You are on page 1of 96

Saver Fares: Differentiation and Potential Deregulation Final Report DfT Rail July 2007

Prepared by: ................................................ Approved by: Jane Cornthwaite Brian Vaughan John Dodgson (NERA)

................................................. Simon Temple

Saver Fares: Differentiation and Potential Deregulation Rev No 1 Comments Date

Marlborough House, Upper Marlborough Road, St Albans, Hertfordshire, AL1 3UT Telephone: 020 8784 5784 Fax: 020 8784 5700 Website: http://www.fabermaunsell.com Job No 52988T Reference Date Created July 2007

This contains confidential and commercially sensitive information, which shall not be disclosed to third parties.
f:\projects\t52988 saver fares\report\19 oct saver report final 4 (5.2.7 graph title added).doc

Table of Contents

1

Introduction ..................................................................................................................... 2 1.1 Background........................................................................................................... 2 1.2 Objectives of the study ......................................................................................... 2 1.3 Policy context........................................................................................................ 3 1.4 Structure of the Report ......................................................................................... 5 Review of existing evidence .......................................................................................... 7 2.1 Introduction ........................................................................................................... 7 2.2 Research on Fares Regulation ............................................................................. 7 2.3 TOC Submissions Regarding Saver Deregulation ............................................... 9 2.4 Evidence on Rail Fares Elasticities..................................................................... 10 2.5 Transport Select Committee Report and Government Response...................... 16 2.6 Evidence from Competition Authorities............................................................... 18 2.7 Competition Authority Views on Fare Levels...................................................... 22 2.8 Conclusions from the Review of Existing Evidence............................................ 23 New Research................................................................................................................ 26 3.1 Survey Objectives ............................................................................................... 26 3.2 Survey Methodology ........................................................................................... 26 3.3 Choice of Routes and Trains .............................................................................. 27 3.4 Questionnaire Design ......................................................................................... 27 3.5 Response Rates ................................................................................................. 29 3.6 Survey Conclusions ............................................................................................ 30 Findings From Questionnaire Responses.................................................................. 32 4.1 Introduction ......................................................................................................... 32 4.2 Number of Saver users....................................................................................... 32 4.3 Respondent income and socio-economic information........................................ 33 4.4 Journey Planning and flexibility .......................................................................... 38 4.5 Access to Sales Channels .................................................................................. 41 4.6 Conclusions from questionnaire responses........................................................ 43 Modelling........................................................................................................................ 46 5.1 Stated Preference Results and Robustness....................................................... 46 5.2 Development of policy application model ........................................................... 58 5.3 Deriving ‘ideal departure time’ ............................................................................ 64 5.4 Elasticity Values and Diversion Factors between Ticket Types ......................... 69 5.5 Overall Functionality of Policy Model.................................................................. 70 5.6 Policy Tests......................................................................................................... 73 Pulling the Evidence Together..................................................................................... 83 6.1 Introduction ......................................................................................................... 83 6.2 Assessment of Market Power ............................................................................. 83 6.3 Implications for Competition Authorities if fare levels increase .......................... 84 6.4 Demand Management ........................................................................................ 85 6.5 Social Inclusion ................................................................................................... 86 6.6 Network Benefits................................................................................................. 86 6.7 Conclusions from the Evidence .......................................................................... 87

2

3

4

5

6

Appendix A - List of trains surveyed....................................................................................... 89 Appendix B – Example questionnaire ..................................................................................... 93

Introduction .1.

DfT Rail has commissioned this study to look again specifically at the Saver fare market and to provide new research to inform decisions about saver fare regulation policy.3 1.1.4 The DfT now has responsibilities for fares policy for the rail industry. DfT Rail have commissioned this study to look at the potential impact of Saver Fares Differentiation and Potential Deregulation.1 To consider the markets within which Saver fares operate. and 1. are captive ie have no realistic travel alternatives.1 Background The nature of how passengers buy and choose rail fares. In this light. Now 64% of travellers in off-peak periods are buying their tickets in advance. subject to any proposal demonstrating benefits for passengers and taxpayers. The study aimed to assess the evidence on whether operators on routes where Savers are available have significant market power and what passengers. The findings of their work are presented in this report. Faber Maunsell. Indeed crowding has sometimes been higher in the ‘offpeak’ than the ‘peak’ due to the distortions introduced by Saver regulation. Undertaken new primary research of the Saver market. did not result in Saver deregulation at the time. It may be possible to introduce a new regime earlier than this. As part of a wider strategic assessment of the direction of the rail industry in the years to come. Train operators generally face a competitive market for this type of travel because passengers can normally choose alternative forms of travel if operators do not offer attractive and affordable fares.2.1. with a view to replacing the existing regime by 2006 with one more suited to passenger needs.2 The study was designed to provide quantitative estimates of the extent to which fares would or would not rise and the consequences to passengers of a relaxation of Saver regulation. internet rail purchase was almost non-existent. and To consider the consequences for consumers of relaxation of regulation 1. working in conjunction with the Department for Transport (DfT) have been pressing for a simpler fare structure.1. ‘The regulation of Saver fares (off-peak for leisure travel of about 50 miles and over) has constrained new and innovative.1 1.1.2. Objectives of the study The high level objectives of the study have been as follows: 1. and the DfT are working with the industry delivering this for passengers.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 2 1 Introduction 1. 2. if any.’ 1. The SRA’s Fare Review Conclusions. At the same time. The SRA will be working with the relevant train operators to develop proposals to address these problems. and there is every indication to assume this trend will continue.2. Only a few years ago. A consequence of both these development is that passengers will have more readily available and clearer information about the different fares available for the journeys they want to make. in conjunction with NERA Consulting. Reviewed and assessed existing evidence and research. but comments contained with the report give a clear indication that the arguments for considering saver Fare deregulation are strong. customer-focused tickets and has led to significant overcrowding on some routes. published in 2003.1. undertook the study between February and May 2007. we have undertaken three main tasks: 1. To achieve this. Saver fares deregulation has been on the industry agenda for some while. ATOC.2 1.2 1.3 . particularly for longer distance travel is changing.5 1.

00 Unlike Open tickets. The Existing Situation 1. fare regulation is in place to : Protect consumers in situations where service providers have significant market power and can use this to raise prices significantly. or in advance. it is worth reminding ourselves of why regulation can be beneficial.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 3 3. Saver tickets are placed in a fares basket. along with other regulated tickets such as Standard Class Seasons. because Saver fare de-regulation is unlikely to be acceptable unless it can be shown that the operation of the market will result in satisfactory outcomes in relation to them. at any time of the day. The time restrictions vary from TOC to TOC.00 £117. but typically passengers are not able to use a Saver ticket on arrivals (and sometimes departures from) London in the morning peak.00 £59.00 £78.00 £125.00 £41.3. The restrictions that TOCs are able to put on tickets are currently regulated and currently only trains out of London may have evening peak restrictions placed on them. Developed a policy testing model using information from both tasks 1 and 2.50 from £25.00 from £20.3.00 £37.3. The total weighted increase in the fare basket is restricted (currently to RPI +1% per annum) and there are also caps on how much any individual fare within the basket can be raised. Saver fares are always cheaper than the equivalent Open ticket fare – usually considerably so.1 Illustrative Fares Flow London to Manchester London to Birmingham New St London to Bristol Newcastle to Edinburgh 1.3 1. 1. which can be used on any train.1 Policy context Why Regulate? Before considering changes to a regulatory system.70 from £20.3 Cost of Return Journey (£) Open Saver APEX £219. the extent to which Train Operators can raise prices is limited. As they are currently regulated.00 £49. Saver tickets have time restrictions placed on them. In the rail industry. The table below provides a summary of current Saver restrictions for the long distance TOCs. They can be purchased immediately before travel. These points are important to bear in mind. Ensure affordable tickets are retained in situations where the use of more than one operator’s services is required or where the constraint of being able to only use one operator’s services unduly restricts travel opportunities. Promote social inclusion. The following table show some selected Open. or on departures from London in the evening. in so far as high rail prices could restrict the mobility of some sections of society.2 Savers fares are available for sale for most longer distance point to point journey combinations on the rail network. to provide insight into the implications of selected policy tests.50 from £18. Saver and APEX fares for selected flows: Table 1. .

These travellers would need to make an additional trip to a station or travel agent to purchase their ticket. First.3.4 It is often possible for passengers to buy a cheaper ticket than a Saver fare.00.7 This study is primarily concerned with assessing the potential impact of changing (or removing) Saver fare regulation. it is likely that fare setting will move towards singleleg pricing in the near future. Advance purchase (APEX) tickets are considerably cheaper than Saver fares. an administration charge is made and allowances are made if the train booked did not actually run.5 1. This makes them a product of choice for passengers who: 1.3. but they limit a passenger to a particular train. they can be bought on the day of travel (as well as in advance). Second.00 on the day before travel. the cheapest fare available will then be the ‘available on the day’ Saver. they are generally charged the difference between the fare they paid and the appropriate fare for the journey being made.3. Have been unable to purchase a cheaper APEX ticket in advance (because APEX quotas have been sold out). However. 3. Second. Train operators offer a quota system.6 If a passenger is travelling on a train for which their ticket is not valid (either on an APEX ticket for the wrong train or a Saver ticket in restricted periods). 4. although the cheapest fares are often not available by then. Do not know in advance that they will be making the journey. At the moment APEX reservations can be taken up to 18.2 Saver Restrictions Operator Virgin Trains West Coast Transpennine First Great Western Virgin Cross Country GNER Restrictions Euston departures before 09:15 or between 15:15 and 18:11. Firstly. 2. For APEX tickets. Are unable or unwilling to ensure they can plan their journey to catch a particular train. 1. . Currently a Saver Single is priced at £1 less than the equivalent Saver Return. arrivals to Paddington before 10:40 Restrictions into and out of Birmingham in the morning peak Departures from Kings Cross before 09:00 or after 14:30 1. which allocates a certain number of tickets at each advance purchase fare. When these tickets have been sold. Exceptions for very long distance journeys Journeys that reach their destination prior to 09:30 Departures from Paddington before 09:00 and between 16:46 and 18:59. Euston arrivals before 11. ticket simplification could have an important impact on making it easier for passengers to choose the best ticket for them. Saver tickets have two main advantages over APEX tickets. Current Policy Considerations 1.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 4 Table 1.3. there are other important changes that the DfT is considering. Single leg pricing would mean that single fares are always priced at approximately half the return fare. Are unwilling or unable to purchase in advance by telephone or Internet. and they must be booked in advance. they can be used for any train which is not in the Saver fare restricted period.

Increasingly sophisticated yield management systems. It covers the responses received from TOCs on their fares proposals if Saver fares were deregulated.3 1. and in also. This provides some helpful insight.9 1.1 Structure of the Report Chapter 2 presents existing evidence relevant to the research.4 . Chapter 3 describes the new research that has been undertaken as part of this study through an on-board passenger survey. this chapter assesses the way that competition authorities have assessed the strength of competition in rail industry markets. In summary the following influences will be in operation on the longer distance rail market: Fares simplification. Chapter 4 presents finding from the questionnaire responses. and Increasing availability and use of internet sales. Chapter 6 pulls all the evidence together and reviews the conclusions that can be drawn from the policy tests undertaken using the application model. the DfT are interested in gaining information on these impacts.2 1.3.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 5 1. 1. passengers are generally likely to buy a return.4. 1. The same TOCs have expressed the opinion that the Saver fare is fundamental to their income stream. and the type of tickets they choose. To maintain ‘revenue neutrality’ in a single leg pricing scenario. Single leg pricing.3.4.4.4 1. Within this context. Possible Saver fare deregulation. With single leg pricing we would expect some passengers to choose combinations of different single ticket types for the outward and return legs of their journeys. Chapter 5 also provides results from the application of the model to test a number of policy options. together with inferences and conclusions that we can draw from this research. It also considers evidence from the Strategic Fares model. This includes a review of existing fares regulation.10 1.4. including material in the PDFH and more recent work from MVA. specifically assessing the impact on captive and/or vulnerable people. and Saver Single fares would decrease a lot.8 The DfT has been looking at the likely impact of single leg ticketing. The DfT has asked TOCs for their proposals on how they would change fares and restrictions if Savers were to be deregulated. It examines existing evidence on fare elasticity work. Finally. At the moment. TOCs would prefer passengers to buy their tickets in advance because it helps them to plan and manage to make sure that capacity and resources are used as efficiently as possible. 1.3. while allowing them the best opportunity to yield maximum revenue. Saver Return fares would increase a little. because of very small difference in price between a single and return. within the existing fare regulation.11 All of these factors will impact (to a greater or lesser extent) on the number of passengers choosing to travel.3. Chapter 5 explains in more detail the stated preference section of the survey and the use of the results from the stated preference responses to develop a policy application model.

Review of existing evidence .2.

Bristol.1 2. including stated preference. gender.2. and a study of the effect of competition from other modes on rail fare elasticities. Journey purpose. It included interviews with TOCs and a major market research exercise. which was used to test policy options. It resulted in around 1.1 Research on Fares Regulation ITS Leeds Study This major piece of research by ITS Leeds is very relevant to the current study on Saver regulation.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 7 2 Review of existing evidence 2. First Great Western . Where tickets were bought (station. Section 2. The respondents’ “desired” departure time.2 2. It covered the following long distance flows and operators: 2. Market Research The market research was conducted in August and September 2004. and Respondent characteristics (age.2. including inter-relationships between ticket types. though it was conducted prior to establishing the principle of single leg pricing and so primarily deals with return fares. and on TOC revenues. 2. values used in the Strategic Fares Model.2 GNER .1 Introduction This chapter of the report provides a survey of relevant previous evidence for the present study of the impact of Saver deregulation. Section 2. . trip frequency).2. and Virgin and Silverlink .7 sets out the competition authorities views on fares. Darlington and Durham to London.3 The research allowed ITS to compile data on.Birmingham.1.2 we review two previous studies of the impact of fares regulation. by ITS Leeds and by NERA.5 summarises the Transport Select Committee report on rail fares and the DfT response. the actual research also concerned short distance trips. The number of days in advance the respondent knew that they planned to travel. on the demand for rail travel on other ticket types.200 completed questionnaires for standard class. Bath and Chippenham to London. In Section 2. we concentrate on the work conducted with respect to long distance trips as these are relevant to Saver fares. amongst other things: Ticket type (and whether a Railcard was used). In Section 2. not single fares. Coventry. Section 2.Peterborough to London. In particular we are interested to know what current evidence can tell us about the impact of changes in Saver fares on the demand for Saver tickets. on passengers’ choice of travel time. Central to the work was development of a model. and Section 2. income.4 reviews relevant evidence on fares elasticities. We consider in particular evidence from the latest version of PDFH.6 we review evidence from competition authority investigations. in particular with whether these investigations shed any light on the question of whether leisure rail fares are constrained by competition from other modes. internet.Newcastle. long distance travel. 2. etc). Rugby and Milton Keynes to London. This chapter is structured as follows: In Section 2. Birmingham International. GNER and WAGN .8 provides overall conclusions in regard to the most relevant evidence that can be used in the present study. In this review. as well as further responses for short distance trips. recent work by MVA.3 we summarise TOC submissions to DfT on the potential effects of deregulation of Saver fares.

The appropriate elasticity was taken from Wardman and Shires’ meta analysis of British data presented in Balcombe et al (2004). Each attribute had between two and four levels.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 8 2.demandforpublictransport. ticket type.uk.co.2. traffic between London and Leeds. Total fare deregulation at current service levels. and Apex had a choice of departure times. The model operated at two levels: at the lower level. . 5 and 10 per cent in real terms. Notably: Fare levels and return journey times were specified. Saver and Apex ticket types for both standard and first class travel.2. a sample of one thousand individuals was simulated. individuals chose between rail travel and not rail travel.2. class of travel and journey purpose. The logit model was then used to allocate these individuals to ticket types. Travel time restrictions were modelled as the divergence of the actual travel time permitted from the preferred time for the passenger (in many cases this would be zero). Fare basket regulation of standard open and standard Saver fares. The lower level of the model was calibrated on the basis of the stated preference research. Each scenario had a choice of between two and four fares. to investigate the potential of various policy options: 2. Class of travel. where the weighted average of fares within the fares basket was permitted to rise by 2. Saver had an earliest departure time. and is superseded by the latest version of PDFH.6 For each journey purpose.5 ITS developed a logit model for this work.2. the weighted average of fares within the fare basket was permitted to rise by 2 per cent. At the upper level. so is not reviewed here. individuals choose between six ticket types: Open. resulting in 44 questionnaire designs (some of which concerned short distance journeys). In-train journey time (presumably for flows served by two TOCs). and Separate model parameters were used for each journey purpose.1 Policy Testing ITS used a long distance case study. each with specific desired departure times for outward and return legs. 1 The Demand for Public Transport: A Practical Guide. ITS estimated the impact on fares – where TOCs were assumed to set fares to maximise revenue.7 Fare basket regulation of all standard class fares. and Welfare maximisation holding revenues and costs constant. subject to the constraints. and Add-ons such as free parking.4 The stated preference research was tailored to be realistic for the flows and trains being surveyed. Fare basket regulation of standard open fares. and for short and long journeys. Model Design 2. This guide mostly focuses on short distance public transport trips. where the fare types were a subset of Open. Each ticket type had between three and five stated attributes that were a subset of the following: Fare level. For each of these options and scenarios. Departure time restrictions on outward and return leg: Open was departure at any time. 2004. 2.2. three scenarios were tested. Saver and Apex for both standard and first class. TRL Report TRL593. The upper level calibration (rail versus not-rail) used base demand data (from LENNON). Each stated preference scenario allowed respondents to select their preferred (return) fare type. but they had to be selected in advance. www. but individual fares could rise by no more than 6 per cent (which is akin to current regulatory practice). It modelled Apex demand according to the varying number of days required for advance booking. sampled from MOIRA. Number of days in advance that the Apex ticket had to be purchased. Fare basket regulation of Saver fares (this is current practice).8 For each of these options. 2. Under an additional scenario.

the model forecast that the standard open fare would increase. As these changes could occur in any case under current regulatory constraints. at least in the short term. increased in real terms by around 2. including commuter fares regulation.2. 2.9 For all options and scenarios tested. Savers increased close to the maximum amount permitted. such as cheap day and supersaver tickets. the report investigated the price changes in Saver fares from 1995/96 to 2000/01. and they responded to this jointly through the Association of Train Operating Companies (ATOC). interaction with the London Travelcard and other multimodal tickets.3. Single Leg Pricing 2.2 Most long distance operators accepted the principle of single leg pricing (where single fares are set at half the return fare). Most long distance operators responded in detail. In addition. It found that.3.7 per cent a year from 19952001. revenue.2.1 2. Note that this study did not look at time -of -day fare differentials. except those under going franchise replacement. and welfare (= revenue plus consumer surplus). As part of this work.Rail Fares Policy: an Evaluation2 2. the model results appear to be counter intuitive. overall.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 9 demand. TOCs answered individually. and did not necessarily accept it.3 2. . A key component of the fares simplification proposal is that long distance single fares would be set at half the return fare. 2.2. This has important ramifications for Saver deregulation. In contrast to the constant real price of Savers. and that this should inform their responses. TOCs with high levels of non-London flows were most concerned with this proposal. use of fares baskets and protected fares. The extent of such revenue abstraction must be a concern to TOCs and others in the industry. particularly for long distance flows for which advance purchase alternatives are available. For example. NERA suggested that Saver fares have similar demand characteristics to these fares so that if Savers had not been regulated they may have risen by similar amounts. anxious to avoid accusations of collusion. NERA . TOC Submissions Regarding Saver Deregulation Background In 2006.3 2 NERA (2002). but the Saver fare would fall in price.11 2. There was some reluctance to accept single leg pricing for medium distance flows where advance purchase is not available. consumer surplus. First class fares were set to be constant because ITS considered the model elasticity with respect to these fares to be spurious. the price of pre-book and promotional tickets fell in real terms. as would the standard Business Saver. Several TOCs noted that single leg pricing could result in significant revenue abstraction. a proposal we refer to as single leg pricing. a typical business day trip currently undertaken using a full open return could be replaced by a full single on the outward leg (in the morning peak) and a Saver single on the return leg (there are fewer Saver restrictions in the evening peak). TOCs were asked to consider DfT’s proposals for fares simplification.3. and adjustment of fares on the basis of train performance. DfT instructed Train Operating Companies (TOCs) to assume that Saver deregulation would only be permitted on the understanding that it was revenue neutral. and produced some analysis supporting this. the Department for Transport asked train operating companies (TOCs) to make confidential submissions setting out how they might respond to Saver deregulation. NERA noted that other walk-on reduced fares that were unregulated. It considered many aspects of fares policy. prepared for the Strategic Rail Authority.10 This study by NERA consisted of an overall assessment of rail fares policy in Great Britain from the time of rail privatisation to the time the study was prepared (2001/02). relative to current prices. Separately.

which tests the effect of competition from other modes on fare elasticities – an issue of particular relevance to this study.4 2. and there remained some flexibility to increase restrictions on use. Thus an elasticity of -1 means that a 10 per cent increase in fare will result in a 10 per cent fall in demand. Fares are said to be inelastic if their elasticities are smaller than -1. A number of TOCs. notably day return fares and season tickets. an increase in the fare should result in an increase in revenue. This included a more expansive definition of the evening peak (where only full fares would be permitted) to include services leaving stations other than London.3. Own elasticities are the ratio of demand and price for the fare concerned. Own elasticities are greater (in absolute terms) than equivalent conditional elasticities since generally.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 10 Saver Deregulation 2.4. Whilst several TOCs argued that there was limited scope to increase prices. We also review here recent work by MVA for the Department for Transport examining the validity of PDFH recommendations. however.2 2. set below maximum permitted levels.1. June 2005.4.3 2. Proposals to introduce advance purchase fares (at higher levels than currently) for some peak trains. The principal source on rail fares elasticities is the Passenger Demand Forecasting Handbook. as well as summer Saturdays on some routes. particularly those with non-London flows.permitted at any time. These proposals are quite difficult to compare with the status quo because they tend to represent large reductions on current single fares. and on a number of Sunday services. and other simplification policies (in line with ATOC’s submission on fares simplification). These included TOCs. the more alternatives that are available. Another TOC suggested making a limited number of services reservation only.4. if a fare is elastic.permitted at any time outside the main morning and evening peaks. They noted that Savers were.4 TOCs overall welcomed the principle of Saver deregulation. to manage crowding more effectively. proposed to have three tiers of walk-on fare for many of their flows. and Super off peak . . which are the ratio of percentage demand for the specific fare 3 2. These would be: Any time . who argued that Saver deregulation would in practice make little difference to their fares structure. Passenger Demand Forecasting Handbook3 PDFH makes recommendations for fares elasticities to be used in analysis of rail demand. There are different types of fare elasticity: conditional fare elasticities are those that reflect changes in demand if all other fares (usually other rail fares. in such circumstances. acceptance of railcard discounts for advance purchase fares.3. in some cases. a reduction in its price should result in an increase in revenue. Concern that Saver fare deregulation would result in less flexibility to increase inelastic regulated fares that share the same tariff baskets. but possibly other public transport fares) change by the same proportion. including the peak. TOCs also mentioned a number of other miscellaneous issues including: 2. And we refer to a technical note prepared by Institute for Transport Studies.6 Willingness to introduce single leg pricing for advance purchase fares.5 TOCs went into details as to their preliminary thoughts on how the restrictions on use should change.4 Version 4. 2. The shoulder peak fare would be the cheapest available fare on certain services either side of the morning and evening peaks.1 Evidence on Rail Fares Elasticities A rail fare elasticity is the ratio of the percentage change in demand for the rail fare to the percentage change in price. instead of the current two. the more elastic will be demand. which itself is based on a wide ranging review of evidence. Table 2. with all other fares being held constant. 2.3. One TOC also suggested introducing restrictions on some popular travel days adjacent to public holidays. Fares are said to be elastic if their elasticities are -1 or more (negative). Off peak .1 shows own elasticities. others made tentative proposals (presumably operating within the constraint of revenue neutrality).only permitted during the off-peak.4.

