Design Biases in Tests for Prospect Theory Using Dutch Value-of-Time Data
M Kouwenhoven, P Bucci, Significance, NL; G de Jong, Significance / NEA, NL / ITS, University of Leeds, UK
Stated Preference data of the Dutch Value-of-Time studies of 1997 and 1988 show some peculiar cost and time coefficients when assumptions based on prospect theory are tested in estimation We have investigated whether this is a result of the design of the
The Stated Preference (SP) data of the Dutch national Value-of-Time surveys of 1997 and 1988 have been analysed by several authors. These studies revealed several indications that are consistent with the predictions of prospect theory: time and cost coefficients are different for time savings and time losses, and these coefficients are also different for small and large changes. In addition, significant inertia terms for time and cost have been found.
We investigated up to which level this unexpected behaviour is caused by the specific SP design of these surveys. This was done by simulating choices by respondents assuming utility functions that are both linear in time and cost (and no prospect theory is assumed). Then, we analysed the simulated data in the same way as the observed data had been analysed. Despite the early results that were presented at the ETC conference, a more careful analysis of the simulated data showed no indication of any significant inertia terms, and no indications of any differences in the estimated coefficients between gains and losses or between small and large changes. Therefore, we have not found any evidence of a design bias in these surveys.
Association for European Transport