Sustainable Transportation Lab

November 16, 2017

TNCs and transit: frenemies? Again

Yanbo Ge

A report that came out recently by researchers from the University of California Davis looks closely into the relationship between Transportation Network Companies (TNCs, also called ride-hailing services) and transit. The article sheds light on several interesting questions related to vehicle ownership and trip replacement of TNC users. By collecting data in several metropolitan areas in the US where TNC services are available, the authors found that 30% of the respondents either use TNC services themselves or have shared a ride with friends through TNC platforms. 6% of the adopters use TNCs every day and another 18% use TNCs at least once a week. Most of the trips are for non-commute trips (75%), which is consistent with the result of a study conducted by the Shared-use Mobility Center.

The study found that among the TNC users, the percentage of respondents that reduced vehicle ownership is positively correlated with the frequency of TNC use, as shown in Figure 1.


                                                Figure 1

Among transit users, there is very little difference in vehicle ownership between those who use TNCs and those who do not (the first two columns in Figure 2). And among those who don’t use transit, there is also little difference between those who use TNCs and those who do not.


                                                       Figure 2
The report also looked into the relationship between TNC adoption and transit use. The respondents were asked to answer the question “since you started using on-demand mobility services such as Uber and Lyft, do you find that you use the following transportation options more or less”. The results show that most respondents indicated no change of transit use. The authors then calculated the difference between the percentage reporting more transit use and the percentage reporting less transit use (Figure 3, below). The authors interpreted the results as “After using ride-hailing, the average net change in transit use is a 6% reduction among Americans in major cities”. I do not think this is a well-supported claim. First of all, I suspect the authors took a summation of the changes in the public bus (-6%), heavy rail (+3%), and light rail (-3%) to get their 6% total transit reduction. However, the overall change in transit use should be a weighted average instead of summation of those numbers, so the net change must surely be smaller than -6%.  Second of all, the report showed that most respondents reported they did not change their transit usage. It would be helpful to know what “majority” means, is it 60% or 90%? Thirdly, 6% more of the respondents reporting some transit usage reduction is completely different from a 6% transit use net reduction, because the former does not account for the size of the change. For example, if the 80% of the respondents reported no change in transit use, 7% reported more transit use and 13% reported less transit use. Assume those who reported more transit use increase their transit trips by 20%, and those who reported less transit use reduce their transit trips by 20%, then the total transit use change will be -1.2%. If the sizes of the change are 50% and 20% respectively, then the total transit use will increase by +0.9%. The point is, the number could vary — a lot. 


                                                         Figure 3
The study also looked into the mode substitution of Uber of Lyft weighted by frequency of TNC use as reported by the respondents and found that “49% to 61% of ride-hailing trips would not have been made at all, or by walking, biking or transit”. To see if this conclusion is justified, I would want to know whether the question allowed multiple choices. If only one option is selected, different respondents could have had various interpretations. Some might have recalled the latest TNC trip or a particular trip that somehow triggered deeper memory than others. For some people who have a good record or memory of all their TNC trips, there could be the confusion of whether they should report according to the number of trips or the total mileage traveled. For one individual, there could be four different answers to this one question:

– my last Uber trip would have been by taxi if not for Uber (coming back from the airport)
– I would not have gone to the bar and got super drunk that time if not for Uber
– by the number of trips, I would guess I’ve replaced something like 20% taxi, 10% bike, 40% transit, 30% drive
– by total miles traveled probably 60-80% of my TNC miles are replacing taxi travel, since those are my trips to and from the airport, where each trip is 20 miles
 
In this case, if asked to provide one answer, some may report “transit” since it is the largest number of trips and the question mentions trips. But the effect on VMT is another story. A better approach would be to ask about the last trip taken by TNC – that would generate a representative mix of responses across a large sample.
 
If the respondents were asked to choose all that apply, I would want to know how the authors split the weight (the frequency) among several choices. If every selected alternative is weighted by the same frequency number, I doubt the accuracy of this claim.


                                                      Figure 4