2 Conditional Fare Elasticity Recommendations for Long Distance Trips First London Travelcard Area to and from rest of country.a.05 -1.4.65 Leisure -1. and PDFH explains how to relate the two using diversion factors.0 Commuting -1.90 Reduced -1. Tables B2. 2.4.6 Full -0.0 -0.50 -1. but must now be regarded as somewhat old. The chapter notes that the way that the fares elasticities in Chapter B2 (and reproduced in Tables 2. This is not the same as overall demand for rail. n.0 Business -0. Fares Elasticity Review. Generally elasticity is higher for standard reduced than standard full tickets. is documented in the fares elasticity review conducted by ITS Leeds as part of the PDFH update.00 -1. The AEAT study of inter-urban fares elasticity in 1999. Table B2. Handbook Update Demand Analysis. with particular emphasis on disaggregation of elasticities by ticket type.50 n. May 2002.and long-run responses and showed overall fare elasticity to be around -1. December 1999.85 -0. 4 5 6 7 .5 Table 2.6 This involved disaggregation by ticket type and journey length.7 It is this update that formed the basis of many of the fare elasticity recommendations contained in Edition 4 of the PDFH.2 shows PDFH recommendations for conditional fare elasticities. Saver ticket research on the Manchester-London and other routes shortly after introduction of Saver products. standard reduced. > 20 miles -1.6 -1.10 Source: PDFH June 2005.05 Apex -1. These AEAT results were revisited as part of the update to PDFH4.4.90 -0. February 2005. namely: 2. ITS Leeds. -0. Review of Fare Elasticity Evidence for Passenger Demand Forecasting Handbook Update.00 -1. standard full. 2.0. These studies modelled short. AEAT. Estimates are provided of fares elasticities in the 100-200 miles and over 200 miles distance bands.5 further work by Phillips.2) were arrived at.6.a. because many passengers would switch to and from other fare categories.85 -0.65 -0.1 and 2.25 -0. ITS Leeds. with the price of all other fares being held constant. Fare elasticities are quoted for first non-season travel.3 and B2. and the subsequent Jones and Nichols study published in JTEP.6 Sources of PDFH recommendations with regard to fare elasticities are considered in Chapter C2 of PDFH.4 There are a number of studies that have particularly considered fare elasticities on longer distance flows. Table 2. ITS Leeds. Inter-Urban Fare Elasticities. Table 2. and standard Apex.0 Other -1. January 2005. which are elasticities when all fares change by the same proportion..5 and for nonbusiness travel of between -1.1 Own Fare Elasticity Recommendations for Long Distance Trips Seasons London Travelcard Area to and from Great Britain excluding the South East Inter Urban.7 The econometric study by Owen and Phillips published in Journal of Transport Economics and Policy in 1986.9 -0. excluding London.a.3 and -1. but PDFH notes that the Apex elasticities are not really meaningful as demand for these tickets is constrained by available quotas.9 -1. and higher for standard full than for first non-season. This showed fare elasticities for business travel of -0.4.65 -0.70 n. excluding the South East Non London over 20 miles Non London over 20 miles (No Apex) Non London over 20 miles (No 1st or Apex) Source: PDFH June 2005.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 11 to percentage change in price. Mark Wardman.

suppose the fare per mile on a long distance London flow is 35% higher than the average. Variations in the fare per mile around the average are used to modify the fare elasticity. 2004: SDG.86.2 are taken to be representative at the average fare per mile. 2003.65. November 2003. 9 .7% higher = -1.350.49 meta analysis values of -0.56.6 is used. and recommend a value of -1.” 8 NERA Analysis of Passenger Rail Demand.8 PDFH also advises on how elasticity varies with fare level (Section B2 p6): “There is little empirical evidence on how fare elasticities vary with the level of fare and the attractiveness of rail.For conditional elasticities for reduced fares they quote NERA 2003 value of -0.83 (Whelan long run) and -1. … The evidence for long distance London. the latest PDFH does not advise on how elasticities should be adjusted on the basis of strength of competition. In other words.0.For Apex tickets they quote ITS values of -1.85. for SRA.3 below).00.37 (which only includes evidence from static studies. Report for PDFC.49. This is the value that they recommend be used. “If allowance is to be made for the fare charged per mile. and recommend a value of -0. In regard to non London long distance overall fare elasticities they show that long run elasticity is closely distributed around -1.82 and meta analysis values of -1. For conditional ticket type elasticities: .5. first class elasticity was 0. July 1999. and recommend a value of -0. and Meta-analysis of longer-distance fares elasticities by Wardman and Whelan. Variation of Fare Elasticity With Fare Level in PDFH 2. The Effect of Road Congestion on Rail Demand.23. and reduced fare elasticity -1.55 and reduced elasticity -1. it is recommended that an elasticity of 0. for OPRAF. and again recommend use of this value. . NERA. long distance Non London and short distance Non London flows indicated that the elasticity of the fare elasticity with respect to the fare per mile charged was around 0.07 and -0. and recommend a value of -1. full fare elasticity -1.22 and -0. NERA Long Term Fares Elasticities.0 from seven studies.38.83 (which includes evidence from dynamic studies) and -0.9 The SRA Strategic Fares Model produced by Jacobs Consultancy and ITS Leeds (see Section 2.6 = 19.5.61.4.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 12 For the 100-200 miles band first class own-fare elasticity was -0. -1. The recommended fare elasticities of Section B2. a 10% increase in the fare per mile would increase the elasticity by 6%. The adjustment to the recommended fare elasticities … would be: 1.20” 2. “As an example. NERA’s two studies of fare elasticities.38 (Wardman and Toner). . Oxera Rail Market Elasticities. SDG. meta analysis values of -1. The Wardman fares elasticity review includes the following information: In regard to flows between the London TCA and the rest of the country they cite evidence on overall fares elasticities from Oxera.0.4.8 Reports by Oxera and by SDG. For the over 200 miles distance band.92. full elasticity -1. This is because the empirical evidence only related to Non London long distance flows from a single study and a considerable element of judgement was required in their application.3. PDFH4 Update and meta analysis studies which have weighted average overall long-run elasticity of -1.For conditional elasticities for first class fares they quote NERA 2003 value of -0.9 In contrast to version 4.6. It states: “The elasticity modifiers relating to the competitive position which were recommended in previous editions of the Handbook have been removed.00p and .10 (Whelan short-run).For conditional elasticities for full fares they quote NERA 2003 value of -0.

PDFH notes “if time of travel restrictions on reduced fares are to be tightened.25.22 to 2.4. The recommendations are set out in Table 2. changes in travel restrictions could be converted into an equivalent change in fare of the restricted ticket using parameters set out in Chapter B3 and the forecasting system set out in Section B2. p.16. Alternatively. Chapter B2.4.3 of PDFH. PDFH.20 that these were just “guesstimates”.10 PDFH also gives detailed advice on calculating demand for one fare as a result of changes to the price of another fare on the same flow.4. For standard reduced fares. PDFH recommends. and 51 per cent to standard Apex. However. some of which concern long distance travel.11. p. for routes from the rest of the country to and from London with weak airline competition recommended diversion factors for standard full fares are 65 per cent to first. 39 per cent to standard full.4.10. that locally derived diversion factors are sought wherever possible. 11 12 13 .13 10 Set out in Section B2. When introduction of new tickets is being considered. as well as the simple spreadsheet to implement them. In regard to modelling changes in time-of-travel restrictions.4. PDFH also considers the relationship between conditional elasticities and ‘pure’ own elasticities. The diversion factors are derived from a single piece of research. recommended diversion factors are 8 per cent to first. This tends to be at odds with observed TOC behaviour. and 8 per cent to standard apex. PDFH can only advise that “expert advice should be sought”.4.3.12. PDFH recommends that.4. diversion factor evidence should be collected through market research related to the particular case under investigation. But see the comment in Section 2. where possible.”12 Rail Passenger Demand Forecasting Research13 2.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 13 We review the document concerned in Paragraphs 2.12 2. The manual provides recommended diversion factors between ticket types to undertake these calculations10. MVA (January 2007) Workstream 1: Interim Demand Forecasting Methodology – draft report for Department for Transport. however.3 Conditional Fare Elasticity Recommendations for Long Distance Trips(All Public Transport Fares Changing Together) 2. this may be due to the effects of other fares in their basket. Chapter B2. PDFH. For example. The PDFH research appears to suggest that Saver fares (ie reduced fares) are elastic. and shows how they are related using diversion factors and relative volumes by ticket type. Inter-Relationships Between Ticket Types and Time Periods in PDFH 2. Where Saver fares are not set at the maximum level. which are the proportion of a change in demand for a fare that are diverted to or from each of the other fares (or other transport modes). 24 per cent to standard reduced. and therefore increasing these fares would not increase revenue (and there may be some financial benefit in reducing them).11 2. where Saver fares are typically set at maximum levels.4.14 MVA’s current work for DfT reviewing PDFH has made certain recommendations on fare elasticities.12. PDFH has an accompanying spreadsheet that makes these calculations. Table 2. This can be calculated by using own-fare elasticities and diversion factors. market research should be conducted to identify switching between ticket types as well as the reduction in demand for rail travel so that demand for each ticket type can be forecast.

but there have been substantial recent increases in full fares which will have increased elasticities.5.1 Note that MVA use “conditional” to mean changes in all public transport fares. and because PDFH recommendations for long distance season tickets are very different from those for shorter distance season tickets. It was probably appropriate over much of the period over which analysis was undertaken. excluding South East Non London: Inter Urban ≥ 20 miles -0.” MVA do not dissent from PDFH recommendations for diversion factors with respect to long distance flows.17 This model was developed for the SRA by Jacobs and ITS Leeds.28): “We accept the leisure own-elasticities in PDFH v4. is that: MVA think long distance season ticket demand will be less elastic than that stated in PDFH.4.1 as reasonable.0 -0. However.55 -0.8 Non London: Inter Urban ≥ 20 miles Source: MVA (2007) Tables 4.9 Own Elasticities (other public transport fares held constant) -0. we consider that presenting fares elasticities according to journey purpose rather than ticket type is problematic for the long distance market.16 MVA argue strongly that forecasts should be made with respect to journey purpose rather than ticket type.75 -0. These price increases have generally not occurred on the non-London flows. they face very different price levels.05 Conditional Elasticities (all public transport fares moving together) -0.7 Business -0. and presumably have very different elasticities for the two fares (with – unregulated . Own-Fare Elasticities in the SFM . but consider the elasticity for business on flows to London to be too low. As such.8 -0.6 -1.15 The basis for MVA’s recommendations.5.25 -1.2 and 4. so we would not have expected them to differ to a substantial degree from those recommended by ITS for PDFH. 2.75 -1. 2. including that for longer distance travel. which has been effective for longer distance journeys.4. even in the business market. The Strategic Fares Model 2.7 -0.3) “There is some research into diversion factors. in so far as they differ from PDFH. the SFM needs to take account of impacts of changes in fares for particular ticket types on demand for travel by that ticket type (own fare elasticities) and impacts of changes in fares for a particular ticket type on demand for travel on other ticket types (cross-elasticities).4. Elasticities of demand are thought to vary according to price level: as business passengers buy both full and Saver fares. MVA considers the elasticity for business on flows to London to be too low. They say (para 4. The elasticities used in the SFM model are based on work by ITS as part of the study.7 -0.8 -1.7 Other -0. in PDFH the term is used to mean changes in all rail fares.3 -0. The SFM can be used to assess the effects of changes in different fare categories on demand and revenue in a number of different parts of the rail sector. Cross-elasticities are derived from own-elasticities using the diversion factor approach.full fare elasticity approaching -1.75 Rest of Country to and from London TCA -0.95 Commuting -0.75 Leisure -1. because informal discussion with TOCs suggest that they are below revenue-maximising levels.” In regard to diversion factors.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 14 Seasons London Travelcard area to and from rest of country. the profit maximising level).05 -0. MVA note (Para 4.

In addition these elasticities were compared against the relative generalised costs of car and coach travel on that journey. the SFM team used survey techniques to assess what passengers would do if their current ticket type became unavailable. namely switch of travel time. 14 2.85 -0. we note its main findings here. however. rather than conditional. though these diversion factors are likely to be higher than we need in the present study because it does not seem that passengers were being offered an alternative choice to switch of ticket type.50 -1. 200+m Non-London interurban with full set of tickets Non-London interurban without full set of tickets Source: SFM. the values used do not differ substantially between flows. 100-200m Rest of country from London. Non-London flows were the most rewarding as these had the greatest variation in strength of competition.60 -5. Diversion Effects Between Ticket Types in the SFM 2.4 reproduces values for different ticket types for eight different inter-urban service groups.60 Standard reduced -1.4.60 -2.12) are “guesstimates”. Results are provided in Appendix H for a number of longer-distance routes.4. Analysis was conducted on six years of CAPRI [sales] data (1986-91) on 764 flows.19 We believe that these are overall.10 -0.4. while they refer to those in the PDFH update (which we refer to above in Section 2. Table 2.23 ITS (University of Leeds) Technical Note 348.4.35 -1.30 Standard full -1.18 Own-elasticities for longer distance flows and for current product types are provided in Appendix A to the SFM final report.85 -0. including journeys on GNER services. Great Western.60 -1. 200+m Rest of country from London. 1000-200m Rest of country to London.4.75 -0.22 Previous versions of PDFH drew on this single piece of research to recommend adjustments to fares elasticities according to the strength of competition from alternative transport modes. Table 2.4.25 -2.75 -0.40 -1.40 -1.25 -1. elasticities. we are surprised to note that standard reduced elasticities are less than the standard full elasticity on each flow.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 15 2.4. Apart from the very high own-fare elasticity for first class travel (which is likely to reflect the very low proportion of nonLondon interurban flows on first class tickets).20 Appendix H to the SFM Final Report notes that the SFM methodology requires information on diversion factors in order to deduce a consistent set of own and cross elasticities across their ticket type demand equations.60 2. Appendix A First class -0.25 -1. Virgin Cross-Country and Virgin West Coast. These results might be used to inform the present study.50 -1. Final Report. These were used to calculate fares and generalised journey time elasticities for non London InterUrban flows.60 -0. 20-100m Rest of country from London. 20-100m Rest of country to London.4 Own-Fare Elasticities in the Strategic Fares Model Route Rest of country to London. It was removed from the latest version of PDFH because it was considered to be too subjective. Trans Pennine.25 -1.35 -1. Given the relevance of the research to the current study. there is no available evidence on diversion factors between ticket types. Effect of Coach and Car Competition on Rail Elasticities14 2. However. November 1993 . Consequently. As far as the SFM authors (and we) are aware.30 -0.21 2.

a study by ITS for the Monopolies and Mergers Commission found an elasticity of -0. Sixth Report of Session 2005 .4. distance divided by generalised speed Q. though we are not aware of previous work on the values that these might take.e. Table 2. Conditional own-fare elasticities for reduced price tickets seem to cluster around -1. Results are summarised in Table 2.9 House of Commons Transport Committee.4.26 The preceding section has summarised evidence from previous work on rail fare elasticities.5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 16 2.6.5 Conditional Fare Elasticity Recommendations for Long Distance Trips (All Public Transport Fares Changing Together) Fare Elasticity -1. and the constant elasticity is preferred. Conclusions on evidence on rail fare elasticities 2.9 for leisure travel by rail – see para 2. Results for flows to and from London were less robust. Generalised Cost of car is defined as the road journey time (as estimated by the computer package " Autoroute") multiplied by 4 pence per minute (at 1993 prices) plus petrol cost based on 4 star leaded petrol assuming 40 miles per gallon and cost per gallon from the publication "Transport Statistics". How this will affect behaviour can be investigated using time-of-day diversion factors. not conditional elasticities). and the overall elasticity for long-distance leisure travel by rail is often found to be elastic15. the results from the latter model were not particularly robust.06. For fares elasticities. previous elasticity estimates do not take account of the possibility that Saver fares could vary over the day (for example. eg a direct motorway link or poor rail service Normal Weak Example Flows 1 2 3 -1. But the analysis did find evidence that fares elasticities were higher on flows with strong competition than on those flows with relatively weak competition.0.5 2. Generalised Cost of coach is defined as generalised time (i. In such circumstances passengers will consider alternative times as to travel. “How fair are the fares? Train fares and ticketing”.0 -0.4. distance divided by generalised speed Q) multiplied by 4 pence per minute (at 1993 prices) plus the average fare paid.1 Transport Select Committee Report and Government Response16 In 2005/06 the Transport Select Committee investigated a range of issues with respect to rail fares: 15 However.25 For an individual flow: Generalised Cost of rail is defined as generalised time (i. because of shoulderpeak pricing).2 Competition Strong.e. 2.24 The work used both a constant elasticity model and a non-linear least squares model. Much of the evidence available gives elasticities by journey purpose rather than ticket type. calculated on the coach timetable) multiplied by 4 pence per minute (at 1993 prices) plus an estimate of the average fare paid. 16 . However.8 4 5 6 7 Leeds to Luton Stoke to Blackpool Oxford to Cambridge Cardiff to Manchester Edinburgh to Aberdeen Southampton to Newport York to Newcastle 2. For the present project we are particularly interested in evidence on fares elasticities by one particular ticket type (Savers) and the values of these when Saver fares change in relation to other fares (that is.5.

But it does not believe that those benefits and costs would remain positive if the level of demand on such routes was being additionally stimulated by price subsidy at levels below a competitive market rate. together with the Department for Transport’s associated response. 74) 2. The management and pricing of rail fares.5. are ring-fenced and protected. which are the only remaining affordable walk-on fare on our railways.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 17 Their strategic significance. The complexity of rail fares.4 DfT response: “The Government does not believe that there is a case for capping Open fares. 2. and that Open full fares should also be regulated. Regulating to reduce such prices would not only add to the underlying subsidy cost for rail but also create additional investment pressures to meet the resultant increases in demand.” (DfT para 52) 2. Overall. It is imperative that Saver fares.3 2. It is the Government’s duty to tax-payers and passengers to provide this protection. and good value for money can be seen to prevent the development of more ‘customer focused products’. These exist at present in the Saver market and both Government and operators see this type of fare as important going forward. …We are therefore deeply concerned about proposals to reform or even abolish regulation of Saver fares.” (para.e.” (para. 64) 2.2 The Committee made a number of conclusions that are pertinent to Saver regulation. They are unregulated.” (DfT para 51) 2. at a level reflecting modal competition). 82) and: “We note the claims of some Train Operating Companies that they would retain Saver fares if they were de-regulated. We set out the basis for these conclusions below. and Regulation. through franchise agreements.5.5. The Committee concluded: “…It is imperative that reasonably priced open walk-on fares are re-instated so that nobody is excluded from using the railway for trips which cannot be booked in advance. universal across the network. We have little faith in these assertions.5. the railways would have no hope of competing with road travel. Without walk-on fares. they considered Savers should be more tightly regulated.6 DfT response: “The Government does agree that there is also a role for reasonably priced walk-on fares. and have been subject to substantial increases on many flows. Timetabling issues (including disruption due to engineering works).5.” (para. to cap open walk-on fares.5 On walk-on fares (as opposed to advance purchase fares) the Committee concluded: “The ability to turn up and go without notice is a vital characteristic of the railway. taking into account the wider market and the cost of providing the necessary infrastructure. The value for money of rail travel. We recommend that the Government take measures without delay.7 On Saver fares the Committee said: “We find it hard to see how Saver fares that are easy for passengers to understand and use. The Government does believe that there are strong economic reasons to invest in rail to respond to demand. . Standard open rail fares are walk-on fares that can be used on all trains.5. …Open tickets are used for fully-flexible travel during the Monday-Friday business peaks and appear to be set at reasonable economic rates (i.

passengers (sic) – need to have confidence in the objectives and credibility of outcome of any changes to ensure they deliver benefit. making these tickets difficult for passengers to use. by the Secretary of State for Trade and Industry). …We therefore recommend not only that the Government retain regulation of Saver fares.” (paras. 61) 2. and hence on the question of whether some Saver fares might rise if deregulated. and .6.3 Two or more enterprises must cease to be distinct...both Government – and most importantly. or because of. in assessing whether mergers should be allowed to proceed the OFT and Competition Commission will apply a “substantial lessening of competition” (SLC) test. Second.1 Evidence from Competition Authorities In this section we consider whether evidence from competition authority investigations can shed light on the factors constraining leisure fares on longer distance rail services. but that it take immediate steps to strengthen the current regulatory regime in order to limit the kind of travel restrictions that train operators are able to impose. Cases are first considered by the OFT.6.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 18 Savers have already been eroded in many cases by ever narrower time-oftravel restrictions imposed by train operators.” (DfT para 60.. 85 and 86) 2. There are three criteria to constitute a relevant merger situation: 2.. A merger may be expected to lead to a substantial lessening of competition when it is expected to weaken rivalry to such an extent that the competitive process would no longer deliver a similar level of customer benefits as it would without the merger . the OFT might expect that product choice would be reduced. decisions will in general be taken by the Office of Fair Trading (OFT) and the Competition Commission acting as specialised independent competition authorities (and not.2 2.5. objectively justified by relevant facts. The merger must not yet have taken place. that there is a realistic prospect that the merger will lessen competition substantially. prices could be raised profitably. The Enterprise Act 2002 made two major changes to the UK system of merger control.8 DfT response: “There are…legitimate questions that can be asked about whether [a cheap turn-up and go ticket] should have the same pricing level whatever the demand for the train (causing very high crowding on the ‘last Saver’ trains. The relevant competition authority investigations are those concerned with mergers or potential mergers between passenger rail service operators. whilst other trains remain under-utilised at. We do agree with the thrust of the Committee’s conclusions that de-regulation on the basis of assertion or speculation would be undesirable and could lead to passenger detriment….6 2.6. Saver fares would eventually be eroded to the point of irrelevance because operators would be tempted to apply ever stricter conditions and restrictions on their use. current Saver pricing levels…) “We believe that it is important such issues are properly and objectively considered as part of an ongoing process of discussion. any of these post-merger effects would be expected to be sustained for more than a short period of time. as previously. output could be reduced and/or product quality or innovation could be reduced . We strongly suspect that in a deregulated market. The Test for reference will be met if the OFT has a reasonable belief. and to ensure that travel restrictions are the same for Saver tickets across all operators on the network. who decide whether the case should be referred to the Competition Commission. or taken place not more than 4 months before the reference is made (unless the merger took place without public announcement and without the OFT being informed). First of all.

