Executive Summary

Transportation Network Companies (TNCs) such as Uber and Lyft began providing on-demand, app-based transportation ride-hail services in California in 2009, and have been required to submit annual reports to the California Public Utilities Commission (CPUC) since 2014. These TNC Annual Reports contain information about a wide range of topics, including, but not limited to, trip requests and completions, collisions, and incidents, assaults and harassment, and miles and hours driven. The CPUC has designated the TNC Annual Reports from 2020 onward as public, and a proposed decision would make all past reports public. The 2020 reports are the first reports made public by the CPUC.

This information is of great interest to cities like San Francisco where TNCs operate. In February 2022, the San Francisco County Transportation Authority (Transportation Authority) requested the 2020 public TNC Annual Reports for Uber and Lyft from the CPUC, which provided the reports later that month. These reports cover the period from September 2019 to August 2020 and have been highly redacted by the CPUC.

The CPUC also regulates the nascent autonomous vehicle (AV) passenger service industry and is developing AV regulations in the very same proceedings as TNC regulations. AV passenger services are like TNCs in many ways, but with the important distinction that they plan to, and in some cases already do, use self-driving cars without any human safety driver. AV passenger service companies submit quarterly reports which, by contrast, are routinely published by the CPUC, but similar to the public TNC Annual Reports, are heavily redacted.

This report analyzes and summarizes the 2020 public TNC Annual Reports, and is intended to inform the Transportation Authority Board, as well as state and local policy-makers, and the public, on general characteristics of the TNC market, and on the performance of TNCs in terms of public safety, labor, the environment, and accessibility. Unredacted TNC public Annual Reports could also be used to validate San Francisco’s Prop D Traffic Congestion Mitigation Tax receipts, which have been irregular.

The following findings summarize the Transportation Authority’s analysis of the 2020 TNC Annual Reports, which cover the six months before the COVID pandemic and the first six months of the COVID pandemic. Transportation patterns changed during the pandemic and continue to evolve. When the 2021 and 2022 Annual Reports are disclosed consistent with the CPUC’s data confidentiality rulings, the Transportation Authority will prepare summaries for these reporting years as well.

Key Findings


The public Annual Reports are incomplete by the standards set by the CPUC.

In the 2020 public Annual Reports, Lyft reported 36% of the required data as measured by the percent of required public fields and records that are present and unredacted. Uber reported 99.99% of the required data.

Uber’s and Lyft’s data is internally inconsistent.

For example, Lyft’s Annual Reports include two different totals for the number of completed trips in the state, differing by 49.7 million trips, or 81%. Uber’s Annual Reports also include two different totals for the number of completed trips, differing by 9.3 million trips, or 6%. As a result, it is not possible to identify basic facts such as the number of completed TNC trips that occurred in California in the 2020 reporting year.

Many reporting requirements are not clearly defined, preventing effective regulatory oversight.

For some types of data — such as collisions, DUI complaints, law enforcement citations, and accessibility data, the CPUC provides examples but not requirements about how to report the data. As a result, the companies report this data differently, preventing effective regulatory oversight.

Due to more extensive redactions in the 2021 Annual Reports, a less extensive evaluation of consistency is possible.

However, where consistency can be evaluated, inconsistencies are reduced in some instances. For example, Uber’s number of completed trips in the Requests Accepted and Aggregated Requests Accepted in their 2021 Annual Reports are perfectly consistent, and Lyft’s number of completed trips in these reports are nearly perfect, differing by 0.004%.


TNC trips are highly concentrated in a few urban areas.

TNCs and ride-hail trips are an urban, not a statewide, transportation issue, as shown in Figure 1.

Figure 1. Trip Density by Zip Code from September 2019 to August 2020

Nearly two-thirds of TNC trips are in San Francisco, Los Angeles, and San Diego counties.

Within these counties, trips are most highly concentrated in just a few areas: San Francisco’s downtown core, Los Angeles’ Westside, and at the San Diego airport, respectively.

San Francisco has 500 times more TNC trips per square mile than the rest of California.


Lyft reports 3 times more total public safety incidents per trip than Uber, and 30 times more assaults and harassments per trip.

