Analysis reveals discrimination by Boston cab drivers, but says some of it may be fixable
Boston Fed paper indicates bias partly based in ignorance about potential earnings
Boston taxi drivers appear to discriminate against female, black, and Asian residents by working fewer hours in neighborhoods more heavily populated by those groups, but that bias is partially based in ignorance about potential earnings, according to a Federal Reserve Bank of Boston paper.
“The Supply Side of Discrimination: Evidence from the Labor Supply of Taxi Drivers” examined the hours taxi drivers spent in various neighborhoods around Boston over a six-year period. It found persistent discrimination in areas with higher female, black, or Asian populations, as drivers failed to supply sufficient work hours to meet local demand.
The study focused on how Boston taxi drivers react to uncertainty about where and when demand for rides will rise and fall. The paper found that when faced with that uncertainty, drivers partly rely on area demographics to determine how much to work in different locations, spending fewer hours in areas with higher percentages of female, black, or Asian residents, even when the local earnings opportunities are, in fact, the same. For instance, all else equal, drivers are less likely to go to those areas looking for new fares after they drop off passengers in a given location. That means drivers were willing to work longer to find the next fare by searching outside those areas, potentially at the cost of time and money.
But the paper also found the discrimination generally decreased as drivers gained experience and learned how to better anticipate the flow of potential business in different areas.
“In a case like this, people can perhaps be given the relevant information to alter their behavior,” said the paper’s author, Osborne Jackson, a senior economist with the Boston Fed’s New England Public Policy Center. He added that the paper doesn’t rule out animosity against certain groups as a possible factor in driver decisions as well, in which case a strategy to reduce discrimination by getting drivers better information about earnings opportunities likely wouldn’t be as effective.
The research focused on labor market supply-side discrimination by workers, in contrast to studies that typically examine labor market demand-side discrimination by employers. Jackson said it’s important to understand these supply-side discrimination patterns, especially given that some part-time, self-employed, and “gig” economy workers—for example, certain freelancers and independent contractors—can choose their own hours and customers. Such choices can have real consequences, Jackson said. For instance, his paper estimated that at least four passengers a day may have gone without taxi rides due to local demographics in an area with the typical (specifically, average) Asian representation in Boston, meaning more people would have had to take time to find alternative transportation at potentially significant costs.
“That’s just one demographic group, in one area, on one day,” Jackson said. “One has to think about the larger impact across groups, areas, and days. And the drivers are, in a sense, leaving money on the table because they don’t have better information.”
The data in Jackson’s study, a sample of more than 14.1 million taxi trips taken in the Boston area between 2010 and 2015, come from the electronic credit card processing devices required in Boston cabs that record various trip details, including fares and when and where trips start and end.
The paper suggested that the drivers most likely to practice the discriminatory behavior found by the analysis are those with less than a few months of experience, some of whom exit the industry within a year. As driver experience increased, the discriminatory behavior decreased.
Jackson said experienced drivers tended to drive in different and fewer areas than newer drivers, as well as work regularly in a smaller set of “core” areas. Such location choices may reflect that drivers stopped profiling areas by their demographic makeup and began to choose routes based on an improved sense of the best earnings opportunities.
“This driver learning happens without any intervention,” Jackson said. “The question is, what could hasten that process?”
For more details, please check out the full paper.