5 2. it is perfectly possible that a merger investigation at present would not consider that there would be a danger that they would be raised in a less competitive situation than the pre-merger one because it would simply conclude that regulation prevents this from happening. The award of a rail franchise can therefore constitute an acquisition of control leading to two or more enterprises ceasing to be distinct. but we should bear in mind that their potential to provide useful evidence is limited by the following considerations: 1. The recent Competition Commission investigation of the joint venture between Stagecoach and Scottish Citylink to operate coach services in Scotland.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 19 Either the ‘turnover test’ or the ‘share of supply’ test are met. 12 investigations involving longer distance TOCs.6 2. The question being asked in these exercises involving market definition is different to the question being asked in the present study of potential impact of Saver fares deregulation: In merger investigations the question is that of whether a rail operator could increase fares once a merger had taken place so that the market structure had changed from the present one In the present study the question is that of whether an existing rail operator could increase certain existing fares (namely Saver fares) in the existing market situation if they were to be deregulated. These are not listed chronologically but according to service groups. as follows: Franchise bids for the InterCity East Coast franchise (4). In regard to the present study on Saver fares.6. Market definition focuses attention on the areas of overlap in the merging parties' activities. evidence from these investigations may have considered whether mergers might have reduced competitive pressure on rail fares. Section 66(3) of 1993 Railways Act (as amended by paragraph 30(8) of Schedule 25 of the Act) provides that where a person enters into a franchise agreement as a franchisee there shall be taken to be brought under his control an enterprise engaged in supplying the railway services to which the agreement relates. Once the overlap of the merging parties products or services has been identified. 2. we only list in detail. The main conclusions are as follows: . However. in Table 2.6. but may still be the competing alternatives to each other. where Saver fares would be of importance. Our review is based on the 26 investigations since rail privatisation. Assessment of the Evidence 2. First’s bid for Scotrail. There have been some 26 decisions on rail industry mergers. Franchise bids for the Greater Western franchise (3). we are particularly interested in any analysis of competitive effects on longer distance leisure fares. The 'turnover test' is that UK turnover being acquired exceeds £70 million and the 'share of supply' test is that the merging enterprises must together supply or acquire at least 25 per cent of all those particular goods or services supplied in the UK or a substantial part of it. Most mergers relating to franchises meet the turnover test. 2. In principle then. 2.4 Franchise agreements in the rail industry qualify as potential mergers.8 Table 2.6. and Arriva’s bid for the Wales and Borders franchise. Monopolies and Mergers Commission investigations of initial National Express franchise awards (3).7 Nevertheless we believe it is important to assess the evidence that exists.6. Given that Saver fares are currently regulated.6 summarises information on the way that the investigations considered overlaps between the rail services concerned and other rail or other transport services. This is particularly the case in differentiated product markets (like travel markets) where the parties' products or services may not be identical.6. the investigation can focus attention on assessment of the competitive situation. The proper frame of reference for application of the SLC test is that of market definition – it is necessary to define the relevant product and geographic markets affected by the market.6.

competition from air would constrain the ability of Virgin/Stagecoach to raise fares on the LondonEdinburgh/Glasgow/Dundee routes. it could be expected that rail fares in Scotland would fall over time (as evidenced. But note that this is exactly the type of evidence we are looking for in looking to see if leisure rail fares may be constrained by competition from other modes. services are clearly seen as in competition for leisure passengers. Conclusions on Competition Authority Investigations In conclusion. in their view. competition authority investigations have reviewed competition and potential competition in the rail industry in considerable detail. the Commission themselves were sceptical.10 We note that all this was somewhat speculative. the main parties considered that it could only be a matter of time until similar fares were introduced by First on its ScotRail franchise and that. 2006 . The main parties argued that this was additional evidence that coach and rail services were in competition. Indeed. Only one set of cases. in the case of the bid by Virgin/Stagecoach for the InterCity East Coast franchise.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 20 2. evidence that leisure rail fares (and particularly Saver rail fares) on longer-distance routes are constrained by competition from other modes is limited. We have not covered these cases in this review. among other things. 2. This means that the time and cost of car travel is not expected to exert significant pressure on rail fares on parallel routes. rather than on whether coach competition constrains rail operator’s ability to increase leisure rail fares.” 2.14 17 Competition Commission Stagecoach and Scottish Citylink TSO. more generally. 23 October. raised issues of the impact of competition from air services on rail services. Generally the competition authorities have not regarded competition from the private car as providing a competitive constraint on rail fares because they have not regarded them as sufficiently close substitutes. by lower rail fare pricing on other franchises elsewhere in the UK on comparable routes). namely the bids for the Inter City East Coast franchise. Nevertheless. While many of the investigations have looked at competition between rail and buses this has involved short-distance journeys where Saver fares do not apply. In addition we have also noted that competition authorities would not be inclined to worry whether mergers would lead to increases in Saver fares given that Saver fares have been regulated.12 2.11 2. Most recently the Competition Commission report on the Stagecoach/Scottish Citylink joint venture has provided a suggestion that coach fares could constrain rail fares.6. The report notes17: “The main parties drew our attention to the introduction of new low-price 'Firstminutefares' by First on its Great Western Rail franchise. However. in this case in the Great Western region.13 2.6. While many of the studies include some fare elasticity evidence. The main parties also noted that First had stated that it had begun to monitor coach fares in Scotland.9 There has been relatively high emphasis on the impact of competition between rail services and coach services. and that ScotRail could be expected significantly to lower its rail fares in the near future. However the emphasis has mainly been on the question of whether rail competition constrains coach fares. Longer distance rail.6. work by ITS Leeds for the Monopolies and Mergers Commission in 1996 produced an estimate of own-price elasticity of 0. Further. and coach.9 for leisure travel by rail. and submitted that these were a reaction to coach fares since.6.6.6. and noted that “the existence of Firstminutefare could at most be interpreted as evidence of coach services constraining rail services rather than vice versa”. The OFT thought that. The report concluded that the merger between National Express coach and rail services could be expected to lead to higher coach fares or higher fares on both coach and rail. relationships between these fares and both megabus and National Express pricing schemes were evident. the combined sum of this evidence is much less than that which is contained in the PDFH.

Issue was competition from bus on nine routes. Merger not expected to lead to SLC. not referred 21/12/04. Virgin/Stagecoach and Intercity East Coast LYNE (DSB/EWS) and Intercity East Coast 08/12/04. Also some coach issues. Rail faced continued competition from air on London Edinburgh/Glasgow/Dundee routes so this would restrict merged entity’s ability to raise rail fares. Undertakings had been offered but were not accepted. OFT thought there were substantial incentives for Stagecoach to raise bus fares (and/or reduce frequencies) and perhaps also to raise unregulated rail fares as a result of the merger. but on short distance journeys and concern was that bus fares might be increased. initially referred since First were unable to offer acceptable undertakings. not referred as OFT was considering undertakings but then the SRA awarded the franchise to GNER so undertakings and possible referral no longer relevant.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 21 Table 2. referred.referred but withdrawn on 22/12/05 after First short-listed National Express and Greater Western Stagecoach and Greater Western First and Scotrail 13/01/04. Investigated by MMC National Express and Midland Main Line National Express and Scotrail Investigated by MMC . 21/12/04.6 Competition Authority Investigations of Impact of Competition on Fares Potential bidder and franchise OFT decision and date Referred Competition for rail and conclusions on impact of overlaps No overlaps identified Concern was rail-on-rail competition from Hull Trains. First and Greater Western 30/09/05. also from Trans-Pennine Express which was also owned by First. and some consideration of car.) Coach/rail overlap was expected to lead to some loss of competition. Also wider network markets. The potential issue here was not competition for passengers but access to light maintenance depots. 30/09/05. Competition between coach and rail services. not referred. 30/09/05. Potential reduction in competition from Virgin Cross Country on certain routes did raise concerns with the OFT. GNER and Intercity East Coast First and Intercity East Coast 21/12/04. referred. (The public interest test was the test that preceded the SLC test. OFT’s concern was National Express’ ability to raise coach fares. Merger was judged to be against the public interest because MMC saw possibility of increased coach and rail fares. and increases in coach fares on nine routes. Bus. CC reported 08/03/06. But franchise was awarded to GNER. referred but withdrawn on 22/12/05 after First short-listed.

ORR will need to take into account the extent to which alternative modes of transport provide a competitive alternative to the rail service. In doing so. Competition between bus and rail. OFT’s concern was that Arriva might foreclose access to rail stations by other bus operators through issue of multimodal (i. In assessing whether or not dominance exists.e bus and rail) tickets that could not be used on other companies’ bus services. the ORR note that.4 2. ORR note that where other modes of transport offer viable alternatives such that they impose a competitive constraint on the TOC. the ORR policy note states that the setting of fares by TOCs is subject to competition law and.7. The note concludes with the following statement: “ORR remains of the view that if there are public interest concerns about fare levels. then these should be addressed by DfT through its fares policy”. Under Chapter II of the Act.5 . But.3 2. National Express and Central Trains Stagecoach and Scottish Citylink Investigated by MMC Competition Commission report published 23/10/06 Arriva and Wales Borders 09/07/04. which prohibits abuse of a dominant position. This was a merger between two coach operations but the issue of competition from rail on coach services was considered. ORR can scrutinise fares that appear to be excessive. ORR could intervene if it had reason to suspect that individual fare levels were incompatible with competition law.7 2.7. not referred because Arriva gave undertaking on third party access to multi-modal tickets.7. At the beginning of their note the ORR say: “Responsibility both for fares policy and for the regulation of those fares subject to regulation lies with the Department for Transport”. the ORR would need to have reasonable suspicion that the TOC holds a dominant position in the relevant economic market and that it is abusing its position of dominance. Nevertheless. the ORR would take account of whether a fare is subject to regulation when considering any complaint of excessive fares under the 1998 Competition Act. it is likely that the TOC would not be found to be dominant. even if dominance were to be established. While the main concern was whether competition from Scotrail constrained coach fares (rather than the other way around) there was some discussion of whether Great Western leisure rail fares had been constrained by coach competition.7. 2.7.6 Competition Authority Investigations of Impact of Competition on Fares Potential bidder and franchise OFT decision and date Referred Competition for rail and conclusions on impact of overlaps Some coach/rail overlaps but there was also competition on these routes from other TOCs. to do so.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 22 Table 2. meeting the test that prices are excessive is difficult: 2. In addition.2 Competition Authority Views on Fare Levels A policy note by the ORR sets out the role of competition law in regard to rail fares. as competition authority for rail services.1 2.

That by ITS concluded that Saver fares would fall rather than rise. Overall TOCs welcomed the principle of Saver deregulation. while the Strategic Fares Model uses overall elasticities by ticket type (see Table 2. The DfT had invited TOCs to comment on both proposals to introduce single leg pricing and to deregulate Savers. Some made tentative proposals to increase prices.There have been no 1998 Competition Act cases brought by competition authorities in the UK that have addressed pricing that was exploitative but not anti-competitive. The study by NERA concluded that Saver fares would rise with deregulation. but not on diversion factors between time periods.6 NERA believe that the ORR view is consistent with their ruling on the only complaint of excessive fares on which they have published a judgement.5 above). on the fact that Saver fares were set at or close to the ceiling permitted by regulation. and some argued that in practice deregulation would make little difference to their fares structure. own elasticities and relative traffic volumes.8 Conclusions from the Review of Existing Evidence Two previous studies. and on evidence on what had happened to the most closely related unregulated leisure fares. based on evidence on overall fare elasticity. A number of TOCs proposed three tiers of walk-on fares. Relationships between elasticities can be derived using diversion factors. This case was that of complaints of excessive unregulated fares on Virgin West Coast services (including London-Manchester) in 2001. There is very detailed evidence on rail fare elasticities. which are those which arise in the present study of longer distance rail services.The European Court of Justice has not upheld any decisions where the European Commission has found a dominant firm to have charged an illegal excessive price. Most long distance operators accepted the principle of single leg pricing for long distance flows. 2. Evidence is also available on fares elasticities by journey purpose. One study has considered how the strength of inter-modal competition will affect rail own-fare elasticities. Conditional fare elasticities by ticket type are included within PDFH (see Table 2. For modelling the potential impact of Saver deregulation we require information by ticket type. The ORR note also points out difficulties in establishing whether individual rail fares are excessive in the circumstances. We have reported the conclusions of the Transport Select Committee investigation into rail fares and the DfT’s response.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 23 As the ORR policy note indicates. The effect of this has been captured in the study through the diversion factor approach where the " would not travel by rail " diversion factor is higher for those services where modal competition is higher. but this result might appear curious in that there is nothing at present preventing train operators from reducing Saver fares. where costs are shared between all train services provided by a TOC and between different passengers on the same train. There is some evidence available on diversion factors between ticket types. The Committee considered that Savers should be more tightly . In this case the ORR conducted a preliminary investigation.2 above). including evidence for leisure travel and evidence by ticket type. and the Regulator ruled that although Virgin might be dominant in one or more of the relevant markets. though there were some concerns in regard to other flows. he did not believe that there were reasonable grounds for suspecting that Virgin had behaved abusively in charging excessive fares and he declined to conduct a full investigation. namely cheap day returns and supersavers. by ITS Leeds and by NERA have considered potential impacts of deregulation of Saver fares.7. 2. except in cases where that price itself distorted competition. The conclusion is that fare elasticities are higher on flows with strong competition than on flows with relatively weak competition. and . there have been no cases either in the UK or at the European Union level where prices have been judged to be excessive in the sense that the impact is solely to harm consumers rather then distort competition: .

As we have just noted in the previous section. while competition authority investigations have reviewed competition and potential competition in the rail industry in considerable detail. the combined sum of this evidence is much less than that which is contained in the PDFH. The Committee noted that they had little faith in claims of some TOCs that they would retain Saver fares if they were deregulated. especially on crowded trains close to the peak. .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 24 regulated than at present and that Open full fares should also be regulated. While many of the studies include some fare elasticity evidence. evidence that leisure rail fares (and particularly Saver rail fares) on longer-distance routes are constrained by competition from other modes is limited. An ORR policy note sets out the ORR’s position in implementing the 1998 Competition Act in regard to complaints about excessive fares. In response to this view DfT suggested that there was a case for considering whether Saver fare levels should be the same on all trains.

3 New Research .

and the requirement to purchase in advance. there is an unavoidable amount of preparation and programming to ensure the software is set up correctly. Paper surveys were printed and handed out on-train to be completed during the journey. and A stated preference section which has been used to determine how people trade off between ticket type characteristics. but the nature of the way information had to be customised for the SP part of the survey to be realistic to specific passengers meant there were 378 unique questionnaire types. The survey was designed to include both passengers who use Savers and those who choose other ticket types. The objectives of the research were firstly to establish information about passengers who do or could potential use Savers.1. and at different times of day. it might have been preferable to use a computer aided survey approach. Further details relating to the complexity of the SP design are provided in Chapter 5. The survey was designed to provide a good understanding of the characteristics of the rail market when Savers are currently available.250 completed questionnaires over a range of routes. The format of the final survey consisted of two parts: 3. Although computer aided design allows extremely focussed and customised questions to be presented to an individual.2. we undertook a Pilot study. Friday 16 March 2007. The second objective was to collect Stated Preference data which could be used in the development of a model which can assess the ability of operators to increase revenue through charging higher fares if regulation was reduced or eliminated. Sunday 18 March 2007. for a study of this complexity. fare.2 A section containing questions relating to the respondents’ travel choices and demographics. Responses rates were high (83%). In order to maximise the likelihood of sound SP results. DfT have been working to a very short timescale which did not allow a computer aided approach. The stated preference (SP) part of the survey was designed to look specifically how people trade-off ticket flexibility. and Tuesday 20 March 2007. 3.2. which would assist in determining their need for protection through the regulation of Saver fares.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 26 3 New Research 3.1.2. For this study. Survey Methodology Introduction The requirements for the fieldwork were to collect 1. the quality of the responses is very much dependent on the characteristics of the options respondents are offered. Questionnaires for the main survey were distributed on the following days. If time had permitted. Monday 19 March 2007. .1 Survey Objectives The overall aim of the survey undertaken as part of this study was to obtain information on the characteristics and preferences of passengers making a range of longer distance rail journeys at times when Saver fares are valid. Minor amendments were made to the first part of the survey.3 The pilot survey was carried out on four trains between Manchester and London on Tuesday.1 3. Preliminary results from both the conventional and stated preference results produced encouraging results.1 3. Because stated preference is a complicated and technical way of conducting surveys.2 3.2 3. and production of the stated preference elements of the surveys was set in place before printing and delivering the main survey questionnaires. Survey staff collected completed surveys from respondents or left by respondents on the train. 27th February.

050 were handed out. In total 9. Newcastle to Birmingham.4 3. We also surveyed the following cross country flows: Edinburgh to York. Stated Preference is a research technique used which presents trade-off situations to people. Newcastle to York. The first section of the questionnaire contains questions relating to the journey that the respondent is making.Manchester/Stockport.3. Usually when exploring a complex research topic such as this. because Sundays are currently un-restricted for Saver fares.2 Trains were chosen to give a good representation throughout the day. the timescales constrained the methodology used.Birmingham/Birmingham International. London Euston . To a large extent. London Kings Cross . and London Paddington – Bristol Temple Meads. These questions were decided in discussions between the DfT. London Marylebone – Birmingham SnowHill/Birmingham Moor St/Solihull.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 27 3. London Paddington – Bath Spa. but often experience some of the heaviest train loadings.4.2.3 3. London Kings Cross – Newcastle. and York to Birmingham. London Euston – Stoke on Trent.4 3. Questionnaire Design An example of the questionnaire is shown in Appendix B.1 London Euston .3.4.2. 3. Each of the flows was surveyed in both directions. The fieldwork was undertaken by the survey company SkyHigh. we chose flows which were expected to have significant volumes of in-scope passengers. Faber Maunsell and NERA. considerable discussion took place to ensure a good range of flows were sampled. In order to maximise the number of passengers on any particular train who were ’in-scope’ (bearing in mind that the questionnaires were customised for specific train and flow combinations). The timescale of the study was exceptionally tight.2 The second section of the questionnaire contains the stated preference exercise.Edinburgh. London Euston – Macclesfield.3 .4. It enables their underlying preferences to be revealed using advanced econometric modelling techniques using specialist software package such as ALOGIT.5 The DfT were particularly interested in carrying out surveys on Sunday.1 What are the characteristics of users when Savers are available? For what purposes are they travelling? What alternatives do they have? Do they need the flexibility that Savers provide? Who is choosing which ticket type? What time would they ideally have liked to travel? 3. The questionnaires were designed and printed by Faber Maunsell. Appendix A provides a list of every train flow surveyed. Their purpose was to provide answers on key questions including: 3. we would carry out qualitative research using focus groups with respondents to get a better understanding of the choice context and explore the factors that influence this choice. Choice of Routes and Trains Prior to the final survey schedule being finalised.650 questionnaires were printed and around 6. their ticket choices and their demographics. Leeds to Birmingham. The final list of London flows surveyed is as follows: 3. As 3. and to allow reasonable survey logistics. London Kings Cross – York.

In the SP experiment respondents were asked to choose their ideal time of train departure and ticket for the leg of the journey they were making when contacted in the survey. whether it would be necessary to buy in advance and the time of travel.4 In the SP experiment respondents were asked to make trade-off between ticket types. This only applies to advance purchase tickets as for Open and Saver tickets it is not necessary to buy in advance. This is used to calculate Displacement time which is the difference between travellers’ ideal travel time given elsewhere in the questionnaire and the actual time of departure. Open tickets in the survey were more expensive than others. Seven train times were offered – three either side of the train time which the respondent was travelling. An example of the type of scenario presented is shown below for a train departing at 19. Providing respondents fill in the questionnaire rationally and have understood the choices they are being asked.4. the times of the available trains were based on the actual timetable.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 28 explained previously. Seven departure times were given. it was decided to concentrate on the single leg of the journey respondents were on when contacted.This is the one-way fare payable Advance Purchase .8 . Savers were available on four and Advance tickets were available on two. Saver and Advance. From this. it is theoretically possible to calculate the respondents implied displacement time from their ideal time of travel.4. Given the increasing use of single leg ticket pricing particularly by Virgin and GNER. 3. This meant that the questionnaire was customised for each train so that we could ensure Saver and Advance ticket option were not available for their chosen train. displacement time is assumed to be zero. We did not allow them to choose Saver or Advance tickets in any of the hypothetical situations we offered for the train they were travelling on. in order to encourage them to trade displacement time and fare. However. Displacement time was a key variable of interest in this study. However the availability of Saver and Advance tickets were varied from reality in order to force people to switch time of travel. They had some flexibility because they were usable on any one of the four trains that were available.4. we expect them to choose the departure time/ticket combination which minimises the disutility (fare + convenience + availability etc) of their journey. but fully flexible and available on all trains. This was a highly complex design since for each of the scenarios presented to them they were faced with a choice of 7 departure times and three types of ticket – Open. Advance ticket choices had no flexibility. given the complexity of what we were trying to present to respondents it would have been advantageous to have used a computer aided (CAPI) approach. 3. Advance tickets also needed to booked in advance. the timescale did not permit these approaches.7 When making a choice. 3. Each type of ticket was described in terms of: Fare .4.5 Accordingly. availability (which trains they were available on) and flexibility (whether they could change trains after the ticket was purchased). In the first part of the survey. respondents were asked to consider ticket price.00: 3. Saver tickets were cheaper than Open tickets but were only available on 4 of the 7 trains.This is the departure time of the train. respondents were asked to make a trade off between the fare payable. In order to be as realistic as possible. They could not be changed once booked. Time of Departure .This is the number of days in advance it is necessary to purchase the ticket. respondents were asked to state their ideal departure time. For Open tickets choosers.4. Open tickets were available on all seven.6 2 Ticket Open Saver Advance Fare(£) 55 25 15 Advance Purchase None None 7 days 1730 1800 1830 1900 1930 2000 2100 If these were the only tickets available I would not travel by train 3. Advance tickets were cheaper than Saver tickets but were available on only 2 of the 7 trains.

Response Rates Main Survey Of the 6. However. for example using www.2 Number of Survey Responses by Route Route Bristol Birmingham Manchester Edinburgh Cross Country All Friday 197 78 112 55 9 451 Sunday 49 0 98 0 0 147 Monday 54 211 78 394 75 812 Tuesday 0 245 77 34 24 380 All 300 534 365 483 108 1.11 3. responses were very low on cross country routes due to low numbers of people on the trains making in-scope journeys and problems with on-train Virgin staff. 1. otherwise the data becomes collinear and estimation problems result.5. The times offered to respondents centred on the time of the train they were actually travelling on. Consequently each design was split into two. Table 3. the layouts were designed to be similar to information people are presented with when they book a ticket on-line.4. A and B. each with nine scenarios. Although complex.1 shows the response rate and Table 3.2 shows the actual number of responses received. The full SP design consisted of 18 scenarios which would have been too much for respondents to be expected to complete through a paper-based unsupervised approach.thetrainline. This is why it was necessary to customise the questionnaire for each train. The design was a fractional factorial design based on the standard reference material for designing SP experiments (Kocur et al (1982)). Significant numbers of responses were received from all of the London routes and across all of the days surveyed.4. This would at least to be familiar to some people.4.10 3. During the survey design process the different variables of interest are combined in such a way that the correlations between them are minimised. These were distributed to respondents at random.790 . and fare options were based on actual fares for the journey (although obviously these were varied as part of the SP experiment).1 Table 3.050 questionnaires handed out.790 questionnaires were returned giving a response rate of 30%.com.9 “Would not travel” was a fourth alternative which respondents could choose if the rail ticket alternatives were too unattractive. In stated preference research it is important that the options given to a respondent correspond to situations that they can easily understand and relate to their journey.1 Response Rate by Route Route Bristol Birmingham Manchester Edinburgh Cross Country All Friday 35% 22% 48% 15% 2% 24% Sunday 41% N/A 42% N/A 0% 25% Monday 19% 57% 33% 50% 10% 33% Tuesday N/A 40% 32% 18% 32% 34% All 31% 40% 39% 36% 7% 30% 3.5 3. Table 3.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 29 3.

