DfT plans maintenance programme to keep time saving values up to date

The DfT is planning to invest in a continuous research programme to ensure that values of travel time savings remain up-to-date and informed by the best methods. Andrew Forster reports

The monetised value of travel time savings often forms the most important component in the economic justification of transport projects. The DfT published the findings of a major research project into the value of time savings in 2015, conducted by the Institute for Transport Studies at the University of Leeds and consultant Arup. The values have since been adopted in appraisal. 

Last year the DfT asked the same contractors to advise on a maintenance and development programme to keep the values up to date. ITS and Arup have prepared two reports, which have just been released by the DfT. The first, completed last November, identified a long-list of possible activities. The second develops a shortlist of priority areas for action, which are likely to inform forthcoming research commissions by the DfT.  

The Department’s current default approach to uprating the value of travel time savings (VTTS) is to adjust the base values for growth in GDP per capita using an income elasticity of 1. ITS and Arup recommend reviewing whether this is appropriate, noting that in most European countries the elasticity value is in the range 0.5 to 1. In addition, they recommend uprating values to reflect changes in the travelling population and characteristics of the trips they are making. 

“In practice this would lead to regularly feeding new National Travel Survey data into the behavioural model (the VTTS Implementation Tool), thereby adjusting the base year VTTS, possibly in combination with an additional income growth adjustment,” say ITS and Arup. “The adjusted base year VTTS would then be projected for future years using a standard income elasticity parameter.”

The validity of the existing framework is likely to diminish with time, say ITS and Arup. “The behavioural model was based on 2014 stated preference data and the more time passes the less likely the behavioural framework matches with current preferences. Waiting 20 years before commissioning a new study, as was done before the 2014/15 update, is not advisable.” 

Societal and technological developments such as autonomous vehicles and Mobility as a Service (MaaS) may influence the value people place on travel time savings in future, says ITS/Arup, though this is not one of the immediate research priorities they recommend. 

The researchers do, however, recommend a monitoring regime to validate the values. They identify two ways of doing this: a meta analysis of academic research literature, or collecting small samples annually using repeats of the 2014/15 study. 

“Meta-analysis is perhaps the more cost-effective option of the two with the potential of obtaining buy-in from other European nations to keep the database up-to-date, although some upfront investment is needed to ensure the database is updated over the 2011-2018 period,” say ITS/Arup. “The annual monitoring scheme using new stated preference (SP) data provides, however, a clearer source of comparison for which the costs are primarily based on data collection.

“The general idea of an annual rolling survey was well-received by the market research companies,” they say. “What would be proposed would be to sample a different mode each year (e.g. year one car; year two rail; year three other PT [public transport] and bus) on all journey purposes. Hence, after three years all mode purpose combinations would be surveyed.”  

Stated or revealed preference surveys?

The reports include a lengthy discussion on the merits and drawbacks of stated preference and revealed preference surveys for eliciting people’s value of travel time. 

Stated preference (SP) surveys involve asking people hypothetical questions about journeys, to tease out the effect of journey costs and time on travel decisions. In contrast, revealed preference (RP) surveys observe what people’s actual travel behaviour was.

ITS and Arup recommend that the DfT continues with SP methods for the time being but says the Department should keep an eye on developments in RP methods. “Whilst there is an obvious attraction to the replacement of SP by RP methods, the case has not fully been made for this in the short-term,” they say. “We suggest that the key issues for taking forward emerging SP methods and the monitoring programme are sampling, timing of survey, and medium of survey.” On sampling, they note that SP surveys face the challenge that “those with a high value of time are much less likely to take part in the survey”. 

Stated preference surveys have some advantages over RP, says ITS and Arup. “SP methods essentially construct a ‘laboratory condition’ whereby journey time and cost are varied independently across choice tasks making it easier, from a statistical perspective, to isolate their impact on decisions. In RP this is much harder, since journey time and costs are highly correlated. Moreover, SP studies allow the study of more travel decisions per traveller, thereby increasing the number of observations in the dataset relative to RP.”

Concerns remain about the validity of SP techniques, however. “The economists’ concern about the validity of non-RP methods has been generalised into a worry about the hypothetical nature of stated choice (SC) and other stated preference methods. More specifically, experience with VTTS estimation using SC data has revealed a wholly unwelcome dependence of the result on the design of the experiment, particularly on the size and sign of the time differences that survey respondents are asked to value.” 

Although there may well be theoretical justification for these also applying in the real world, the researchers say the extent to which this is true remains unknown. But they add: “Certainly, size and sign differences in VTTS are unacceptable to governments seeking to appraise transport policy and projects.

“Size differences cannot be eliminated at all easily and an assumption of a specific amount of savings needs to be made; this is frequently set at ten minutes, despite evidence that time changes brought about by transport policy are usually considerably smaller than this.”

ITS/Arup predict a growing interest in RP data. “Traditional travel survey data, but also new data sources becoming available in this era of ‘big data’, such as smart-ticketing or GPS-based data, combined with new modelling approaches, are expected to revive interest in RP data collection methods in the near future. 

“The original issue that RP data volumes were too small to make reasonably accurate estimates has been overtaken by the collection of much larger home interviews and other RP datasets. These describe mode, destination and other everyday choices made by travellers and are routinely used to estimate large-scale travel demand forecasting models; good estimates are generally made of the time and cost coefficients, despite the correlation in the relevant variables.”

ITS and Arup note RP surveys present their own set of problems. For instance, how much does the traveller know about the alternatives and the time and cost of using each of them?

“It may be questioned whether it is necessary to consider these issues at all,” say ITS/Arup. “[After all] They do not form part of a classical economic analysis of a market, which works on the basis that consumers behave ‘as if’ they have perfect information. At least it seems we are moving closer towards a situation of perfect information with travel times from Google Maps nowadays being accurate for most modes, and travel costs can also be obtained accurately by automatic queries to e.g. public transport web pages.”

Another drawback of RP methods is that the detail and context that SP surveys elicit is lost. “SP surveys that are used to estimate VTTS collect a wealth of information from each survey respondent, including: mode of travel; journey purpose; travel costs and travel time by mode (preferably what ticket each user is on); other travel conditions (e.g. crowding and congestion); socio-economic characteristics (e.g. income, gender, age); household characteristics (e.g. car ownership, household size).”

With RP it can also be harder to survey a “truly random and representative set of travellers”. 

ITS/Arup recommend that the DfT takes steps to investigate the potential for RP surveys, including working alongside journey planning applications and other online providers to determine how suitable their data is to determining how people choose journeys based on travel time and cost. “It may be that, although the ‘standard’ version of the app cannot be used for VTTS, a variant version may be designed at small cost, or that future versions of the app may allow more data to be gathered about the user and their actual choices.”?

In addition, they say that “macro-economic-style research into the impacts of Crossrail and HS2 (the latter in particular) could provide good RP case studies in the near future”.