not been well represented in BAA s forecasts. DfT found that it had to overlay low fare airline demand on the base model outputs within the White Paper forecasts in order to replicate the specific market generative effects of the existence of low fare services at particular airports. 6 York Aviation LLP

Previous Stansted Forecasts 2.4 We understand that, whilst assumptions underlying the forecasts may change over time, BAA s forecasting methodology has been fundamentally the same since the original forecasts relating to the development of the new terminal were put forward at the time of the Public Inquiry in 1981-3. Indeed, BAA has suggested to the ACC 1 that many of the assumptions upon which today s forecasts are based rely on the judgement and experience of the forecaster built up over the intervening time period. 2.5 The approach adopted, as we discuss below, is basically a top down approach with forecasts prepared for BAA s London area airports in total, then Stansted s share of that market determined by reference to a view about which airports passengers prefer to use and capacity constraints at Heathrow and Gatwick. Essentially, BAA s approach to modelling demand at Stansted is one which is founded in the assumption that growth is driven by spill from Heathrow. This is clear from its response to the CAA s consultation on the Airports Review Policy Issues 2. Growth at Stansted is seen simply as a function of the scale of the London market as a whole, with growth at Stansted occurring principally because Heathrow is full. This completely ignores the fact that low fare airline growth at Stansted has stimulated the market independent of capacity constraints at Heathrow. BAA contends that: First, the nature of the London system is that the traffic volumes for Gatwick and Stansted (..) are driven to a large degree by the spill of Heathrow traffic to those other airports. 3 2.6 A consequence of this approach is that demand is assumed to be significantly less influenced by price, specifically at Stansted, leading to BAA s assumption that charges at Stansted will be able to rise simply because there is excess demand in the system as a whole, giving rise to monopoly profits to the operator of the three main London airports. However, there is no evidence to suggest that this is what will actually happen in practice as we explore below. 1 At the ACC Meeting of 17 th January 2006 2 March 2006, as published on the CAA s website. 3 Ibid, para 58 York Aviation LLP 7

2.7 The concept of growth at Stansted being driven by overall demand within the system exceeding the supply of airport capacity elsewhere has been implicit in BAA s approach from the outset as, when the development of Stansted as the third London airport was first proposed, Heathrow was operating under an annual cap on air transport movements of 275,000 air transport movements (atms). Initial forecasts for Stansted were, hence, that it would develop as a result of airlines being unable to gain slots at Heathrow. In practice, this did not happen independently of the attraction of low fare services through heavily discounted airport charges in the 1990s. We illustrate the accuracy of BAA s forecasts for Stansted since the early 1990s in Figure 2.1. Figure 2.1: BAA s track record of Passenger Forecasting at Stansted 30 mppa 25 20 15 10 5 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Actual BAA 1991 BAA 1995 BAA 2002 CIP 1998 CIP 1999 CIP 2002 CIP 2003 BAA G2 2.8 It is clear that early growth expectations were not fulfilled as airlines operating at Heathrow and Gatwick were unwilling to transfer services, or growth in services, to Stansted. BAA s early forecasts for Stansted based on an overspill role were 58% too high. Instead, both the other airports have attained growth above expectations and there is no evidence that network carriers are willing or able to develop services at Stansted in parallel with the low fare carriers. Indeed, those that attempted to develop services at Stansted in parallel with operations to the other London airports and in competition with low fare airlines, such as Lufthansa, have demonstrated that Stansted does not provide a viable alternative for such airlines. In fact, growth at Stansted has been almost entirely due to new and innovative types of operation by carriers with a fundamentally different business model. 8 York Aviation LLP

2.9 By the mid-1990s, BAA adopted a marketing strategy for Stansted which sought to attract services by the low fare segment of the market. This involved heavily discounted airport charges which allowed these airlines to grow and exploit the market opportunity at Stansted. It is true that, in part, growth in low fare airline services (LCCs) has displaced some growth in the market for the network carriers short haul services at Heathrow and other airports as indicated by BAA in its response to the CAA: Put simply, the LCCs have been substitutes for part of the economy sector of the flag carriers short haul market. 4 2.10 However, as we go on to discuss later in this report, low fare airlines have created new market opportunities to destinations as varied as Bergerac to Bydgosz. This is not simply a substitution of traffic from Heathrow but a new market enabled through low fare offers, underpinned by low airport charges at Stansted and the destination airports. 2.11 That BAA has failed to understand the dynamics of this new market is evident in the systematic under prediction of demand at Stansted over the last decade. Forecasts have been up to 63% too low. Despite this, BAA trumpets the role of new services and low fare airlines in growing traffic at Stansted but fails to understand the consequences of this pattern of growth for the future market potential: "Stansted again reported strong performance, up 9.1%, reflecting both the continuing growth of existing and new carriers and the development of new routes across the UK and Europe. The airport served 23 new routes during the year." 5 2.12 Despite evidence that growth at Stansted is subject to its own unique set or drivers arising from low fare airlines attracted to Stansted by low airport charges and only influenced in part by any ongoing capacity shortfalls at the other London airports, BAA persists in putting forward a view that in future, growth will be driven by overt spilling of services from Heathrow and Gatwick to Stansted, resulting in a changing mix of traffic over time. It is instructive to look at the forecasts submitted in 2001 in support of the previous substantive planning application for the expansion of Stansted, where it is stated: 4 Ibid, para 127 5 The UK Airports Industry, Airport Statistics 2004/2005, Centre for the study of Regulated Industries, The University of Bath, Author: Alan Cruickshank, Aviation Policy Director, BAA. York Aviation LLP 9

It is anticipated that some of the traffic which cannot be accommodated at Heathrow and Gatwick will be picked up by other airports and modes. As its network of routes develops, Stansted will be increasingly well placed to attract some of this demand that would otherwise be lost. In particular, opportunities will increasingly emerge to develop long haul services from Stansted. 6 2.13 On this basis, it was projected that 17% of passenger demand at Stansted would be on long haul services by 2010, which is simply not credible on current performance. 2.14 This change in traffic mix has simply not taken place and, as we go onto discuss, no new evidence is adduced by BAA as to why it should be different in future. Despite the evident discrepancies in earlier forecasts and the lack of justification for repeating the same erroneous assumptions, stakeholders are being asked to accept that, in future, such changes in the mix of traffic will indeed take place. This has profound implications for the future development of Stansted in terms of the scale and nature of facilities needed to meet demand. 2.15 Taken in combination, there is clear evidence that BAA s track record of forecasting demand at Stansted is unreliable, suggesting that BAA s ability to correctly interpret market signals is very poor. We are concerned, therefore, that the airlines, the regulator and the planning authorities are being asked to make judgements about substantial and expensive development at Stansted based upon these forecasts, which are clearly not robust. The key issues, which we explore later in this report, are: overestimation of the extent to which Stansted can act as an alternative to the other London airports; underestimation of the importance of low airport charges in attracting airlines to develop low fare services. 2.16 Furthermore, if it is assumed that growth at Stansted will be driven by spill from Heathrow and Gatwick to a greater extent than it has been in the past, it is essential that: realistic assumptions are made of the capacities of the other London airports in future; and 6 Environmental Statement Volume 2, BAA, August 2001 10 York Aviation LLP

clear explanation is given as to what types of services would spill and under what circumstances. 2.17 It is for this reason that the ACC has posed questions seeking to understand what BAA assumes to be the key drivers of demand at Stansted in the future and how varying assumptions give rise to alternative projections for Stansted, with different proportions of natural growth or traffic assumed to be spilled from the other airports. York Aviation LLP 11

3 BAA S FORECASTING METHODOLOGY London Area Forecasts 3.1 BAA produces forecasts of passenger demand for the London area using econometric models similar to those used by DfT. The principles are set out in BAA Passenger Forecasting Paper 1 7 with the assumptions set out in BAA Forecasting Paper 2 8. Assumptions underpinning the forecasts were revised and issued as BAA Passenger Forecasting Paper 6 9. We comment on BAA s methodology based on the presentations given and these 3 papers. These papers and presentations concentrate almost entirely on the methodology by which the overall London area forecasts are produced and give little explanation of the more critical, to the ACC s view, assumptions regarding Stansted s specific pattern of growth. 3.2 We understand that BAA uses only econometric methods for long term (10-30 year) forecasts but that shorter term forecasts are informed by a mix of methodologies. In essence, forecasts of demand for BAA s 3 London airports combined are produced, assessed against capacities available at the 3 airports and demand allocated to the airports accordingly. The rationale given for this approach is that macro-economic drivers of passenger demand are most clearly seen at an aggregate level. 10 3.3 In terms of producing these aggregate forecasts, markets are segmented by passenger type and geography, with different explanatory variables and elasticities applied to each segment. Sensitivity tests are undertaken to reflect uncertainties, which inevitably are greater for longer term forecasts than for short term projections. The June 2005 presentation to the ACC contained substantial information about the current market segmentation at the BAA airports as well as examining the relationship of demand growth in the London area as a whole to factors such as GDP, trade, consumer expenditure etc. Rates of decline of average air fares in real terms across the market segments were also noted. 7 July 2005 8 September 2005 9 November 2005 10 BAA presentation June 2005 12 York Aviation LLP

