Air traffic forecasts for the United Kingdom 1997

Table of contents

• Acknowledgements
• I Introduction
• II Interpretation of forecasts
• III 1997 Forecasts of air traffic demand
• IV International air traffic
• V Diversion of air passengers to Channel Tunnel rail services
• VI Domestic air traffic
• VII Split of traffic between London area and regional airports
• VIII Sensitivity tests
• IX Performance of past forecasts
• X Comparison with other forecasts
• Annex 1: Methodology
• Annex 2: Assumptions
• Annex 3: Summary of 1997 air traffic forecasts

Acknowledgements

The CAA assisted in the preparation of these forecasts by providing the split of the passenger forecasts between London and regional airports. The CAA, BAA and British Airways provided their own forecasts of air traffic growth for comparison with the Department's forecasts. The Department of Trade and Industry provided advice on fuel price assumptions used for projections of air fares. The Department of National Heritage provided detailed information on passengers and fares from their International Passenger Survey data base. The Department would like to record its gratitude for these contributions.

I Introduction

1.1 This document sets out the Department of the Environment, Transport and the Regions' (DETR) 1997 forecasts of air passenger traffic at United Kingdom airports. These forecasts supersede those published in 1994. They are the sixth set prepared by DETR (formerly the Department of Transport - DOT) since 1984 and reflect the intention to monitor air traffic developments and to keep assumptions and methodologies under review, as set out in the 1985 White Paper 'Airports Policy' (Cmnd 9542). The forecasts serve a number of purposes:

• they take a view on expected demand in order to inform policy on airport capacity
• they are a key input to studies on future capacity requirements such as RUCATSE
• they can be used to examine the economic implications of policy alternatives
• they are used by other Government departments

1.2 The forecasts are for the demand for air travel by UK and foreign residents to and from UK airports up to the year 2015, extending the forecast period 5 years beyond that in the 1994 forecasts.

1.3 The forecasts are used by Government departments, the Civil Aviation Authority (CAA), airports and airlines, together with their own forecasts, for planning and appraisal purposes. They are an input to the assessment of the need for runway and terminal capacity and the impact of capacity constraints on passengers, airlines and airport operators. The forecasts are also important to ensure that the environmental and development impacts of growth in air traffic can be assessed and taken into account in developing future Government policy on aviation.

1.4 Previous sets of forecasts were used in the 1993 RUCATSE study of potential sites for additional runway capacity in South-East England and more recently considered as evidence to the public inquiry into a fifth terminal at Heathrow Airport.

1.5 The rest of this report is structured as follows:

• Section 2: covers the basis of the forecasts and a number of key issues relevant to their interpretation;
• Section 3: presents a summary of the main forecasts, with a more detailed break down of International traffic in Section 4 and domestic traffic in section 6;
• Section 5: sets out how diversion of air passengers to Eurostar services using the Channel Tunnel was calculated;
• Section 7: gives a split of total passengers between London and regional airports;
• Section 8: sets out and discusses the results of a number of sensitivity tests which were conducted;
• Section 9 and 10: consider the accuracy of previous DOT forecasts compared with observed demand and compares these forecasts with the most recent sets produced by other organisations (such as CAA and BAA plc)

The annexes contain details of the methodology used to produce the forecasts and the assumptions about key variables as well as a summary of the forecasts.

II Interpretation of forecasts

Basis of forecasts

2.1 The underlying approach to the 1997 forecasts is similar to that used to prepare the 1994 forecasts. The forecasts are based on econometrically derived equations which specify a relationship between passenger traffic and a number of explanatory variables which determine it, using data from the early 1960s to 1995. The key variables determining air traffic were found to be, in broad order of importance: domestic and foreign economic growth (principally GDP); air fares; trade and exchange rates. This is consistent with the results of other research conducted over many years by the former DOT and others into air passenger demand and is also what would be expected from economic theory.

2.2 The relationships derived from past years' data are applied to projections of future year values of the explanatory variables to give forecasts of air traffic. Forecasts in this report follow the established pattern and are given for 5 yearly intervals from 1995, the base year, to 2015. Further details of these assumptions and of the methodology used to produce the forecasts are given in Annexes 1 and 2.

2.3 The total forecast is built up from individual forecasts of 16 international market segments and 3 domestic markets. The international markets are disaggregated by UK and foreign residents, business or leisure trip purpose and four geographic areas: Western Europe; other OECD countries; the newly industrialised countries of East and South-East Asia; and others. Forecasts for passengers travelling to and from Western Europe are adjusted to take account of the diversion of passengers to Eurostar services using the Channel Tunnel. There are three UK market segments: London to and from regions, the UK to and from the Channel Islands and between regional airports.

2.4 In addition to the 16 markets which are directly modelled, assumptions are made about the future course of a number of other categories of passenger. The most significant of these is 'interliners', international passengers who change flights at UK airports, but do not pass through customs control. Others include travel to and from oil rigs and by diplomats.

2.5 Air passenger demand is forecast in this disaggregated way for two main reasons. First, separate forecasts are required by the Government for some purposes. Secondly, the precise relationship between passenger traffic and the factors which determine it differ between sectors. Modelling individual segments is likely to produce a more accurate overall picture than treating the demand for air travel as a whole.

2.6 The forecasts are long term forecasts and show the trend growth in demand for air traffic to the year 2015. As noted above, a main determinant of air traffic is economic growth. Because of the difficulty in accurately predicting turning points in the economic cycle, short term deviations from the long term growth rates of air traffic in a particular year are to be expected. For example, the forecasts produced in 1994 underestimated demand between 1992 and 1995, partly due to higher than anticipated growth in the UK for those years as the economy recovered strongly from the recession of the early 1990s. But this does not necessarily effect the validity of the forecasts in the longer term. Demand for air traffic in the short term may also be subject to factors which no econometric forecasting model could accommodate, such as wars or terrorist action. Strategic behaviour by airlines can also have a significant short-term effect on demand, but in the long run may have little impact on the trend in growth. Trends in air traffic between 1992 and 1995 are discussed further below.

