Back in April, it seemed the City of Seattle had finally made peace with Uber and Lyft; after years of legislation and lawsuits, they settled in court and all parties agreed to work together to determine a fair compensation standard for TNC drivers. Sadly, it was not to last: earlier this month a new skirmish broke out, with dueling studies, academic cat-fighting, and some big policy questions coming to the forefront.
The war started in 2015, when Council member Mike O’Brien spearheaded a legislative effort to grant collective bargaining rights to Uber and Lyft drivers. The companies, joined by the U.S. Chamber of Commerce, tried to stop it, both in Council Chambers and in court. That worked its way through the court system until in May 2018, the 9th Circuit handed the city a major setback in ruling that the city is not exempt from federal antitrust laws. In January of 2019 the Council amended the collective-bargaining ordinance to strip out compensation as a topic for collective bargaining in order to stay out of antitrust trouble, but nevertheless the companies pressed on with their lawsuit. However, in a clear shift of strategy, last September Mayor Durkan proposed her “fare share” package, increasing the tax on TNC rides and proposing a path for directly establishing a minimum wage — following the lead of New York City.
Since then, the city won a battle in the lawsuit, gaining the right to conduct discovery on Uber and Lyft. Shortly after that, the sides laid down their weapons, settled the case, and issued joint statements agreeing to work together on a compensation model for drivers.
In public, the truce has continued: just last month the City Council stripped mention of TNC drivers out of a bill requiring “gig economy” companies to pay premium pay to their drivers, so that both sides could stay focused on the bigger fish, the minimum wage. But behind the scenes, the war has raged on over how to calculate the minimum wage for drivers. At this point the argument is no longer about whether to regulate the TNC businesses; that ship has sailed. The debate is over what the minimum wage should be. The companies, of course, want it to be as low as possible. And in labor-friendly Seattle, the city wants it to be as high as possible.
Last fall’s “Fare Share” ordinance called for a task force to be set up under the supervision of the Office of Labor Standards (OLS) to study and recommend a minimum compensation standard for drivers that is comprised of at least the equivalent of the hourly minimum wage in the city’s existing minimum wage standard schedule, plus “reasonable expenses.” But the scope of “reasonable expenses” turns out to be surprising. It’s split into two types: “mileage expenses” and “non-mileage expenses.” Mileage expenses are what we would expect: gasoline, car maintenance and repairs, license and registration fees, etc. But “non-mileage expenses” is a broad, eyebrow-raising list:
- the amount of employer-side payroll taxes that TNC drivers must pay;
- business license fees that TNC drivers must pay;
- compensation for meal periods and rest breaks;
- compensation for paid sick and safe time;
- cost of workers’ compensation insurance;
- cost of paid family medical leave insurance;
- cost of medical, dental and vision insurance.
So “non-mileage expenses” means “compensation, benefits, and the expenses associated with being an independent contractor.” The list makes it very clear what the city’s intent is here: it wants Uber and Lyft to treat their drivers as employees.
One of the long-standing issues in the relationship between the companies and the city is access to data about Uber’s and Lyft’s business. The city would like data about driver hours, payments, routes, and everything else that would help them determine what fair compensation might be and other pertinent regulations. But Uber has given little, and Lyft has given almost none. Their stated reason is confidentiality: since the City of Seattle is a government agency, any data given to them is subject to release under the state Public Records Act — release to the public, but also to each other. Both companies (probably rightly) believe that the data the city is asking for reveals confidential information about their company that they don’t want their competitors to see. However, both companies have given the same data to New York City’s Taxi and Limousine Commission (TLC), the regulatory board there that imposed the minimum wage. But according to a the companies, the public records and data protection rules there allow for appropriate confidentiality to be preserved. Here in Seattle there have been conversations over the years about setting up a secure data repository at the University of Washington’s school of public policy, where both companies could securely deposit their data and UW researchers could analyze all of it together on behalf of the city. But UW too is a government agency and is also subject to the Public Records Act, a deal never came together, and last fall when OLS’s efforts to study and recommend a minimum compensation standard began, it had no data to work with. So it commissioned a survey of TNC drivers through a company called PRR Inc., and it contracted with Dr. James Parrott of the Center for New York City Affairs at the New School and his colleague Dr. Michael Reich of U.C. Berkeley to analyze the data and recommend a minimum compensation standard for Seattle TNC drivers.