3.4 BAA presented its most likely scenario based on the assumptions of: falling air fares in real terms; declining elastisticies of demand over time; Channel Tunnel improvements; limited rail competition or impacts from video-conferencing. 3.5 A number of sensitivity tests were also outlined relating to inbound or outbound market saturation, lower or higher economic growth, or increased environmental taxes. 3.6 Summary demand forecasts for the London system as a whole were presented, showing that under BAA s most likely and higher growth scenarios, it expected there to be surplus demand over available capacity at the London airports, even if Stansted is developed to the full extent envisaged by BAA s G2 scheme (76mppa). This is used to justify the approach to spill and the assumption that airport charges will be able to rise without impacting on demand levels at individual airports as low fare demand is simply priced out of the system. 3.7 In Forecasting Paper 2 and in its presentation to the ACC of September 2005, BAA set out in more detail the assumptions underpinning its total London area forecasts. These were discussed and revised, as set out in Forecasting Paper 6, with the ACC s at Heathrow and Gatwick. In general, the does not dissent from the views on key econometric assumptions relating to GDP, consumer expenditure and trade agreed with the other ACCs except in relation to fuel prices, where BAA had originally assumed $55 per barrel and falling in real terms, which was revised to remain at $50 per barrel over the period to 2015 in Forecasting Paper 6. As a result, apart from short term rises in average air fares assumed as a result of fuel price rises in 2005, fuel prices are assumed by BAA not to be a substantial factor influencing air fares in future. Where there is particular disagreement with BAA is in respect of the price sensitivity of demand to rises in airport charges and the price of fuel assumed as we discuss below. 3.8 Forecasting Paper 2 also sets out the price elasticities assumed by BAA to apply to the different market segments. Although the specific price elasticities vary in the range -1.2 to -1.8 across individual segments, they are generally in the range to be expected in terms of overall market forecasts. What is unclear is how these elasticities have been applied and the extent to which dramatic changes in airport charges at Stansted have been factored into the overall demand base for the London area as a whole, taking into account that the impact will not be averaged over the whole system but impact on a specific segment of the market. York Aviation LLP 13

Traffic Allocation Catchment Area 3.9 In its June 2005 presentation, BAA explained that it allocates forecast passengers to each of its London airports based on the share which that airport is expected to achieve of demand in its own catchment area. This catchment area has been derived for each market segment from CAA 2003 survey data, which shows for example that Stansted presently captures 42% of UK leisure demand from within its own catchment area as defined by BAA but captures a lower proportion of demand from within the Heathrow catchment area. We understand that BAA uses these defined catchment areas as a surrogate for the surface access variable which would normally form part of a traffic allocation model. 3.10 The ACC was supplied with full details of these catchment area matrices in April 2006 11 in terms of the proportions of passengers in each of the 3 airport s catchment areas for each passenger type which are assumed to use each of the 3 airports in the base year and how this is assumed to change in future (2015 and 2030) in the maximum use of a single runway scenario, Stansted G2 development only, and the scenario where Heathrow moves to mixed mode operations and with a third runway. These 3 scenarios are the core scenarios for which BAA has provided information, although information for the latter scenario was not forthcoming until relatively recently. 3.11 The first difficulty which the ACC has is that there has been no adequate explanation of what is assumed to be Stansted s catchment area or how it has been derived, nor how this will change over time, merely that each zone is allocated to an airport s catchment area if the proportion of demand using that airport is more than the sum of the demand using the other two BAA London airports. The lack of adequate justification is of particular relevance given the nature of the current low fare offer in stimulating the market over a wide area, the implications for the catchment area of Stansted and the Airport s market penetration if this low fare offer is reduced due to price increases there and the effect of increasing competition on the margins of the catchment area from other airports offering low fare services, such as Coventry or Nottingham East Midlands or, indeed, in the core of the catchment area from Luton. 11 Letter to the ACC of 6 th April 2006 14 York Aviation LLP

3.12 For this approach to modelling demand at Stansted to be valid, it is important to understand the dynamics of the catchment area and how it will change as a consequence of the Airport losing competitiveness due to higher airport charges resulting in airlines switching generative aircraft capacity elsewhere. BAA has provided no explanation other than that districts are assigned to each of the London airports based on the current market shares and then adjusted over time based on experience and taking into account changes in the relative attractiveness of each airport induced by additional runway capacity at one or more locations 12. Allocation 3.13 Although the ACC has recently been supplied with information on the percentages assumed to choose the different airports for each catchment area and passenger type, no explanation is given of how this is estimated. So for example, we are told that Stansted attracts 44% of the UK business passengers from within its own catchment area now and that this will increase to 50% in the maximum use scenario, increasing to 74% with the G2 only development. This is an input to the BAA s traffic forecasting based on judgement, rather than an output as a result of modelling the competitive interaction between services offered at the 3 airports, air fares and access time/costs as would normally be the case in a fully developed traffic allocation model. This is a fundamental weakness in BAA s approach. 3.14 Put simply, BAA appears to assume that there will be a fixed volume of demand in each market segment in future years and that if Stansted attracts x, y and z% today from each defined airport catchment area for Heathrow, Gatwick and Stansted today, it will attract the same percentages in future (subject to judgemental adjustment) and that airlines will automatically put on services to meet this demand. This view is simply not consistent with the performance of a market where much of the demand is stimulated by low fare offers and airlines may be unwilling to initiate services in the face of higher airport charges than have applied in the past. In these circumstances, past observed market shares may not be a good guide to future behaviour. Even for conventional scheduled services, it may not be realistic to assume that airlines will simply meet demand, either because they are unwilling to compete with the existing low fare offers or because yields from the Stansted catchment area would not sustain viable operations. These issues need to be tested in significantly more detail than BAA has done to date. 12 Letter to the ACC of 16 th February 2006 York Aviation LLP 15

3.15 We are surprised that BAA s approach to top down modelling relies so heavily on observed patterns of demand adjusted by the subjective judgements of the forecaster. A more normal approach to top down modelling would be to derive a passenger allocation model in which demand in each district is allocated to airports based on an explicit trade-off between flight frequency, access time/cost and air fare. This would be assumed to be broadly constant over time but the attributes of each airport and the services offered there can be varied over time. Such an approach enables the sensitivity of the allocation to changes in flight frequencies or air fares offered by the airlines to be tested more explicitly, giving greater confidence in the results. This is the approach adopted by the DfT in its SPASM model. It should be noted, however, that this form of traffic allocation model still leaves unresolved the problem of how to predict the specific generative nature of low fare airlines. 3.16 BAA s approach is very static and deterministic and does not lend itself to explicitly or transparently testing the impact of varying assumptions. In particular, there is no apparent linkage between the allocation of passenger demand and the willingness of airlines to meet that demand through the provision of services. BAA has supplied no explanation as to the basis of these assumptions and we are greatly concerned that they may not be valid, particularly given the lack of past experience to draw on. We do not know how sensitive the results are to the judgement made. In our view, this approach to guesstimating the allocation of passengers between airports rather than modelling the interaction of demand and airline capacity is not robust. Spill 3.17 We understand that following this initial allocation of demand to airports, BAA checks demand against capacity and, if capacity is exceeded, assumes that some demand will be displaced to its second or even third choice airport. Having determined which passengers are displaced, these passengers are reallocated to a second choice airport and so on. 16 York Aviation LLP

3.18 In Forecasting Paper 1, BAA explained that it has algorithms to determine which market segments will be displaced by capacity constraints generally it is assumed that business passengers will be less likely to be displaced and leisure or transfer passengers more so. Although the ACC has been supplied with BAA s assumptions as to which type of passengers will spill, the derivation and validation of the assumptions are unclear. It would appear that having determined how many passengers need to be spilled to bring demand and capacity into line at an airport, it then divides this fixed amount of spill into different passenger market segments based on units of relative propensity to overspill 13. The propensities differ for an airport s own catchment area and the wider area. We understand also that judgemental adjustments may be made over time. 3.19 Significantly, no account is taken of how airlines would respond to capacity constraints. This appears to us to be fundamental as passengers cannot spill to an alternative airport without an airline willing to put on the relevant services from there. 3.20 We remain concerned that this deterministic approach may not be reliable given changing circumstances, particularly if the judgements are not well founded in what actually occurs in the face of capacity constraints. This appears to be another major flaw in BAA s approach to passenger forecasting for the London airports. Clearly, examination of BAA s early forecasts for Stansted (see Figure 2.1) would suggest that these spill algorithms may not be robust and tend to seriously overstate the willingness of passengers and airlines to relocate from Heathrow to Stansted. This is implicit in all past projections for Stansted which have shown dramatic increases in long haul traffic, assumed to spill from a constrained Heathrow, which have simply not materialised. 3.21 Having completed its allocation of passenger demand across its three airports, we understand that BAA then applies the effect of increased airport charges at Stansted to the demand assumed to want to use the Airport. However, to the extent that projected demand exceeds assumed available capacity at each of the BAA s airports, the effect is merely to moderate growth on the margin but with no material impact on 2030 demand levels as low fare demand priced out is assumed to be replaced by higher value traffic spilled from Heathrow, resulting in demand always being assumed to match capacity in the longer term. 13 Letter to the ACC of 16 th February 2006 York Aviation LLP 17