Capacity

2.7 For the purpose of the total UK air traffic forecasts, it is assumed that additional airport and airspace capacity is made available as necessary to accommodate growth in passenger numbers, which broadly represents the experience of the past 30 years. In this sense they are unconstrained forecasts of the underlying demand for air passengers, independent of any supply side limitations. Unconstrained forecasts are necessary to identify where and when the need for additional airport capacity will arise and to inform decisions about the provision of such capacity. Forecasts constrained by capacity would not give any useful indication of demand that would exist for new facilities. In order to make an assessment of the case for permitting additional capacity to be provided, it is necessary to consider underlying demand independent of any supply constraints.

2.8 In recent years, Heathrow, and to a lesser extent Gatwick, have experienced excess demand for airport facilities at certain times of the day and year. But capacity constraints at individual airports do not necessarily result in any suppression of total demand. Passengers can respond to constraints at particular sites by switching to a different airport or travelling at less busy times. This will be at a 'cost' to passengers, but there is little evidence at present to suggest that, to date, this effect has had any significant impact in depressing overall traffic growth. However, if there were significant shortages in airport capacity over the forecast period at London area and other airports, this would be likely to have an impact on total demand. For the purposes of these forecasts, it is assumed that at the UK level, capacity is sufficient to meet total demand, but an element of capacity constraints is incorporated in the split of the forecasts between London area and regional airports. An estimate of unconstrained demand at London area airports has also been made.

2.9 This division of total traffic between London area and regional airports is modelled using the CAA's passenger allocation model, which requires assumptions on the future capacities of airports. The only additional runway assumed is the second runway at Manchester Airport, for which planning permission was given in January 1997. The methodology, capacity assumptions used and results are discussed further in section 7 below.

2.10 The response to capacity constraints is particularly important when converting passenger numbers into estimates of landings and take-offs, air traffic movements (ATMs). Forecasts of ATMs are sensitive to runway constraints and trends in aircraft size. Airlines are likely to respond to runway constraints by deploying larger aircraft, giving rise to slower growth in ATMs than would otherwise have been the case. On the other hand, increased airline competition as a result of liberalisation and the development of new services will tend to result in the use of smaller aircraft, with a corresponding faster growth in ATMs. The Department does not produce forecasts of ATMs, which are made by the CAA. They take into account the factors mentioned above together with possible changes in route structure and airlines' aircraft purchasing decisions.

Treatment of uncertainty

2.11 All econometrically based forecasts depend on a number of different factors, assumptions and judgements which mean they are subject to a greater or lesser degree of uncertainty. There are three main sources of uncertainty:

a) the future path and values of explanatory variables; GDP, trade volumes, exchange rates and air fares, including oil prices

2.12 Producing robust, long term annual forecasts of factors such as GDP is difficult. Generally, more confidence can be placed in forecasts of long term average growth rates than shorter term forecasts of turning points in the economic cycle. For example, the long-term average annual growth rate of the UK economy has been broadly stable over the post-war period. Judgements are also required when considering the future path of air fares, regarding the effects of liberalisation and withdrawal of duty-free allowances on intra-EU flights for example. Where possible, quantitative forecasts produced by the Department and other organisations were used as a basis for the range of assumptions chosen, but the process inevitably involves a number of judgements. These have been taken on the basis of available evidence, expert opinion and make use of statistical data, where available. Further details of the assumptions referred to above and the others used are in Annex 2.

b) the specification of the statistical relationships used to produce the forecasts and the stability of these relationships over time

2.13 The specification of the statistical models on which the forecasts depend is also in many respects a matter of judgement, over such factors as the time period on which the relationships are to be estimated, explanatory variables to include and the functional form chosen to represent the relationship (i.e. exponential, logarithmic, linear etc.) In reaching decisions on the above, the statistical performance of models, analysis of past growth rates and income elasticities and evidence on market maturity were all considered.

c) the omission from the model of factors that may in future be important determinants of air traffic

2.14 A further problem with any forecasting model is that relationships based on past behaviour may not always be accurate for predicting behaviour in the future. In the case of air transport for example there may be changes in national and international policies or attitudes which will affect demand. By their nature, it is difficult, if not impossible, to predict when these might occur and what their impact might be. Demand for travel by air has tended to bounce back after one-off shocks such as the Lockerbie bombing and the Gulf War, although this may not necessarily be the case in the future. The impact on demand of the introduction of a tax on aviation fuel is discussed in section 8, on sensitivity tests. The analysis assumes that the response to such a policy would follow that which could be expected from relationships based on past behaviour in reaction to increases in air fares.

2.15 To reflect this uncertainty, the Air Traffic Forecasts are presented as a range based on 'high' and 'low' scenarios of demand. Generally, the wider the range presented, the greater the degree of confidence that actual traffic growth will lie within the forecast range, although this makes the forecasts less useful for planning any enhancements to capacity, which typically have long lead times.

2.16 The scenarios are based on different sets of assumptions about the future path of the explanatory variables which identify a span of possible outcomes which to be taken into account when planning long term capacity. The high demand scenario is based upon assumptions which tend to increase demand, and the low demand on assumptions which tend to decrease demand. The table below summarises the assumptions on which the scenarios are based.

Table 2.1 Summary of inputs for scenarios

2.17 In previous Air Traffic Forecasts, the Department has given a Demand Planning Value (DPV), as a point estimate to use for planning purposes. The DPV was two-thirds of the way up the forecast range for the short and medium term (to 2000 in the 1994 forecasts) and then the mid-point for the remainder of the forecast period. For the 1997 forecasts however, it should be the mid-point of the high and low scenarios that is used as the point estimate throughout the forecast period, bringing the approach more into line with that used by other forecasts produced by Government. In interpreting the forecasts, greater weight should be attached to the mid-point estimate than the high and low ends of the range.

