On the cost of misperceived travel time variability

Recent studies show that traveler’s scheduling preferences compose a willingness-to-pay function directly corresponding to aggregate measurement of travel time variability under some assumptions. This property makes valuation on travel time variability transferable from context to context, which is ideal for extensive policy evaluation. However, if respondents do not exactly maximizing expected utility as assumed, such transferability might not hold because two types of potential errors: (i) scheduling preference elicited from stated preference experiment involving risk might be biased due to misspecification and (ii) ignoring the cost of misperceiving travel time distribution might result in undervaluation. To find out to what extent these errors matter, we reformulate a general scheduling model under rank-dependent utility theory, and derive reduced-form expected cost functions of choosing suboptimal departure time under two special cases. We estimate these two models and calculate the empirical cost due to misperceived travel time variability. We find that (i) travelers are mostly pessimistic and thus tend to choose departure time too earlier to bring optimal cost, (ii) scheduling preference elicited from stated choice method could be quite biased if probability weight- ing is not considered and (iii) the extra cost of misperceiving travel time distribution contributes trivial amount to the discrepancy between scheduling model and its reduced form.