3.22 If low fare traffic is priced out and there is no evidence what spill actually takes place in the manner postulated by BAA, then the assertion that capacity at Stansted will be always be taken up appears subject to the same flawed logic as underpinned the original forecasts for Stansted in 1981-3, which proved to be 58% too high. We discuss these issues further below by reference to the forecasts disclosed by BAA to the ACC. Effect of Charges 3.23 In considering the impact of increased airport charges at Stansted, BAA made a number of assumptions: varying elasticities of demand by market segment, around a central assumption of around -1.3; that 50% of the increase in airport charges would be passed on by the airlines to passengers; an assumed average one way air fare of 50 (for short haul services) at Stansted. 3.24 Based on the results of this analysis, BAA suggests that, even with a 6 per passenger real increase in airport charges, demand at Stansted might only be reduced by 8.8% over the base Stansted forecast. However, whilst the range of price elasticities assumed by BAA, in terms of passengers response to changing air fares, may be robust, there are a number of fundamental flaws in this approach: it is unrealistic to assume that the airlines will absorb 50% of the increase in airport charges; from information supplied in later discussions, BAA has assumed an unrealistically high achieved one way fare for low fare services at Stansted 14 ; the approach fails to take account of the impact of increased airport charges on airlines fleet deployment plans in that loss of passengers will be more than incremental if services are simply not put on due to the effect of increased airport charges on airlines decisions as to which routes to serve over the network as a whole. 3.25 We explore these issues further in Sections 4 and 5. 14 BAA s data related to summer season fares only. Both Ryanair and easyjet, who account for the vast majority of traffic at Stansted have provided actual average fare data. 18 York Aviation LLP

Comments on the BAA Methodology Traffic Allocation 3.26 We have discussed in the paragraphs above a number of concerns regarding BAA s approach to modelling the allocation of passengers across the London airports. There appear to us to be substantial weaknesses in BAA s approach to traffic allocation modelling founded mainly in the deterministic and subjective approach to deriving future allocation parameters. There are other concerns which we discuss in more detail below. Treatment of Market Stimulation 3.27 In terms of an approach to forecasting passenger demand at an aggregate level and over the long term, the approach adopted by BAA is similar to conventional top down approaches albeit substantially reliant on deterministic judgements rather than the more usual dynamically modelled approach. We have discussed these issues above. 3.28 A fundamental flaw in these top down approaches is the treatment of market stimulation. This was evident in the adjustments made to the DfT models to overlay demand stimulated from by low fare airlines 15. This is of particular concern in terms of modelling the future growth at Stansted as the market across the London airports is not homogeneous, with traffic growth around the three airports likely to vary over time as a result of a number of factors, not least the availability of market generative low fare services. 3.29 Hence, as well as the shortcomings of the traffic allocation process described above, a further critical issue which BAA appears to ignore is the extent to which the low fare airlines have stimulated demand at Stansted (also Luton and more recently Gatwick). It is important to consider whether the specific nature of this demand stimulation and the future drivers for such stimulation are captured within an aggregate approach which assumes that the demand drivers for each market segment are constant across the whole London airport system. In our view, this market does not necessary derive from a consistent set of demand drivers and elasticities across the London system as a whole. It does, to an extent, have unique characteristics in terms of catchment area and price. We consider that, even within a top down approach to modelling demand in the London airport system, clearer account would need to be taken of how this generative effect would arise across the airports and in particular the effect of price at the individual airport level. 15 Passenger Forecasts: Additional Analysis, Department for Transport, December 2003. York Aviation LLP 19

3.30 By effectively ignoring the impact of price increases at Stansted and the resultant reduction in demand stimulation at the individual airport level and at the aggregate level, we consider that overall future demand in the London system is likely to have been overstated by BAA as the price increases at Stansted to pay for new capacity would not impact just on Stansted demand but the total levels of traffic in the system as a whole to be allocated across the airports. Higher airport charges at Stansted and the loss of low fare offers will adversely impact on airline competition and result in higher air fares overall, with a consequent loss of user benefit. 3.31 To the extent that BAA has used demand elasticities derived from recent market performance including this generative effect driven by ultra low fares made possible at Stansted as a result of low airport charges, it will overstate long term demand growth in the system as a whole. If air fares rise at Stansted as the increases in airport charges are passed on this will impact on: the willingness of airlines to initiate new services and the viability of existing services resulting in constraints on the market; loss of passengers on those services operated as a result of higher air fares; higher fares at the other London airports due to a loss of downward competitive pressure on fares overall. Taken in combination, these effects could be substantial. 3.32 We asked BAA to test the effect of applying the increased charges at the system level rather than the airport specific level. As we report below, BAA claimed that this would have little effect. However, the ACC has not been provided with any details as to how this result was derived and we thus have little confidence that BAA has properly modelled the impact of increased charges at Stansted under either methodology. We still consider that the impact of increased airport charges will have a more substantive effect on overall demand levels in the London system than estimated by BAA on either approach, particularly when taken in combination with other factors discussed elsewhere in this Report. 20 York Aviation LLP

Fuel Prices 3.33 As we indicated above, the ACC believes that it would be imprudent to assume that oil prices will fall and believes, at the very least, future projections should be based on $60 per barrel. Once this has been factored into the assumptions about future air fares and when coupled with the effect of increased charges at Stansted, the overall effect on demand in the London system, and Stansted specifically, is likely to be far more significant than illustrated by BAA s sensitivity test, which we outline below. Airport Capacity Assumptions 3.34 Of equal significance to the forecasts, particularly given BAA s methodology, are the capacity assumptions made for the other airports. We are aware that the Heathrow and Gatwick ACC s have undertaken detailed work with BAA to agree working assumptions as to the passenger and movement capacities of those two airports on the basis of existing infrastructure (including Terminal 5). We have not examined the detailed capacity issues at Heathrow and Gatwick in detail in the course of preparing this Report but have relied on the validity of the work undertaken by the ACCs at those two airports. However, it is important to remember that the work at Heathrow relates only to the use of capacity without any assumption as to the impact of mixed mode operations or of a third runway. 3.35 Initial forecasts presented by BAA did not take fully into account the additional capacity being planned by Luton and London City Airports in their Master Plans but we were informed that this was addressed in the more recent forecasts, alongside the revised capacities at Heathrow and Gatwick. 3.36 Rather, the critical issue relates to the assumptions made regarding the provision of additional runway capacity, particularly at Heathrow. Whereas it is reasonable to ignore the implications of additional runway capacity at Heathrow for the purpose of the next quinquennial review at that airport, it is unreasonable to base the long term planning for Stansted, which impacts on the next quinquennium, on the assumption that the only additional runway capacity will be at Stansted as this could result in costly overprovision of capacity ahead of demand. This is simply not consistent with Government policy as set out in the White Paper. Yet, forecasts presented by BAA in the context of the quinquennial review and in respect of the G1 and G2 developments are all presented on this inappropriate basis, albeit that some sensitivity tests have been presented demonstrating BAA s view of the impact of mixed mode and a third runway at Heathrow, as we will discuss below. York Aviation LLP 21

3.37 Whereas BAA argues 16 that it is preferable to provide capacity ahead of demand, this does not seem prudent when the costs of providing that capacity, when passed through to users, could result in capacity being unnecessarily provided and demand being priced off with net disbenefits to users, whether airlines or passengers. Given that it is well known that the Government is considering the introduction of mixed mode operations at Heathrow, upon which it may consult later this year, and that the White Paper sets out the conditions for another runway to be provided at Heathrow, or possibly Gatwick after 2015/2019, it is simply unrealistic to put forward demand forecasts as a basis for planning and regulatory decisions at Stansted which ignore those developments. We believe that the base case for Stansted should be a scenario where mixed mode is introduced at Heathrow in c.2010 and with an additional runway in c.2020, in line with the core White Paper forecasts, which we discuss in Section 3. BAA Projections 3.38 In this section, we set out the various demand forecasts for Stansted produced by BAA since June of last year. Our comments on these forecasts need to be read in the light of the foregoing commentary on the methodology. 3.39 In June 2005, two initial scenarios of demand at Stansted were tabled by BAA. The first of these assumed that runway development would take place at Stansted only and the second on the basis of what we believe to be the more realistic underlying capacity scenario, namely the introduction of mixed mode at Heathrow in 2010 and a third runway in 2020. These projections are set out in Table 3.1 (the period 2004/5 to 2009/10 is taken from the 2005 CIP Document 17 ). 16 in its response to the CAA Review Policy Issues Paper 17 CIP Capital Investment Programme 22 York Aviation LLP

Table 3.1: BAA s 2005 Forecasts for Stansted (mppa) Stansted Only Stansted + Heathrow 2004/5 20.9 20.9 2005/6 22.2 22.2 2006/7 23.4 23.4 2007/8 24.7 24.7 2008/9 26.3 26.3 2009/10 28 28 2010/11 29.6 29.6 2011/12 30.9 30.9 2012/13 32.3 32.3 2013/14 33.7 33.7 2014/15 35 35 2015/16 41 38 2016/17 45 39 2017/18 49 41 2018/19 52 42 2019/20 56 44 2020/21 60 45 2021/22 62 47 2020/23 64 50 2023/24 65 52 2024/25 67 55 2025/26 69 57 2026/27 70 60 2027/28 72 63 2028/29 73 66 2029/30 75 69 2030/31 76 72 Source:BAA 3.40 In the first instance, it is difficult to interpret these forecasts given the lack of detail supplied. Aggregate passenger numbers are meaningless in the absence of detail about the mix of traffic and how passenger numbers relate to natural demand growth at Stansted and how much is assumed to be displaced from Heathrow and/or Gatwick. Two key features do stand out, namely that: there is a sudden jump in traffic in the assumed opening year of the new runway (2015/6) which needs to be explained; if Heathrow is assumed to have additional runway capacity through mixed mode, demand growth at Stansted is much slower and only reaches 72 mppa 18 through accelerated growth towards the end of the period to 2030; 18 million passengers per annum York Aviation LLP 23

