What is the elasticity of sharing a ridesourcing trip?

•The willingness to share varies spatially and temporally.•The willingness to share is sensitive to price per mile, total price, and trip duration.•Price per mile is the most important predictor in a random forest model.•The total price of a trip is more important for prediction than trip duration.•...

Full description

Saved in:
Bibliographic Details
Published in:Transportation research. Part A, Policy and practice Policy and practice, 2021-11, Vol.153, p.284-305
Main Authors: Wang, Sicheng, Noland, Robert B.
Format: Article
Language:English
Subjects:
Elasticity
Random forest
Ride-splitting
Ridesourcing
Transportation network companies
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•The willingness to share varies spatially and temporally.•The willingness to share is sensitive to price per mile, total price, and trip duration.•Price per mile is the most important predictor in a random forest model.•The total price of a trip is more important for prediction than trip duration.•Trip duration is positively associated with the willingness to share. Transportation network companies (TNCs) offer a ride-splitting option for ridesourcing trips, allowing users to share the vehicle with others at a lower fare. While encouraging shared rides has environmental benefits, little is known about how price affects the decision to share. Using TNC trip data from Chicago, we investigate the temporal and spatial distribution of authorized ride-splitting trips in 2019. We found that the willingness to share TNC trips differed across neighborhoods with different demographics, socioeconomic status, and built environment characteristics. The willingness to share was related to price and trip duration. We estimate logistic regression and random forest models to determine the marginal price and time effects on the decision to share. The results indicate the probability of authorizing a ride-splitting trip is highly elastic to the price per mile and the random forest model had better predictive accuracy than the logistic model. Additionally, we examine the importance and marginal effects of total price and trip duration. We use two data preprocessing methods to address rounding errors in the price and demonstrate the robustness of the results. Policy implications for increasing shared trips are discussed based on the findings.
ISSN:0965-8564
1879-2375
DOI:10.1016/j.tra.2021.09.008