If passengers see that train staff have shown a negative or uncooperative attitude towards survey staff. Some passengers may chose not to fill in the questionnaire (after they have accepted it). stated preference is a more complicated technique than a conventional survey.g.6. Piloting is an important way of testing the survey approach. Several respondents gave up on the stated preference section (around 33% of otherwise completed questionnaires). because it takes them longer than they had anticipated. fewer flows).6. It is possible that staff were able to spend more time explaining the purpose and nature of the questionnaire to every person it was handed to. There are risks associated with all surveys (and particularly SP) surveys.6.3 .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 30 3. In this particular case. In this case. These might have been minimised with either more review time scheduled into the process and/or a somewhat reduced survey scope (e. survey staff received little and at times hostile co-operation from on-train staff. This meant the survey logistics were extremely tight and allowed very little review time between Pilot results and main survey printing. 3. the complexity of the trade-offs we required for the study meant the design of the SP was always expected to be complex. The number of different survey questionnaires was also an important factor for the fieldwork part of the study. As we have discussed. On some trains.2 3. and there are trade-offs to be made between complexity of design and quality of response. the Pilot survey did not indicate any major design problems and the overall timescale for the study limited the time available for review.6 3. It is not entirely clear why this is so.1 Survey Conclusions The main survey response rate of 30% is reasonably good for a self completion survey. The high response rates for the Pilot study suggested that the complexity of the design was not unduly affecting response rates. Pilot study survey administrators were only distributing one type of survey form. However it was considerably lower than the Pilot survey. it could reduce the likelihood of passengers filling the surveys in. and permission had been received from all. All TOCs whose trains were being surveyed were informed of the trains affected.

Findings From Questionnaire Responses .4.

and Lack of access to booking channels other than at stations.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 32 4 Findings From Questionnaire Responses Introduction The survey collected a large volume of information about the characteristics of Intercity rail users travelling outside peak periods. In this Chapter we focus on providing information which assists in answering a series of questions which are of particular relevance to understanding the potential vulnerability of passengers to any possible increase in walk-up fares due to: 4. Between 37% and 39% of passengers are using Saver tickets. Need for flexibility in departure times. it is a high proportion. In the non-peak periods. A lack of alternative modes.2 4.1 What type of ticket are you using for your journey? Outward Return Open Saver Advance First Advance Season Other Total 25% 37% 18% 8% 2% 10% 100% 25% 39% 17% 7% 2% 9% 100% This table is corroborated by information we have summarised from the LENNON ticket sales data.1 4.1 Income and socio-economic information.1 Number of Saver users First we look at the number of respondents who are using Saver tickets. 4. . Table 4.2.1.

This may be because APEX purchases are increasing year on year (and LENNON data is historic in this respect). both sources of data clearly show the significance of the Saver product in the current long-distance rail market.3.501 to £35. The table shows that Saver respondents have very slightly lower household incomes than the average. Table 4. which is surprising.500 £12. Using the categories attributed within LENNON.3 4.500 £17.3 Total Household Income Overall Under £7. We might expect that the total proportion from the LENNON data to be lower than in the survey.501 to £17. but not significantly so. In any event.2 2006 LENNON Revenue by Flow Ticket types Flow Birmingham to London London to Birmingham Manchester to London London to Manchester Edinburgh to London London to Edinburgh Bristol to London London to Bristol Birmingham to Newcastle Newcastle to Birmingham Edinburgh to York York to Edinburgh All 4.001 to £75. Respondent income and socio-economic information On average non-peak Intercity rail users are relatively affluent. Over 50% of those surveyed had a household income of over £35.000 £50. In reality it is likely to be a combination of both factors. the categorisation of ticket types within LENNON ‘reduced’ may include some APEX tickets.2.000 £7.000 More than £75.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 33 Table 4.000 Prefer not to disclose Don't know Total 7% 5% 7% 20% 13% 14% 17% 12% 5% 100% Saver d t 9% 7% 9% 20% 13% 13% 13% 12% 5% 100% 4.000 to £12. APEX is very small share – smaller than we have been finding from the survey respondents.1 . which was conducted at non-peak (Saver valid) times only.000. because the LENNON data includes sales from all time periods.001 to £50.000 £35. Alternatively.000.2 Standard Full 50% 44% 39% 34% 35% 19% 47% 36% 43% 53% 11% 11% 37% Standard Reduced 15% 13% 11% 13% 14% 24% 1% 0% 17% 12% 44% 35% 13% Standard Reduced Advance Purchase 1% 1% 2% 1% 8% 11% 13% 16% 1% 1% 9% 7% 6% Standard Season Ticket 6% 0% 1% 0% 1% 0% 7% 1% 1% 0% 3% 4% 2% Saver Standard 29% 42% 47% 51% 42% 46% 31% 48% 38% 34% 33% 43% 42% The LENNON data gives a slightly higher proportion of Saver fares than the survey. ONS statistics state that average UK household income is £27.

3. the proportions are very similar – although there are slightly more students than average travelling on Saver tickets. Table 4.4 8% 41% 4% 3% 3% 23% 2% 8% 4% 3% 100% Saver respondents 6% 35% 5% 4% 4% 30% 2% 6% 5% 3% 100% The high proportion of business travel leads us to look in further detail at who is paying for nonpeak tickets. and a slightly lower proportion travelling on business.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 34 4.3.3. First. Nevertheless. than respondents overall. as we would expect.2 Most respondents are employed – over 75% are employed in some way. 45% of passengers overall do not pay for the ticket themselves. But the proportion of passengers not paying is still high. there is a high proportion of passengers travelling on business.4 Main Occupation Overall Work full time Self-employed Homemaker Work part-time Student Retired Other Total 4. a higher proportion of passengers using Saver tickets are paying themselves than on average.regular commuting On business/in connection with work To/from school/college/university Personal business Sport or entertainment Visiting friends/relatives at their home Shopping Holiday/short break Other social/recreational Other Total 4. .3 57% 11% 2% 7% 12% 9% 1% 100% Saver respondents 54% 11% 1% 8% 15% 10% 1% 100% Looking at journey purposes. There is a higher proportion of Saver fare users visiting friends and relatives. For Saver fare users. Table 4. compared to the average.5 Journey Purpose Overall To/from work . the journey purpose profiles are remarkably similar for Saver fares. and 36% of Saver fare passengers.

we can look at the split of journey purpose by income.3. Unsurprisingly. and that employers and employees are taking advantage of cheaper fare alternatives in non-peak times. business.5 £35K 13% 5% 5% 4% 5% 3% 7% 3% 7% 49% 100% Over £35K 12% 14% 3% 9% 4% 4% 5% 4% 18% 27% 100% Holiday/short break On business/in connection with work Other Other social/recreational Personal business Shopping Sport or entertainment To/from school/college/university To/from work . These results show that the traditional correspondence is changing.3.8 : Journey purpose by income band Under £17.3. 4.5 .5K 10% 10% 7% 6% 2% 5% 4% 9% 4% 44% 100% £17. This leaves 32% of Advance purchase tickets being bought by an employer. more. leisure and commuting roughly corresponding to full.7 To try and isolate how many of these lower income respondents may be particularly vulnerable (i. or someone else – a surprisingly high proportion.7 Income splits for passengers paying for their own tickets Paying Respondents themselves overall Under £17.5 55% 15% 5% 19% 5% 100% Saver respondents 64% 17% 5% 12% 5% 100% The picture is similar to Savers for Advance purchase tickets. making non-discretionary trips). proportionally. given the traditional view of journey purposes.6 Table 4.regular commuting Visiting friends/relatives at their home Total .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 35 Table 4.£35K Over £35K Total 33% 27% 41% 100% 23% 23% 54% 100% 4.5K £17. of those respondents who did pay for their own ticket. Table 4.e.6 Who paid/will pay for ticket Overall I have paid myself I have paid myself. were from were from lower income household than average. reduced and season ticket types. but will claim the cost from my employer I am self-employed and the ticket will be a business expense My employer bought the ticket Other Total 4. with 68% of Standard Advance tickets being purchased by individuals.

e. Table 4. not all APEX tickets permit railcard reductions. Looking specifically at railcard use. The evidence we have from this survey does not suggest that Saver fare users have particularly different characteristics to other ticket types.3. Third. the profiles are fairly consistent across all income bands. more passengers are using railcards with Saver tickets than overall – with a much larger proportion using a Young Persons card with Saver tickets than overall. .9 Did you use a railcard? All respondents No Young persons Senior Family Other Total 78% 11% 8% 2% 1% 100% Saver respondents 68% 18% 11% 2% 1% 100% 4. because the fares are already significantly discounted. This suggests one of three things.10 The use of railcards is particularly interesting to note in relation to how passengers choose ticket types.3.3. it reduces their price sensitivity. but this will account for a substantial number of lower income respondents paying for their own tickets. Senior/Young person railcard users have a higher need for the flexibility offered by a Saver ticket over an APEX than average.3. which would be surprising. and there is some evidence to suggest that this is large enough to make them less likely than other users to use APEX products. Perhaps what is most noticeable is the high proportion of student trips in the low income bands.11 One area of concern to the DfT is whether changes to Saver fare regulation would particularly impact on vulnerable users.10 – Railcard user buy ticket type Choice of ticket for railcard users Advance Open No railcard 32% 31% Senior Railcard 28% 16% Young Person's Railcard 16% 20% Total 30% 29% Saver 37% 57% 64% 42% Total 100% 100% 100% 100% Conclusions on income/socio-economic section 4.8 Again. Other economically vulnerable passengers (senior and young people) already have access to substantial ticket discounts. although there are differences that one would expect – for example a higher proportion of long distance commuting amongst high income respondents. Second.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 36 4. Passengers who are particularly price sensitive are likely to already be taking advantage of much cheaper APEX fares. Again not surprising. Table 4. First. Although these passengers are vulnerable by virtue of their low income. they also have access to reduced ticket prices through a young persons railcard. The following table shows that Senior and Young person railcard users are much more likely to use Saver tickets than respondents overall.9 4. Railcards provide a sufficiently large discount to allow these passengers to access the flexibility of a Saver ticket – i.

. to try and get a feel for how passengers perceived the relative costs of their alternatives. 16% were students. The following table shows the alternative modes that respondents stated that they would use if the rail service they were on. Table 4. Alternatively respondents may have assumed that they only discovered the service was unavailable on arrival at the station. open or season) which might give a better indication of the proportion of Saver users who would consider alternative modes if a Saver ticket was not available.11 Alternative options that customers perceived available to them Short Term Car Coach/Bus Would not have made this journey Plane Train Other 4.3.12: Ticket Type (Outward Leg) for ‘Would Not Travel’ respondents Short Term 25% 37% 17% 6% 3% 12% 100% Long Term 21% 45% 23% 3% 1% 7% 100% 4.15 Of these Saver users who would not travel in the long term. Unfortunately.3.3. but this falls to 10% if rail was unavailable in the longer term. 45% of people who would not travel if their rail service (eg London to Edinburgh via York) was not available in the longer term were using Saver tickets. it is difficult to gain a clear view of what passengers estimate in their costs for something as complicated as total travel costs. This is slightly higher than expected by the overall proportion of Saver users (between 37% .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 37 Availability of alternatives 4.39%). because this was not how the question was presented to respondents. The passengers who would not travel if rail was unavailable in the long term are more likely to be genuinely captive to rail. Responses to the cost of alternative were too erratic to gain any useful information from. It is difficult to tell how many of these passengers would have chosen a different ticket before an alternative mode. Table 4.12 The second thing to look at is the availability of feasible alternatives for passengers who are currently using Saver tickets. We did include a question asking which tickets respondents would have chosen if their chosen ticket was not available.14 Open Saver Advance First Advance Season Other Total 4. by which time it was too late to make alternative arrangements. and 14% were travelling on business. was unavailable for a short period or long term.3. if their particular service was not available. 6% chose ‘other’ (rather than advance. 43% were visiting friends and relatives. Possible explanations for the high proportion who stated that they would not travel if their rail service was unavailable in the short term include the possibility that they are making discretionary journeys that can be re-arranged for another day.13 22% 12% 41% 10% 13% 1% Long Term 30% 25% 10% 7% 15% 12% % Difference 8% 13% -31% -3% 2% 11% The questionnaire did ask passengers to quantify the total cost of alternative means. For Saver ticket respondents. 40% of passengers would not travel if the rail service was unavailable for a short period.

3.1 4. Conclusions on alternatives available 4. Almost 40% of respondents were making a day trip with a further third staying away for one or two nights. will be small. including some business passengers.14 Length of Stay Overall Day trip 1 or 2 nights away 3 or 4 nights away five or more nights away Total 38% 34% 17% 11% 100% Saver respondents 33% 39% 15% 13% 100% . but could reduce the amount they pay if the fare structure offered a 'super off-peak ticket'. Conventional wisdom states that high income travellers are more likely to use car and less likely to coach/bus than lower income travellers. ticket type or time of day.500 £17.13 Alternative Mode in longer term All Under £17.3. This is not borne out for existing long-distance rail travel in this survey. or chose to plan their journey ahead.4 4.4. Curiously. and the extent to which respondents could. Saver users were slightly less likely to be making a day trip than overall. data on what alternative mode would be chosen if a particular rail service were not available does not give us a picture of what passengers would do if their fares were to rise materially.5K Car 30% 39% Coach/Bus 25% 16% Plane 7% 6% Other train 15% 7% Other 12% 12% Don’t make this journey 1% 0% often Would not have made 10% 20% this journey Lower income bands are household income less than £17. it is likely that the number with no alternative either in terms of mode. Although we do not have conclusive proof that all passengers have alternatives. In the first instance. we would expect them to consider the consequences of choosing an alternative ticket. we find that lower income respondents are more likely than average to consider car an alternative to rail.5 .2 Journey Planning and flexibility The third consideration is to look at the need for flexibility in tickets.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 38 4. A captive passenger may be defined as one who must travel by a specific rail service at a particular time but as a consequence would face by a higher fare. 4. However the proportion which would not travel is twice that for the sample as a whole. high income passengers are more likely than others to use coach. Such timesensitive passengers.17 While confirming that alternative modes are available.16 If we look at alternative modes by income groups. in the event of the train not being available.4. Table 4. Table 4. including an advance purchase ticket. although.3.18 Our evidence shows that many passengers do perceive there to be alternatives to travel. again the differences in profile are not great. would face a higher fare if they had to travel at a peak shoulder time.£35K 36% 21% 6% 13% 12% 0% 12% Over £35K 25% 32% 7% 20% 12% 1% 4% 4.

Faber Maunsell

Saver Fares: Differentiation and Potential Deregulation

39

4.4.3

Of those who are staying away more than one night from home, unsurprisingly most knew well in advance about their trip.

Table 4.15 No of days in advance know about trip vs length of trip Length of Trip Days in Advance Less than 1 day 1-3 days 4-7 days 1-2 weeks More than 2 weeks All Day 11% 17% 19% 19% 34% 100% More than 1 day 5% 9% 12% 20% 55% 100% All 7% 12% 14% 20% 47% 100%

4.4.4

The following table shows how journey purpose and ticket type are strongly related to the duration of the trip. The leisure advance purchase tickets are much more likely to be for a multi-day trip (17% compared to 1%), whereas the business advance purchase trips for APEX are more evenly split (6% compared to 5%). This table shows clearly that the leisure market is much more likely to be a multi-day trip across all ticket types. For business trips, the split for multi-day trips is even for Advance and Saver fares, but dominated by open tickets, when a trip is being carried out in a day. Table 4.16: Market splits by ticket type and trip length Day trip More than 1 day Business 5% 6% Advance Leisure 1% 17% Business 14% 6% Open Leisure 2% 7% Business 8% 8% Saver Leisure 6% 20% Total 36% 64% Total 11% 18% 19% 10% 16% 26% 100%

4.4.5

The majority of people (69%) caught the train that they intended. This is an important finding, because it shows strong awareness of journey planning. Looking at this finding split by ticket type, we find that planning is strong across all ticket types, but the more flexible tickets allow a higher degree of flexibility. What this demonstrates is that passengers behave according to the flexibility of their ticket. This table does not provide us with any insight into whether those who are travelling on a Saver ticket (particularly) could have been less flexible in their train choice. This is why we require further information from the SP part of the survey. What is somewhat surprising is the number of APEX ticket users who ‘just turned up’. Considering the difference in cost of an APEX ticket to a standard fare, this suggests there are some passengers who are buying APEX tickets but perhaps can claim the cost of a standard single in the event of a change of plan. What is also not available from the questionnaire is the proportion of these responses affected by train–related delays. If we assume at least some passengers were not on their chosen train because of train delays, the proportion of passengers intending to catch a specific train is higher than stated across all ticket types.

Faber Maunsell

Saver Fares: Differentiation and Potential Deregulation

40

Table 4.17: Did you plan to catch THIS particular train? Open Saver Yes No planned other No turned up All 50% 16% 34% 100% 68% 13% 18% 100%

Apex 88% 7% 6% 100%

Total 69% 12% 19% 100%

4.4.6

To get a better feel for whether Saver passengers particularly could be more flexible in their planning, we can look at how many days in advance they knew of their trip. For well over 60% of journeys, the trip was planned over a week in advance. Again the Saver responses are slightly, but not much different from the overall results. Table 4.18 How many days before travelling did you know that you were going to make this journey Overall Saver respondents Less than 1 day 1-3 days 4-7 days 1-2 weeks More than 2 weeks Total 7% 12% 14% 20% 48% 100% 8% 13% 17% 20% 42% 100%

4.4.7

When asked about the time of day of travel, the difference between Saver respondents and overall results do start to differ. Saver respondents seem to know less in advance than average about their time of travel. We cannot tell from this table whether this is because respondents had decided to use a Saver ticket, and therefore knew they were able to retain flexibility around their time, or how many genuinely could not have been less flexible. In practice, it is probably a combination of both.

Table 4.19 How many days before travelling did you know the TIME OF DAY at which you would like to travel? OUTWARD Overall Saver respondents Less than 1 day 1-3 days 4-7 days 1-2 weeks More than 2 weeks Total 18% 16% 13% 15% 38% 100% 21% 20% 17% 15% 28% 100%

4.4.8

For the return journey, the time of day of travel is slightly less known in advance, and Saver responses tend to knowing less far in advance about their time of return.

Faber Maunsell

Saver Fares: Differentiation and Potential Deregulation

41

Table 4.20 How many days before travelling did you know the TIME OF DAY at which you would like to travel? RETURN Overall Saver respondents Less than 1 day 1-3 days 4-7 days 1-2 weeks More than 2 weeks Total 23% 13% 13% 14% 37% 100% 25% 17% 16% 15% 27% 100%

4.4.9

Where differences do show between Saver respondents and the average picture is how long in advance that saver tickets are purchased. Saver users are more likely than average to purchase on the day of travel. Again, this is likely to reflect both a necessary and discretionary preference for flexibility. Table 4.21 How long before your OUTWARD journey was your ticket purchased? Saver Overall respondents On day of travel 1-3 days before travel 4-7 days before travel 1-2 weeks before travel More than 2 weeks before travel Total 36% 15% 12% 13% 24% 100% 46% 17% 13% 13% 13% 100%

Conclusions on journey planning and flexibility 4.4.10 The survey results show that most passengers do know in advance about their trip and that they exercise a high degree of planning and intent about the train they use. Open tickets are much more likely to be used for business day trips. The dominant market segments are daytrip business using open ticket and multi-trip leisure APEX and Saver ticket users. Although Saver fares users are less likely to know or plan than journey in advance than other users, but the differences are not so marked to suggest that Saver users are substantially different from other passengers – they may just be opportunistically flexible – particularly as we know a higher than average proportion of Saver users are railcard users.

4.5 4.5.1

Access to Sales Channels Our final consideration is the extent to which passengers have access to sales channels. In the past, ticket reservation systems were slightly onerous, involving a phone call and the postal service, or a trip to an agent, or to the station. Nowadays access to tickets is extremely convenient, if you have internet access and a credit/debit card. Furthermore, they have changed rapidly over the last few years. In this section we see whether Saver fare users differ in their access to sales channels than other average passengers.

even though they did not frequently use it. . Some passengers may have done this for convenience (to obtain a seat reservation. at least some must still use the internet infrequently. Since telephone ownership is close to 100% in the UK and most people have a payment card of some sort. so we can conclude that their propensity to buy at the station is related more to choice than opportunity. and we would expect it to increase further for the long-distance market. Saver ticket users do not differ from the average respondent in their internet usage. Interestingly. or to avoid queues at the station). most respondents had access to at least two booking channels for advance purchase. Saver fare users do not differ from passengers overall in their access to the internet.5.5. The wording of the internet question was ‘Do you frequently use the internet?’ Of the 9% who said they did not. 29% of respondents bought their tickets by internet with a further 8% using tele-sales.the proportion of internet and tele-sales is already substantial.23 Ticket Purchase Location Overall At Station Internet Bought by Others Tele-sales Other Total 45% 29% 12% 8% 6% 100% Saver respondents 59% 23% 7% 8% 3% 100% 4. Although fewer Saver fare users bought their fare over the internet. a reasonably high proportion did (23%).3 When it comes to ticket purchasing. This suggests that the major barriers to advance purchase are likely to be lack of familiarity with internet and tele-sales and concerns about card security.5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 42 4. but it is likely that at least some may have been searching for the cheapest ticket. 64% bought their ticket at the station.2 Over 90% of respondents have frequent access to the Internet. elsewhere No Internet Note: Multiple responses permitted 74% 44% 7% 9% 6% Saver respondents 73% 40% 7% 9% 7% 4. at home with broadband Yes. Conclusions on access to sales channels 4. but found no availability for Advance purchase fares at the time they wanted to travel.5 Although the most popular way to purchase a ticket continues to be at the station – particularly for saver fares . We would expect this proportion to continue to increase over time as passengers become more familiar with the internet systems.4 Of those passengers who responded that they did not regularly use the internet. and 11% bought it over the telephone. other than going to a station.5. 7% of people still bought their ticket via the internet.22 Frequent Internet Usage Overall Yes. Table 4. Table 4. at work Yes at home (not broadband) Yes.

We conclude that rail passengers have higher incomes than the national average.6 Conclusions from questionnaire responses In this chapter we have looked at the characteristics of survey respondents. A reasonably large proportion of passengers can plan their journeys in advance.6.2 4. to see if we can identify whether Saver fare users are likely to be vulnerable in the following respects: • • • • Income and socioeconomic data A lack of alternative modes Need for flexibility in departure times Lack of access to booking channels other than at stations 4.3 Table 4.6. There does not seem to be evidence from this research that suggests that there are vulnerable captive Saver fares users. Saver and Advance offer passengers a varying levels of price. .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 43 4. most did have an alternative and that small proportion would be expected to have no ticket choice or alternative mode. and Saver respondents in particular. such as fare. and many take advantage of APEX tickets. There will be some ‘captured’ Saver fare users. Price sensitive passengers are likely to prefer the cheapest tickets.4 In conclusion.1 The results of the survey show that the characteristics of Saver fare users are not significantly different from passengers as a whole.6. or already have access to a discounted fare using a railcard. these are already available to them. flexibility and the need to purchase in advance. Open. availability and time of travel. passengers are already taking advantage of mix and match ticket opportunities to give themselves the best deal. For those passengers with lower incomes. At the moment 18% of respondents are using two single tickets. We would expect this to increase for the following reasons: • • • • Single leg pricing (changing the relative value of return Saver fares compared to others) High internet access Evidence of intention to plan by rail users (intending to catch the train they choose) Evidence of the ability to plan journey within the timeframe for purchasing tickets in advance 4. A reasonably large proportion of these are travelling on business (and therefore not paying for their ticket). The impression from both the questionnaire and the stated preference questionnaire is that there are a number of complicated decisions that a passenger makes in relation to their journey all of which combine with the supply side issues. the picture is that Saver fare users are not very different from passengers as a whole.6. The three main ticket products.24 What ticket(s) are you using for your journey? (B3a) Percent Single Return Two Singles Total 11% 70% 18% 100% 4. Given that only 6% of passengers are making single trips. and with sufficient planning. The results of the survey indicate that the picture of how passengers choose tickets will change with single leg pricing.