In addition, some impact of the opening of the third runway at Heathrow would have been expected the converse of the step change assumed when additional runway capacity opens at Stansted, in the G2 + Heathrow scenario. 3.41 In combination, these figures would suggest that BAA s projections of growth at Stansted are highly dependent on the assumed rate of spill from Heathrow. Given BAA s past track record of projections in this area, we do not consider these forecasts robust as a basis of planning the development of Stansted in the absence of substantial additional justification by BAA, particularly on the basis of the Stansted G2 only scenario. 3.42 Because of the apparent dependence of Stansted s growth on spill from Heathrow and the criticality of this issue for the timing of the need for additional capacity, the ACC requested further explanation of the assumptions underlying the specific forecasts for Stansted, including the breakdown of demand by traffic type and the dependency of each type of traffic on spill under varying assumptions of capacity elsewhere. The aim was to test the robustness of the projections as a basis for future physical and financial planning. 3.43 In December 2005, the ACC was supplied with a more detailed breakdown of the passenger forecasts underpinning the 2005 CIP, with a similar breakdown of passenger demand for the G2 (Stansted only) growth scenario for 2015, 2020 and 2030. This breakdown is set out in Table 3.2. 3.44 A number of features of BAA s projections are evident from this table. Dealing first with the period to 2015 (CIP and G1), there are 3 key issues: i) there is a rapid growth in long haul scheduled services expected at Stansted from 2007/8. Whilst Maxjet and Eos are successfully developing a niche market to the USA, BAA has been unable to provide an explanation of what additional long haul markets it expects to see served; 24 York Aviation LLP

Table 3.2: BAA Passenger Market Segmentation at Stansted (mppa) 2005 CIP G2 2004/5 2005/6 2006/7 2007/8 2008/9 2009/10 2010/1 2011/2 2012/3 2013/4 2014/5 2015 2020 2030 Low Cost Carriers 19.2 20.4 21.5 22.3 23.3 24.4 25.1 25.8 26.7 28.0 29.2 31.7 42.6 47.0 Other Short Haul 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.4 1.5 3.1 6.8 14.1 Long Haul Scheduled 0.0 0.0 0.0 0.3 0.7 1.2 1.9 2.3 2.6 2.6 2.7 3.5 7.7 13.2 Charter 1.1 1.0 1.0 1.1 1.2 1.3 1.5 1.6 1.7 1.7 1.6 2.7 2.9 1.7 Total 20.9 22.2 23.4 24.7 26.3 28.0 29.6 30.9 32.3 33.7 35.0 41.0 60.0 76.0 Source: BAA Table 3.3: BAA Passenger Market Segmentation at Stansted (mppa) with G2 + Mixed Mode and a third runway at Heathrow 2015 2020 2030 Low Cost Carriers 31.7 35.3 50.0 Other Short Haul 0.9 2.0 8.2 Long Haul Scheduled 2.4 4.8 11.7 Charter 3.0 2.9 2.1 Total 38.0 45.0 72.0 Source: BAA York Aviation LLP 25

ii) iii) BAA has not been able to explain the build up of other, non-low fare, short haul scheduled services within its forecasts, particularly given past performance by conventional scheduled carriers; growth in charter markets is unexplained given recent stagnation in that market in the UK generally. 3.45 Turning to the G2 forecasts, it is clear that there is an unexplained step jump between the 2014/5 G1 forecasts and the 2015 G2 forecasts. In discussion, BAA has asserted that there will be a rapid take up of new peak period slots when the new runway comes on stream. However, this would only arise to the extent that there is suppressed demand due to lack of capacity at Stansted in the year prior to the runway opening. This is not the case BAA is making. In practice, experience at Manchester, when it opened its second runway, would suggest that growth will still continue incrementally with at least a proportion of new peak period capacity being taken up by retimings of existing flights. Whilst the UK Government is considering mechanisms for allocating slots provided by new airport capacity in the context of the revisions to the Slot Regulation, there can be no certainty that the new capacity provided would be safeguarded for new demand and it is more likely that it will, at least in part, be filled by existing carriers retiming into the peak. 3.46 Of particular significance is the sudden increase in other scheduled services and charter services, as well as step changes in the growth of the low fare and long haul services projected by BAA. It has been unable to provide any justification for these step changes in growth, particularly in markets where, in the former case, Stansted has not been strong and in the latter case, the market is not currently showing growth and there are no substantive indicators of future accelerated growth. In this context, the rapid growth in charter services at Stansted to 2020 followed by a decline to 2030 remains unexplained other than BAA expects the charter market to decline in the longer term. It appears to us that the decline may already have begun. 26 York Aviation LLP

3.47 As can be seen the fastest growing markets between 2015 and 2030 are other scheduled services and long haul, both markets which can only rely on assumed spill from Heathrow. We discuss further in Section 5, how likely this growth is to materialise. Subsequently 19 the ACC was supplied with some information about the mix of traffic at Stansted in 2015, 2020 and 2030 assuming mixed mode and a third runway at Heathrow. This is shown in Table 3.3. 3.48 It is clear that the provision of additional capacity at Heathrow makes a material difference to the mix of traffic assumed at Stansted, with less other scheduled and long haul scheduled traffic projected. Furthermore, it is clear that in BAA s 2030 forecast low cost and charter traffic is being squeezed out from Stansted by traffic displaced from Heathrow. It is not clear how this is in the wider interests of users. 3.49 BAA also supplied the ACC 20 with the mix of traffic in the 40 mppa maximum use of the single runway scenario, on the basis that this level of demand would be reached on a single runway between 2020 and 2025. This is shown in Table 3.4. Once again it is evident that mix of demand assumed differs from either the G2 2015 projection when the Airport is expected to be handling 41 mppa or the 2020 projection. The G1 maximum use scenario shows significantly less full service scheduled demand and less long haul demand. BAA has not provided any explanation for the differences but it would appear to be at odds with the assumption that when capacity constraints bite business passengers are less likely to be displaced than leisure passengers. Table 3.4: BAA Passenger Market Segmentation at Stansted (mppa) in the G1 Maximum Use Scenario Low Cost Carriers 33.9 Other Short Haul 1.6 Long Haul Scheduled 3.3 Charter 1.2 Total 40.0 Source: BAA 19 Letter to the ACC of 4 th April 2006 20 Letter to the ACC of 16 th February 2006 York Aviation LLP 27

3.50 BAA was asked to provide information specifically about how much of the demand in each market segment was expected to be natural to Stansted in response to its own catchment area market and how much would be directly due to spill from a constrained Heathrow, and how these figures would be affected by the introduction of mixed mode and/or a third runway at Heathrow. Only headline figures for 2015 and 2020 have been supplied 21. These are set out in Table 3.5. We understand that the Throughput numbers include demand spilled from Heathrow, with the combined total capped by available capacity, which may result in spilled demand being assumed to replace Own Demand. Table 3.5: Throughput and Natural Demand projected at Stansted 2015 2020 Maximum Use of Single Runway G2 Own Demand 34.0 36.5 Throughput 35.0 41.0 Own Demand 41.5 52.5 Throughput 37.0 22 60.0 Source: BAA 3.51 From this analysis, it is clear that spilled traffic makes up an increasing proportion of demand at Stansted in BAA s projections. Given the uncertainties about whether the forecasts of spill are robust and the strong possibility of additional capacity at Heathrow, which would reduce the scope for spill, there have to be substantial question marks about the BAA throughput forecasts for Stansted. As indicated earlier in this section, we also have concerns about the robustness of the natural or own demand forecasts due to the lack of adequate explanation of their derivation and a concern, which we expand on later, that BAA may have underestimated the price sensitivity of demand to increases in airport charges. 3.52 At the ACC s request BAA tested a number of different assumptions through its forecasting model. These alternative assumptions were: the assumption that oil prices would remain at $60 per barrel; a lower base one way fare of 33 23 against which to test the price sensitivity of 100% pass through of airport charges increases; 21 Letter to the ACC of 6 th February 2006 22 Capacity constrained 23 Based on actual Ryanair and easyjet fares in 2005 see Section 4 28 York Aviation LLP

mixed mode and a third runway being constructed at Heathrow. 3.53 In addition, BAA also tested the effect of the price sensitivity of demand to increases in airport charges being modelled at the aggregate demand level as opposed to being applied to Stansted only. These results are set out in Table 3.6. 3.54 We note that the BAA column in these forecasts closely resembles the forecasts which BAA submitted to the CAA at the end of February for the purpose of the quinquennial review. These included further sensitivity tests showing the effect of 100% pass through of airport charges but are otherwise based on BAA s underlying assumptions, with the sensitivity of the results to a higher proportion of low fare traffic in the mix also given. Projections are given for intermediate years of 2020 and 2025. We have not repeated these forecasts as submitted to the CAA here as generally they reveal the same pattern of demand growth. 3.55 It should be noted that the final column in Table 3.6 represents BAA s modelling of the ACC s alternative assumptions set out above. Given our general reservations about BAA s approach to modelling demand at Stansted and the lack of adequate explanation, we have no confidence in the validity of these figures. These forecasts are not accepted by the ACC. 3.56 Whilst we continue to have significant reservations and substantial uncertainty about much of BAA s approach to modelling demand at Stansted, it is interesting to note that BAA s own model would suggest that demand at Stansted would only reach 32 mppa by 2015 and 66 mppa by 2030 if more appropriate assumptions are made regarding fuel prices, fare levels, the pass through of airport charges increases into air fares and more reasonable capacity assumptions at Heathrow. At these levels of demand, the G2 development could be deferred for a number of years (accepting that delaying the increases in charges would lead to higher demand in the short term) and could be more limited in scope to meet demand by 2030. York Aviation LLP 29

