Seminar: The Pure Characteristics Approach to Travel Cost Models and Implementation of Benefit Transfers
Date(s): May 14, 2009, 10:00-11:30 am
Location: Room 4144 EPA West
Contact: Carl Pasurka, 202-566-2275
Presenter(s): Christopher Parmeter (Virginia Tech)
Description: Standard travel cost models rely on a priori assumptions about the distribution of preferences. The resulting estimates reflect uncertainty relating to unobserved characteristics and unobserved preferences on observed characteristics. Currently, few methods have challenged the distributional assumptions applied in empirical settings. This paper argues that the pure characteristics approach, applied to travel cost models provides bounds on benefit estimates that reflect uncertainty on unobserved preferences on unobserved characteristics. Thus, this method is a semiparametric alternative to traditional travel cost models. The ability to relax distributional assumptions on both unobserved characteristics and preferences may provide bounds that are tighter than traditional confidence intervals, improving the quality of policy analysis.
The authors extend the existing pure characteristics methodology to the case where different people face different prices (travel costs). The new estimator uses the additional information in the variation in prices to relax the need for independence between observed and unobserved characteristics of the object of choice. The variation in prices acts as a type of instrument. It provides a new way to deal with endogeneity in recreation demand applications that is less data intensive than the combined stated / revealed preference approach proposed by von Haefen and Phaneuf (JEEM, forthcoming). Additionally, by identifying bounds on welfare measures, rather than point estimates, the PCM provides a way to explicitly address uncertainty when conducting benefit transfers—one of EPA’s guidelines. Thus, this work fills in two gaps in the literature. First, it provides a theoretically sound approach to estimating travel cost models without the use of distributional assumptions. Second, it accounts for preference uncertainty in benefit transfers, which have hindered past methodologies.
Seminar Category: Environmental Economics