4. that APEX ticket availability (or lack of it) meant that passengers were using more expensive tickets that they would have preferred.6 4. There is a small suggestion the discount offered by railcards might be biasing railcard users to use Saver tickets. This is also suggested from the numbers of Saver ticket users who have purchased their tickets in advance. Passengers themselves have found it difficult to answer some of the questions (for example.6. show how difficult it is to isolate the combined influences affecting a particular individual’s trip choices. We also have some indication (from the comments in the survey as to why respondents were not travelling at their ideal time). This is a complex area of research. For price sensitive passengers.6. or the ideal time of departure). the responses we have had from both parts of the questionnaire. the Saver product is the next best ticket if APEX quotas have been used. but not comprehensive insight.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 44 4.7 .5 It appears that passengers are already making trade-offs between the different ticket products so that they optimise their choice for their particular journey. The relationship between APEX availability and Saver fare level will be an important consideration in deciding Saver fare policy. the cost of alternative modes. of which this study provides a helpful. Finally.6. where otherwise we might expect them to choose APEX tickets in the same way as average users.

Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 45 .

that the minimum value of Rho Squared varies with the proportion of individuals choosing each alternative.96. which is comparable with the R2 statistic in a regression model. ASC advance.1. The T Ratio indicates whether a variable adds to the explanation of the model.6 . The goodness of fit of the model is given by the Rho Squared statistic.1. The meaning of these parameters is discussed further below.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 46 5 Modelling 5. for example the fare parameter shows the change in utility resulting from a £1 change in fare. d. c.3 ADVANCEfare + c ADVANCEdisptime + dADVANCEadvpurch + ASCADVANCE UWouldNotTravel = Where: Ut1 Faret1 DispTimet1 ASC saver ASC advance ASCWNT = = = = = = ASCWNT the Utility of using ticket type 1 the Fare level of ticket type1 the displacement time of ticket type 1 Alternative Specific Constant for Saver compared with Open Alternative Specific Constant for Advance compared with Open Alternative Specific Constant for Would Not Travel b. one for each ticket type alternative: Uopen Usaver advance = = = b* b* b* OPENfare SAVERfare + c SAVERdisptime + ASCSAVER 5.1 5.1. and ASCWNT are parameters to be estimated.1. Relative valuations described in money units are also shown. It has been shown .25 with equal shares. However. the interpretation of what constitutes a high goodness of fit is less clear with Rho Squared. Associated with each parameter is a T Ratio which is the ratio of the standard error and the parameter estimate.2 The SP modelling attempts to estimate utility functions of the general form below: Ut1= ASCsaver + ASC advance + b Faret1 + c DispTimet1 + d Adv Purchase A utility function shows how important different variables are in determining the overall utility/disutility of different alternatives.1 Stated Preference Results and Robustness Introduction 5.5 5. For example a model estimated with market shares of 90%/10% giving a Rho squared of 0. Each parameter measures how utility changes given a unit change in that variable. These models are described in terms of parameter estimates. 5. Within our SP models we have four utility functions. The standard error indicates the accuracy with which the coefficient is estimated.4 5.55 would be much weaker than a model with a Rho Square of 0.1. In order to be 95% confident that the parameter is not zero the minimum value of the T Ratio is 1.1. T Ratios and Rho Squared statistics. ASC saver. Model Interpretation In the following sections a number of SP models are outlined for both the Pilot and Main survey.Ortuzar and Willumsen(1994) . The parameter estimates indicate the change in utility arising from a unit change in the variable concerned.

business meeting or catch a plane. Approximately 230 questionnaires were handed out. The survey was conducted on the following four trains. 5. This expectation was borne out by the results of the Pilot study which gave a value of 10p per minute for early and 20p per minute for late displacement time. reducing the number of observations to 788. e. commuting. However. each containing 9 scenarios. Faber Maunsell professional staff administered the survey. 5.e.1 can be regarded as good.g. There were two versions of each questionnaire A and B.05 .10 5. In some cases. longer journeys and overnight stays there is likely to be less of a difference between early and late displacement time.40 or higher.1.10 is generally regarded as good. than a later one. the perception of early and late time has often been viewed in the context of outward travel where there is commonly a need to be at the destination at a fixed time and there is a penalty of being late.1. On all trains saver fares were valid: Manchester to London departing 0845 London to Manchester departing 1235 Manchester to London departing 1445 London to Manchester departing 1835 5.449 SP choices.0. whereas other choice situations can be modelled better.1. outward and return leg journeys. This gave 161x 9 =1.12 .1.1. For the pilot survey people who did not vary their choice of ticket throughout (so called non-traders) were excluded.11 Separate questionnaires/designs were produced for each train surveyed. In the case of inter-urban travel with a mixture of purposes. The resulting model is shown in Table 5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 47 5. Generally it is hypothesised that people would rather get an earlier train than their ideal departure time.9 The Pilot survey was carried out on Tuesday 27th February. From our experience of developing this type of model over a large number of years a Rho Square of 0. low values (0.10) can be considered to be good. do they accord with expectations/other evidence? The goodness of fit of the model Does the model fit the data well? The interpretation of Rho Squared has already been discussed and interpreting Rho Squared is not straightforward. i. although 16% of these (31) did not complete the SP section. the smaller the number of alternatives. This was about 37% of the sample. giving Rho Squared values of 0. However.96)? Are the relative valuations sensible? For example are the values of time sensible. Travellers making a journey between London and Manchester or vice versa were asked to complete the questionnaire and leave it on their seat to be collected.8 Manual makes it clear that interpretation of Rho Square can be difficult: It is difficult to give 'rules of thumb' for comparing Rho Squared values between differing models. For the Pilot model displacement time was broken down into early and late displacement time. In general. Pilot Survey 5.1. Experience is the only reliable guide. the higher the Rho Squared that can be expected.7 The following extract from the ALOGIT V3. the simpler the choice that is to be explained.8 In assessing the robustness of a model the following need to be taken into account: The absolute and relative size and significance of the parameter values Firstly are the parameters of the expected sign? Are they significantly different from zero (T Ratio > 1. a value of 0.1. 192 completed questionnaires were collected.

1.8 -2. given the very short time available in the programme schedule between the Pilot and main surveys. We were aware that the complexity of the exercise might affect the response rate.1 -1. One change made was to increase the price of Saver fares in order to encourage people to choose the ‘would not travel’ alternative.1.005 -0. Given the Pilot survey seemed to work well we decided to base the main survey SP designs on the Pilot design and these were developed according to the same principles.063 T-ratio -5.4 -6.12 788 0. All the parameters are of the expected sign and their relative values seem sensible and intuitively correct.1. and Advance tickets which were always cheaper. we expected that with a bigger sample it would be significant.12 more – all other things being equal .14 Not sig Not quite sig Main Survey Design Issues 5.2 -0. Although the ASC for Advance over Open was not significantly different from zero (the T Ratio was 1.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 48 5.12 means that people would be prepared to pay £16. However given the reasonable model developed from the pilot data and the short timescale available it was decided to continue the main survey based on the approach and designs used for the Pilot survey.010 -0. It did not seem sensible to have Saver fares cheaper than Advance tickets. the survey design was reviewed. In order to try to reduce the proportion of non-traders we restricted the availability of Saver and Advance tickets so that every respondent was presented with at least one scenario which would not allow them to pick a train at the time before or after the train they were travelling on.1.744 9. As in the pilot a different design was produced for each flow which again was split into an A and B version in order to present the respondent with nine rather than eighteen scenarios.to have the flexibility of being able to use more than one train. The high relative proportion of non-traders was an issue and this was a consideration for the main survey.1. Fieldwork Issues 5.4 5. Nevertheless only 16% of those respondents in the Pilot study did not complete the SP part of the questionnaire.041 -0.046 -0.17 .59 88.1 Pilot SP Model results Fare DisplacementTime (Early) DisplacementTime (Late) ADV Purchase ASC (Saver) ASC (Advance) VOT (DTE) pence per min VOT (DTL) pence per min VOT (Avd Purch) pence per day VOT (ASC Saver) Pence cf Open VOT (ASC Advance) Pence cf Open N Rho Squared Estimate -0.81 20.96).13 The value of advance purchase was 88p per day and the alternative specific constant (ASC) for Saver over Open was effectively zero. Producing this many designs and questionnaires was an challenge. Overall 139 flows were covered. Table 5. There were therefore 378 questionnaire variants.4 against a minimum of 1. 5.16 5.028 -0.1. Although Saver fares were higher than in the pilot they were still bounded by Open tickets which were always more expensive.15 After the Pilot study had been undertaken.18 The main survey covered a larger number of flows than the pilot survey.04 0.61 16.4 -2. The value of £16.

given the time constraints the survey was generally carried out well.650 questionnaires were printed. Punching and Data Checking 5. Data Coding.1. 6059 questionnaires were handed out and 1790 were returned. there were some minor logistical difficulties.110 SP observations given that each respondent was offered 9 scenarios.1.23 Table 5. The questionnaires were coded and punched. defined as odd. A problem with survey logistics meant that some questionnaires ended up in the wrong place so a handful of trains were not covered. For example.1. 5.1. In order to check the quality of the data inputting.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 49 5. However. This is equivalent to 9 x 1790 = 16. This check also revealed that there were a number of questionnaires which contained illogical choices. if the respondent’s ideal time was 1000 and they had chosen Saver.2 shows an audit trail of the number of cases rejected from the SP data.19 In total 9. This rationality test was carried forward to the computerised data checking phase. Fieldwork took place on : Friday 16th March Sunday 17th March Monday 18th March Tuesday 19th March The survey schedule is shown in Appendix 2.2 SP data Main SP Interviews complete No SP No ideal time Odd choices rejected from manual check Excluded Therefore: Main Survey SP Data Pilot Survey SP Data Total SP Data Not Odd by Rationality Test Observations 16110 5276 1963 174 7413 Interviews 1790 586 218 19 824 33% 12% 1% 46% 8697 1266 9963 7312 966 141 1107 812 54% 100% 73% . and later excluded see Table 5. we defined as illogical anyone choosing the 0900 saver ticket. 5. there were problems with co-operation on some Virgin trains particularly on the Friday and Sunday. a small proportion of the data (approximately 200 questionnaires) were re-punched to check accuracy and few problems were discovered (less than 5% contained errors). For example. This represents a response rate of 30%.22 5. For their chosen ticket respondents had not picked the train closest to their ideal time.20 With a survey of this size organised in the short time available.21 During the survey.2 Table 5.1. and they were being offered a Saver ticket on both a 0930 and 0900 train. In total 1790 questionnaires were completed.

it can be shown that if all individuals have all alternatives available and if the model has a complete set of mode specific constants an unbiased model may be obtained just by correcting the constants according to the following expression: Ka=Ku.27 5. equivalent to 812 questionnaires. Traderonly SP models were tested. keeping non-traders in the model is valid.1. and it is acceptable to develop models based on the whole dataset which includes non-traders and to apply these models unadjusted (since the captivity of certain people to particular types of tickets is allowed for by the parameters and values derived). The presentation of an ideal departure time was essential for calculating the extent to which a respondent had been displaced from their ideal departure time. A further 1% of respondents were identified as containing irrational choices during the manual data check as described above. it is appropriate to adjust the ASCs in order to reflect actual shares as suggested by Ortuzar and Willumsen (1994): ‘If the model is estimated with information from a sub area or with data from a biased sample. They were: Overall. Given the Pilot survey had worked well we added back in the Pilot SP data. However.1. Non-Traders Some of the respondents did not change their ticket choice throughout the SP experiment and these people are referred to as non-traders. as a result of ticketing considerations.1. Ka – adjusted ASC Ku – unadjusted ASC Ss – Share of Alternative in sample Sp – Share of Alternative in Population References: nd Ortuzar and Willumsen (1994) Modelling Transport 2 edition John Wiley & Sons p216 5. indicating that some people found the exercise too difficult to carry out.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 50 5. However.29 A number of different models were estimated. In which case. Just over half of the SP data from the Main Survey was carried forward.28 Main SP Model Results 5. when the computerised rationality test described above was applied. but they produced relative values of displacement time implausibly higher than the all other data models. Ideal time was a requirement for calculating displacement time. Where the modelled choices are different from actual choices as shown by the relative ticket market shares. no SP exercise was completed. Employers Business v Non Employers Business. So these models were rejected on the grounds that the valuations were implausible. Low Income v High Income.7. This is a known difficulty of SP application. ASC Adjustment One further issue is whether the modal constants (ASCs) needed to be adjusted to take into account actual shares compared with modelled shares. and Short Distance v Long Distance. a sizeable proportion of respondents chose the cheaper tickets (Saver and Advance) throughout the SP exercise.1.1. Usually non-traders are excluded from modelling because they do not tell us much about how they choose between the different tickets.25 5.log (Ss/Sp) Where. A further 12% did not give an ideal travel time.3 to 5.1. In our survey. This may be seen as rational for respondents on a tight budget.26 5. this reduced the number of observations to 7312. Excluding non-traders would have reduced the amount of data for modelling to 3254 observations (362 interviews). These are shown in tables 5. .24 For a third of questionnaires. if they are excluded any application model based on these models should make an allowance to take into account that some people are captive to particular types of ticket.

98 -30. The goodness of fit as shown by the rho squared statistic is on the low side 0.1.02960 5.73230 7.31 5.3 Final Model (ALL) Fare (b) T Ratio Advance Purchase (d) T Ratio Displacement Time ( c ) T Ratio ASC Saver T Ratio ASC Advance T Ratio ASC Would not Travel T Ratio Rho Squared Nobs VALUES Displacement time per min Advance purchase per day ASC Saver compared with Open ASC Advance compared with Open Would Not Travel Constant Adjusted ASC Saver Adjusted ASC Advance ASCA V ASCS £ 0. longer journeys and overnight stays there is likely to be less of a difference between early and late displacement time.1.16 0. commuting. Given the only difference between Saver and Advance is that Savers can be used on any of the trains shown in the SP experiment whereas Advance tickets could only be used on the service booked.89700 31. Unlike the Pilot. the relative value of Advance over Saver is -£9.30 Table 5.4 -0. Therefore.03022 19.87 -9.0 -0.87. Compared with Open.32 .02 and compared with Open Advance is valued at -£36. if it was necessary to buy the advance ticket 3 days in advance.23 -195. The ‘would not travel’ ASC is modelled as an alternative and is valued at £195.26. The overall model gives a value of displacement time of 16p a minute. The value of Advance Purchase is 98p per day.84. However. Consequently.03070 7312 5.14 -27. we can assumed that people would be willing to pay £9. we were unable to distinguish between early and late displacement time separately therefore they have been combined into one single displacement time value.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 51 5.20 -24. but the parameter values are plausible compared with other evidence. In the case of inter-urban travel with a mixture of purposes.1.g. this penalty would be valued at £2. and the price of the Open Ticket was £195 the model would predict an equal split between the two alternatives. business meeting or catch a plane. we conclude that there is no bias in displacement time when considered for long distance inter-urban trips. The SP data here includes both outward and return trips.0 -0. outward and return leg journeys.94.3 -0.03.00480 9.7 0.3 shows the overall model. This means that if there was the choice between an Open ticket and ‘would not travel’. e. Saver is valued at -£27.91270 12. All the parameters are of the expected sign including the ASCs for Saver and Advance which have been adjusted using the Ortuzar and Williamson method described in 5.02 -36. Table 5. So.84 in order avoid having to stick with a particular train. the perception of early and late time has often been viewed in the context of outward travel where there is commonly a need to be at the destination at a fixed time and there is a penalty of being late.84 Parameter -0.5 -5.1.

However.02880 3856 .10. The value of being restricted to one particular train as revealed by the difference between the ASCs for Advance and Saver is £20 for business compared with around £1.1.25400 11.04031 4.05900 29.1 0.4 Final Model (Employers Business v Non Employers Business) Non Business Business Fare (b) T Ratio Advance Purchase (d) T Ratio Displacement Time ( c ) T Ratio ASC Saver T Ratio ASC Advance T Ratio ASC Would not Travel T Ratio Rho Squared Nobs -0. We would therefore not have captured some of the high value peak business travellers in our sample which could explain why our values are lower.6 -0.36 Table 5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 52 Note.53510 10. The goodness of fit for the business model is higher than that for the non business model but both are on the low side.99.33 Table 5.1.8 -0.1 as below Ut1= ASCsaver + ASC advance + b Faret1 + c DispTimet1 + d Adv Purchase Ut1 Faret1 DispTimet1 ASC saver ASC advance ASCWNT = = = = = = the Utility of using ticket type 1 the Fare level of ticket type1 the displacement time of ticket type 1 Alternative Specific Constant for Saver compared with Open Alternative Specific Constant for Advance compared with Open Alternative Specific Constant for Would Not Travel b. ASC saver. The value of ‘would not travel’ is higher for business (£232) compared to non business (£140). and ASCWNT are the estimated parameters Business v Non Business 5.10 and for leisure between 61p and £1.4 shows the models split by journey purpose – Business v Non Business (there were a low proportion of commuters in the data set). ITS calculated a value for business between £3.03496 15.1.35 5.15 and non business value slightly lower at £1. ITS included peak trains in their survey. c. For Advance Purchase our business value per day is £1.00269 3. The respective values found by ITS were between 50p and 71p for business and between 5p and 21p for leisure.03982 5.08 and £7.0 -5.00000 0.01253 10. whereas we only surveyed off peak trains on which savers tickets were available. d.3 0.7 0.30 for non business.5 -1. Displacement time is valued at 36p a minute for business compared with 7p a minute for non business. Our non-business values are comparable to the values obtained by ITS but our business values are lower.1 -8. ASC advance.1.1.1 -0. All the parameters are of the expected sign and as expected people on business have much higher values of all parameters than other respondents.34 5.61700 11.05500 3456 -0.2 -0. 5.03620 26. The parameter values in the above table are used in the utility equation from 5.10000 18.6 -0. the absolute and relative parameter values are sensible and accord well with other evidence.6 -1.

69 -139.1.05092 0. All the parameters are of the expected sign.36 0. They value the restriction imposed of having to use a particular train at £16.10 ASC Saver compared with Open -35. c.5 shows models split by income.03321 10.77 for low income travellers.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 53 £ £ VALUES Displacement time per min 0.15900 -11.64 -14.3 -5.14).26500 -9. The cut off between low and high was a household income of £35.38 .10500 17.04240 3564 5.03797 4.70 ASCA V ASCS -20.64100 -23.25 0.02750 2694 High Income -0.35 Note.75 Adjusted ASC Saver -35.01020 -9.3 0.8 -0.07 Advance purchase per day 1.0 -6.6 -0.43890 3. High Income travellers value displacement time at 32p a minute compared with 6p a minute for low income travellers.00212 2.15 1.8 -0.2 -0. One would expect high income travellers to have higher values than low income travellers and this is borne out for all the parameters except Advance Purchase.1.000.3 -0.5 -0. ASC saver.87 -14.0 0.1. and ASCWNT are the estimated parameters Income Models 5. Table 5.34 Adjusted ASC Advance -55.5 Final Model (Low v High Income) Low Income Fare (b) T Ratio Advance Purchase (d) T Ratio Displacement Time ( c ) T Ratio ASC Saver T Ratio ASC Advance T Ratio ASC Would not Travel T Ratio Rho Squared Nobs -0.1 as below Ut1= ASCsaver + ASC advance + b Faret1 + c DispTimet1 + d Adv Purchase Ut1 Faret1 DispTimet1 ASC saver ASC advance ASCWNT = = = = = = the Utility of using ticket type 1 the Fare level of ticket type1 the displacement time of ticket type 1 Alternative Specific Constant for Saver compared with Open Alternative Specific Constant for Advance compared with Open Alternative Specific Constant for Would Not Travel b. They seem to value advance purchase slightly lower than low income travellers (94p v £1.4 -1.03015 -3.00 Would Not Travel Constant -231.49 -13.15 -1.58 compared with £3.78 ASC Advance compared with Open -46. d. ASC advance.37 Table 5.03205 -14. The parameter values in the above table are used in the utility equation from 5.0 -1.

065 -49. c.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 54 VALUES Displacement time per min Advance purchase per day ASC Saver compared with Open ASC Advance compared with Open Would Not Travel Constant Adjusted ASC Saver Adjusted ASC Advance ASCA V ASCS £ 0.94 -36.586 Note.16 -39.1.14 -13.22 -1. ASC saver.77 £ 0.32 0. d.06 1.651 -16.47 -207.72 -14. and ASCWNT are the estimated parameters .53 -153. ASC advance.21 -33.1 as below Ut1= ASCsaver + ASC advance + b Faret1 + c DispTimet1 + d Adv Purchase Ut1 Faret1 DispTimet1 ASC saver ASC advance ASCWNT = = = = = = the Utility of using ticket type 1 the Fare level of ticket type1 the displacement time of ticket type 1 Alternative Specific Constant for Saver compared with Open Alternative Specific Constant for Advance compared with Open Alternative Specific Constant for Would Not Travel b. The parameter values in the above table are used in the utility equation frpm 5.20 -17.97 -3.

61 -28.32130 -2.1. ITS work which took short distance to be for Southend. The valuations for short distance journeys are higher than for long distance journeys. and ASCWNT are the estimated parameters .00564 -5. Upminster. Very few short distance journeys were contained in our survey data.11 5.03274 -3.3 -1.11 -198.8 0.98700 -22. Table 5.00700 -9.6 Final Model (Short v Long Journey Distance) Short Fare (b) T Ratio Advance Purchase (d) T Ratio Displacement Time ( c ) T Ratio ASC Saver T Ratio ASC Advance T Ratio ASC Would not Travel T Ratio Rho Squared N observations VALUES Displacement time per min Advance purchase per day ASC Saver compared with Open ASC Advance compared with Open Would Not Travel Constant Adjusted ASC Saver Adjusted ASC Advance ASCA V ASCS £ 0.43 -37.7 -0. c.14 0.1 as below Ut1= ASCsaver + ASC advance + b Faret1 + c DispTimet1 + d Adv Purchase Ut1 Faret1 DispTimet1 ASC saver ASC advance ASCWNT = = = = = = the Utility of using ticket type 1 the Fare level of ticket type1 the displacement time of ticket type 1 Alternative Specific Constant for Saver compared with Open Alternative Specific Constant for Advance compared with Open Alternative Specific Constant for Would Not Travel b.92050 -6.05556 0.26 -10.4 -3. The threshold between short and long distance was just over 2 hours (125minutes). which would generally be an hour or less).1.19 Long -0.0 0.57 -26.21 -28.39 Table 5.02138 -2.00481 -7.52 -27.2 0.58 3. therefore breaking down the models in this way may not be sensible. ASC advance.03525 -16. Milton Keynes and Peterborough to London.9 -0.00970 -3. although our definition of short distance is longer than that for other studies (e. d.73 -409. ASC saver.g.01010 2875 £ 0.03610 4437 Note.37 -33.4 -0.4 -6.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 55 Distance Bands 5.62 -9.97100 -16.2 -0. The parameter values in the above table are used in the utility equation from 5.6 shows models split by distance band.34 -0.86 -17.2 -0.9 -0.