Parameter Value (BAA) Table 3.6: BAA Sensitivity Tests (mppa) Value (ACC) $ oil price $50 $60 Fare / Passthrough % (@ $50 oil) Fare / PT% (@ $60 oil) 50/50 % 50/50 % 50/50 ACC method % (@ $50 oil) ACC method (@ $60 oil) 50/50 % 33/100 % 33/100 % 33/100 % 33/100 % Charges Level Capacity Scenario Year BAA assum -ptions ACC assum - ptions 24 Max Use 2015 35 35 2030 40 40 Current 2015 46 46 SG2 2030 76 76 SG2 & LHR 2015 43 41 MM/R3 2030 76 76 5.50 SG2 2015 46 44 2030 76 76 7.00 SG2 2015 46 41 2030 76 76 8.50 SG2 2015 45 39 2030 76 76 5.50 SG2 & LHR 2015 41 39 MM/R3 2030 76 76 7.00 SG2 & LHR 2015 41 36 MM/R3 2030 76 74 8.50 SG2 & LHR 2015 40 34 MM/R3 2030 76 70 5.50 SG2 2015 43 42 2030 76 76 7.00 SG2 2015 44 39 2030 76 76 8.50 SG2 2015 40 37 2030 76 76 5.50 SG2 & LHR 2015 38 37 MM/R3 2030 76 75 7.00 SG2 & LHR 2015 37 35 MM/R3 2030 74 70 8.50 SG2 & LHR 2015 35 32 MM/R3 2030 70 66 8.50 SG2 2015 46 45 2030 76 76 8.50 SG2 & LHR 2015 41 40 MM/R3 2030 76 76 8.50 SG2 2015 44 43 2030 76 76 8.50 SG2 & LHR 2015 40 39 MM/R3 2030 76 76 Source: BAA 24 These projections are BAA s interpretation of the ACC s revised assumptions and do not represent forecasts agreed by the ACC. 30 York Aviation LLP

3.57 Within the long haul forecasts, a particular issue of concern to the ACC has been the projection of demand for services using A380 or equivalent aircraft as these aircraft are driving several costly elements of the G2 plan. In our view, the likelihood of such aircraft using Stansted and in what numbers cannot be assessed without some estimation of the routes upon which they may be operating. BAA has been unable to provide any information on the long haul markets expected at Stansted. We are aware of the likelihood of A380 type demand from the cargo integrators at Stansted but it is not clear to us, as we will discuss in Section 5, that the likely long haul markets at Stansted would sustain very large aircraft to any substantial extent, certainly not to the extent of the 3 million passengers using such aircraft by 2030 as asserted by BAA in planning documents and in correspondence. 3.58 Most recently, BAA has issued its Capital Investment Programme for Stansted for the next 10 years. This includes updated forecasts for the 10 year period. There are some fairly substantial changes in the rates of growth assumed in the short to medium term. We illustrate these in Table 3.7. The changes to the short term forecasts have not been explained by BAA. Table 3.7: BAA s Stansted CIP Forecasts 2005 2006 2005/6 22.2 22.2 2006/7 23.4 23.0 2007/8 24.7 23.8 2008/9 26.3 25.2 2009/10 28.0 26.8 2010/11 29.6 29.0 2011/12 30.9 31.0 2012/13 32.3 32.2 2013/14 33.7 33.4 2014/15 35.0 34.4 2015/16 n/a 40.5 Source:BAA York Aviation LLP 31

3.59 It is clear that BAA is now expecting somewhat slower growth in the short term than indicated by earlier forecasts. However, there is an acceleration of expected growth in 2010/11 which cannot be explained by any assumed boost to traffic arising from new capacity. It also appears perverse to expect acceleration of growth at that time as the increased charges which BAA hopes to levy on Stansted users would come into force in the next quinquennium and might be expected to lead to a slowing of growth. It appears on the surface as if BAA has undertaken a market driven analysis of its short term growth prospects at Stansted, which has resulted in a slowing in growth expectations, but has then simply interpolated these back to its long term trend; continuing to show demand at 40.5 mppa in 2015/6. We would have expected these adjustments to the short term forecasts to have had some impact on the longer term forecasts, even if merely in terms of a slippage in the projections. 3.60 In addition, BAA has recently submitted its G1 planning application to Uttlesford District Council. The forecasts accompanying the application appear, in overall passenger mix terms, to be consistent with the information supplied to the ACC, as given in Table 3.2. We do not comment in this Report on the more detailed implications of these forecasts in terms of aircraft movements and patterns of traffic implied by Stansted operating at 35 mppa, which will be the subject of a separate response by the ACC to the planning application. Conclusions 3.61 In the absence of adequate explanation from BAA, it is not clear the basis upon which growth at Stansted has been forecast, in particular what is driving natural demand growth in the catchment area and what sort of traffic is expected to spill from Heathrow. The varying mixes of traffic under the different scenarios do not appear logical and have not been explained by BAA. 32 York Aviation LLP

3.62 Generally, our view would be that top down modelling, assuming a full tradeoff model is developed, may be appropriate for modelling growth of conventional scheduled services at an airport but that bottom up supply side approaches are more suitable for markets dominated by low fare and charter air services. In both cases, the forecast for an airport cannot be made without reference to the expected pattern of routes and services. It is simply unrealistic to forecast future demand at an airport on an aggregate basis without taking account of the likelihood that the airlines will meet that demand, i.e. whether services are likely to be viable. 3.63 Furthermore, the assumptions made by BAA about what traffic will spill from Heathrow and how that will be taken up at Stansted appear to be entirely judgemental without supporting evidence. This is not a sound basis for making such critical assessments. 3.64 Given the extent to which the ACC is being asked to rely on unexplained judgements made by BAA and BAA s evident poor track record in projecting demand growth at Stansted, we cannot recommend BAA s projections as being a robust basis upon which to consider the business case for the provision of additional capacity at Stansted. 3.65 The particular areas of concern, which we will explore in Section 5, are: market stimulation and market maturity; the price sensitivity of demand; the market for long haul services; spill. York Aviation LLP 33

4 COMPARISON TO OTHER FORECASTS 4.1 In order to understand the forecasts produced by BAA, we have first compared them to the forecasts presented by DfT as part of the preparation of the Future of Air Transport White Paper. We draw some inferences from these comparisons. 4.2 In making this comparison, it should be noted that both the Department for Transport (DfT) and BAA produce their airport passenger forecasts using a top down methodology. To that extent, we would have the same concern about the appropriateness of this approach to modelling the generative effects of the low fares market in relation to the DfT approach as to that of BAA. 4.3 In the case of DfT, passenger forecasts are produced for the UK as a whole using econometric models which project future demand by market segment 25 based on assumptions about economic growth, trade growth, exchange rates, air fares and the rate at which the market will mature (a slowing of growth relative to growth in GDP). The current forecasts, underlying the Future of Air Transport White Paper, were produced in 2000 26 and we understand that these may be subject to review for the White Paper Progress Report due at the end of 2006. It should be noted that this view of the long term market share of the low fare airlines may be understated based on more recent market trends. 4.4 In preparing the White Paper, DfT made two adjustments to the forecasts produced in 2000. The first was to take into account assumed increases in air fares arising from the internalisation of the industry s environmental costs (emissions trading or otherwise) and the second was to take into account the accelerated reductions in air fares brought about by the low fares sector. In overall terms, the effect was broadly net neutral on the total volume of air passenger demand expected for the UK as a whole to 2015 and to 2030. The details are summarised in Annex A to the White Paper. 25 Business, leisure (and IT), UK or foreign resident by world region 26 Air Traffic Forecasts for the United Kingdom 2000, Department for the Environment, Transport and the Regions 34 York Aviation LLP

4.5 In essence, the modelling approach assumes that demand is fixed and seeks to allocate all demand across all airports using a model called SPASM, which estimates passengers choice of airport by examining the surface origin or destination of the trip, air fares, flight frequencies and surface access costs using a probability function 27. However, as the costs (to the passenger) of using an airport rise, demand is priced off. As capacity fills up at airports, these shadow costs rise to the point where some passengers are expected not to travel at all. Passengers using low cost and charter services are expected to be particularly affected by rising costs, hence these passengers are the first to be displaced away from congested airports. 4.6 Prior to the White Paper, DfT produced a revised version of the SPASM suite of models, which separately identified a block of low cost airline traffic, defined on the basis of airlines deemed to be low cost, including Ryanair, easyjet etc but excluding airlines such as Flybe and emerging foreign low fare airlines such as Air Berlin. The DfT estimated the size of the low cost market in 2030 to be 103 mppa out of their central national passenger forecast of 501 mppa. This implied a small reduction in the share of the total market over the longer term, on the basis of assumptions that, in the long run, longer haul markets would be the faster growing. Growth in the low cost market was projected to be only 5% p.a. from 2005 and 2% p.a. from 2010. These market growth estimates may be unrealistically low in the short term, leading to an inadequate recognition of the extent to which low fare services are replacing conventional scheduled services in short haul markets. 4.7 Whilst working within an overall controlled total of low cost demand for future years, DfT created a mechanism (NODDI) for overlaying newly stimulated low cost demand around airports offering low cost services currently (2003 in the case of the White Paper forecasts). This had the effect, within the passenger allocation model, of allowing pro-rata growth of low cost traffic at each airport in proportion to its UK share of that market in 2003 in broad terms. Thus, within the White Paper forecasts, low cost traffic at Stansted is shown growing roughly in line with the assumed national average growth rate. More detail on the approach is given in Passenger Forecasts: Additional Analysis, issued by the DfT at the same time as the White Paper. 27 In other words, passengers will choose which airport they use based on a weighted combination of these factors which will vary according to journey purpose etc but the model reflects that not all passengers will weigh the factors the same or make the same choice of route/airport. Hence 50% of demand from a given point may be assumed to use one airport, 20% another and so on. York Aviation LLP 35

