Occupational licensing is often intended to protect consumers by ensuring that workers meet minimum standards of quality and safety. More than 22 percent of US workers now need a government-issued license to work in their occupation. Licensing requirements can include training, examinations, background checks, fees, and administrative procedures. They can also restrict entry into an occupation and raise costs for workers and consumers. Whether these requirements actually improve the services consumers receive remains an important policy question.

Ridesharing provides a useful setting for examining that question. In most licensed occupations, the quality of service is difficult to measure, but ridesharing generates detailed records of every trip, making it possible to measure service quality and safety-related driving behavior. Uber and similar platforms also already use several mechanisms to monitor drivers, including background checks, rider ratings, and performance-based deactivation. These systems may incentivize safe and high-quality service even when the government does not require drivers to hold an occupational license. Our research examines whether licensing adds meaningful consumer benefits in a market where private platforms already monitor performance.

We used proprietary trip-level Uber data from 2017 and early 2018 that linked individual rides with driver and rider characteristics, star ratings, and telematics-based measures of driving behavior. The telematics measures captured hard braking and hard acceleration, behaviors associated with a greater risk of crashes. We studied two settings in which licensed and unlicensed drivers could serve similar trips: a cross-border market involving New York City and New Jersey, and Houston after the city’s rideshare driver licensing rules were eliminated.

In both settings, Uber’s dispatch system assigned riders to nearby drivers regardless of licensing status, and riders could not see whether their driver was licensed. We focused on trips where the two closest available drivers differed in licensing status, so whether the rider got a licensed or unlicensed driver depended largely on which one happened to be slightly closer. This let us compare otherwise similar trips, even though drivers differed in experience and other characteristics. 

In the New York City–New Jersey setting, New York City drivers needed a Taxi and Limousine Commission license, while New Jersey drivers did not. The New York City license required a 24-hour course and examination, a defensive driving course, and a fingerprint-based background check. Although unlicensed New Jersey drivers could not pick up riders in New York City, both licensed New York City drivers and unlicensed New Jersey drivers could serve trips in New Jersey.

The New York City–New Jersey analysis includes more than 213,000 trips. Licensed drivers received slightly lower rider ratings than unlicensed drivers: Their average ratings were less than 0.5 percent lower, and they were about 0.9 percentage points less likely to receive five stars. We also found no difference in how often licensed and unlicensed drivers received one-star ratings. Because licensing sets a minimum standard for entry, the lowest-rated trips are where its effects would be most expected to appear.

We found little consistent evidence that licensed drivers drove more safely. On average, they did not have lower rates of hard braking or hard acceleration. One measure indicated that licensed drivers were 15 percent less likely to have trips with a particularly high share of hard-braking events. However, that result weakened and was no longer statistically significant under several alternative ways of analyzing the data. Overall, the New York City requirements did not produce economically meaningful improvements in either ratings or driving behavior.

We also examined Houston, where drivers had previously needed a local license and a fingerprint-based background check but did not need substantial training. In 2017, Texas eliminated local licensing requirements for rideshare drivers. The change allowed drivers who had not obtained a Houston license to pick up passengers throughout the metropolitan area.

This policy change allowed us to compare previously licensed drivers with drivers who entered Uber after deregulation. These new, unlicensed drivers differed substantially from previously licensed drivers: They had much less platform experience, were about five years younger on average, and were more likely to be women. Despite those differences, they were statistically indistinguishable from previously licensed drivers in both rider ratings and measures of hard braking and acceleration. Our estimates were precise enough to rule out anything more than a small safety advantage for previously licensed drivers on these measures.

We performed a second examination of Houston, this time comparing previously licensed drivers with drivers who had been active before deregulation but had not held a Houston license, meaning they could pick up riders only outside city limits. Rider ratings did not differ between the two groups, but previously licensed drivers had higher rates of hard braking and hard acceleration on several measures. These results do not show that licensing made drivers less safe, since the groups differed in prior experience and other characteristics. But they provide no evidence that Houston’s licensing and screening system selected drivers with safer observable driving behavior.

Our findings suggest that occupational licensing did not consistently improve the service quality or safety of Uber trips in these settings. The results do not mean that every form of occupational regulation is unnecessary. The study examines one platform, two metropolitan areas, and licensing regimes that ranged from screening requirements to moderate training requirements. It also cannot directly measure rare outcomes, such as crashes. Additionally, rider ratings are heavily concentrated at five stars.

Still, the evidence is consistent with the view that the marginal benefits of licensing are limited when other quality-control mechanisms already exist. Uber’s monitoring systems, rider feedback, and threat of deactivation give drivers continuing incentives to perform well. In such markets, policymakers should weigh any additional consumer-protection benefits from licensing against its costs, including reduced entry, higher barriers for workers, and—as other research finds—higher prices for consumers.

Note
This research brief is based on Jonathan Hall et al., “Occupational Licensing of Uber Drivers,” National Bureau of Economic Research Working Paper no. 35635, August 2026.