Willingness to Pay for Car Safety: Sensitivity to Time Framing
H Andersson, Swedish National Road and Transport Research Institute (VTI), SE; J Hammitt, Harvard University, US; G Lindberg, VTI, SE; K Sundstrom, SLI, SE
The aim of this study is to examine the effect of time framing on respondents' WTP for car safety. This research question is of major policy relevance since if estimates from SP studies depend on the time frame, they may not reflect preferences.
Stated preference (SP) surveys attempt to obtain monetary values for non-market goods that reflect individuals' true preferences. Evidence against well-defined preferences among respondents in surveys have been found in numerous empirical studies, however, which would suggest that monetary values from SP studies are sensitive to survey design and may not always reflect respondents' true preferences.
The aim of this study is to examine the effect of time framing on respondents' willingness to pay (WTP) for car safety. We expect, on theoretical grounds, WTP per unit of risk reduction to be insensitive to the time framing. When asked to state their WTP (or WTA) for a non-market good in an SP study, respondents are faced with a time frame. That is, respondents need to consider when and how often payments are going to be made, and for how long the good (or bad) will be provided. The objective of this study is, therefore, to examine whether respondents' preferences are sensitive to the framing of the scenario. This research question is of major policy relevance, and of general interest to the evaluation of non-market goods, since if the values differ, it shows that values in use in benefit-cost analysis depend on the time frame of the survey format and may not reflect actual preferences. To estimate WTP and study the effect of time framing, we conduct a contingent valuation survey using a double bounded (DB) dichotomous choice format on a Swedish sample of ca. 900 individuals.
We find that estimated WTP is sensitive to the time framing of the question. Hence, in empirical applications it is not unproblematic to treat a one period risk reduction as equivalent to a series of risk reductions. Both non-parametric and parametric techniques are employed when analyzing the data, and in the non-parametric analysis estimates of the weighted average value of a statistical life (VSL) based on the single bound (SB) and DB formats were 69 and 28 percent higher in the annual than the monthly scenario. However, only the SB estimates were statistically significantly different. Also in the parametric analysis the results were mixed, with one of the regressions of the mixture models suggesting that WTP in the annual scenario was ca. 40 percent higher than in the monthly scenario, but with the other regressions not finding any statistically significant relationships. Overall, we conclude that our results suggest that preference elicitation for car safety is sensitive to time framing, but that the problem is not too problematic considering other aspects of preference elicitation for mortality risk reductions, i.e. especially scale insensitivity.
Association for European Transport