2.18 The scenario approach aims to capture the uncertainties inherent in long term forecasts. In addition, a number of sensitivity tests have been conducted to test the impact on the forecasts of varying in isolation a number of key input variables. These are discussed in section 8 and include varying assumptions about market maturity, airport capacities and Channel Tunnel diversion rates. The forecasts have been prepared on the basis of no radical change in government policy but the effect of policy changes is illustrated by the aviation fuel tax sensitivity test.

Market maturity

2.19 A key issue in long term air traffic forecasts is the degree to which a particular market is 'mature', that is whether or not there will be a significant fall in annual growth rates over time. The growth in demand for products and services can be characterised into three phases: low growth following its initial introduction with limited consumer awareness and supply; followed by rapid growth as it achieves greater awareness and acceptance; and finally slower growth once the product has become established and the market approaches saturation. The phase in the above cycle will vary by market segment. In the longer term, time-constraints are also expected to limit demand for air travel, particularly for leisure purposes. Market maturity implies falling growth rates in traffic and declining income elasticities over time, but not zero growth or complete saturation.

2.20 A study commissioned by the Department in 1991 on maturity in the UK leisure market concluded that there was some evidence of market maturity but that there was still considerable potential for growth. There had been some decline in growth of the well developed mass European holiday markets, but growth had continued to be strong in newer markets such as some long-haul destinations, winter holidays and short breaks. Even though growing, these markets remain relatively underdeveloped, with relatively few people taking winter holidays or short breaks abroad. The propensity to fly, the number of trips per capita, was still relatively low in the regions compared with South East England and in low income groups compared to high.

2.21 The forecasts incorporate a degree of market maturity and hence declining growth rates for the major leisure market segments. However, within the leisure market some newer sections are likely to be experiencing constant or increasing growth rates, such as weekend city breaks for example, while others decline more quickly, for example, summer charter flights to the Mediterranean. A more modest reduction in growth rates is assumed for business segments. The impact on demand of greater or lesser maturity than in the main forecasts is examined in section 8 on sensitivity tests

Supply side factors

2.22 Evidence suggests that to date supply side improvements, that is better frequencies and increases in the number of destinations, have had only a small impact in generating extra growth in demand for air travel. As the forecasts are based on past data, the trend impact of supply side improvements should be picked up by the equations, and therefore reflected in the forecasts. However if growth at regional airports results in further improvements to frequencies and the number of destinations to which flights are available, further generation of traffic may take place, although this effect, if it does occur, is likely to be very small in relation to total demand.

2.23 Previous forecasts have included a small downward adjustment to growth rates, to reflect the potential impact of tourist infrastructure constraints, such as a lack of hotel rooms particularly in London. However, in recent years the hotel sector appears to have been able to increase capacity to meet demand. There is little other evidence to suggest that infrastructure constraints are likely to have any more than a short term impact in reducing growth rates in tourist numbers in London or elsewhere in the UK, so no adjustment has been made to the 1997 forecasts.

Air passenger demand 1992 to 1995

2.24 The total number of terminal passengers at UK airports grew strongly between 1992 and 1995 at an average annual rate of 6.9%, compared with an annual average of 5.2% for the 1980 to 1992 period. Growth was particularly strong in 1994, at 9.0%.

2.25 Of all international passenger market segments, growth was generally stronger in foreign resident market segments, and for business as opposed to leisure purposes. The UK short haul leisure market also experienced very strong growth in 1994 compared with 1993, increasing by about 13% as a result on increases in capacity and lower fares. However growth was much lower in 1995, at close to 1%, suggesting that charter companies were not able to maintain the high level of capacity and low fares seen in 1993 and 1994.

2.26 Limited Eurostar services between London, Paris and Brussels were introduced in Autumn 1994 with service frequencies gradually increased over the first half of 1995. The impact of Eurostar services on air terminal passenger numbers will therefore be reflected in observed data for 1995 and to a lesser degree 1994, in all the short haul markets. In 1995, it is estimated that approximately 1.2 million passengers who would have travelled by air diverted to Eurostar services. The likely effect of Eurostar services on air travel between London, Paris, Brussels and other destinations over the forecast period is discussed further in section 5.

2.27 Air Passenger Duty (APD) of £5 on domestic and EU departures and £10 on other international departures was introduced in 1994. The effect of APD on demand is difficult to isolate because of the large number of other factors which influence demand. The impact of a flat rate tax is likely to be greatest on domestic and short haul services where it forms a greater percentage of the fare, and in leisure markets where price sensitivity is higher than in business segments. Analysis suggests that the initial level of APD reduced demand in 1995 by less than 2%. The increases in APD announced in the 1996 Budget have been incorporated into the fares assumption for 1997 onwards. These are likely to have a similar effect on demand as the first round of APD.

III 1997 Forecasts of air traffic demand

3.1 A summary of the main traffic forecasts is presented in Table 3.1 below.

Table 3.1: Forecast terminal passenger numbers at UK airports, 1995 to 2015 (millions)

Notes: 1. On round trips each person is counted twice, departing and arriving. Passenger movements at Northern Ireland and Isle of Man airports are included. International traffic at Channel Island airports is not included. Forecasts are net of diversion from air to rail services through the Channel Tunnel.

2. Domestic passengers are counted at airports at both ends of a journey, with the exception of flights to the Channel Islands which are only counted at mainland airports.

3. CAA figures are used as the basis for the forecasts, 1995 figures are actuals

4. Totals may not sum due to rounding

3.2 Figure 3.1 shows past (actual) growth in air traffic and the Department's forecasts to 2015. The forecasts are shown for both low and high scenarios, which reflect the uncertainties inherent in producing these forecasts, as well as for the mid-point.

Figure 3.1 Actual and forecast passenger numbers at UK airports, 1975 to 2015

3.3 Air traffic at UK airports is expected to grow at an average of 4.5% per annum between 1995 and 2015 under the mid-point of the forecasts (3.2% and 5.5% in the low and high demand scenarios respectively). Total air traffic using the mid-point forecast is expected to increase by 139% over the forecast period. The average annual growth rates are very similar to those from the 1994 Air Traffic Forecasts, which were 3.1% (low demand scenario) and 5.5% (high demand scenario). However, the strong growth in the intervening period between 1992 and 1995 mean that the 1997 set start from a higher base in 1995 than would have been expected from the 1994 set. This means that the absolute number of passengers forecast in a given year is higher than that in the 1994 forecasts.