White Paper Forecasts 4.8 In the White Paper, forecasts were presented for the following range of options: Maximum use of existing runways (Stansted at c.35 mppa and Heathrow at c.90 mppa on the assumption of continued segregated mode operations); 1 new runway at Stansted in 2011; 1 new runway at Stansted and 1 new runway at Heathrow: a short runway provided in 2020; a short runway provided in 2016; a long runway provided in 2020; 1 new runway at Stansted and 1 new runway at Gatwick post 2019; 1 new runway at each of Stansted, Heathrow and Gatwick. The option highlighted in bold is generally considered the core White Paper scenario. Our understanding is that forecasts underpinning this scenario did not assume the use of mixed mode at Heathrow 4.9 In Tables 4.1 and 4.2 below, we compare the DfT White Paper forecasts for Stansted with those tabled by BAA in connection with the G2 Master Plan (Tables 3.2 and 3.4). In Table 4.1, we have used BAA s assumption, set out in the draft Interim Master Plan, that the capacity of the Airport with one runway would eventually reach 40 mppa (as at Gatwick) and we have grown the G1 traffic from 2015 pro-rata to 40 mppa by 2025 in line with BAA s working assumptions. DfT took 35 mppa as the maximum capacity of the single runway. Where precise figures are not given, we have interpolated. 36 York Aviation LLP

DfT Table 4.1: Comparison of DfT and BAA Passenger Forecasts for Stansted with maximum use of a single runway (mppa) LCC Short Haul Full Service Long Haul Charter Total 2015 21 4.4 7 0.4 32.8 2020 21.1 6 7.9 0.2 35.1 2025 20.2 6.6 8.9 0.1 35.8 2030 18.4 8 10 0 36.5 BAA 2015 29.2 1.5 2.7 1.6 35 2020 31.5 1.6 3.0 1.4 37.5 2025 33.9 1.6 3.3 1.2 40 2030 33.9 1.6 3.3 1.2 40 4.10 Clearly, the mix of traffic at Stansted expected by BAA is materially different from that projected by DfT at the time of the White Paper, with materially higher levels of low fare traffic, for which increases in airport charges necessary to fund the cost of development will be much more significant. This also casts doubts upon the conclusions reached by DfT as to the fundability of the Stansted runway, which is relied upon by BAA in making the case for the additional runway. 4.11 As discussed in Section 3, BAA s core G2 Scenario is one where the only additional runway developed in the London system is at Stansted, giving a throughput of 76 mppa in 2030, whereas the equivalent White Paper scenario shows only 73.2 mppa at Stansted in 2030. The comparison is shown in Table 4.2. Again, the higher dependency on low fare traffic is evident, albeit that long haul traffic is shown to be moving closer to the levels assumed in the DfT s White Paper work. York Aviation LLP 37

Table 4.2: Comparison of DfT and BAA Passenger Forecasts for Stansted with two runways on the assumption of no development at Heathrow (mppa) DfT LCC Short Haul Full Service Long Haul Charter Total 2015 28.1 16.6 11 1.4 57.1 2020 30.7 21.6 12.7 1.2 66.2 2025 30.3 24.2 14.7 0.9 70.1 2030 28.9 26.9 17 0.3 73.2 BAA 2015 31.7 3.1 3.5 2.7 41 2020 42.6 6.8 7.7 2.9 60 2025 44.7 9.8 10.1 2.2 66.8 2030 47 14.1 13.2 1.7 76 4.12 In Table 4.3, we set out DfT s White Paper projection for Stansted for the case where a third runway is also built at Heathrow in 2020. This shows traffic at Stansted materially lower at 70 mppa, consistent with some of the BAA sensitivity tests set out in Table 3.6, but with the effect of the introduction of mixed mode at Heathrow included. Whilst BAA s forecasts show greater reliance of Stansted on growth in the low cost market, we remain concerned that the combined impact of mixed mode and a third runway may not have been adequately taken into account in BAA s projections. The impact is unclear as BAA has not supplied the ACC with projections isolating each effect. 38 York Aviation LLP

DfT Table 4.3: DfT Passenger Forecasts for Stansted with two runways on the assumption of a third runway at Heathrow in c.2020 (mppa) LCC Short Haul Full Service Long Haul Charter Total 2015 28 17.2 11 1.2 57.4 2020 32.2 16.8 12.2 1 62.2 2025 32.2 19.6 13.5 1.1 66.5 2030 32 21.9 15 1.1 69.9 4.13 The projections presented in Table 4.3 did not allow for the impact of mixed mode operations on the current runway in addition to the provision of a third runway there. We are not aware of any recent forecasts published which show the specific impact of mixed mode at Heathrow on the build up of demand at Stansted. Work undertaken as part of SERAS would suggest that the impact of mixed mode at Heathrow would increase demand at Heathrow by 5 to 15mppa in the longer term (4 to 14mppa in 2015) 28. The difference in the amount of additional capacity provided depends on whether mixed mode is assumed to operate for part or all of the day. 4.14 Provision of this amount of additional capacity at Heathrow would inevitably impact on any assumption which BAA has made regarding displacement or overflow of traffic from Heathrow to Stansted. This would suggest that, if mixed mode was introduced at Heathrow, most, if not all, of the incremental full service traffic (G2 over G1) would not use Stansted at 2015 and at a substantially lower volume than projected by BAA at 2030. BAA s current projections of 73-76 mppa at Stansted (dependent on the assumed increase in airport charges) with both a mixed mode and third runway at Heathrow do not look credible. More realistically, it would be expected that there would be a reduction of around 9 mppa in the other scheduled and long haul categories at Stansted in 2015 and 10 mppa in 2030 (taking an mid point figure for the impact of mixed mode at Heathrow). 28 SERAS Stage Two: Appraisal Findings Report (Halcrow for DfT) York Aviation LLP 39

4.15 Clearly, the assumptions made regarding the introduction of mixed mode at Heathrow will be critical to the timing of the need for additional capacity at Stansted and to the affordability of its provision. If we assume that the introduction of mixed mode at Heathrow in 2010 absorbs much of the increase in other scheduled and long haul scheduled traffic expected to spill to Stansted, then it would be expected that demand at Stansted would be significantly lower. In the absence of any relevant detail from BAA, we have made a tentative estimate of what the Stansted G2 forecasts might look like with the provision of mixed mode by 2010 and a third runway in c.2021 at Heathrow on the basis of DfT s estimates of incremental capacity at Heathrow (assuming a 10 mppa capacity gain) and BAA s traffic mix projections set out in Tables 3.2 and 3.3. These are set out in Table 4.4. Table 4.4: Estimated Impact of Mixed Mode and a Third Runway at Heathrow on G2 demand at Stansted (mppa) LCC Short Haul Full Service Long Haul Charter Total 2015 31.7 1.2 1.9 2.7 37.5 2020 42.6 1.8 2.8 2.9 50.1 2025 44.7 2.7 4.1 2.2 53.7 2030 47 4.1 5.8 2.1 59 Source: York Aviation based on BAA, DfT and SERAS work 4.16 Whilst the figures set out in Table 4.4 need to be treated with some caution, they present a sighting shot of possible demand at Stansted once the full impact of expanded Heathrow capacity is taken into account. Our analysis would suggest that the subsequent build up of traffic could fall far short of that necessary to justify the full two-runway layout stated as the preferred option by BAA in its G2 Master Plan. At the very least, there are high risks that the levels of demand implicit in the G2 scenario as currently presented will not be realised. We explore an alternative approach to assessing future demand at Stansted in Section 6. 40 York Aviation LLP

4.17 To the analysis in Table 4.4 needs to be added the full impact of the pass through of G2 development costs set against more realistic assessment of the actual fares charged by the low fare airlines, making up the vast majority of Stansted demand. We discuss this issue in more detail in the next section but present an initial adjustment here to demonstrate the sensitivity of BAA s forecasts to the critical issues of Heathrow capacity and airport charges. 4.18 Assuming a 6 per passenger increase in airport charges passed on to passengers against an average 1 way fare of 33, this would imply a 15% increase in fares. Using BAA s central price elasticity of demand to rises in air fares of -1.3, this would imply a reduction in LCC demand at Stansted of almost 23%. Applying this to our adjusted Stansted with Heathrow development scenario as set out in Table 4.4, this results in revised Stansted projections as set out in Table 4.5. In making this assessment, we have assumed that the increased charges would have no effective impact on the level of demand for full service scheduled flights and no further impact on the already declining charter market. This may, therefore, be conservative. Table 4.5: Possible Impact of Stansted Charges Increases, couples with development at Heathrow on G2 demand at Stansted (mppa) LCC Short Haul Full Service Long Haul Charter Total 2015 24.4 1.2 1.9 2.7 30.2 2020 32.8 1.8 2.8 2.9 40.3 2025 34.4 2.7 4.1 2.2 43.4 2030 36.2 4.1 5.8 2.1 48.2 Estimates made by York Aviation based on BAA, DfT and SERAS work Conclusions 4.19 By examining BAA s scenario forecasts for Stansted with earlier work undertaken by the DfT, it is clear that BAA is assuming a much less significant impact on demand at Stansted from the introduction of additional capacity at Heathrow than would be suggested by earlier analyses. In the absence of more detailed explanations from BAA regarding how it has taken increased capacity at Heathrow into account in its projections, it is at least reasonable to test what the impact on the projections for Stansted would be if the earlier impact assessment proved to be of the right order of magnitude. York Aviation LLP 41

