My highlights at the ETC

Several people from the Centre attended the European Transport Conference - I think we were 10 people there. I thought it was a great conference this year - lots of really interesting papers. Some of my highlights were (I'll link to the papers once they are up - note that the conference nowadays make the papers available, which is a huge improvement!):

• Robert Bain presented a survey on optimism bias when forecasting toll road traffic demand. In short, forecasts overpredict traffic volumes with 30% on average - mostly due to the incentives present in toll road contexts (i.e., the overpredictions aren't present to the same extent for public non-tolled roads)
• Pedro Abrantes presented an extension of Mark Wardman's meta-analysis of UK value-of-time studies (Mark and Jeremy Toner were co-authors). Lots of interesting results, and a huge material.
  
• Andrew Daly showed that when computing the willingness-to-pay from mixed logit models in the "standard form" (i.e. separate, random parameters for cost and time, for example - not the "WTP-space method" of Mogens Fosgerau and others), then without additional contraints on the distribution of the cost parameter, the WTP isn't even defined. In particular, normally distributed cost coefficients give undefined WTP:s... Bad news for all you out there estimating normally distributed cost coefficients (if you want to infer values of time etc., that is - forecasting is OK, of course)
• Andrew also showed a set of nice ideas on how to transform the deterministic part of the utility (the "V part") to improve fit. I guess others have tried some of it before, but it was a good summary and organisation of different methods to do it.
  
• Mogens Fosgerau presented a nice and simple derivation of the valuation of headway. The expression results in a decreasing marginal disutility of headway, just like the functions people often have estimated. Good to have a theoretical base for this. The paper is already published in Transportation Research B.