Figure 2 shows the incident rate per hundred thousand trips and suggests that the companies may be reporting public safety incidents differently, pointing to the need for increased review by regulators.

Figure 2. Incidents per 100,000 trips from September 2019 to August 2020


California law limits drivers providing passenger transportation to “10 hours in any 24-hour period unless 8 consecutive hours off duty have elapsed.” The Annual Reports include 1.3 million days during which drivers drove more than 10 hours. While this report alone cannot confirm that a drive-time violation has occurred, the reports do not account for additional factors like drivers who may be in violation due to driving for both services, or whose shifts straddle 2 or more calendar days. No public enforcement actions have been taken regarding possible violations of legal drive-time limits.

Figure 3. Driver Days by Hours Worked from September 2019 to August 2020


Lyft’s redacted reports prevent environmental oversight.

Lyft’s annual report withholds key data items necessary to estimate emissions: vehicle miles traveled (VMT), geographic trip origin and destination data, and vehicle make, model, and year.

Uber produced 494,000 metrics tons of CO2 in reporting year 2020, based on a Transportation Authority estimate. Almost 30% of those emissions occurred with no passengers in the vehicle.

This is comparable to the CO2 emitted by the 2020 Caldwell Fire in northern California, which burned 81,000 acres.


Less than half of all Wheelchair Access Vehicle (WAV) trip requests are served.

Under the TNC Access for All Act (Senate Bill No. 1376), the CPUC established a program where TNCs collect a fee from riders for every TNC each trip, which is then used to subsidize on-demand transportation for persons with disabilities, including wheelchair users who need a WAV. But even with this additional financial support, less than half of WAV trip requests are fulfilled.

Uber provides nearly all TNC WAV trips in California.

Uber provided 16 times as many WAV trips as Lyft.

WAV Requests217,93511,605229,540
Completed WAV Trips101,5946,158107,752
Completion Rate47%53%47%


The 2020 public TNC Annual Reports reveal numerous issues related to basic compliance with data reporting requirements, and the integrity of the data itself. At the most basic level, Lyft’s 2020 Public Annual Reports are incomplete according to the rules adopted by the CPUC: 8 of their 19 public reports are missing required data fields, and 64% of all Lyft’s required public data items are missing. By contrast, Uber’s 2020 Public Annual Reports contain all but one of the required public fields. This suggests that reporting rules are applied or enforced inconsistently.

The data contained within the 2020 TNC Public Annual Reports is often self- contradictory and internally inconsistent. For example, Uber’s total number of trips differs by more than 9 million from one report to the next, while Lyft’s differs by nearly 50 million trips. In some cases, the data submitted is erroneous or unreasonable: Lyft’s reports indicate that it accepted 100% of trip requests received across vast swaths of California. These issues are exacerbated by, if not directly caused by, data reporting requirements that are, at times, unclear; lack of quality assurance or enforcement of quality standards; and application of confidentiality standards that are not consistent with the CPUC’s orders.

The lack of accurate, timely and transparent data has left localities without necessary information to support a basic understanding of TNC operations in their jurisdictions or their potential impacts. Timely and accurate data is fundamental to developing sensible public policy and to identify where it is appropriate to seek improved oversight. The pervasive data quality issues suggest the need for quality control, greater adherence to CPUC direction regarding disclosure of data, and enforcement of reporting requirements.

TNCs operate almost exclusively in dense urban areas and during the busiest times of day, where they have been shown to exacerbate congestion and reduce transit ridership. As the reports show, there may be public safety risks, environmental harm, and issues of equitable access to TNC services. California cities, which have no regulatory authority over TNCs, rely on the CPUC to manage impacts, enforce regulations, and provide relevant, timely, thorough, and quality data to support the effective development of informed public policy. Cities face similar regulatory reliance on CPUC regarding AV passenger services. CPUC’s public AV reports are following a similar pattern to the public TNC reports of redacted data. Timely, thorough, quality data reporting is essential to effective research and policy-making for both TNC and AV ride-hail passenger services, and effective regulation is critical as these new services become more widely available.