Personalized Pricing and the Value of Time: Evidence from Auctioned Cab Rides

Abstract

We recover valuations of time using detailed data from a large ride-hail platform, where drivers bid on trips and consumers choose between a set of rides with different prices and waiting times. We estimate demand as a function of prices and waiting times and find that price elasticities are substantially higher than waiting-time elasticities. We show how these estimates can be mapped into values of time that vary by place, person, and time of day. We find that the value of time during non-work hours is 16%lower than during work hours. Most of the heterogeneity in the value of time, however, is explained by individual differences. We apply our estimates to study optimal time incentives in highway procurement. Standard industry practices, which set incentives based on a uniform value of time, lead to mis-priced time costs by up to ninety percent.

The Demand for Mobility: Evidence from an Experiment with Uber Riders

“…Our estimates are larger than recent private travel elasticities from the United States gasoline market, which are larger than had been found in prior studies with aggregate data and cross-sectional designs Levin et al (2017). They are also larger than those found in the United States taxi market (Rose and Hensher, 2014) However, they are consistent with recent estimates from ride-hail services in Prague (Buchholz et al, 2020). Our estimates may differ with the earlier literature for several potential reasons: (1) Prior studies have typically examined the effects short-run price changes; (2) Whereas prior studies have typically focused on transport markets with higher-quality substitutes, this study specifically focuses on a transit-constrained city; (3) The large price changes examined in this study may induce significant substitution from lower quality substitutes.…”

The Value of Time in the United States: Estimates from Nationwide Natural Field Experiments

“…One shortcoming with this first 56 These results may be sensitive to the choice of features used to define groups of sessions. 57 For example, Buchholz et al (2020) utilizes a hierarchical model with individual-specific coefficients to estimate the entire distribution of VOT by assuming (i) a particular parametric form to the distribution of individual-specific heterogeneity and (ii) that heterogeneity in time and price responses not captured by additive effects for a few locations and time categories is individual-specific; the full model is then specified by a (relatively) small number of parameters, which can be estimated by Bayesian or maximum likelihood methods.…”

“…1 Our research is complemented by two contemporaneous studies that use very different identification strategies and data sets (Castillo, 2019;Buchholz et al, 2020). Both studies observe market wait times and prices but use econometric structure as opposed to experimental variation to solve the identification problem for Houston and Prague consumers.…”

The Value of Time in the United States: Estimates from Nationwide Natural Field Experiments

“…Our approach allows us to recover an estimate of the VOT over wait time gradients, providing insight into the shape of the VOT function. 1 Our research is complemented by two contemporaneous studies that use very different identification strategies and data sets (Castillo, 2019;Buchholz et al, 2020). Both studies observe market wait times and prices but use econometric structure as opposed to experimental variation to solve the identification problem for Houston and Prague consumers.…”

“…These results may be sensitive to the choice of features used to define groups of sessions.57 For example,Buchholz et al (2020) utilizes a hierarchical model with individual-specific coefficients to estimate the entire distribution of VOT by assuming (i) a particular parametric form to the distribution of individual-specific heterogeneity and (ii) that heterogeneity in time and price responses not captured by additive effects for a few locations and time categories is individual-specific; the full model is then specified by a (relatively) small number of parameters, which can be estimated by Bayesian or maximum likelihood methods.…”