75.2 -0.40 Table 5.07 -25.18290 -1.02530 2801 £ 0.4 -5. d. Day trippers value displacement time at 25p per minute compared with 12 per minute for non day trippers.4 0.12 1.0 -0.02520 4511 Note.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 56 Trip Duration 5.3 0.90 -45.71 -17.03015 -4.00595 -7. The parameter values in the above table are used in the utility equation from 5.5 -1.8 -7.57 compared with £11.4 -0.7 Final Model (Day Trip v Non-Day Trip) Daytrip Fare (b) T Ratio Advance Purchase (d) T Ratio Displacement Time ( c ) T Ratio ASC Saver T Ratio ASC Advance T Ratio ASC Would not Travel T Ratio Rho Squared Nobs VALUES Displacement time per min Advance purchase per day ASC Saver compared with Open ASC Advance compared with Open Would Not Travel Constant Adjusted ASC Saver Adjusted ASC Advance ASCA V ASCS £ 0.00335 -4. They value advance purchase at £1.32 -32.7 shows models split by whether people were making day trips or longer trips.47 -179.7 -0.51 -11.07300 -23.1.1 as below Ut1= ASCsaver + ASC advance + b Faret1 + c DispTimet1 + d Adv Purchase Ut1 Faret1 DispTimet1 ASC saver ASC advance ASCWNT = = = = = = the Utility of using ticket type 1 the Fare level of ticket type1 the displacement time of ticket type 1 Alternative Specific Constant for Saver compared with Open Alternative Specific Constant for Advance compared with Open Alternative Specific Constant for Would Not Travel b.0 -0. ASC advance.14 -34.0 -0.04236 -4.75 Non Daytrip -0.07.02829 -13.02351 -9. Table 5.50000 -12. c.36 -6. ASC saver. and ASCWNT are the estimated parameters .71750 -7.76 -44.1.80 compared with £1.01 -17. they are likely to have higher values than people making journeys of longer than a day.75000 -6.06600 -6.80 -31. Because people making day trips are likely to be more time constrained. which is found to be the case with our data. The value of having to use a particular train is £17.3 -0.57 -0.34 -319.25 1.

This is particularly the case for the journey purpose models.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 57 Conclusions 5. This may be down to the definition of short distance which in our models is anything up to 125 minutes. . When split by distance ITS found longer distance values to be higher than shorter distance. whereas the ITS definition was for much shorter trips. whereas our shorter distance values were higher than our longer distance values.1.41 The models developed during this study have parameters which are of the expected sign and their relative values are sensible and compared well (where they are comparable) with the previous results obtained by ITS. We found the values from people making day trips to be higher than those making non-day trips and generally high income travellers had higher values than low income travellers.

2.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 58 5. and An application of the stated preference model that comprises a multi-nomial model at the lower nest to predict ticket choice and time of travel and a higher level that applies purpose based elasticities.1 Development of policy application model In order to test a range of alternative ticketing scenarios a policy model has been developed that can produce estimates of demand and revenue by ticket type for the following changes in the rail market: Introduction of a shoulder saver fare on specified trains. However. Ideally the stated preference work would have been extended to incorporate non rail users in the markets of interest in order to gain a fuller understanding of the potential generation effects of reduced fares in the off peak. and their ability to integrate information and outputs. The three approaches used in the policy model are as follows: 5.2 5. the in built desire for total flexibility that certain travellers have. however.2 In order to assess the implications of what are very complex fare options with the traveller having to choose a ticket type based on the fares available on trains that are either at or close to their ideal time of travel we require a model that involves parameters relating to fares.2. the cost of displacement time if travelling at a time other than ideal time. The survey work undertaken in the study was designed to provide passenger valuations of each of these elements such that a modelling system could be established that would provide the allocation of passenger demand over the day based on the respective fare and availability options presented.3 5. The model converts the changes in displacement time and advance purchase effects into fare equivalents using the weights derived from the Stated Preference models. to weighted fare changes to reflect generation and suppression of rail trips. . The stated preference model can also provide estimates of the distribution of the existing rail demand as the fares and ticket availabilities change in each scenario. Consequently. provide the estimates of the importance of the individual parameters which can be imported into existing approaches as described in PDFH. This enables an element of sensitivity testing of the policy model outputs as the three versions will produce varying results. could be compared. They do. Changes in saver fares and the trains on which they are available. 5. the stated preference models described in the preceding sections cannot on their own be directly applied to predict the overall change in rail demand from alternative fare strategies. and the impedance of having to book in advance. One aspect of the stated preference work that did not result in sensible outcomes was the attempt to understand the extent to which changes in fares and availability would lead to a reduction in travel by rail. and diversion factors to derive the cross elasticities of demand. NERA elasticity model – this develops from first principles an application to predict choice of train time within a ticket type based on assumed own price elasticities. this was not a practical option due to the tight study timescale as identifying and then interviewing a suitable sample would have significantly extended the study time frame which was not an option. Changes in the level of APEX fares and the quotas available on each of the trains for which APEX are provided.4 PDFH based model – uses PDFH recommended conditional elasticity values and diversion factors derived from market research survey data. drawn directly from PDFH. In view of the above issues three separate approaches have been adopted in developing the policy model so that the relative merits of each approach. The model has been enhanced so as to reflect the effect of APEX availability and quotas on choice of ticket type.2.2. and Changes in open fare levels.

the availability of each ticket type. F are changes in fare and f denotes own or cross elasticity. . Denoting conditional elasticities by ticket type: Eopen = foo + fos + foa Esaver = fso + fss + fsa Eapex = fao + fas + faa This means that if the conditional elasticities are known then it is sufficient to know two of the three terms on the right hand side of the equation since we can use the diversion factor relationship to help deduce the own and cross elasticities.2. the base ticket type shares. The diversion factor relationship is: Fos = -fss * (Ss / So) * Dso Where S denotes the demand by ticket type and D the diversion factor between ticket types.6 The PDFH approach considers changes in demand by ticket type as fares change differentially through the use of own and cross elasticities for demand. by ticket type.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 59 PDFH Based Model Application 5. the conditional elasticities are available and the diversion factors defined. Therefore the above equations can be solved to provide the own and cross elasticity values for each specific corridor or flow group where the base market shares by ticket type are known. The application used in the policy model has some differences to the basic PDFH model which are possible due to the additional information provided by the stated preference models and the profiling and inputs on ticket type availability. and the diversion factor method. The basic functional form for predicting the demand for open. This approach is embedded in the application model developed for this study. The PDFH approach is to use a set of conditional elasticities by ticket type. into a fare change equivalent. and 7 pence per minute for other travellers. The valuations of displacement time used from the SP data are 36 pence per minute for business travellers.5 The PDFH based model uses the recommended approach in the PDFH manual for the derivation of changes in ticket type usage as fares change at differential levels for each ticket type and the overall availability of each ticket type is amended. The overall weighting is then derived based on the proportion business and other travel that is entered into the model. 5.2. The enhancements to the PDFH approach are: To use the stated preference estimates of the value of displacement time to convert the calculated change in displacement time to alternative ticket types as availability by train changes. In terms of f the subscripts identify whether it is an own or cross elasticity and the ticket type combinations. and a set of diversion factors that dictate in what proportion each user of a specified ticket is likely to transfer to alternative tickets or to not travel by rail. and Calculate from input quotas by train the base and test scenario availability for each ticket type. As recommended own and cross elasticities are not provided by ticket type in PDFH they have to be determined by using the conditional elasticities provided in PDFH. saver and APEX fares is: VOpen = AOpen * FOpen^(foo) * FSaver^(fos) * FApex^(foa) VSaver = ASaver * FSaver^(fso) * FSaver^(fss) * FApex^(fsa) VApex = AApexr * FApex^(fao) * FSaver^(fas) * FApex^(faa) Where A are constants representing the underlying different sizes of each market.

This is particularly important in deriving the overall effect of changes in fare structures on the generation/suppression of rail trips.4 1. in accordance with general advice on the application of the PDFH 5. the proportion of the market that is business and other trips.6 2.8 2 2.50% of the base fares. Indeed the limited information on diversion factors provided in PDFH reflects these factors with differing values depending on the strength of air competition for example. The diversion factors are influenced by factors such as the strength of competing modes (air/coach/car). It is stressed therefore that wherever possible some market research is undertaken for the flow groups and corridors under examination to ascertain the appropriate diversion factors for use in the model. This is shown in the following plot that shows all ticket type demand changes for different proportional changes in saver fares for a typical corridor.7 In applying the approach detailed in PDFH it is very important to understand the constraints that exist within the mathematical approach adopted in that it is applicable for marginal changes in fare only and where very significant increases or decreases in fare are being considered then considerable care has to be taken in interpreting the results. diversion factors and base market shares the change in demand predicted will begin to increase as saver fare is increased once fare increases are very large.8 3 Proportional Change in Saver Fare Illustration of the necessity of limiting fares changes tested to no more than +/.4 2.6 1.8 1 1.6 0. . and the length of the journey. 140 Percentage Change in Overall Demand 130 120 110 100 90 80 70 60 0. in excess of 100% increase.2.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 60 5.2 2.50% of the base fares.2 1.8 It is extremely important to note that the outputs from the PDFH based model are directly influenced by the assumptions on the diversion factors between ticket types and would not use rail. PDFH emphasises that diversion factors are not constant across all rail trips and indeed they can be flow group and corridor specific. The functional form has the property that under certain combinations of conditional elasticity. In this study therefore the range of fare changes tested has been restricted to +/.2.

This is the diversion factor between time i and j . where i ≠ j it can be shown that the cross price elasticity of demand in time period i with respect to price in period j is denoted by ε ij : ε ij = This can be equivalently written as: dqi p j ⋅ dp j qi ε ij = Note that dq j p j q j dq i dq i p j ⋅ = ⋅ ⋅ ⋅ dp j q i dp j q j q i dq j dq j p j ⋅ = ε jj . This corresponds to the dq j proportion of changed demand in time period j that diverts to time period i .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 61 NERA Elasticity Model Application 5. which is the own price elasticity of demand at time j . . Also denote dp j q j dqi by Dij .10 In the case where there are only two time periods: i and j .9 The NERA based elasticity model adopts the same basic principles that are recommended in PDFH for the modelling of mode or ticket type switching and applies them to the derivation of time switching effects within a ticket type as described below.2. it can be shown that: [cross-elasticity of demand for mode i with respect to the price of mode j] [own price elasticity of demand for mode j] [Journeys on mode j] [‘diversion factor’ = proportion of changed demand on mode j that diverts to mode i] = X ___________________ [Journeys on mode i] X The equivalent equation for time-shifting is: [cross-elasticity of demand for travel in time period i with respect to the fare in time period j] [own price elasticity of demand for travel in time period j] [journeys in time period j / journeys in time period i] [‘diversion factor’ = proportion of changed demand in time period j that diverts to time period i] = X X This can be shown mathematically as: dqi p j dq j p j q j dqi ⋅ = ⋅ ⋅ ⋅ dp j qi dp j q j qi dq j 5. In the modal choice case.2.

The diversion factors for the time shifts are then derived using the stated preference model application which provides estimates of the probability of changing to alternative time periods based on the displacement time from the current time period to the alternatives and the differential fare values.11 Suppose that the price of travelling changes in both time periods i and j .2. where all prices are allowed to change.2. within a particular ticket type where fares are differential by time period. . the new level of demand is the same as the last equation above. The percentage change in demand at time period i is given by the following: ⎛ dq p dq p q dq ⎞ %∆qi = %∆pi ⋅ ε ii + %∆p jε ij = %∆pi ⋅ ε ii + %∆p j ⎜ i ⋅ j = j ⋅ j ⋅ j ⋅ i ⎟ ⎜ dp q ⎟ ⎝ j i dp j q j qi dq j ⎠ ^ ~ Denoting the new level of demand at time i by qi and the old level of demand by qi . the new level of demand is then: qi = qi + qi ⋅ %∆qi In general. which are detailed later. This is an example of how the different theoretical approaches are integrated to make best use of the varying strengths of each aspect of the data and techniques available. and draws on the same input values. for fifteen time periods.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 62 5. There are two effects on demand at time period i .12 The above described approach to the use of own and cross elasticities with appropriate diversion factors to derive the change in time of travel. but with % ∆qi defined as: ^ ~ ~ %∆qi = ∑ %∆p j ⋅ ε ij j =1 15 5. has been extended to incorporate the potential change in demand across ticket types using a further set of cross elasticities and diversion factors between the ticket types within each time period. The model has been implemented within the same overall framework as the PDFH based model described earlier.

The own price fare elasticities used to derive generation/suppression effects at the upper nest of the stated preference model application were -0.25 for leisure travellers. which have already been scaled to reflect the ticket type shares in the overall rail passenger survey data. 5. to predict the distribution of rail demand over ticket type and time of day. The model uses a multi-nomial approach that calculates for each ideal time of travel the utilities of travelling by each ticket type in each time period in relation to the ideal time of travel. the model takes into account the displacement time to other time periods. Hence. The default purpose-based elasticity values used in the model are taken from PDFH. The SP model is undertaken for business and other travel separately as these were shown to be distinct market sectors with significantly different values of each of the core trip attributes. The probability of each option is then derived and the rail demand for that ideal travel time period distributed across the ticket types and time periods.2.14 Policy Model Inputs 5.13 The final modelling approach used in the policy model is the application of the stated preference model parameters. This is repeated for each ideal travel time period and then the demands are accumulated across all periods to give an overall profile that reflects the availability and price of ticket types.2. As the stated preference model does not directly enable the generation/suppression of rail trips to be derived it is supplemented by the use of purpose based elasticities applied to the weighted fare changes once all time of travel changes have taken place. the fare differential. and the value of flexibility and advance purchase.65 for business travellers and -1.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 63 Stated Preference Model Application 5.15 The policy model requires a number of parameters which dictate the responsiveness of the model to changes in inputs as shown by the following: Inputs PDFH Model Approach NERA Elasticity Stated Preference Yes Ideal time of travel profiles Yes – to enable displacement times to be derived Yes – based on PDFH Yes Conditional elasticities by ticket type Own price elasticities by ticket type Diversion factors between ticket types Own price elasticities by purpose Yes Yes Yes Yes – to provide rail generation and suppression effects Displacement time valuations Yes Stated preference model parameters .2.

Week day / Saturday / Sunday. Week day / Saturday / Sunday. .3 The origin / destination categories are: to London. 5. we needed to develop an ‘ideal departure time demand profile’. in either direction. For example. Length of trip away. The profiles are VERY OLD (dating from early 1990s we believe). to Blue Stations.8 Critique of Data Sources ORCATS Sample quality Good – detailed profiles. Good – fully consistent Compatibility with modelling framework Review of ORCATS Data 5.3. All blue stations have a “blue” index. Bradford Foster Square = 617. which is equivalent to ideal departure time demand. Concept of “blue” stations potentially useful for applying our model to flows more widely. then flows between the station pair are classified as “other”.3. This is potentially useful because the non-London profiles could be said to represent departure time demand in the absence of ticketing restrictions. journeys from Bradford FS to Liverpool are classified as “to blue”. If the ratio of blueness for one station to its pair exceeds 10/7. as the ratio of Liverpool to Bradford is 6716 / 617 = 10. Flow. Blue stations are non-London attractor stations. But data are for “ideal” departure times. and sample not representative: eg only surveyed trains for which Savers were valid. ORCATS profiles differentiate by: Ticket type. and make up less than 10 per cent of all stations on the network.9 which is greater than 10/7 = 1. If the ratio is less. Liverpool = 6716. Our survey data differentiate ideal departure times by. Data Sources 5. with the exception of the stated preference model data which has already been described in some detail.3. For example. rather than those observed. then the flow is classified as “to blue”. this may be an advantage: less distorted by ticket restrictions or crowding No disaggregation by journey purpose or duration of trip away.3. 5.3. 5. Outward / return legs. season) for flows to and from London: it seems to assume that there are no ticket restrictions for reduced fares for other flows. from London.2 We have two sources of data: ORCATS departure time profiles and survey data on “ideal” departure times. Origin / destination type. Survey Data Smaller – potential for market segmentation is limited.4 Journey purpose. amongst other things: 5.3. Manchester = 11584. reduced.4.5 Strengths and weaknesses of both data sources are set out in the following table Table 5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 64 The source of the above are described in the following sections. from Blue Stations and Other.1 Deriving ‘ideal departure time’ So that we can represent passengers’ preference for departure time within a model that is looking at different time availability for tickets.3 5. Journey duration.6 ORCATS profiles only appear to vary by ticket type (full.

Some weekday example profiles are shown below. because they are profiles in the absence of ticketing restrictions. 101-140 Minutes. 5. 181-270 Minutes. profiles from blue stations have marked evening peaks. These peaks are less pronounced for longer trips. 21-50 Minutes.7 As we would expect. 141-180 Minutes.3. 51-100 Minutes. 271-360 Minutes.8 . the survey data show markedly different profiles according to whether the journeys are for the outward or return leg of a trip. where the journey bands are: 0-20 Minutes. and the very long flows allow for sleeper train usage. and >360 Minutes.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 65 “Blue” profiles are more relevant to “ideal” departure times than to / ex London profiles.3. profiles to “other” stations (ie not strongly directional) have morning and evening peaks. Review of Survey Data As we would expect. Profiles vary markedly according to journey length. This is illustrated below. and whether they are day trips or more than one night is spent away. Departure Demand Profiles From Blue Stations by Journey Duration 50 45 40 Passenger Demand 35 30 25 20 15 10 5 0 0000 0500 1000 1500 Time of Day 2000 2500 3000 51-100 Minutes 101-140 Minutes 141-180 Minutes 181-270 Minutes 271-360 Minutes >360 Minutes 5. profiles to blue stations have marked morning peaks.

broadly split by {day trips. possessions typically last all day on Sunday). Recommended approach 5.3.18 Restricted Sunday trading hours affect travel times.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 66 Ideal Departure time by Trip Duration and Journey Leg 40% 35% Proportion of Passengers 30% 25% 20% 15% 10% 5% 0% 04:48 Day. particularly on Sundays. RTN 07:12 09:36 12:00 14:24 16:48 19:12 Ideal Departure Time 5. we do not think that they are significant factors for the flows concerned for the following reasons. return leg} x {shorter trips. completed in 2007. We use weekday profiles and argue that they are equally valid for weekends: weekends have more stay-away trips. and therefore there is no a priori reason to suppose that the ideal demand profiles within individual days are affected by this. Given the limited sample size (less than 2. OUT Stay. but while this may strongly influence the timing of short distance shopping trips. it will have a very limited effect on the long distance flows for which Saver fares are regulated. Passenger perceptions that services may be disrupted due to engineering works (as well as actual service disruption – but we are concerned with modelling normal service patterns) are thought by the industry to affect demand on Sundays in particular.000 observations). the profiles are not fully representative: for example a disproportionately large number of return trips from London were sampled. and once this is taken into account it is not clear that there is a systematic difference between journey purposes. stay away trips} x {outward leg. Reduced service levels. For the purposes of this study. compared to only 19 per cent of leisure trips. and asymmetric patterns of out and return legs. 18 .9 It is less clear that the profiles vary systematically by journey purpose: 57 per cent of business trips were day trips. It may be argued that weekend trips have different ideal departure times from week day trips for a variety of reasons. OUT Day. In addition.3. NERA examined and modelled the demand effects of possessions as part of a study for ORR. the survey data are not readily amenable to analysis by journey length.10 For the policy modelling work we will use up to eight profiles of ideal demand. and service limitations would be taken account separately through measuring divergence from ideal time. certainly influence travel patterns: but the profiles concern ideal departure times rather than actual departure times. RTN Stay. These are based on ORCATS profiles with some minor adjustments. but this is taken into account in the proposed disaggregation. It is also clear that the profiles are not strictly “ideal departure times” because they reflect Saver restrictions to some degree (there is a dip in demand during the middle of the evening peak because we did not survey at this time). on efficient engineering access. But they are not thought to affect the time of day that people travel (under the current regime. longer saver trips}.

The following table sets out proposals for the source of ideal departure profiles. shorter trips be those that can be made to arrive by mid to late morning (for which morning peak travel may be convenient). But although more people will travel late on Friday and Saturday evenings than during the remainder of the week. whereas longer trips require half a day travel. so in practice this means trips of at least 30 minutes length. 40 Orig / Dest To blue station From blue station To blue station From blue station To blue station From blue station To blue station From blue station Duratio n 51 to 100 minutes 51 to 100 minutes 101 to 140 minutes 101 to 140 minutes 141 to 180 minutes 141 to 180 minutes 271 to 360 minutes 271 to 360 minutes Take profile to 14:00 hours. longer trips to be those of more than 100 minutes. thereafter assume zero Take profile from 14:00 hours only. shorter trips to be those of less than 100 minutes. .3. and illustrated in the following charts. For day trips. because they are not working the next day. For stay-away trips. prior to that assume zero Take profile to 14:00 hours. We are only interested in trips that have regulated Saver fares. prior to that assume zero No adjustment Modification “Ideal Departure Time” Profile Description Leg Outward Time away Day trip Journey length <100 minutes <100 minutes >100 minutes >100 minutes <180 minutes <180 minutes >180 minutes >180 minutes Return Day trip 38 Outward Day trip 18 Return Day trip 18 Outward Stay away 19 Return Stay away 19 No adjustment Outward Stay away 6 Remove trips departing between 20:00 and 02:00 hours (sleeper service) Remove trips departing between 20:00 and 02:00 hours (sleeper service) Return Stay away 6 The proposed profiles for ideal departure times are set out in the table below. Table 5.9 Description of ORCATS Profiles Used ORCATS Profile Profil e Ref. thereafter assume zero Take profile from 14:00 hours only. this will tend to affect shorter distance trips rather than Saver flows.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 67 Friday and Saturday evenings are characterised by late night leisure activities. 5.11 The journey length varies according to whether they are day or stay-away trips.

Out. Long 1% 5% 15% 18% 17% 15% 11% 9% 8% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% Day Trip. Long 0% 0% 0% 0% 0% 0% 0% 0% 0% 8% 13% 18% 20% 21% 11% 5% 3% 1% 0% Stay Away Short 0% 1% 6% 9% 9% 8% 7% 6% 5% 6% 7% 9% 10% 8% 5% 3% 1% 1% 0% Stay Away Long 0% 1% 3% 5% 8% 11% 11% 9% 8% 8% 8% 8% 7% 5% 4% 2% 0% 0% 0% Outward Trips 30% Proportion of passengers 25% 20% 15% 10% 5% 0% 00:00 Day. Out. Rtn. Rtn.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 68 Table 5. short Stay. short Day. Short 0% 0% 2% 7% 17% 28% 20% 16% 8% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% Day Trip. long 04:48 09:36 Hours 14:24 19:12 00:00 . long Stay. Short 0% 0% 0% 0% 0% 0% 0% 0% 0% 5% 7% 11% 19% 25% 17% 8% 4% 2% 1% Day Trip.10 Proposed Profiles for Ideal Departure Times Hour Ending 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 Day Trip.

Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 69 Return Trips 30% Proportion of passengers 25% 20% 15% 10% 5% 0% 00:00 Day.4. 5. It is therefore very important to bear this factor in mind when interpreting results for the models in that the outputs are very much influenced by the core judgements on diversion factors. These values can be amended by the user in the policy model if more up to date information is available. It is important to highlight that the diversion factors are strongly influenced by the proportion of business travel that exists in different markets as these trips are in themselves not elastic and as such they will.2 for saver tickets and -1. These show how much variance can occur in the diversion factors due to flow and corridor specific factors. In the following tables the diversion factors are quoted as from a current ticket type to another but the effects of the diversion factors is bidirectional when applied as a decrease in fare would cause a diversion to the ticket type where the fare changed in a similar manner. short Stay.1 Elasticity Values and Diversion Factors between Ticket Types The conditional elasticities adopted as the default in the policy model are derived from PDFH and are -0. are less likely not to travel by rail. -1. The default diversion factors for choice between ticket types used in the model are based on the same principles as those used in PDFH but modified to reflect information from the rail passenger surveys undertaken in this study and previous work by ITS that looked at diversion factors by flow group as part of the Strategic Fares Model development which show alternative tickets considered. The following tables show diversion factors for two corridors and in one corridor for flow groups over 200 miles and the other for flow groups of 100-200 miles. short Day. long 04:48 09:36 Hours 14:24 19:12 00:00 5. pay higher fares and have a greater propensity to switch to full or first tickets as saver fares approach full priced tickets.4 5. and two.4.85 for open tickets. long Stay. as well as the conditional elasticity values.5 for APEX tickets.2 . one.

22 0.3 The issue of the business and other traveller composition and their effect on diversion factors is a particularly important one as one would obtain markedly different responses if the market was dominated by business travel.12 Diversion Factors : Corridor A flows 100 . Trains on which each fare type is available in the base and test scenarios. Consequently.13 Diversion Factors : Corridor B flows 100 .200 Miles Ticket Type (From) Full Saver APEX First 0.43 5.4.29 Table 5.42 APEX 0. APEX fare levels by time of travel.38 APEX 0.40 0.11 0.09 0.23 0.11 Full 0.00 0. shoulder saver and saver fare levels. The proportion of seats available for APEX purchase on each train.36 0. as opposed to one dominated by leisure travel. and Alternative values for the default elasticity values and diversion factors between ticket types can also be tested by the user.1 Overall Functionality of Policy Model The policy model has an input sheet format that enables the user to specify: Open.40 0. The model enables the user to construct a complex set of scenarios where the availability of a ticket in any time period can be specified its fare and where necessary the proportion of tickets of that type which are available.38 Not Rail 0.21 0.19 0.19 Full 0.23 0.08 Ticket Type (To) Saver 0.13 0.18 0.05 Full 0.5.08 0.5 5.11 Diversion Factors : Corridor A flows Over 200 Miles Ticket Type (From) Full Saver APEX First 0.33 0. 5.38 0.200 Miles Ticket Type From) Full Saver APEX First 0.24 Ticket Type (To) Saver 0.38 0.38 Not Rail 0.39 0.25 APEX 0.15 0.56 Table 5.21 0.39 Not Rail 0. .00 Ticket Type (To) Saver 0. there would be significant differences in the diversion factors for movements to/from London (with high business use) which the above figures represent.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 70 Table 5. and to/from South West (with high leisure use).19 0. The base demand by ticket type either over the whole analysis period or by time of day if the information is available.

2 The input screen for the policy model is shown in Figure *. The primary outputs reported in the model are the base and test demand and revenue estimates by ticket type. 5.5. The main limitation of the NERA and SP models is their current inability to model the effect of changes in APEX quota availability. The policy model currently presents outputs for each of the three modelling approaches that have been described in the earlier sections but it is necessary to understand that there are some limitations with the NERA and SP model approaches.* and provides a relatively clear description of the input data that can be entered and how alternative policy scenarios can be constructed.5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 71 5.3 . availability and quota by each ticket type to be modelled effectively. The PDFH based model is therefore the only approach that enables all combinations of fare.

aber Maunsell Saver Fares: Differentiation and Potential Deregulation 72 .

5 Corridor A – an Intercity corridor of approximately 270 miles.3 5.4 5.30 £10. The model has been enhanced so as to reflect the effect of APEX availability and quotas on choice of ticket type.6. Where one of the models is not applicable to the test being undertaken the relevant cells in the tables are shaded grey and the results omitted. it is unlikely that a policy model could also function as an effective yield management tool. are indicative. Base Inputs Two corridors have been tested: 5. and Corridor B – an Intercity corridor of approximately 190 miles. after 19:00 10:00 to 16:00 .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 73 5. Three different models have been used: PDFH based model – uses PDFH recommended conditional elasticity values by ticket type and diversion factors derived from market research survey data. and An application of the stated preference model that comprises a multi-nomial model at the lower nest to predict ticket choice and time of travel and a higher level that applies purpose based elasticities. after 19:00 10:00 to 16:00 Corridor B £109.70 £67. but others.2 While the models are separate they have linkages so that features of one model that can enhance the outputs of another are used to provide the best estimate of demand and revenue over a range of scenarios.6. drawn directly from PDFH. At the moment the PDFH model best incorporates the availability of APEX tickets .although theoretically. The model converts the changes in displacement time and advance purchase effects into fare equivalents using the weights derived from the Stated Preference models.6 The following base fare inputs have been used: Ticket Availability Full Fare Saver Fare Shoulder Saver Fare APEX Fare Saver Availability APEX Availability Corridor A £107.6.50 £45. In terms of the scenario tests described below the NERA and SP model currently have the limitation that they cannot take account of the quota of APEX tickets by train.00 09:00 – 15:00. although they do reflect the availability of APEX on a train but with the assumption that they are as fully available on those trains as Open and Savers.50 10:00 to 16:00.6.6 Policy Tests Introduction 5. and diversion factors to derive the cross elasticities of demand. It is important to note that the two corridors are in some senses notional corridors in that the some of the input assumptions are drawn from actual corridors. NERA elasticity model – this develops from first principles an application to predict choice of train time within a ticket type based on assumed own price elasticities.50 £29. 5.6.75 £44. 5.6.63 £19. No attempt has been made in either case to calibrate to actual base ticket types and hence the description of the model as a policy model to enable alternative assumptions to be tested.1 This note provides the results of policy tests carried out using the Saver Fares application models. though. to weighted fare changes to reflect generation and suppression of rail trips. this functionality could be extended to the other versions if required in the future. In practice. such as fares and ticket types. such as demand by ticket type and purpose splits.

6. Saver 46% and APEX 22%. In reality an APEX quota is available on most trains and there is a range of different APEX prices available each with its own quota. The base shares for the various ticket types used in the model tests are Open 32%.8 .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 74 5.7 It should be noted that this is a considerable simplification of the actual situation.6. An assumption has also been adopted for the corridors that the business / non business split is 44%/56% which is derived from the surveys undertaken during the course of the study. This should be borne in mind when interpreting the results. 5.

-0.05 for commuting. 5.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 75 5. 4. In our survey we had the following responses by journey purpose and ticket type: Full (Open) Commuting Business Leisure Total 24 255 129 408 Reduced (Saver) 37 210 346 593 APEX 27 153 243 423 Total 88 618 718 1424 5.12 For Corridor A and Table 5.11 This produces weighted elasticities for full. and -0. Shoulder Saver plus offpeak reduction As 1.11 and 5. Saver fare + 20% plus increased availability of APEX As 4 plus APEX availability increased by 20% in time periods where available. plus 20% reduction in offpeak saver fare.75. 1000 and 1200 (Corridor B) 5. 3.92 respectively. Saver + 50% Saver fare increased by 50%.6. and leisure respectively.6.10 Recent work by MVA indicated that elasticities should be lower than PDFH recommends.9 Policy Tests Eight discrete policy tests have been undertaken: 1.6. 7. 5. 2. Saver.13 and Table 5. Shoulder Saver A shoulder Saver has been implemented between 0900 and 1000 for Corridor A. and -1. Saver fare + 20% Saver fare increased by 20%. business.12 .70. Reduction in Saver availability Saver tickets no longer available between 0900 and 1100 (Corridor A). 6.92. We have therefore re-calculated the conditional elasticities used within our PDFH-based model using those quoted within MVA’s report and our survey responses.84.6. The MVA report quotes elasticities of -0. Saver + 50% plus increased availability of APEX As 6 plus APEX availability increased by 20% in time periods where available. and APEX of -0. ShoulderSsaver plus offpeak reduction plus increased availability of APEX As 2 plus APEX availability increased by 20% in time periods where available. 8.14 for Corridor B. 1000 and 1100 for Corridor B. -0. Test Results Results of the Policy Test are set out in Tables 5.

4% -10.6% -1.6% -4.11 Corridor A Results: % Change in Demand Between Base and Test PDFH NERA Full Saver APEX Total Full Saver APEX Total Full SP Model + Elasticity (V2) Saver APEX Total 1. Reduction in saver availability Test No Test No 1.5% 1.2% 14.3% 34.3% -3.6% 4. Saver fare + 20% plus increased availability of APEX 6. Saver+ 50% 7.0% 2.9% -57.0% -53. Reduction in saver availability 1.3% -9.8% -10.1% -6.8% -12.4% -12.7% -40.7% -15.3% 15.0% 9.9% -0.5% -8.0% 1.4% -5.3% 3.1% -11.6% -6. Shoulder saver 2.0% 17.3% 12.4% 57. Shoulder saver plus off peak reduction 3. Shoulder saver 2.4% 1.4% 0.9% -35.5% 1.1% -1.5% -33.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 76 Table 5. Saver+ 50% 7.4% 9.1% -11.7% 0.8% 0.0% -8.5% -20.6% -3. Shoulder saver plus off peak reduction plus increased availability of APEX 4.0% 2.4% 15.0% -5.3% -9.2% NERA Total Full Saver APEX Total Full SP Model + Elasticity (V2) Saver APEX Total Saver -12.4% 57.0% 0.9% 5.4% 1.7% -3.2% 12.5% -12.2% -10.9% 8.9% -4.6% -2.7% -28.6% 0.9% .3% 7.6% 21.0% -1.6% -1.3% 53.8% APEX 0.3% 53.2% 6.5% 0.9% 37. Saver+ 50% plus increased availability of APEX 8.7% 15.5% 13.8% 34. Saver fare + 20% plus increased availability of APEX 6.8% 2.1% 2. Saver fare + 20% 5.8% 1.12 Corridor A Results: % Change in Revenue Between Base and Test PDFH Full 1.3% 15.2% -14.4% 12.4% 9.9% 3.8% 3.7% 2.0% 10.4% -5.8% 4.7% 32.0% 9.6% -16.4% 8.1% 2.8% 3.2% -8.2% -10.6% -3.9% 36.9% 2. Shoulder saver plus off peak reduction 3.6% 0.1% 2.0% 5.7% -14. Shoulder saver plus off peak reduction plus increased availability of APEX 4.5% -8.4% -5.0% -29.8% 1.9% -5.7% -40.3% 13.7% 13.8% 4.3% 7.0% -0.7% 21.8% 3. Saver fare + 20% 5.1% -6.0% 0.7% -14. Saver+ 50% plus increased availability of APEX 8.9% 5.8% -10.1% -11.8% -10.6% -20.2% 46.2% 48.6% 3.7% -10.7% -3.2% 6.7% 7.0% Table 5.7% -16.6% 2.6% -6.6% -2.

Reduction in saver availability 2. Saver+ 50% 7.0% -10.7% 22.7% -10.7% 8.7% 7.0% 3.6% 11.7% 22.1% 3.1% -3. Saver fare + 20% 5.4% 8.8% 12.1% 3.0% 0.8% -66.2% 0.2% 2.1% 6.1% 2. Shoulder saver plus off peak reduction plus increased availability of APEX 4.8% 14.4% 53.2% -16.9% -40.2% -16.8% -16.7% -6.9% 2. Shoulder saver plus off peak reduction 3.1% 61.8% 19.4% 3.3% -21. Saver fare + 20% plus increased availability of APEX 6.8% 12.3% -2.3% -5.3% -52.0% 3. Shoulder saver plus off peak reduction 3.3% -2.9% 22. Saver fare + 20% plus increased availability of APEX 6.9% -5.3% 4.0% -23.6% -9.8% 21.7% -0.3% 1.3% 2.0% 3.1% -29.1% 61. Saver+ 50% plus increased availability of APEX 8.3% -21.8% 2.0% -2.4% 12.3% 1.9% -4.6% 26.9% -23.7% 2.0% -5.0% 8.4% 6.1% 3.2% -48.8% 1.3% -10.0% -7.7% -10.8% 14.7% 7.4% 8.0% -5.6% 9. Shoulder saver plus off peak reduction plus increased availability of APEX 4.6% -9.0% . Saver+ 50% plus increased availability of APEX 8.7% -9.13 Corridor B Results: % Change in Demand Between Base and Test PDFH Full Saver APEX Total Full Saver NERA APEX Total Full SP Model + Elasticity (V2) Saver APEX Total Test No 1.2% 2.4% -5.0% 3. Shoulder saver 2.5% -7.7% -0.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 77 Table 5.5% -9.0% 48.4% 9.8% 5. Reduction in saver availability 1.8% -2.5% -3.5% -23.1% -34.2% -5.9% -10.3% 9.6% 4.0% -10.1% 3.3% 0.5% 9.8% -19. Shoulder saver 2.4% 11.8% -6.9% -4.3% 2.5% 22.4% -17.14 Corridor B Results: % Change in Revenue Between Base and Test PDFH NERA Full Saver APEX Total Full Saver APEX Total Full Test No SP Model + Elasticity (V2) Saver APEX Total 2.0% 48.4% 11.8% 1.4% -10.7% -2.1% 19.3% 0.5% -24.6% 0.4% 53.6% -5.1% -16.8% -19.6% -4.6% 8. Saver fare + 20% 5.6% -6.2% -13.8% 2.1% 4. Saver+ 50% 7.0% Table 5.5% -49.9% 22.2% 0.0% 9.6% 26.5% 0.6% 6.0% -5.4% -25.0% 1.8% 21.7% -0.9% 10.

These are summarised below and the discussion uses the PDFH based results as the basis in order to provide clarity and as this is the only approach that covers all the various test scenarios.2% compared to Test 1.5% decrease in revenue.6. was applied to this test and the impact was to marginally reduce the demand and revenue effects. The saver fare reduction has a significant overall effect on passenger numbers and revenue and also the distribution by ticket type.7% increase in demand by corridor but with an attendant decrease in revenue of around 6%.3% increase in demand and a 2. In revenue terms the overall effect is positive with up a 0.13 The first point to note from the results is that in scenarios where the individual models are considered applicable there is a reasonable degree of consistency between the scenario outputs. However. The results of this . Shoulder Saver plus offpeak reduction As 1. plus 20% reduction in off-peak saver fare. and hence in the peak. 5. the revenue would drop by 10.10.4. The introduction of the higher shoulder saver fare means that some travellers will make the decision that the saving to travel at that time rather than their ideal time is no longer worth while and hence will resort to full priced tickets and hence be able to travel at their ideal time. The impact of the change in conditional elasticity values. The overall effect of the test fare strategy is to result in an 12. This provides the change in demand by ticket type by time period as a result of the test fare strategy.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 78 5.6.6. Test 3.4. The combined effect of the shoulder saver and a saver reduction is an increase in demand of 9.1% increase in revenue.4% and 2. Limited evidence from PDFH indicates that at higher average fare per km the elasticity should be increased using an elasticity modifier based on the equations described in section 2.6% to 3.0% by corridor. The test results indicate that the majority of the demand that moves away from the shoulder peak time period travels later (44%) to take advantage of the saver fare and only a small proportion (4%) would travel earlier. The incremental effect of reducing the saver fares by 20% whilst introducing the shoulder saver fare is to significantly increase demand in both corridors. There is also a significant difference in the average open fare per kilometre between the two test corridors and this has a potential impact on the elasticity value to be applied in each case. In terms of the impact of the tests there are differences in the sensitivity of response between Corridors A and B and the primary reasons for this are the differences in fare differentials between Open.14 5. 1000 and 1100 for Corridor B. The distribution of the ticket types is however significantly different with a substantial increase in APEX with these being drawn from both full and saver tickets. Saver and APEX tickets and their availability profiles. The effect of the 20% saver reduction is to increase demand over the Test 1 results by around 7% to 10%. This provides some degree of confidence that the overall results are likely to be robust as three methods provide similar outcomes. A further set of sensitivity tests were undertaken on this test which related to examining the impact of the valuation of the displacement time on demand by ticket type.8.8% and some transfer from saver tickets to full and APEX.6.4% and a consequent reduction in revenue of 1. There are some specific points that need to be highlighted with respect to the individual test results which go some way to explaining the main drivers of the outputs. The effect of the shoulder fare is similar in each corridor with an overall small reduction in rail demand of 1. described in section 5. The incremental effect of the 20% quota increase of APEX tickets compared to Test 2 is a 3. Test 2.3% and 5.6% to 8.15 Test 1.3% .7% and 8.6% to 1. Shoulder saver plus offpeak reduction plus increased availability of APEX As 2 plus APEX availability increased by 20% in time periods where available. This finding is based on an analysis of the SP model application results which predict the probability of using each train. Shoulder Saver A Shoulder Saver implemented between 0900 and 1000 for Corridor A.

the different revenue effects despite similar overall demand changes. Overall the demand for rail reduces by 6% but the revenue increases by 2% to 4%. It should be noted that the nature of the test is such that only small impacts in terms of changes in displacement time exist in the test scenario. Test 6. In corridor B the ratio of APEX fare to saver fare is 0. The overall effect of the saver fare increase with enhanced APEX availability is to reduce demand from the base case by 3% but increase revenue by 1% in corridor B yet reduce revenue by 4% in corridor A. and hence as the saver fare increases the value of delaying their travel time becomes less attractive and hence they may as purchase the flexibility of the full ticket. Saver + 50% Saver fare + 50%. Saver fare + 20% plus increased availability of APEX As 4 plus APEX availability increased by 20% in time periods where available. Saver fare + 20% Saver fare increased by 20%.6 46.0 Displacement Time * 1. The incremental effect of adding 20% to APEX quotas was to add 3% to demand and reduce revenue by 3.2 50. compared to the central test in shoulder saver demand as the reduced displacement time value means more people would divert to other time periods.2% by corridor.3 28. The reason for the differential effects on revenue is the quite different ratio of the saver to APEX fares in the base data between the two corridors. The demand for savers decreases by 40% to 53% by corridor and there are significant increases in full fare demand as saver travellers transfer to full fares as APEX quotas remain unchanged and hence can only accommodate .3 32.7 32.5 Open Saver Shoulder Saver APEX 28.3 3. The transfer from saver to full fares is consistent with the diversion factors adopted for the model run and is a reflection of the relatively high proportion of business travellers who are in the saver market.4 The above shows that decreasing the displacement time value results in a reduction.9 3.22. The 20% increase in saver fare results in just over a 20% reduction in saver demand but with a significant proportion of this moving to other ticket types with a 7% to 8% increase in full fare demand and about a 3% increase in APEX demand.4 48.0 32. This test represents a significant increase in saver fares and the overall demand effects are significant with a 12% to 13% reduction in demand.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 79 test are shown below in absolute changes in demand between three scenarios where the displacement time valuation is increased and decreased by a factor of 50%. The inelastic nature of the business market makes this a reasonable outcome from the policy test. and conversely an increase in displacement time valuation causes an increase in shoulder saver demand and reduction in saver demand as less people divert from the shoulder saver time period.67 whereas in corridor A it is 0. and for whom the ticket is paid for. namely.5 3. Hence. Test 4.4% – 6. Test 5. Ticket Type Displacement Time * 0. the replacement of one time period saver with a shoulder saver. The effects are similar to those noted in test 3 when the incremental effect of APEX quotas is examined.2 28.5 Test 3 Displacement Time * 1.

The inelastic nature of the business market makes this a reasonable outcome from the policy test.7 42. The results of this test are shown below in absolute changes in demand between three scenarios where the displacement time valuation is increased and decreased by a factor of 50%. The significant transfer to full fare from saver fare is consistent with the diversion factors adopted for the model run and is a reflection of the relatively high proportion of business travellers who are in the saver market. A further set of sensitivity tests were undertaken on this test which related to examining the impact of the valuation of the displacement time on demand by ticket type. There is no reason to expect peoples’ propensity to divert to alternatives will differ in terms of the proportion which divert to each alternative. Test 7. 1000 and 1200 (Corridor B). input diversion factors would be different.5 Open Saver APEX 32. as would the conditional elasticity values. Ticket Type Displacement Time * 0.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 80 moderate increased APEX demand. There is no similar evidence on the impact of higher fares on diversion factors. Theoretically it could be hypothesised that the diversion factors would be relatively stable at the higher fares. compared to the central test. PDFH evidence indicates that at higher fares. Clearly if the model was able to separate out effects by the business and leisure markets then quite different impacts would be seen from the two tests.7 The above shows that decreasing the displacement time value results in a 8% increase in saver demand. . Test 8.5 Test 8 Displacement Time * 1. Increasing the APEX availability enables more demand to use APEX in this scenario and consequently the demand impacts are reduced compared to Test 6 and the revenue gain is reduced. The reduction in saver availability reduces demand by 4% . Note that this test does not include a shoulder-peak Saver option. Saver + 50% plus increased availability of APEX As 6 plus APEX availability increased by 20% in time periods where available.2 36. However.3 22. There is a further issue in this test that relates to the stability of the elasticity and diversion factors as fares increase significantly. and conversely an increase in displacement time valuation causes a 7.5 34.1 22. and hence as the saver fare increases the value of delaying their travel time becomes less attractive and hence they may as purchase the flexibility of the full ticket. fare elasticities will be higher. Revenue would increase by 6% to 11% by corridor and again this is a function of the relative levels of saver and full fares in the two corridors in the base case.3 33. the absolute level of diversion away from rail would be expected to increase. and for whom the ticket is paid for.5 39. The reasons for the different responses between the corridors is a direct function of the base fare differential between saver and APEX fares which is much higher in Corridor A. Reduction in Saver availability Saver tickets no longer available between 0900 and 1100 (Corridor A). In this case. as the reduced displacement time value means more people would divert to other time periods rather than not travel by rail.7% by corridor but with marginal changes in overall revenue as Saver passengers transfer to full fare tickets.2 22.4% decrease in saver demand as less people divert to other saver time periods once the saver availability is restricted.0 Displacement Time * 1. Indeed the impact on revenue in corridor A would be a reduction compared to the base of -2%.

. 5. for corridor A in Test 3. For example. some general conclusions could be inferred as follows: overall Saver fares increases would lead to an increase in operator revenue after taking account of impacts on revenue from other ticket types.7 per cent and a revenue increase of 1 per cent.6. and shoulder peak fares combined with either Saver fare reductions in the off-peak. increasing APEX fares by 20 per cent would lead to a demand increase of 5. or increased APEX quotas can be used as a demand management tool with potentially revenue neutral effects.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 81 All the tests could be fine-tuned by adjustments to inputs which result in additional patronage and a flattening of the demand profile.16 For each corridor.

