Value of travel time estimation from DCE data

Value of travel time is one of the most important outputs of transport SP research - and one of the most misunderstood. This requires careful model specification and honest reporting.

This article explains how to estimate value of travel time from DCE data, the factors that affect VoTT estimates, segmentation by trip purpose and mode, and how VoTT is used in transport appraisal.

Ben White, 07.07.2026

What value of travel time means

Value of travel time (VoTT) is the amount of money a traveller would pay to save one unit of travel time - typically expressed per minute or per hour. It is derived from the ratio of the travel time and cost coefficients in a choice model: VoTT = -(beta_time / beta_cost).

VoTT estimates from SP studies are used directly in transport cost-benefit analysis. Every UK transport appraisal uses official VoTT values derived from SP research - DfT's WebTAG. Getting VoTT estimates right matters not just scientifically but economically: over-estimating VoTT overstates the benefits of time-saving infrastructure investment.

What drives variation in VoTT

VoTT varies systematically by trip purpose (business travel has higher VoTT than leisure travel), mode (car has higher VoTT than bus for the same trip), income (higher income implies higher VoTT), and journey length (VoTT may be lower for very long journeys where in-vehicle activities reduce the disutility of travel time).

The model specification affects VoTT estimates. Including in-vehicle activities, crowding, or seat availability as separate attributes partitions the disutility of travel time across multiple parameters. VoTT from a model with crowding is lower than from a model without crowding because some of the disutility of slow travel is attributable to crowded conditions rather than time itself.

Pivot designs produce VoTT estimates that are better calibrated to realistic trade-off ranges than fixed-level designs. SurveyEngine's pivot design approach is the standard for transport VoTT studies.

TLDR Quick links

Transport resources

Forum: modelling questions

Open a free account

Estimating VoTT in Apollo from SurveyEngine data

Step 1: Export the combined RP/SP dataset from SurveyEngine. For VoTT studies, joint RP/SP estimation is typically used. The export includes RP mode and attributes alongside SP choices and attribute levels.

Step 2: Specify the utility function carefully. In the MNL or mixed logit specification, include travel time and travel cost as separate continuous attributes. The ratio of their coefficients gives VoTT.

Step 3: Segment VoTT by trip purpose. Interact the time and cost coefficients with a trip purpose dummy (business vs leisure/commuting). This produces separate VoTT estimates for each purpose segment.

Step 4: Account for the value of reliability. If your design includes a reliability attribute, estimate the value of reliability separately. The value of reliability is the WTP for a reduction in the standard deviation of journey time.

Step 5: Compare against official benchmarks. Compare your VoTT estimates against the DfT WebTAG values (UK) or equivalent national guidance. Explain any differences in terms of sample characteristics, corridor, or model specification.

Worked example - business travel VoTT by mode

A joint RP/SP study of intercity business travel estimates VoTT by mode using a mixed logit model. Business travellers by car: VoTT = £18.40/hour (SD = £6.20). Business travellers by train: VoTT = £22.10/hour (SD = £8.40). Leisure travellers: VoTT = £8.60/hour (SD = £4.10).

The higher train VoTT for business travellers reflects a combination of productivity during train travel (which reduces the disutility of journey time but is not captured in the model) and the higher opportunity cost of business time. The estimates are within 15% of DfT WebTAG 2023 values, which provides confidence in the model validity.