Does Taxing TNC Trips Discourage Solo Riders and Increase the Demand for Ride Pooling? A Case Study of Chicago Using Interrupted Time Series and Bayesian Hierarchical Modeling

This paper studies the tax intervention applied to transportation network company (TNC) trips starting on January 6, 2020 in the City of Chicago. An interrupted time series (ITS) with an autoregressive integrated moving average (ARIMA) methodology is employed to infer the causal impact of the intervention on the percentage of shared trips and the counts of shared and private trips. Analysis is conducted at a community area level, either as pickup or drop-off. The results show a significant but small increase in the share of shared trips as well as the count of shared trips, specifically on weekends because of the intervention. Private trips, on the other hand, are found to have decreased on the weekdays, but potentially increased on the weekends. A Bayesian hierarchical model is then employed to combine information across community areas, examine a pos-teriori if there are significant spatial differences, and estimate the common treatment effect. The analysis suggests minimal spatial differences across community areas. The common treatment effect on weekdays ($1.75 tax difference) is a 3.78 percentage point increase in the share of shared trips, a 27% increase in the count of shared trips, and a 12% decrease in the count of private trips (at an approximate base of 10% market share of shared trips). Thus, the intervention likely shifted demand toward pooled rides, reducing congestion caused by TNCs. However, there is little evidence that this shift is sufficient to offset or reverse the systematic trend of declining use of shared rides.