6 Pulling the Evidence Together .

4. 6. These results are reasonably consistent across both the NERA and PDFH approaches and in relation to both corridors tested. To promote social inclusion.2 The policy tests also show that there is potential to increase demand while retaining a broadly revenue neutral position through a combination of shoulder peak pricing and increased APEX availability.2 6. although this may be flow specific and would require careful optimisation of the absolute and relative fare levels. 6. A policy of increasing shoulder peak fares with compensating reductions in the off peak leads to lower revenue and higher ridership than simply increasing shoulder peak fares. we summarise the results of the tests that were undertaken using the policy application model described in Chapter 5 to determine whether operators have significant market power. In Section 6.2.1 6.2. we find that Saver fare increases will in general increase revenue while reducing ridership.1. 6.2 In this chapter we seek to combine existing evidence with the findings of the new research carried out during this study to assess whether these arguments apply to the regulation of Saver fares and whether abolition of Saver fare regulation is desirable. Assessment of Market Power Results of Policy Tests The results of the policy tests reported in chapter 5 suggest: 1.3 6. in so far as high rail prices could restrict the mobility of some sections of society.1. combining one of these strategies with increasing APEX availability leads to an increase in ridership and a reduction in revenue relative to the equivalent test where APEX availability is not increased. The tests which increase shoulder peak fares lead to more modest increases in revenue and reduced ridership. On the basis of the two major routes modelled. The magnitude of the changes do vary depending on the initial fares and the model used. and To ensure affordable tickets are retained in situations where the use of more than one operator’s services is required or where the constraint of being able to use only one operator’s services unduly restricts travel opportunities.3 . This applies both to Test 1 (Shoulder peak fares) and Test 8 (reduced Saver Availability – in effect raising shoulder peak fares to peak levels).2.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 83 6 Pulling the Evidence Together 6.1. We also consider alternatives to full de-regulation. In each case. 3.1 2. We also discuss other factors that may affect operators’ pricing strategy in practice before considering the social inclusion and retention of network benefits issues. It has to be recognised that the policy model results are very much driven by two main factors with regard to their sensitivity in terms of rail trip generation/suppression and allocation between ticket types and these are the conditional 6. This implies that if TOCs were to maximise revenue given the capacity they operate they would increase Saver fares.2.1 Introduction In Chapter 1 we noted three main arguments for fare regulation in the rail industry: To protect consumers in situations where service providers have significant market power and can use this to raise prices significantly. so we can have more confidence in the direction of the changes than in their magnitude when applied nationally.

and the current evidence base on elasticity and diversion factors and it is .1 Would fare levels be considered to be excessive? . the hypothesis could be that demand will be more elastic as a result of the composition of the passenger base at those time periods with lower proportions of business travellers and higher proportions of leisure travellers and long stay trips. and is based on all day travel.6 6. namely excessive fares arising from abuse of a dominant position.2. when and how they are travelling. and to a lesser extent on journey purpose. As the proportional mix changes then the overall ‘elasticity’ by time period could be expected to change and become more elastic towards the middle of the day when business travel is at its lowest as a proportion of overall demand. The current all day ticket type conditional elasticity values by ticket type reflect purpose splits that were inherent in the market base from which the values were derived and over time these may change and as such the elasticity values vary. separately from all the other factors that also influence their decision. We have tried to address this in the stated preference exercise we have undertaken.3.4 The existing elasticity evidence from PDFH on elasticities and diversion factors types is based primarily on ticket type. The former are more inelastic and the latter more elastic.2. in particular. the current journey purposes. More generally.5 6. Existing general evidence on this is extremely limited. the long-distance traveller has many contributory factors into why. they are in reality flow group specific in many cases and are judgmental in nature.in reality – elastic in parts of the deeper off peak. where the market is generally fairly homogenous. As the diversion factors reflect differing levels of competition and are influenced by journey purpose composition and journey length. The diversion factors are an important feature of the model structure and determine the proportion of passengers who will transfer to/from other ticket types or to would not travel by rail if changes are made in fares. It is possible that the finding that demand is inelastic is influenced by the shoulder peak. It has proved difficult to pin down the extent to which passengers have adjusted their time of travel in response to ticket constraints. With regard to the deeper off-peak. 6. this study has highlighted the complexity of the relationships between ticket types and. and we must bear this in mind when offering recommendations. the relationship between time of day trade-offs. the results are by no means definitive. This will influence the degree of certainty that can be placed on conclusions from the work. Increasing the diversion factor for ‘would not use rail’ increases the elasticity for the specific ticket type to which it applies. We consider the implications of this for competition authorities under two headings. If this is the case operators would have an incentive to reduce fares at these times. 6.3 6. It does not directly provide evidence on the elasticities for different time periods.2. Although the study has contributed to the knowledge of time-of day trade-offs. Implications for Competition Authorities if fare levels increase This section of Chapter 6 considers the implications of our findings on the impact of deregulation of Saver fares for the competition authorities. Unlike commuting into London. only by implication if the journey purpose splits are known by time of day can an approximation be made within the period when Savers are currently valid and this would still have weaknesses.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 84 elasticity values and the diversion factors adopted. We have concluded that deregulation of Saver fares could lead TOC’s to increase Saver fares either by means of general increases in Saver fare levels or by means of the introduction of shoulder peak Savers during certain parts of the day. and investigations of future mergers in the rail industry.

Faber Maunsell

Saver Fares: Differentiation and Potential Deregulation

85

6.3.2 6.3.3

In section 2.7 above we described the ORR’s position in regard to fare levels as set out in their policy note on the role of competition law in regulating rail passenger fares.19 Any future complaint about excessive prices that arose in the event that Saver fares rose would obviously be treated by ORR on its merits, and according to the relevant law. However, NERA’s own view, based on the advice that we provided to ORR in the Virgin fares case, is that the ORR would be likely to take the following factors into account in coming to a view on whether an individual fare could be considered excessive for the purposes of competition law:

The circumstances in which longer-distance TOCs now operate (and which were relevant factors taken into account in the Regulator’s decision), in regard to competition in the market for franchises, with common costs spread across different services and in particular across different passengers on the same train, are the same as they were in 2001; and The level of Saver fares relative to the level of Open fares. The overall range of prices available to passengers. Relatively few passengers are ' captive ' ie with no realistic alternatives such as travelling at a different time of day or buying an advance ticket. This is against the background where the railway as a whole far from covers its costs overall; Government pays Network Rail a grant even on passenger franchises which pay Government a premium.

How would future merger investigations be affected?
6.3.4 We set out in section 2.6 of this report how the competition authorities have investigated rail mergers. In any future investigations that took place in an environment where Saver fares had now been deregulated, the investigation would need to take account of both the higher Saver fares now in operation in the market and the fact that, if the merger reduced competition, Saver fare increases would not be constrained by regulation. Therefore, for example, if a longer distance rail franchise were to be awarded to a coach operator with services in the same corridor, then the investigation would need to consider whether the merger might lead to a substantial lessening of competition such that leisure rail fares might increase further.

6.4 6.4.1

Demand Management The policy model simulates the impact of alternative pricing strategies on largely unconstrained demand and revenue. Although the model incorporates ticket availability by time periods and an estimate of APEX quota’s in each time period, it has not been designed to assess capacity constraints. The model implicitly assumes the supply (capacity) is unlimited. This is clearly not the case for rail. Firstly, as in other service industries, the product (in this case a seat on a specific train on a particular day) is produced and consumed simultaneously and cannot be stored. Secondly there are significant and growing constraints on the number of trains that can be operated on the network at certain times of day, while overcrowding is becoming increasingly common. The first of these factors encourages operators to maximise load factors, while the second leads to demand management measures to encourage customers to shift their time of travel to periods when more capacity is available. In order to meet these requirements, operators have developed, and are continuing to enhance, yield management strategies. In order for these strategies to work effectively, operators need to know how many passengers will arrive for a particular train. While Open and Saver ticket users may reserve seats there is no requirement for them to travel on the particular train where these have been made. Accordingly operators have to make allowance for the number of Open and Saver ticket users who may arrive. Since demand can vary significantly and unpredictably from day to day (due to factors such as the weather) this results in a lower average load factor than might be achieved with an all advance purchase strategy, combined with episodes of overcrowding, which leads to significant customer dissatisfaction. A shift from flexible to restricted tickets would enable operators to allocate higher quotas to restricted tickets and thereby help to achieve higher average load factors. Accordingly there is an opportunity cost
19

6.4.2

ORR, Regulating Rail Passenger Fares – The Role of Competition Law. Available at: http://www.railreg.gov.uk/upload/pdf/competition-briefing-note-fares.pdf

Faber Maunsell

Saver Fares: Differentiation and Potential Deregulation

86

associated with flexible tickets. This may justify pricing them somewhere above the point where they become price elastic, as the loss of revenue from this segment is more than compensated by the ability to sell more restricted tickets, albeit at lower fares. 6.4.3 We would expect operators to increase Saver fares in shoulder peak periods to assist them in managing demand more effectively and increasing average load factors, in some situations above the level at which demand becomes elastic. At particularly busy times, advance purchase ticket prices may also rise to maximise revenue within capacity constraints. This may explain why operators have not reduced Saver fares in the way that the Leeds ITS study, discussed in Chapter 2, suggested that they should. Finally, it is important to recognise that much of the revenue benefit from yield management systems comes back to DfT either through increased value when franchises are re-let or through the revenue sharing mechanisms in recent franchise agreements. This may be used either to reduce overall subsidy to the industry or to invest in increased capacity. Social Inclusion An important theme of government policy has been to promote social inclusion for disadvantaged groups. In the case of Saver fares de-regulation, the group who could be significantly disadvantaged if de-regulation led to higher fares are people:

6.4.4

6.5 6.5.1

On low incomes; With no alternative mode Paying their own fares.

6.5.2

If we define low income as those with a household income of less than £17,500, and those with no alternative mode a those who responded ‘other train service’ or ‘would not travel’ to the survey question about long term alternatives for their journey, the proportion of our sample at risk is only 5%. If the definition of low income is increased to those with incomes under £35,000 (which is in fact significantly above the national average), this proportion rises to 9%. In some cases, the fares of people on low incomes are paid by others (apart from employers). The person who pays may also be on a low income, for example another family member. If we include this group, the potentially disadvantaged proportions rise to 6% or 10% depending on the income threshold chosen. Some of these passengers, in practice, would first choose to use an alternative ticket type, including advance purchase. Others may be able to change their time of travel to allow them to access cheaper ticket alternatives which may be available at less busy times. Overall then, we conclude that a fairly small proportion of passengers travelling at times when Saver fares are valid are potentially at risk from deregulation. However, it is important that any changes are communicated effectively to users so that they are aware of the fares that are available. Analysis of the survey data suggests that 56% of respondents with an income of less than £17,500 are either under 26 or over 60 years of age and could make use of Young Persons or Senior Railcards20. Some other respondents may be entitled to a Family Railcard. This demonstrates that a high proportion of lower income passengers do have access, via a railcard, to reduced fares. . Overall 22% of respondents were using a railcard. Network Benefits An important feature of Saver (and Open) fares is that they are inter-available between all operators serving a flow. This is not the case with APEX tickets, which are generally operator specific. This gives rise to two potential areas of concern:

6.5.3

6.5.4

6.5.5

6.6 6.6.1

A loss of affordable through ticket opportunities where the journey requires the use of more than one operator’s services; A loss of choice where more than one operator serves the same flow.

6.6.2

The current fare setting arrangements within the industry make one operator responsible for setting fares for any flow, even where the use of another operator’s services is required. The fare setting operator typically offers their own range of discounted tickets for the full journey. For
20

Average UK household income is ££27,000 (source ONS)

Faber Maunsell

Saver Fares: Differentiation and Potential Deregulation

87

example Tonbridge to Edinburgh fares are set by GNER, even though passengers would also need to use SouthEastern services in the course of their journey. The full range of GNER discounted tickets is available at prices which are typically below the cost of buying separate Tonbridge – London, London Underground and London – Edinburgh fares. There is no reason why Saver de-regulation would change this situation. 6.6.3 There is only one operator serving the majority of long distance Intercity flows. Accordingly, loss of inter-availability is not a significant issue on these routes. There are, however, a number of flows where different operators provide parallel services at present, or will do so in future. Examples include GNER and Virgin between York and Edinburgh; GNER and First Scotrail between Edinburgh and Aberdeen; Virgin and Transpennine Express between Preston and Glasgow/Edinburgh; and Virgin and First Great Western between Taunton and Penzance. Many of the journeys on these routes are relatively short and may be made at shorter notice than longer distance journeys. The absolute difference between Saver and Advance Purchase fares is also likely to be less. It might be thought that there would be scope for the fare setting operator (normally the one with the greatest number of services) to increase Saver fares in the expectation that this would lead to a disproportionate shift to his restricted tickets. This might be the case with walk-up fares, as choosing the operator with the higher frequency maximises departure time choice, but advance purchase passengers are likely to split between operators in much the same proportions as with inter-available tickets. It is understood that franchised operators are not permitted to sell operator specific walk-up fares. If this was allowed, there could be a benefit to some operators in providing them, at the expense of a loss of network benefits to passengers. However, the Passenger Licence and the Ticket and Settlement Agreement in place protect network benefits. Conclusions from the Evidence The evidence presented above indicates that, in the absence of regulation, a revenue maximising operator could increase walk-up fares in at least part of the period when Savers are currently available. However the extent to which fares would rise is uncertain. It is also uncertain, if TOcs were free to choose, whether fares would rise throughout the entire period for Saver fares are currently valid, or whether there are time periods in the deep off-peak where demand is more elastic. The current cap on Saver fares also constrains the level of APEX fares, so, after de-regulation, these might also rise during any times of day where demand is less elastic. In practice, given the uncertainties in the evidence and the risk of adverse customer reaction, operators might be expected to proceed cautiously and change fares in relatively modest increments, while monitoring the effect on ridership and revenue. A move to single-leg pricing could enable passengers to shop around for the cheapest fare on each leg of the journey. There would be benefits in raising shoulder peak walk-up fares to reduce the growing problem of crowding at this time and to encourage a further shift to advance purchase tickets. A higher proportion of APEX tickets would make it easier for operators to manage demand and make best use of available capacity. This has benefits both in terms of revenue to operators and in making best use of capacity. At the moment, APEX availability is limited by the requirement on TOCs to provide sufficient capacity for walk-on fares. However, placing obligations on TOCs to increase the availability of APEX fares would be a further layer of regulation. The social inclusion arguments for Saver fare regulation do not appear to be strong given that only 5% of our respondents had a household income under £17,500, no alternative mode and were paying their own fares. Relaxing these criteria would only increase the proportion that might be considered vulnerable up to 9%. Accordingly Saver fare regulation is not an effective tool for encouraging social inclusion given that it has significant costs (in increased franchise subsidies or reduced premia) and most of the benefit does not flow to disadvantaged groups. Over half low income respondents were eligible for either a Young Person’s or Senior railcard. Overall, all ‘captive’ passengers i.e. those who have no real choice but to pay any higher fare are a small proportion of the total.

6.7 6.7.1

6.7.2

6.7.3

6.7.4

Appendix A .List of trains surveyed .

00 12.Bristol TM Paddington ./New St.London Train 18.00 09.45 09.Edinburgh Edinburgh .05 16.Manchester Manchester .Edinburgh Kings Cross .00 12.Bristol London .00 18.Edinburgh Kings Cross .Edinburgh Edinburgh .52 10.Kings Cross Edinburgh .Kings Cross Edinburgh .30 09.03 09.Birmingham Edinburgh .Stockport/Manchester Manchester/Stockport .30 15.00 10.00 10.Bath London .05 16.50 12.York Newcastle .London London .Leeds Birmingham .Solihull.Edinburgh Kings Cross .Birmingham Euston .London London .London Stoke .Manchester Euston . Birmingham New St/Int .London Birmingham .05 18.Birmingham Edinburgh .Paddington Euston .Kings Cross Edinburgh .00 16.London Newcastle .Bristol TM Bristol TM .Bristol TM Paddington .Edinburgh Birmingham .Euston Manchester .London Stoke .35 18. Solihull .35 18.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 89 Appendix A .Euston Paddington .Manchester Euston .Birmingham Leeds .Newcatle York .Edinburgh Kings Cross .Euston Manchester .00 16. B'ham Moor St/Snow Hill B'ham Snow Hill/Moor St.Kings Cross Edinburgh .Bristol Bristol .List of trains surveyed Table of train flows and trains surveyed Flow London .London Bath .Bath London .35 14.Edinburgh Birmingham .Euston Marylebone .Stoke London .35 18.Edinburgh York .47 14.York Newcastle .00 09.Edinburgh Edinburgh .Marylebone Kings Cross .47 Euston .Birmingham Birmingham .00 10.00 16.Manchester Euston .40 12.Edinburgh Birmingham .Newcastle York .Euston Manchester .London London .00 15.Edinburgh Edinburgh .Bristol TM Paddington .05 16.Paddington Bristol TM .London London .05 16.35 15.03 09.00 18.45 15.Newcastle York .Euston Manchester .Macclesfield London .Manchester Euston .Manchester Manchester .47 14.London Macclesfield .Birminghamham Int.Edinburgh Birmingham .Birmingham Edinburgh .03 09.00 10.00 16.London York .Paddington Bristol TM .York Newcastle .Stockport/Manchester Manchester/Stockport .45 15.Birmingham London .Birmingham York .York London .York Edinburgh .Birmingham Birmingham .Newcastle London .Paddington Bristol TM .Stoke London .London Macclesfield .London Bath .03 16.35 18.Macclesfield London .London Bristol .York Birmingham .Euston Day Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Friday Sunday Sunday Sunday Sunday Sunday Sunday .03 09.Birmingham Edinburgh .00 09.Kings Cross Birmingham .00 16.

Edinburgh Edinburgh .London London . Solihull .Bristol TM Paddington .York Birmingham .York Birmingham .00 15.Manchester Euston .30 11.London Birmingham .25 17.Newcastle Birmingham .00 07.00 15.Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 90 Flow London .Newcastle London .07 19.Birmingham Birmingham ./New St.03 13.York Newcastle .London London .Leeds Birmingham .Bristol TM Paddington .03 17.Paddington Bristol TM .London Macclesfield .London Stoke .Kings Cross Edinburgh .00 15.00 07.Euston Manchester .Leeds Birmingham .20 18.York Birmingham .Leeds Birmingham .Kings Cross Edinburgh . Birmingham New St/Int .07 16.00 13.Marylebone Kings Cross .Newcastle York .00 12.Birmingham London .Edinburgh London .Newcastle London .London York .25 17. B'ham Moor St/Snow Hill B'ham Snow Hill/Moor St.Stockport/Manchester Manchester/Stockport .Euston Marylebone .03 12.London Bath .Birmingham Leeds .Birminghamham Int.Edinburgh Birmingham .Bristol TM Bristol TM .Edinburgh Kings Cross .London Birmingham .London York .Edinburgh Birmingham .Edinburgh Kings Cross .Edinburgh Birmingham .03 15.00 09.Edinburgh York .Edinburgh Kings Cross .Edinburgh Birmingham .00 19.Edinburgh Edinburgh .00 15.Kings Cross Edinburgh .Kings Cross Edinburgh .Edinburgh Kings Cross .05 10.03 Paddington .Stoke London .Paddington Birmingham .London Bath .Manchester Manchester .Edinburgh York .Newcatle Train 16.Birmingham Euston .Euston Manchester .03 13.Edinburgh Kings Cross .Newcastle Birmingham .00 13.Edinburgh Day Sunday Sunday Sunday Sunday Sunday Sunday Sunday Sunday Sunday Sunday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday .03 13.03 12.Paddington Bristol TM .45 11.00 09.Euston Paddington .Macclesfield London .05 10.Birmingham York .Edinburgh Birmingham .Birmingham Newcastle .Kings Cross Edinburgh .Kings Cross Edinburgh .York Edinburgh .12 13.00 13.05 13.Edinburgh Birmingham .Edinburgh Birmingham .00 15.Bristol Bristol .03 15.00 07.Paddington Euston .Bath London .London London .Newcastle York .York Newcastle .Kings Cross Edinburgh .30 09.45 13.00 13.25 10.Newcastle York .Edinburgh Birmingham .Newcatle Newcastle .Kings Cross Birmingham .London London .Birmingham Newcastle .Birmingham Birmingham .00 09.03 13.00 15.00 13.London Newcastle .Kings Cross Edinburgh .00 07.00 07.Edinburgh Kings Cross .03 15.Manchester Euston .Edinburgh Kings Cross .00 09.00 09.Edinburgh Kings Cross .York London .Newcastle Newcastle .10 18.York London .Bristol TM Bristol TM .Bristol Bristol .York Edinburgh .Newcatle York .00 15.45 13.Edinburgh Kings Cross .Bath London .Kings Cross Edinburgh .London Newcastle .Solihull.00 09.London Edinburgh .

Birmingham York .Birmingham Edinburgh .London Birmingham New St/Int .Newcastle Birmingham . London .00 13.05 13.Birmingham York .Euston Manchester .Birmingham Birmingham .Birmingham Leeds .00 13.Birmingham Edinburgh .Newcastle Edinburgh .03 07.Euston Birmingham .Leeds Birmingham .Newcastle Edinburgh .Birmingham Edinburgh .Newcastle Edinburgh .45 08.52 19.00 13. Solihull ./New St.Manchester Manchester .Newcastle Birmingham .Kings Cross Birmingham .Edinburgh Edinburgh .03 15.05 09.Birmingham Euston .London Birmingham .Edinburgh Birmingham .Birmingham London .Kings Cross Edinburgh .Euston Manchester .Birmingham Euston .Birmingham Edinburgh .45 08.00 19.Euston Euston .03 19.York Newcastle .Birmingham Leeds .Birmingham York .Birmingham Day Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Monday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday Tuesday . B'ham Moor St/Snow Hill London .00 13.00 19.Edinburgh Edinburgh .05 08.Marylebone Kings Cross .05 Birmingham .00 13.Manchester Euston .Newcastle Birmingham .Birmingham Birmingham .London London .50 13.Macclesfield London .York Newcastle . Solihull .Stoke London .York Newcastle .Birminghamham Int.Newcatle York .Newcastle Edinburgh .05 09.Kings Cross Edinburgh . Birmingham New St/Int .Birminghamham Int.London London .Birmingham Edinburgh .York Edinburgh .Birmingham Edinburgh .London York .Manchester Euston .Newcastle Kings Cross ./New St.Marylebone Birmingham .Newcastle York .London London .York Newcastle .05 13.Birmingham Edinburgh .03 19.03 19.Solihull.05 07.London Stoke .20 12.London Newcastle .York Birmingham .10 09.52 15.05 13.30 15.10 13.Birmingham Edinburgh .Birmingham Edinburgh .05 09.Birmingham Train 15.York Newcastle .05 07.45 10.Birmingham Edinburgh . B'ham Moor St/Snow Hill B'ham Snow Hill/Moor St.Birmingham Edinburgh .Kings Cross Edinburgh .05 09.Birmingham Edinburgh .London B'ham Snow Hill/Moor St.Newcastle York .05 13.Newcastle Kings Cross .Birmingham Edinburgh .05 09.05 07.00 19.Birmingham Leeds .Birmingham Marylebone .Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 91 Flow York .30 10.Birmingham Edinburgh .Kings Cross Edinburgh .03 13.05 12.York Newcastle .London Macclesfield .York Newcastle .Stockport/Manchester Manchester/Stockport .05 12.York London .Euston Marylebone .05 07.05 12.Solihull.

Appendix B .Example questionnaire .

Faber Maunsell Saver Fares: Differentiation and Potential Deregulation 93 Appendix B – Example questionnaire .