4.20 Clearly this analysis has substantial health warnings attached to it, not least as the projected composition of demand at Stansted is now radically different from that projected in the White Paper work. However, we have considered the impact of Heathrow capacity only in so far as it might impact on full service long and short haul scheduled services and considered the effect of price rise adjustments only in relation to the attractiveness of Stansted to low fare airlines. If BAA is right that the development of the low fare offer at Stansted has in part been motivated by lack of capacity at Heathrow and taking into account the high risk to conventional scheduled airlines of initiating services at Stansted, both these assumptions may be conservative. 4.21 It should be noted that we are not presenting the analysis in this section as our projection of demand at Stansted over the period to 2030 but use it to illustrate how sensitive the projections are to varying assessments of the impact of critical assumptions. Given that the BAA forecasts are heavily dependent upon unvalidated judgements, it is important to understand how sensitive its projections are to different interpretations of how passengers and airlines might respond. In Section 6, we set out a bottom up projection of demand growth at Stansted. 42 York Aviation LLP

5 KEY ISSUES 5.1 In this section, we address a number of key issues affecting demand growth at Stansted. The analysis in this section underpins our growth projections set out in the next section. Market Stimulation and Market Maturity 5.2 Our hypothesis is that the market is now largely supply rather than demand driven in terms of growth of short haul services, and that top down modelling is no longer appropriate as a basis for forecasting demand in a largely low fares environment. 5.3 It is illuminating to examine the actual pattern of growth which sits behind recent high overall passenger growth at Stansted. In Appendix B, Table B.1, we set out the growth in scheduled traffic route by route for the last 10 years taken from CAA Annual Airport Statistics. In Table 5.1, we summarise the key features of this growth. 5.4 There is clear evidence emerging of market maturity at Stansted on an individual route basis. In particular, growth at Stansted in 2005 was almost entirely sustained by the introduction of new routes. These routes have been made possible by discounted/low airport charges. Examining the pattern of route growth at Stansted shows substantial route churn over time. Table 5.1 shows that typically 41% of growth at Stansted each year has come from the introduction of new routes. However, in 2005, the volume of passengers on new routes accounted for almost all the growth achieved at the Airport, indicating a maturing of the market on existing routes coupled with some route cancellations 29. In 2005, Ryanair accounted for 78% of the passengers on new routes, with easyjet accounting for ¾ of the remaining new route growth. The analysis also revealed that only around 58% of newly introduced routes since 2000 have survived longer than 3 years and that, even when routes sustain beyond the first 3 years, some are eventually withdrawn due to ongoing viability issues or other factors. 29 To an extent, this pattern will be a function of new route discounts time expiring and routes ceasing to be viable in the absence of discounted charges York Aviation LLP 43

Table 5.1: Sources of Growth at Stansted 1996-2005 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Total Scheduled 3,480,363 3,932,364 5,245,600 8,033,627 10,508,162 12,469,210 14,824,408 17,498,311 19,722,698 20,947,731 Other and Charter 1,327,637 1,431,636 1,584,400 1,375,373 1,349,838 1,184,790 1,224,592 1,217,689 1,184,302 1,044,269 Total 4,808,000 5,364,000 6,830,000 9,409,000 11,858,000 13,654,000 16,049,000 18,716,000 20,907,000 21,992,000 Scheduled growth 452,001 1,313,236 2,788,027 2,474,535 1,961,048 2,355,198 2,673,903 2,224,387 1,225,033 Growth Rate 13% 33% 53% 31% 19% 19% 18% 13% 6% Total growth 556,000 1,466,000 2,579,000 2,449,000 1,796,000 2,395,000 2,667,000 2,191,000 1,085,000 Growth Rate 12% 27% 38% 26% 15% 18% 17% 12% 5% Proportion of scheduled growth from new scheduled services 34% 63% 28% 36% 51% 36% 34% 36% 86% Proportion of all growth from new scheduled services 27% 56% 30% 36% 55% 35% 34% 37% 98% 44 York Aviation LLP

5.5 Examining the pattern of growth across the low fare routes started since 1996 shows a pattern of rapid route growth then a levelling off/reduction in volumes as passengers switch to new destinations. This is illustrated in Table B.2 in Appendix B, where we show how routes have performed since inception/1996. The growth profiles are averaged for the first year after introduction, second year, etc. This reveals a pattern of market maturity at Stansted as shown in Figure 5.1. Figure 5.1: Market Maturity Profile of Scheduled Routes at Stansted 120% Average Annual Growth Rate 100% 80% 60% 40% 20% 0% -20% 2 3 4 5 6 7 8 9 10 Year from1st Operation Source: CAA Statistics/York Aviation York Aviation LLP 45

5.6 It is clear from Figure 5.1 that, on average, new routes are introduced and show rapid initial growth but that these rapid growth rates are not sustained as the market matures at the route level. There is even evidence that in the longer term demand on an individual route basis levels off and can start to decline as passengers switch to new destinations when other new routes are introduced. The key reason behind this is that new low fare scheduled services will stimulate new travel between city pairs, giving rise to rapid initial growth in demand, but this demand base is not stable in that it is willing to switch to alternative destinations when new low fare offers are made available. This type of behaviour is virtually impossible to model in traditional top down passenger forecasting approaches which rely on consistent and predictable patterns of demand which can be allocated between airports based on stable relationships. 5.7 We use the market maturity pattern, illustrated above, in the next section to develop an initial bottom up projection of demand at Stansted. 5.8 There is a body of evidence for the stimulative effect of new low fare services, most notably in the Low Cost Monitor Reports produced by the Route Development Company (rdc) 30. We illustrate four routes in Figures 5.2 to 5.5. Clearly the pattern of market stimulation is variable, dependent in part upon the maturity of the city-pair market at the outset. It is clear that some passenger demand has simply shifted from conventional services to low fare offers at Stansted, whilst other routes have shown very rapid growth, more than conventional market analysis would indicate. In our view, these are material considerations to both the development of Stansted s own market and the characterisation of spill. To the extent that growth at Stansted has replaced growth at the other airports, spill is already taking place. 30 We have used the 2003 Report as the 2006 update is not yet available 46 York Aviation LLP

Figure 5.2: Market Stimulation: London-Aarhus Omitted Source: rdc Figure 5.3: Market Stimulation: London-Faro Omitted Source: rdc York Aviation LLP 47

Figure 5.4: Market Stimulation: London-Frankfurt Omitted Source: rdc Figure 5.5: Market Stimulation: London-Shannon Omitted Source: rdc 48 York Aviation LLP

5.9 On the other hand, in circumstances where the growth in demand is largely generative, adequate levels of demand do not exist, other than can be satisfied through an airline being willing to service it through low fares. In other words, the market has a very strong supply characteristic, in addition to that of demand. 5.10 Our analysis would suggest that existing markets at Stansted are showing signs of maturity and that growth is now to a large extent sustained by the introduction of new routes. The introduction of such new routes has been fundamentally underpinned by marketing support payments and discounts which BAA is intending to eliminate, as well as proposing substantial increases to the published airport charges. Furthermore, there is strong evidence that, at Stansted, routes are often introduced and then terminated over a 1-3 year timescale, to be replaced by other new routes. Given this high level of route churn and the dependence on new services, it is difficult to see how growth at Stansted can be expected to continue to any great extent when airlines are faced with substantially increased airport charges to a level which would make the introduction of new routes too great a risk for the airlines to bear. Contribution of Low Fare Airlines to Short Haul Growth in the London System 5.11 It appears to us that it is important to explicitly consider the impact of market stimulation and market maturity when forecasting passenger demand both in the whole London system and at the airport specific level. The pattern of demand is not independent of the service offered by the airlines, which in turn is heavily influenced by the cost of operating to an airport as we discuss below. 5.12 In Figure 5.6, we illustrate the share of 2004-2005 growth in domestic and intra-eu markets across the London area airports. Assuming that most of the short haul scheduled growth at Gatwick is also coming from the low fare sector, it is clear that in terms of short haul markets, most growth is coming from the low fare sector. This reinforces our view that effective spill in the short haul sector is already occurring but it is not occurring independently of low airport charges to attract new services. York Aviation LLP 49

Figure 5.6: Share of London Area Growth in Domestic and EU Scheduled Markets 2004-2005 Stansted Luton London City Heathrow Gatwick -20% -10% 0% 10% 20% 30% 40% 50% Source: CAA Statistics/York Aviation 5.13 It is also significant that in relative terms, growth at Stansted was less than at Luton and Gatwick. This may indicate that the Stansted market is becoming saturated in terms of both the existing routes, as discussed above and in terms of potential new city pairs. Stansted already serves more short haul scheduled destinations than any other London airport by a factor of 31%. The number of short haul destinations 31 served in 2005 from each of the London airports was: Stansted 142 Gatwick 108 Heathrow 87 Luton 55 London City 24 31 CAA Airport Statistics, all routes to Western or Eastern Europe carrying more than 1,000 passengers, excluding routes to Uzbekistan and including all domestic routes 2005 50 York Aviation LLP

