TNCs 2020: A Profile of Ride-Hailing in California

The previous section evaluated the completeness and integrity of the 2020 public TNC Annual Reports, revealing extensive data quality issues. This section explores the reports, in order to identify general characteristics of TNC activity, where possible, and acknowledge limitations and uncertainty otherwise. In some places, this section reveals additional data quality issues. The 2020 public TNC Annual Reports, and the figures presented in this section, cover the period of September 2019 through August 2020.

How many TNC trips were taken?

Due to internal inconsistencies in the reports noted in the prior section, the number of TNC trips taken vary from 218 million and 277 million trips, a range of 59 million trips (27%). Table 18 shows the reported trip totals by company. Uber’s reported trips range from 157 million to 166 million and Lyft’s range from 61 million to 111 million; the total ranges from 218 to 277 million.

Where were TNC trips taken?

TNC trips were highly concentrated in urban areas.¹ Figure 5 shows total trips and trips per square mile by county for the 10 counties with the most TNC trips. Nearly two-thirds (64%) of all TNC trips in California occurred in just 3 counties: Los Angeles, San Francisco, and San Diego, which collectively contain only 5% of its land area. While Los Angeles has the most trips of any county, San Francisco has by far the greatest concentration of TNC trips, with nearly 500 times more TNCs per square mile than the rest of the state.

Figure 6 shows trip density by zip code tabulation area (“zip code”). It illustrates the extreme concentration of trips within a few small areas, most prominently San Francisco. Within San Francisco, trips are further concentrated within the downtown core on the city’s most congested streets where the city prioritizes sustainable, space-efficient modes of travel, such as transit, bicycling and walking.

When were TNC trips taken?

This section is limited to Uber because Lyft’s 2020 TNC Public Annual Reports are missing required data and time information necessary for temporal analysis.

Figure 7 shows the average Uber trips by day of week for the 6 months prior to the COVID pandemic and the first 6 months during the COVID pandemic. The figure shows that Uber trips steadily increased from Monday to Friday, are at their highest on Friday and Saturday, and their lowest on Sunday. It further shows that trips declined by 80% during the first 6 months of the COVID pandemic.

Figure 8 shows Uber trips by time of day for a typical weekday and average Friday before and during the COVID pandemic.² Prior to the COVID pandemic, trips had a diurnal distribution during typical weekdays: low trip volumes during late night, peaks of activity in the morning and early evening when roadway congestion is most severe, and sustained but lower volumes throughout the midday. Fridays had a similar morning peak, but higher trips throughout the midday, a much larger evening peak, and a third late-evening peak. During the COVID pandemic, Uber trips decreased substantially and time- of-day profiles were flatter, and peaked earlier, in the mid-afternoon.

How many miles did TNCs drive?

VMT is a measure of the total amount of travel. It is used in environmental analysis to calculate emissions and is a key indicator of driving demand.

Table 19 shows the VMT reported by each company. Uber’s reported VMT ranges from 662 million to 1.6 billion, a difference of 960 million. The CPUC redacted VMT data from Requests Accepted and reported 1.1 billion VMT in Number of Miles. Fleetwide VMT is unknown due to internal inconsistencies and data redacted from Lyft’s reports. Fleetwide VMT could range between 1.7 billion and 2.7 billion, or even exceed these figures.

How many total hours of service does each TNC provide?

Total hours of service is a measure of the service provided, and when compared with completed trips or VMT can give insights into service efficiency. The number of hours worked are reported for each driver on each day worked by that driver in the Number of Hours report.

Table 20 shows the total and share of driver hours reported by each company. Uber reports 46.9 million hours and Lyft reports 52.4 million hours. Uber reported 47% of the total hours, which is much lower than their share of trips presented in Chapter 3 where, depending on the report, Uber’s share of trips could be as low as 60% or as high as 72%. This could either mean that Lyft drivers log many more hours for each trip they provided, effectively parked or driving empty more of the time than Uber, or Uber and Lyft are not reporting trips or hours the same way.

How many TNC trips are "pooled"?

A “pooled” TNC trip is a trip when a passenger indicates they are willing to share a ride with another passenger in exchange for a reduced cost. A pooled trip is “matched” when two or more passenger requests are put into a single driver itinerary that results in the passengers sharing some portion of their trip. In theory, if pooling led to sufficiently high vehicle occupancy rates, it could reduce VMT enough to compensate for the increased VMT due to TNC deadheading and due to shifts to TNCs from lower VMT modes such as transit, biking, and walking.

Figure 9 compares shares of pooled trips out of all completed trips, based on the Requests Accepted and Requests Not Accepted reports. About 31 million (14%) of all completed TNC trips were requests to be pooled. Only 16 million were successfully matched with another passenger. In other words, more than half of pool-requested trips are functionally solo TNC trips.

Pooling services were suspended starting in March 2020 due to the COVID pandemic. Lyft’s reports withheld trip dates and times, so the effect of the pandemic on Lyft’s overall pooling rates cannot be evaluated. Uber’s data indicates that 85% of all their trips during the reporting period of September 2019 to August 2020 occurred before shelter-in-place orders went into effect on March 17, 2020. Figure 10 shows that 15% of Uber’s pre-COVID trips were requested to be pooled, and 10% were successfully matched.

Figure 11 shows the pooled requests received by each company. Uber receives more total pooled requests, accepts more, and matches more of them than Lyft does. Uber received 20.7 million requests for pooled trips, of which 20.0 million (96%) were accepted, and 12.7 million (61%) were matched. Lyft received 12.4 million requests for pooled trips, of which 11.3 million (91%) were accepted, and 3.4 million (27%) were matched.

Where are requests not completed?

Requests for TNC trips may not result in completed trips for a number of reasons. For example, a request may not be successfully matched with an available driver, or may be accepted by a driver and then cancelled, or a passenger may cancel their request after some time has passed. The TNC company, the driver, and the prospective passenger each play a role in whether a request results in a completed TNC trip. The trip acceptance rate is the number of trip acceptances divided by the number of trip requests, expressed as a percentage. Trip acceptance rates may reveal implicit or explicit biases if, for example, drivers are less likely to accept trip requests from some areas compared to others.

Extensive discrepancies in Lyft’s aggregated request data make it impossible to perform meaningful analysis of trip acceptance rates. Figure 12 shows areas where Uber and Lyft have reported completing 100% of trip requests. Uber has perfect trip completion rates in only a handful of zip codes, within which it received fewer than 400 total trip requests. Lyft reports perfect trip acceptance rates in half of the zip codes where it provided trips, including all of Sacramento County, and most of San Diego and Santa Clara counties. This implies, for example, that of the 4.2 million trip requests received in Sacramento County alone, not a single one was ever cancelled by a passenger, or not accepted by a driver, or not matched with an available driver. Across all of these zip codes Lyft received more than 26 million trip requests. It’s extremely unlikely that Lyft’s reported trip completion rates in these zip codes are accurate.

< 2. Reporting Compliance & Integrity 4. Public Safety >