3.4 In producing the forecasts, consideration was given to adjusting the 1995 base year figures downwards to compensate for the stronger than average growth between 1992 and 1995. However such an adjustment was not made for two main reasons: it could only be on the basis of an ad hoc estimate of what trend demand was in 1995 rather than on observed data; and the GDP assumptions used already allow for the short-term effects of the economic cycle on demand. Evidence from data for the first three quarters of 1996 also shows growth to be broadly in line with that in 1995. The sensitivity of the forecasts to the level of demand in the base year is discussed further in section 8, as is the effect of reducing UK GDP growth rates up to 2005 to the long term average of 2.3% p.a.

3.5 The greater than forecast growth in demand between 1992 and 1995 was mainly due to higher than expected economic growth in the UK as the economy recovered strongly from the recession of the early 1990s. There was also a substantial level of fare discounting in the short-haul leisure markets, which boosted demand in those sectors. The forecasts of diversion to Channel Tunnel rail services in the 1994 forecasts proved too high for the period 1992 to 1995, mainly due to the services commencing later than expected, which also partly accounts for the difference between forecast and actual traffic in 1995. Estimates of diversion to Channel Tunnel rail services used in the 1997 forecasts are discussed in section 5 and Annex 1. The differences between the 1994 and 1997 forecasts are discussed further in section 9.

3.6 Figure 3.2 below plots past and forecast average annual growth rates of terminal passengers at UK airports. Average annual growth rates have generally been declining over time as the market moves towards maturity, falling from above 12% in the early 1960s, to close to 5% between 1990 and 1995. The latter figure was influenced by the reduction in traffic during the Gulf War, although this was followed by a period of stronger 'catch up' growth in 1993 and 1994. The average annual rate from the mid-point of the two scenarios of 4.5% over the forecast period continues the long term trend of gradual reduction in growth rates.

Figure 3.2: Average annual growth rates, actual and forecast, 1960 to 2015

3.7 The forecast growth rates reflect the market maturity effects in some markets i.e. the UK leisure market, and strong growth in other markets such as travel between the UK and South East Asia.

IV International air traffic

4.1 The chart below shows the composition of International Passenger Survey (IPS) international traffic in 1995 disaggregated by business and leisure trip purpose, UK and foreign residents and Short Haul (West Europe) and Long Haul (everywhere else) destinations. Interliners and miscellaneous traffic categories are not included in the IPS but are also shown to provide a complete picture.

Figure 4.1: International air traffic, 1995, by market segment (millions)

Notes:

4.2 As can be seen, the UK short haul leisure market is by far the largest segment, accounting for 35% of all international passengers. Trips for leisure purposes account for 68% of all traffic with business trips comprising 21% of the total. The majority of trips, 59%, made in 1995 were to and from Western Europe. The composition of international traffic in 1995 has changed little since 1992, the most noticeable difference being the growth of the share of interliners, from 5% to 9% of total international traffic.

4.3 The growth in interlining traffic has occurred almost entirely at Heathrow airport where airlines and the airport operator have sought to increase the number of interliners, partly in response to the excess capacity due to the reduction in 'point to point' traffic during the Gulf War. For forecasting purposes, interlining air traffic is estimated to grow at the same rate as for total foreign air traffic (business and leisure). However, the actual level of interliners in future years will depend largely on airline and airport strategies, particularly their response to capacity constraints at airports. Heathrow also faces strong competition for interliners from other European hub airports such as Amsterdam's Schipol Airport.

4.4 Table 4.1 shows the forecasts of traffic by trip length (short haul or long haul). These forecasts are for IPS traffic only and therefore exclude the interliner and miscellaneous categories. The forecasts incorporate estimates of diversion of passengers from air to rail services using the Channel Tunnel (see section 5 for details).

Table 4.1: Forecasts of international terminal passengers, 1995 to 2015 (millions)

Note: 1. The above passenger forecasts differ from those presented in Table 3.1 as they exclude the Interliner and Miscellaneous air traffic categories.

2. Totals may not sum due to rounding

4.5 IPS international traffic is expected to grow at a similar rate to total traffic, at an average annual rate of 4.5% between 1995 and 2015 under the mid-point of the two scenarios. Short haul traffic is expected to grow at 4.0% per year over the same period, considerably less than long haul traffic, which is forecast to grow at an annual rate of 5.5%. The more rapid growth for long haul traffic is mainly driven by trips to and from newly industrialised Asian countries and less developed countries, reflecting the higher rates of economic growth assumed for those market segments.

4.6 Growth rates for leisure and business traffic by UK and foreign residents are presented in Table 4.2 below. These figures clearly show the market maturity effects applied to travel for leisure purposes, as mid-point growth rates decline over the forecast period for UK and foreign residents. Growth rates for leisure purposes are higher for foreign residents than for UK residents, partly reflecting higher economic growth rates for countries outside Western Europe, and also the much larger population base.

Table 4.2: Forecast growth in international business and leisure travel by residence of passengers - average annual growth rates (%)

4.7 Business markets are not expected to mature to the same degree as leisure markets over the forecast period. It is likely however that towards the end of the forecast period there will be some decline in traffic growth rates. This has been allowed for through a small downward adjustment to business traffic growth over the forecast period.

4.8 The forecast shares of business and leisure markets for UK and foreign residents is plotted in Figure 4.2, below using the mid point of the low and high scenarios.

Figure 4.2: Forecast shares of business and leisure markets for UK and foreign residents,
1995 to 2015

4.9 The chart shows the dominant, but declining position of the UK leisure market, whose share of total IPS traffic falls from 53% in 1995 to 47% in 2015. This is due to market maturity effects, as annual average growth rates decline over time. All other segments gradually increase their share of the total over time.