5.14 In these, circumstances, it is not clear what routes are left to spill from Heathrow or Gatwick to Stansted. We discuss this point further below. 5.15 In any event, there may be a limit on the opportunities for new short haul routes at Stansted. An examination of the remaining cities in Europe with population in excess of 150,000 and not otherwise served by a neighbouring airport (excluding the UK and the former USSR) would suggest that there may only be around 70 cities which are not currently being served from Stansted but with an underlying population base necessary to sustain low fare services. This is around 50% more European routes than are being served today. It is notable that some of these, such as Dortmund or Charleroi, have sustained services in the past but are no longer operating. Others may be sufficiently close to places with direct services that additional services are relatively unlikely. 5.16 This analysis, taken in combination, paints a picture of growth at Stansted being driven by the introduction of new routes, with rapid growth through a combination of market stimulation through price and displacement of demand from Heathrow. The opportunities for continued rapid growth at Stansted in short haul markets are now much more constrained and are likely to be much more marginal, relying to an even greater extent on low airport charges to allow operations to be viably sustained. There is little to suggest that the provision of additional capacity would lead to an acceleration of growth, whether by low fare or conventional scheduled carriers, given the relative saturation of the market. We deal with the potential for long haul services from Stansted below. Price Elasticity of Demand 5.17 A fundamental issue is whether Stansted will be able to sustain the growth outlined under any of BAA s scenarios with the increased airport charges implied by the cost of development. In Section 4, we outlined an initial view of how BAA s scenarios might be adjusted if the full impact of additional capacity at Heathrow and increased airport charges, with 100% pass through into air fares are taken into account. 5.18 It is important to recognise that increases in airport charges will operate at two levels: York Aviation LLP 51

these will be passed onto passengers as part of the airport charge element of the fare and will have an impact on whether the passenger flies at all or switches to another lower cost service. In circumstances where prices rise across the board at Stansted, passengers may switch to alternative services from other airports still operating at a market competitive level of airport charges; to the extent that some element of the increased charge has to be absorbed by the airline, in the form of consequential lower yields, or reduced demand results in services becoming unviable, there will be a loss of routes and services with a much more than incremental effect on passenger volumes at an airport. 5.19 Almost all of the work on the price elasticity of demand has looked at the first element only. However, it is obvious that where the growth in demand is largely generative, demand cannot be assessed independently of whether airlines are willing to operate services with sufficiently low fares to stimulate the market. This is the model through which demand at Stansted has grown and through which it has enabled spill from Heathrow to take place that is spill of passengers to a more attractive low fare offer, not spill of services as envisaged by BAA. This reinforces our view of the inappropriateness of top down modelling as used by BAA as a basis for forecasting demand in a largely low fares environment. It is important that more detailed consideration is given as to how the generative effect of low fare services affects both demand levels overall and the specific demand at Stansted. 5.20 Even when BAA has assessed the impact of increased airport charges, it has understated the base fare against which to consider the impact. BAA assumed an average 1 way fare to/from Stansted of 50, based on a trawl of web fares during summer 2005. This is obviously wholly inappropriate given that fares in the winter season are much lower as airlines struggle to achieve acceptable load factors. We have been able to analyse confidential fare data from the airlines in more detail, some of which is being made available to the CAA directly by the airlines. Average one way fares at Stansted for 2005 were 33. There is also evidence that yields have been declining over time. It is this lower fare level against which any increase in airport charges needs to be considered. 52 York Aviation LLP

5.21 In 2004/5 average charges actually paid by airlines to BAA, after any discounts, were 2.61 32 per passenger. The most recent indications are that BAA will be seeking an increase to around 8.50 per passenger, assuming no cross subsidy is allowed from Heathrow or Gatwick. This would constitute an increase of around 5.50-6 over current levels, equivalent to an increase of around 17-18% of the air fare. Taking a central elasticity to price of around -1.3, this would imply demand being some 22-23.5% less than would be the case if the actual price paid by the airlines remained constant. In these terms, BAA s plans to charge up to the cap of over 5 at current prices in 2007/8, when current airline discount agreements come to an end, will in itself be a contributory factor to a slow down in growth. 5.22 An examination of traffic growth and airport charges across UK airports based on data from the Centre for Regulated Industries 33 (CRI) shows that there is a clear correlation between changes in airport charges and growth in air traffic, with the downward pressure on airline yields reflected in downward pressure on the level of airport charges revenue per passenger realised. Generally airports with higher than average growth in passengers have been those with real reductions in actual airport charges revenue per passenger over recent years. In Appendix B, Figures B.1 to B.6 we set out six different regression equations looking at the relationship between change in airport charges and levels of growth in air passenger demand relative to the average across the group of airports, which includes most of the UK airports handling commercial air services. The fastest growing airports are those which have been reducing their charges in real terms. 5.23 We would accept that, with Regression Coefficients (R 2 s) in the range 0.31 to 0.66, these relationships explain only some of the variations in growth rate across UK airports, with other factors coming into play. Furthermore, whilst CRI attempt to ensure that the data is consistent, it is possible that in some cases marketing support payments are shown separately and not netted off the reported airport charges revenue. Despite these shortcomings there is clear evidence of a relationship between changes in airport charges in real terms and the extent to which an airport will be able to sustain higher levels of passenger growth. 32 CAA Airports review policy issues Consultation Paper, December 2005, Annex B 33 CRI, University of Bath, Annual Airport Statistics Series York Aviation LLP 53

5.24 We use the relationships derived from this work in moderating our projections in Section 6 to take account of increases in charges at Stansted. In view of the uncertainties about the extent to which the derived relationships fully explain differential growth rates, we have deliberately chosen to be conservative and used the relationship which gives the least impact on demand at Stansted. We have taken the increase in charges as 16% per annum on average over the period from 2004/5 to the end of the next quinquennium in 2012/3 ( 2.61 in 2004/5 rising to 8.50 by 2012/13) as proposed by BAA in order to pay for the first phase of the G2 development. 34 This would suggest that, if airlines are faced with increases in charges to the extent planned, Stansted could see reductions in demand of the order of 6.3% per annum, compared to what it would otherwise attract for each year that the charges increase is applied. Beyond that, growth would potentially continue but only at national average growth rates. 5.25 To reflect the fact that the derived relationship may only explain about 32% (R 2 of 0.31) of the changes in growth rates (accepting that other factors such as capacity constraints overall may be at work in the London system over the period), we have also looked at a more moderate partial impact of charges assuming a slowing of demand growth of only 50% of that derived from the analysed relationship to allow for other mitigating growth factors, such as shortage of capacity at other airports in the London system or increases in prices at Luton to pay for development there. Spill 5.26 We have seen no supporting evidence and cannot conceive that full service airlines will transfer services from Heathrow or Gatwick to Stansted, a view shared by the carriers at those airports, for example, in its response to the CAA s consultation on regulatory policy issues, British Airways made clear that it did not see the various London airports as interchangeable in terms of providing capacity to meet demand: In competing for the business of airlines, it is important to note that different airports have different characteristics and may not be substitutable. 35 34 We have used a polynomial relationship excluding Blackpool as illustrated in Figure B.3 as the inclusion of Blackpool in the data suggests a more exaggerated relationship between reductions in charges and growth than is the norm as a result of both rapid passenger growth and a dramatic reduction in aeronautical revenues per passenger as a result of Ryanair commencing services. 35 Para 4.2 54 York Aviation LLP

5.27 BAA offers no explanation as to why additional full fare services would spill to Stansted in the future. Indeed, history would suggest that full service airlines have been reluctant to relocate away from Heathrow to any great extent, offer the low fares necessary to stimulate growth or to otherwise compete with the low fares market at Stansted, although recent fare reductions by British Airways suggest that it may be responding to competition. 5.28 Rather, our analysis would suggest that spill is already occurring as passengers from within Stansted s catchment area switch to using capacity offered there. However, as we have explained, this is not independent of the provision of low fare capacity and hence is not independent of the charges at Stansted. Furthermore, to the extent that spill is already occurring, it is effectively part of the natural demand growth at Stansted and not a new phenomenon to be added in at a later date if Heathrow becomes capacity constrained. Heathrow is already capacity constrained, albeit these constraints will reduce in the short term with the opening of Terminal 5. 5.29 We have examined the catchment area for the London airports. This is illustrated in Figure 5.7, taken from CAA provisional survey data for 2005 36. We have assigned each district to one of the London airports based on whether the share of the market captured by that airport exceeds 25% of the total, i.e. it captures more than an even share of demand from that district. The data is given in Appendix B, Table B.3. This is a broader definition than applied by BAA and there are clearly areas of overlap. The approximate Stansted catchment area is outlined in red in Figure 5.7. 5.30 As can be seen, Stansted s principal catchment area remains bounded in North and East London and the East of England region itself. This area accounts for approximately 70% of Stansted s terminating passenger demand. Stansted handles approximately 20% of all air passenger demand from within this area currently. However, it should be noted that over 60% of this demand is from Greater London, where competition with Heathrow and Gatwick is greatest and Stansted s share of the market lower. 5.31 In terms of short haul markets alone, Stansted captures 27% of demand from the whole of the wider catchment area in London and the wider South East and East of England, although the same pattern of catchment area concentration is clear. Further detail is given in Table B.4 in Appendix B. 36 Note: London City Airport was not surveyed in 2005 York Aviation LLP 55

Figure 5.7: Stansted Catchment Area 2005 Source: Mappoint/York Aviation/CAA Surveys 56 York Aviation LLP