100 likes | 200 Views
Transport CBA– valuing the benefits. Peter Mackie. Principles of CBA (1). There are two competing conceptual versions of cost-benefit analysis. Version 1 is from Harberger, Sugden et al. They say CBA is a version of commercial appraisal expanded to account for market failures, unpriced goods
E N D
Transport CBA– valuing the benefits Peter Mackie
Principles of CBA (1) • There are two competing conceptual versions of cost-benefit analysis. • Version 1 is from Harberger, Sugden et al. They say • CBA is a version of commercial appraisal expanded to account for market failures, unpriced goods • So we should discover willingness to pay for benefits in relevant market segments by studying actual or stated behaviour (RP, SP) • We should use the resulting values directly in modelling, forecasting and appraisal • Just like commercial appraisal, we should add up individual (or group) WTP to get market WTP, and compare market costs and benefits to help decide what to do. UNWEIGHTED ADDING UP OF WTP
Principles of CBA (2) • Version 2 follows authors such as Pearce, Nash, Bohm, Bergson • CBA is a form of social appraisal—we are interested in the social valuation of costs and benefits. Society is buying a road, what should we collectively be willing to pay for the benefits? • Again, discover individual (or group) WTP and use directly for modelling and forecasting---behaviour is based on perceived cost, time…. • But when it comes to appraisal we want the marginal social utility of doing something ; sum of individual WTP might not be a good proxy. • Might want to adjust WTP of rich vs poor people ; rich vs poor regions ; cities vs rural…. Utility of 1 SEK to a poor roadsweeper might not equal utility of 1 SEK to a rich transport planner.
Standard values of time • A convenient and common form of adjustment is to use a standard value of non-working time savings. This effectively applies a set of social welfare weights to behavioural values. • Suppose a very simple case--- individual values of time savings are directly proportional to income and welfare weights are inversely proportional to income. The combination of the two produces a standard or ‘equity’ value of time. • Actually in other contexts like value of life and safety, this is not so controversial. The concept that for appraisal we should use a standard value of saving statistical life (at a given age) is probably more acceptable than using differentiated values of life by income. • But overall, it’s a political matter, NOT a technical matter, whether version 1 of CBA is preferred to version 2. In Britain, version 2 was decided on by a Minister in 1968 so we use standard values of time.
From concept to practice • Standard values were invented in the context of road investment appraisal. So in this case, it was public investment, users were not paying tolls or fees, majority of the benefits were time savings. • But in context of user pays schemes the case is much less clear. Standard values of time mean that we reweight the travel time benefits of public transport schemes, road charging schemes but we don’t reweight the money flows. • So, for a tram scheme from a poor section of town to the centre, the appraisal value of time savings may exceed the behavioural wtp. The travellers might prefer a low fare slow bus to a higher fare faster tram. • Conclusion--- there are some logic issues with using a standard vot for appraisal.
Practical considerations • Most practical appraisal is bound to require a significant amount of averaging. It is not practical to assemble bespoke WTP data at the scheme level. So national VOT studies are used. • Highly differentiated VOTs would create problems for modelling especially assignment modelling • In appraisal, matrices are unlikely to be supported by local evidence on income, journey purpose splits etc • It’s noticeable that countries which in principle use behavioural values in appraisal in practice do a lot of averaging
Business Travel Time Savings • There are numerous big issues with the VBTTS. They account for a large proportion of the benefits and rely on some version of the Cost saving approach which is not thoroughly substantiated empirically. • However the benefits from EB time savings are generally assumed to flow through to firms and the wider economy rather than accruing to the individual. • So there is not really any case for using standard values in principle, some form of behavioural values are wanted. In UK for EB traffic we have different values for professional drivers, briefcase travellers by mode etc.
Non-work time appraisal value • In UK, we differentiate --commuting vs all other non-work • We have multipliers for walk, wait, interchange etc • But we do not differentiate by income, journey length, sign/size of time saving or mode • Journey length – VOT is higher because incomes of people who make longer journeys is higher, boredom and discomfort set in, there can be time constraints at destination. Probably we ought to take the income effect out and differentiate by journey length for the other factors.
Modal values • Probably not a good idea to use differentiated modal values • Which modal values would you use—mode used or mode valued (see handout)? • Clearly raises issues in context of mode switching ; desirable to respect the rule of half in computation of benefits when people switch from choice A to choice B • Preferable to use a characteristics approach valuing the attributes of modes (eg freedom to read, do wi-fi, phone etc) rather than the mode itself.
Conclusion—what should we do? • For appraisal work a high degree of averaging will be needed • There is clear evidence that VTTS varies by journey purpose and journey length independent of income • There is clear evidence about multipliers for walk, wait, interchange and attribute values like crowding, wifi availability etc • For a wide range of appraisal work it is probably not sensible and practical to go further than that. • However for very big projects and for user pays schemes such as toll financed schemes, both modelling and appraisal require greater market segmentation by willingness to pay. • Where the user is paying for the benefits, the case for using behavioural values for both modelling and appraisal is stronger because the risks of internal inconsistency are greater.