V Diversion of air passengers to Channel Tunnel rail services

5.1 Forecasts of diversion of air passengers to Channel Tunnel inter-city rail services have been included in the previous two sets of Air Traffic Forecasts (those published in 1991 and 1994). Diversion forecasts are required as the relationship between air traffic and its determinants was derived from data covering the period 1960 to 1995. This means that the forecasts for short haul market segments will not reflect the impact of the introduction of rail services through the Channel Tunnel which began in late 1994. It is therefore necessary to produce 'off-model' estimates of the diversion of passengers from air to rail over the forecast period, and to deduct these from the modelled forecasts of air traffic demand.

5.2 Unlike in previous forecasts, the diversion forecasts in the 1997 forecasts are able to take account of two years experience following the commencement of Channel Tunnel rail services in late 1994. It must be noted however that services were built up slowly in terms of destinations and frequencies and it was not until May 1995 that a full service was in operation. More up to date information about future extensions to the high speed rail network is also now available. This has been used to revise the relationships used to produce diversion forecasts for the 1994 set of forecasts. Further details of how diversion forecasts were arrived at are given in Annex 1.

5.3 Table 5.1 sets out the estimates of diversion used in the 1997 forecasts. The figures are slightly lower than those forecast in the 1994 Air Traffic Forecasts. This was mainly due to a revised view of the relationship between rail journey times and rail/air market shares.

5.4 The analysis used to calculate the diversion rates applied to the 1997 forecasts indicates that, with rail journey times of 3 hours, rail travel should account for approximately 60% of the business market and 70% of the leisure market. This suggests that there will be further diversion on the London-Brussels and London-Paris markets even in the absence of further rail journey time reductions. This has been allowed for in the calculation of diversion rates in the year 2000 in addition to diversion occurring as a result of the Brussels high speed link, expected to be in operation by 2000. Diversion is expected to increase substantially in 2005 with the completion of the CTRL resulting in significantly reduced journey times on many routes.

Table 5.1: Channel Tunnel rail service diversion forecasts (millions)

Notes: 1995 figures are estimated actuals

5.5 Diversion on the regional routes is overall expected to be higher than in the 1994 forecasts, due mainly to the revised view of the relationship between rail/air market shares and rail journey times. While diversion on routes to more distant destinations is not expected to be significant, diversion on routes from the regions to destinations such as Amsterdam and Cologne is expected to be higher than previously estimated.

5.6 Actual air and rail traffic figures were used to derive an estimate of diversion in 1995 and 1996. However it must be noted that while it has been possible to make some calculations on experience to date there may well be future developments in rail/air diversion that have not been foreseen. As Channel Tunnel rail services have been only fully operational for less than two years, it is rather too soon to be confident that initial experience will provide an accurate guide to subsequent years.

VI Domestic air traffic

6.1 Domestic air traffic comprises flights between London and regional airports, between intra-regional airports and between all UK airports and the Channel Islands. Forecasts of domestic traffic are shown in Table 6.1 below.

Table 6.1: Forecast passenger numbers in the domestic market (millions)

Notes:

1. Domestic passengers are counted at the airports at both ends of the journey, with the exception of flights to the Channel Islands where passengers are only counted at mainland airports.

2. 1995 figures are actuals

3. Totals may not sum due to rounding

6.2 Domestic traffic over the forecast period is expected to grow on average by 3.7% using the mid-point of low and high demand scenarios, or by 3.2% in the low demand scenario and 4.2% in the high demand scenario. In absolute terms, traffic is forecast to slightly more than double between 1995 and 2015. This is lower than the growth in total traffic from UK airports due to market maturity assumptions. The Channel Islands market in particular has seen declining growth rates in the past, and is expected to continue to do so over the forecast period. Market maturity effects have also been incorporated for the London-regional and the intra-regional markets, although the decline in growth rates is not as strong as that for the UK-Channel Islands market.

VII Split of traffic between London area and regional airports

7.1 The split of traffic between London area and other UK airports is produced by CAA using their Passenger Allocation Model (PAM). The model takes the forecast of total traffic and allocates it to UK origins and destinations on the basis of passenger surveys. The distribution of passengers between the main airports is then determined by passenger preferences, which include the range and frequency of services at an airport, the accessibility of the airport, fare differentials and other quantitative factors. Separate relationships are calculated for long haul, short haul and domestic services. Further details of PAM are provided in the RUCATSE report.

7.2 In allocating forecast total traffic growth to UK origins and destinations, assumptions are required on whether the existing pattern of origins and destination of passengers will be replicated in the future, or whether some regions may grow more rapidly than others. The propensity to fly in the regions is still far below that in South East England, which would indicate greater potential for growth in those areas. Over the longer term, there is evidence that the gap in propensities to fly in South East England and other regions is narrowing. To reflect this, it was assumed that traffic in the regions would grow 1% p.a. faster than in the South East over the forecast period with the differential growth adjusted to ensure that overall growth is consistent with the total forecast. However, the propensity to fly to and from the South East is likely to continue to remain higher than in the regions.

7.3 Other important inputs into PAM are expected future capacities of airports. A number of elements need to be taken into account when assessing these. The total passenger capacity of an airport is determined by its runway capacity, the average number of passengers per aircraft and the number of passengers that can be handled by the terminal facilities. All of these are subject to a considerable degree of uncertainty. Comparison of previous assessments with actual passenger throughput show that at certain airports capacity has exceeded what was previously considered as a maximum, often by a significant amount. Taking the above considerations into account and using the most recently available data, small upwards adjustments have been made to some of the London area airport capacities assumed for the RUCATSE study and used in the 1994 forecasts, although no additional runways have been assumed. It should be noted that the airport capacities are assumptions for modelling purposes only. They do not represent the Department's policy and do not take into account planning considerations.

7.4 Capacity at Heathrow airport was assumed to rise to 84 million passengers per annum (mppa) by 2015, compared to 80 mppa used in RUCATSE. The revised assumption is consistent with an annual increase in passengers per ATM of 1.5% over the forecast period, reflecting the use of larger aircraft by airlines in response to the limited runway capacity at Heathrow, together with some increase in runway utilisation to 480,000 ATMs per annum.

