How Well Do Laboratory-Derived Estimates of Time Preference Predict Real-World Behaviors? Comparisons to Four Benchmarks

Abstract

A large literature implicates time preference (i.e., how much an outcome retains value as it is delayed) as a predictor of awide range of behaviors, because most behaviors involve sooner and delayed consequences.We aimed to provide the most comprehensive examination to date of how well laboratory-derived estimates of time preference relate to self-reports of 36 behaviors, ranging fromretirement savings to flossing, in a test–rest design using a large sample (N = 1,308) and two waves of data collection separated by 4.5 months. Time preference is significantly—albeit modestly—associated with about half of the behaviors; this is true even when controlling for 15 other demographic variables and psychologically relevant scales. There is substantial variance in the strengths of associations that is not easily explained. Time preference’s predictive validity falls in the middle of these 16 possible predictors. Finally, we ask time preference researchers (N = 55) to predict the variation in the relationship between time preference and behaviors, and although they are reasonably well-calibrated, these experts tend to overestimate the predictive power of time preference estimates. We discuss implications of invoking time preference as a predictor and/or determinant of behaviors with delayed consequences in light of our findings.