Transport Welfare Benefits in the Presence of an Income Effect
J Laird, ITS, University of Leeds, UK
A key application of discrete choice models within the transportation field is the calculation of welfare changes as a result of a policy intervention. Discrete choice models assist this process in two ways: firstly in the development of demand forecasts and secondly through the derivation of marginal values for policy variables of interest (e.g. the value of travel time savings). This paper focuses on the latter application and the circumstances in which the marginal values derived from the models can lead to a serious bias in the welfare benefit estimate of a transport intervention.
Economic appraisal procedures utilise the Marshallian measure of consumer surplus to calculate the economic surplus of a transport intervention. This is estimated for an average trip, and then aggregated over all trips made in a year through a series of annualisation factors to give the total welfare benefit in each of the forecast years. Often the true Marshallian consumer surplus is approximated through the use of a linear demand curve in what is known of the ?Rule-of-a-Half?. Aside from the well known limitations to the Rule-of-a-Half, this method only gives a precise estimate of the economic benefit when there are zero income effects. With zero income effects the Marshallian consumer surplus measure is exactly equal to the Hicksian Compensating Variation measure, otherwise it overestimates it (with transport acting as a normal good).
Within a transport appraisal context it is implicitly assumed that income effects are negligible. In many applications this may not be the case as for example a 20 minute time saving on a commuting journey may lead to total levels of benefit that are a significant part of an individual?s disposable income. The more extreme cases of congestion charging and the provision of fixed links to islands as a replacement to ferry services also challenge this assumption. There is however a paucity of evidence as to the size of the income effect necessary to make the use of Marshallian measure indefensible.
This paper aims to start to fill this evidence gap by drawing off research conducted in the Outer Hebrides on the economic value local residents place on fixed links as a replacement to a ferry service. Householders responded to stated preference questions drawn from two designs ? one couched in terms of willingness to pay for an individual trip and another in terms of willingness to pay over the course of the year ? as well as three contingent valuation questions. Estimates of economic welfare derived with methods consistent with industry practice for model estimation and appraisal realise estimates that are approximately double the compensating variation measure. This is a serious bias and arises as a consequence of the Marshallian measure giving an estimate of economic welfare that is more than 25% of household income. The implications for modelling and appraisal practice are discussed before future research directions are identified.
Association for European Transport