7.5 At Gatwick, a maximum capacity of 40 mppa was assumed to have been reached by 2010, 5 mppa higher than in RUCATSE. This is consistent with an increase in passengers per ATM of 1.8% p.a. between 1995 and 2010 and increased utilisation of the runway to 240,000 ATMs per annum. Growth in passengers per ATM is greater at Gatwick than at Heathrow as the former starts from a lower absolute number of passengers per ATM. Over the forecast period smaller aircraft presently using Gatwick are expected to be progressively 'squeezed out' as the traffic mix becomes more like that at Heathrow, dominated by larger aircraft types. In absolute terms, passengers per aircraft at Gatwick are predicted to remain lower than those at Heathrow between 1995 and 2015.

7.6 Stansted was assumed to reach the same capacity as at Gatwick, but 5 years later in 2015. Capacity at Luton airport (20 mppa) and all regional airports were as in the RUCATSE study, which include a second runway at Manchester Airport.

7.7 Table 7.1 shows the expected demand at the London area and regional airports.

Table 7.1 Forecast split of traffic between London area and regional airports

Note The second row for 2015 is an estimate of underlying demand in the absence of capacity constraints. See para 7.10 for details.

7.8 The number of terminal passengers forecast for London airports in 2015 is identical in the mid-point and high demand scenarios because all London area airports are assumed to have reached capacity by that date. Under the mid-point forecasts, these capacity constraints are not reached until 2015, the very end of the forecast period. Under the high demand scenario capacity constraints are reached at all airports in the London area from around 2010 onwards. This is illustrated by the sharp drop in the share of total passengers using London area airports which falls from 57% to 49 % between 2010 and 2015. Under the low demand scenario, there is still some spare capacity at London airports in 2015. It is only in the high demand scenario that throughput of the London airports is constrained by passenger handling capacity before 2015.

7.9 In the high demand scenario, growth at regional airports is forecast to increase sharply between 2010 and 2015. This arises because of the way in which the PAM works once airports reach capacity. Passengers are assumed to redistribute amongst other UK airports still offering spare capacity whilst incurring additional surface access and other costs. Total numbers are reduced by applying price elasticities in the same way as if these additional costs were caused by air fare increases, but the net effect in reducing the total number of passengers is comparatively small. In practice, there would be pressure at London airports to introduce further capacity increments and to utilise existing capacity more intensively. Some passengers unable to use London airports may decide not to fly at all. Others might divert to competing airports on the continent. These are likely to be mainly interliners, but could also include 'point to point' passengers using Channel Tunnel rail services or other modes to access continental airports. The above factors would in practice act to reduce the extent of this overspill of London area passengers to regional airports under the high demand scenario.

7.10 As the note to table 7.1 explains, the second set of figures given for 2015 are an estimate of what demand would be in the absence of any capacity constraints. This is necessary if the forecasts are to be used to help assess the need for additional airport capacity. When capacity constraints are met the forecasts become projections of throughput with no indication of what underlying demand in the absence of such constraints would be. In these forecasts, it is only under the high demand scenario after 2010 that capacity constraints are met at the London area airports. Assuming that the London area airports share of total UK traffic under the high demand scenario declined in line with that in the two other scenarios to 56% in 2015, underlying demand would be 212 million passengers in that year, 28 million greater than under the capacity-constrained forecast. The forecasts for regional airports are reduced by an equal amount to 166 million.

7.11 To reflect the uncertainties inherent in arriving at future estimates of airport capacities and the tendency to underestimate future capacities a sensitivity test was carried out assuming increased capacity in 2015 at London area airports for the mid-point forecasts. The results are discussed in section 8.

VIII Sensitivity tests

8.1 A number of sensitivity tests were carried out on the main forecasts to identify the impact on demand of changing key input variables. The range in which the Air Traffic Forecasts are presented, reflecting 'high' and 'low' demand scenarios, covers a number of different assumptions, primarily about GDP growth. This section looks more closely at specific scenarios which alter the mid-point traffic forecasts. The results of these sensitivity tests are discussed below and summarised Table 8.1.

Adjusting the base year

8.2 To test the impact of adjusting the observed base year to a trend value, modelled growth rates were applied to a provisional estimate of 1996 outturn for the UK short haul leisure market. This approach aims to remove the effect of any short term factors outside those included in the forecasting equations which may inflate or depress the base year data and thus affect the forecasts over the entire period. The UK short haul leisure market was chosen due to its volatility in terms of average annual growth rates. It is also the largest market segment. The strong growth seen between 1992 and 1994, thought to be due to substantial increases in airline capacity and lower fares for charter flights, might suggest that the segment was above trend in 1995. On the other hand, growth in 1995 was well below average and initial estimates for 1996 show negative growth, indicating that traffic levels in 1995 might be reasonably 'on trend'.

8.3 To reflect this uncertainty, the impact of reducing the volume of charter traffic in the base year was considered by using the provisional 1996 outturn as the base year for the UK short haul leisure market as opposed to observed demand in 1995. This reduced total terminal passengers by 4.6 million in 2015 under the mid point forecasts, a reduction of 1.5%, well within the range of the high and low demand forecasts.

Lowering medium term trend GDP growth

8.4 The UK economic growth assumptions for the years 1996 to 2005 used in the forecasts are based on a view about the position of the economy in the cycle and other short term factors which influence growth. Economic growth is expected to be above its long run annual average trend throughout this period. From 2006 onwards, growth is assumed to return to its long run trend (see Annex 2 for details).

8.5 The sensitivity test assumed that the long run growth rate applied throughout the forecast period. This had the effect of reducing the average annual growth rate using the mid point forecasts from 4.5% to 4.2%. Total UK traffic in 2015 is reduced from 310 million passengers to 296 million passengers using the mid-point forecasts, a reduction of about 4.5% or approximately one years growth.

