Can time-inconsistent preferences explain hypothetical biases?

Abstract

We investigate whether the value of time (VOT) depends on when the corresponding preferences are measured: in advance, just before, or after the time period for which the time preferences are being evaluated. We find that the VOT is highest when elicited just before the time period. This is an indication of the VOT being affected by time-inconsistent, and more specifically, present-biased preferences. We argue that this result may explain why time valuations based on stated preference (SP) data are typically found to be lower than those based on revealed preference (RP) data: most RP surveys evaluate the preferences of respondents close to the time period for which the preferences are being measured, whereas the time instances for which preferences are evaluated in SP surveys tend to be more abstract, or referencing past or future time periods.