Introduction of aviation fuel tax

8.6 Aviation fuel is currently exempt from taxation because of international agreements under the Chicago Convention not to tax fuel used for international air travel. However, the possibility of removing that exemption has recently been discussed in a number of fora in response to concerns about the environmental impact of air travel, in particular the contribution of emissions of carbon dioxide and oxides of nitrogen from aircraft to global warming.

8.7 The sensitivity test assumed that such a tax was introduced in 2000 and had the effect of raising the projected price of fuel in that year by 50%. Fuel prices then increased in line with the assumptions used in the main forecasts until the end of the forecast period in 2015.

8.8 A number of simplifying assumptions were necessary:

1. the fuel tax was introduced globally, in a way which did not affect the existing fuel price differentials between countries, thereby eliminating the scope for 'leakage' through increased fuel tankering.
2. all the increase in fuel prices were passed through to fares. In practice airlines might absorb some of the increase through lower margins, or increase business fares more than leisure fares due to the generally lower fare elasticities for business passengers.

8.9 The impact of the fuel tax on passenger demand is not substantial, with fuel prices representing between approximately 8% to 13% of airline costs, a 50% rise in fuel prices would be expected to lead to an increase in fares of between 4% and 7%. If the full extent of the rise in costs is passed onto passengers in higher air fares and the own price elasticity of demand for air travel is assumed to be -0.8 for all markets, then the resulting decrease in demand will be between 3% and 6%, approximately one year's growth. In the year 2000 this would account for approximately 5 to 10 million passengers. Using the same growth rate that occurs under the central case this leads to a mid-point value of 296.1 million passengers in 2015, a reduction of 4.5%. These results can be scaled up or down to reflect alternative assumptions about the rate of any tax on aviation fuel.

Increases in airport charges

8.10 The impact on demand of higher airport charges or introducing other taxes which affect airline costs was also considered. Airport charges represent a slightly greater proportion of airline costs than aviation fuel, between 10% and 18%. Assuming airport charges were increased by 25% which was all passed on to pasengers through higher air fares and a price elasticity of -0.8, demand would be reduced by 2.0% to 3.5%, between approximately 6 and 9 months growth under the mid-point forecasts.

Market maturity

8.11 As discussed in Section 2, the forecasts incorporate the concept of market maturity. A sensitivity test was carried out to examine the impact on demand of greater or lesser market maturity than assumed in the main forecasts. For greater market maturity, forecast growth rates were adjusted by a factor to increase their rate of decline over time, depending on how quickly traffic was expected to grow. For a slower move to market maturity, growth rates from the first 10 years of the forecast period were applied throughout the entire period 1995 to 2015.

8.12 The results show the significance of market maturity effects on the demand for air travel. By 2015 mid-point growth under the sensitivity test was 12% below or above that in the base case, with demand at 273 million for increased market maturity or 347 million for a slower move to maturity. Average annual growth rates in the sensitivity test were 3.8% and 5.0% respectively, compared to 4.5% in the main forecasts.

Channel Tunnel diversion

8.13 The main inputs required to produce forecasts of diversion of passengers from air to rail services using the Channel Tunnel are information about the impact on journey times from future improvements to the high speed rail network and estimates of market shares for business and leisure passengers for a given rail journey time. Of these two inputs, the estimates of market share are subject to the most uncertainty. The sensitivity test varied the market share estimates, by using the relationship derived for leisure passengers for business passengers as well. Leisure passengers are assumed to be significantly more likely to divert than business passengers in the main forecasts, so the sensitivity test had the effect of increasing the amount of diversion. This might occur if, for example, airlines were unable to keep fares broadly comparable to those on rail services.

8.14 In fact, the effect on numbers diverting was not great and the impact on total demand almost negligible. Even arbitrarily doubling the diversion forecasts in table 5.1 would reduce total demand by only 3% in 2015 relative to the base case, less than a year's forecast growth. The impact is limited because diversion is likely to occur on a relatively small number of short haul routes served by UK airports. Halving the level of diversion in the forecasts would increase traffic under the mid point scenario in 2015 by 1.5%

Capacity assumptions

8.15 The effect of increasing London area airports capacity on the London- regional split of traffic under the mid-point forecasts was considered. Increasing capacity in 2015 did not significantly alter the amount of traffic using London's airports because the mid-point traffic forecast indicate that capacity constraints at London area airports are not reached until 2015. The main effect of assuming higher capacity would be to delay the date at which capacity constraints at London's airports bite. For example, an additional 10 mppa of capacity would provide for slightly less than 2 years further growth before constraints were reached, using the average mid-point growth rates between 2010 and 2015.

Summary of sensitivity tests

8.16 Table 8.1 below lists the results of the sensitivity tests on the mid point air traffic forecast for 2015. Where the sensitivity tests reduce traffic in a given year, such as 2000 in the case of a fuel tax, this figure has been increased at the same rate as the mid-point traffic forecast to obtain the 2015 value.

Table 8.1: Results of sensitivity tests.

IX Performance of past forecasts

9.1 Figure 9.1 and Figure 9.2 compare the performance of the former DOT forecasts produced in 1986, 1988, and 1991 with actual traffic for all UK airports and London area airports respectively. As mentioned in section 2 above, the forecasts are essentially long term forecasts of the trend in growth. Actual traffic may well deviate from them in any particular year due to the uncertainties in making accurate assumptions about short term economic growth and other factors necessary to produce forecasts. This needs to be borne in mind when making comparisons of actual and forecast growth over comparatively short periods of time.

Figure 9.1: Performance of 1986, 1988 and 1991 forecasts at all UK airports.

9.2 The past forecasts of total demand have performed reasonably well, with actual traffic usually falling within the specified forecast range. Of those considered in this section, the 1986 set appear to underestimate growth, particularly for total demand, with actual traffic either above or close to the top of the range of forecasts.

9.3 The sharp short-term fall in demand for air travel in the 1991 arising from a combination of recession and the Gulf conflict resulted in actual traffic below the range for the 1988 and 1991 forecasts as might be expected, although it was within the range for the 1986 forecasts. The impact of the Gulf war illustrates the problems in producing forecasts due to the wide range of uncertainties about the impact of unpredictable factors which influence future demand.

9.4 The forecasts for demand at London area airports appear to follow a similar pattern to those for total demand, although the 1986 forecasts look more accurate than those for total demand and appear to be less prone to underestimating demand.

Figure 9.2 : Performance of 1986, 1988 and 1991 forecasts at London area airports.

Comparison of 1991, 1994 and 1997 Air Traffic Forecasts.

9.5 As discussed in section 2, total air traffic in the 1997 forecasts is expected to be higher than in the 1994 set. Table 9.1 compares mid-point estimates for 1995, 2000 and 2010 for London area airports and total demand from forecasts published in 1991. 1994 and 1997. As noted in section 3, average annual growth rates between the two most recent sets of forecasts are similar, although the level of demand forecast in any particular year is higher in the 1997 forecasts than in the 1994 set.

Table 9.1: Comparison of 1991, 1994 and 1997 Air Traffic Forecasts, mid-points (millions).

Notes: 1995 figures are actuals for 1997 column, and forecasts for 1991 and 1994 columns

9.6 The 1997 forecasts are close to those produced by the Department in 1991 in terms of the mid point demand predicted for the year 2005, which was 208 million (the end of the forecast period for the 1991 forecasts). The annual average growth rates over the entire 1991 forecast period was 4.8%, above those from the 1997 and 1994 sets, of 4.5% and 4.4% respectively.

X Comparison with other forecasts

10.1 The Department's forecasts of total UK demand can be compared with similar forecasts produced by the CAA and for the RUCATSE study, while the London area airports forecast from the London-regional split are also comparable with those produced by British Airways and BAA. For ease of comparison, single 'mid-point' estimates of forecasts are presented below.

10.2 Table 10.1 compares the mid point from the DETR 1997 forecasts for all UK airports with the most recent set produced by the CAA and the forecasts used in the RUCATSE study published in 1993. The 1997 DETR forecasts are rather higher than the CAA and RUCATSE forecasts. Part of the difference is due to the dates at which these forecasts were made and the corresponding different base years of air traffic.

Table 10.1: Comparison of DETR, CAA and RUCATSE forecasts for all UK airports (millions).

Notes: 1995 figures are 1995 actuals for DETR, 1994 actuals for CAA and 1995 forecasts for RUCATSE

10.3 Table 10.2 gives forecasts for London area airports, namely Heathrow, Gatwick, Stansted and Luton. It should be recognised that these forecasts differ in the extent to which they are constrained by capacity limitations at the airports and in the dates they were produced, making direct comparisons difficult. For example, in calculating the split of total UK air traffic into London and regional traffic the Department uses capacity assumptions slightly higher for Heathrow and Gatwick airports than those used in RUCATSE and which differ from those assumptions used by other organisations.

10.4 The CAA forecasts, like those of the Department, are based on unconstrained demand at all UK airports and assume traffic is constrained by London area airport capacities while those of BAA and BA assume no changes in the effect of airport constraints as compared to the base year.

10.5 Notwithstanding the difficulties in making direct comparisons, the forecasts give quite a wide range. The DETR forecasts (again the mid-point) are consistently the highest with those from CAA the lowest. As noted above, the Department's latest forecasts use higher base year air traffic than the forecasts of other organisations which inevitably affects the level of forecast traffic. Sensitivity analysis carried out on historical air traffic data indicates that traffic in 1995 was more in line with long term trend than was traffic in 1992 or 1993 when the 'catch-up' period of air traffic growth following low economic growth and the Gulf War was not complete.

Table 10.2: Comparison of forecasts of passenger numbers for London area airports (millions)

Notes: 1995 figures are 1995 actuals for DETR, 1994 actuals for CAA, 1995 forecasts for RUCATSE and not available for BAA and BA.

Annex 1: Methodology

Annex 1: Methodology (PDF, 64KB)

Annex 2: Assumptions

A2.1 Assumptions about future economic growth, trade volumes, real exchange rates and air fares are required to produce the Air Traffic Forecasts. A range of assumptions are presented for each variable, reflecting uncertainties involved in looking up to 20 years ahead. The ranges are expressed as 'high' and 'low' relating to the 'high demand' and 'low demand' scenarios of air traffic in which they are used, as discussed in section 2 of the main report, and set out in table 2.1.

Economic growth

A2.2 The central assumptions for UK economic growth are based, to the year 2000, on the forecasts in the 1996 Financial Statement and Budget Report (FSBR). The economy is expected to grow at rates above trend between 1996 and 2000 as the gap between actual and trend output close. Between 2000 and 2005 growth has been assumed to be slightly above its historic trend, due to increasing participation rates and the effects of supply side reforms on the labour market.. From 2005 to 2015, growth is assumed to revert to longer term trend of 2.3% p.a. UK consumer expenditure is expected to grow in line with GDP over the medium to long term, but is expected to grow more quickly in the short term, as savings decline from their current high level.

A2.3 Assumptions for economic growth in Western Europe, Other OECD countries, Newly Industrial Asian Countries and Less Developed Countries are based on the May 1996 IMF 'World Economic Outlook'. Growth assumptions for Newly Industrial Asian Countries are significantly greater than those for used for non-OECD countries in the 1994 forecasts.

A2.4 The high and low ranges are constructed around the central assumptions. Ranges are wider in the short term than in the medium and longer term reflecting greater uncertainty about the position of the economy in the cycle. Longer term trend growth rate can be predicted with more certainty, and hence the ranges progressively narrow into the future. This is a departure from the 1994 forecasts in which GDP ranges were assumed to be constant throughout the forecast period. Ranges around the central assumption are 0.6 percentage points between 1996 and 2000, 0.4 percentage point between 2001 and 2005 and 0.2 percentage points from 2006 to 2015.

Table A2.1: Real GDP growth assumptions (% change p.a.)

Trade volumes

A2.5 The growth rates used for trade volu