OLS knew what they were buying: Parrott and Reich had done the same study for New York City’s TLC and had produced a very labor-friendly recommendation, in line with their reputation. In fact, OLS circumvented the City of Seattle’s rules for bidding out contracts in order to make sure they could hire Parrott and Reich. The city’s procurement rules dictate that any contract of $54,000 or more must be bid out; so instead OLS signed a contract in November 2019 for $53,000 (you will no doubt be unsurprised to learn that the city signs a lot of $53,000 contracts). The contract specified that a draft report would be delivered on March 13, with the final report delivered on March 31 (OLS’s deadline as specified in the Fare Share ordinance). Then, on March 23, OLS signed an addendum to the contract, which kept all the same work items and deliverables, moved the final report deadline out eight days to April 8, and added a whopping $33,000 to the payment for a new total of $86,000. A 62% increase in the contract payment for eight days more work, and a blatant end-run around the no-bid contract rule. When OLS was asked for an explanation for the addendum, a spokesperson for the Mayor’s Office replied by email: ” In short, the issues ended up being more complex and the economists needed more time and resources to finish the study.” That answer is unconvincing: the economists had done the same study previously for New York City and were intimately familiar with the issues going into the contract with Seattle, and the addendum only gave them eight more days.
Nevertheless, with their preferred consultant signed up the city gave Parrott and Reich the survey results from PRR and sent them off to do their analysis. However, unbeknownst to them Uber and Lyft had commissioned their own study by a team of researchers at Cornell University’s School of Industrial and Labor Relations led by Dr. Louis Hyman. Cornell, located in upstate New York, is subject to the same data protection rules as New York City, and in fact has a Restricted Access Data Center to support exactly the kind of research study that Uber and Lyft wanted them to do: each company could securely deposit its data, neither could see the other’s data, and the researchers could access all of it in a closed, secure environment. So while Parrott and Reich were working with survey data, Hyman and his team were working with actual, logged data from a week of Uber and Lyft operations in Seattle.
Anyone who has watched the TNC drivers lobby the Seattle City Council over the last five years can testify to the fact that they are well organized. The individuals who drive for TNCs as a full-time job are particularly well organized, and have the backing of the Teamsters in demanding that the city get involved to help them get a better deal with the companies. But the full-time drivers care about a different set of issues than those who drive less than full-time, and that makes for a challenging political split. The full-time drivers are trying to live off of their earnings: they care about wages, but also about benefits, working conditions, sick leave, etc. The part-time drivers of course want to make more money, but flexible hours is also a priority for them. Part time drivers include retired people and students looking to make some extra money on the side, as well as working people who drive as a second job to supplement their income. The part-time drivers, as a group, aren’t nearly as well organized as the full-time drivers when it comes to lobbying for their interests, and they don’t have a powerful union like the Teamsters backing them up (though Uber and Lyft jump in to help from time to time — we’ll discuss why in a moment).
So it should be no surprise that PRR’s survey results don’t look like a random sample of Seattle’s TNC drivers; the organized full-time drivers took it seriously and responded in number. In fact, when looking simply at the demographics in the survey results, as reported by Parrott and Reich, they don’t match up well with the U.S. Census Bureau’s numbers. According to the Census Bureau, 50% of the taxi and TNC drivers in King County are black, and 27% are white non-Hispanic; but PRR’s survey, of which 3,469 out of 6,554 gave race/ethnicity information, had nearly the reverse: 23.1% black and 44.6% white non-Hispanic. In their report, Parrott and Reich did some cherry-picking and sleight of hand to make the data look like it matched other demographic data more than it really did. For example, they looked at percentages of different types of cars driven, which closely matched between the survey and the King County’s TNC license data –but only if you combine categories together. Sea-Tac Airport requires TNC drivers picking up passengers there to have a hybrid or electric vehicle, and full-time drivers often buy a hybrid sedan as their “work vehicle” so they can work the Airport. The survey data shows substantially more hybrid cars as a percentage than King County’s license data, more evidence that full-time drivers are over-represented in the survey data.
But Parrott and Reich are fine with that, because it skews the numbers in the direction they like: their policy goal is to maximize the number of individuals driving full-time for Uber and Lyft — at the expense of part-time drivers. They don’t hide this; they spelled it out in their recommendations to the New York City TLC, and in their report to the City of Seattle. And that’s why the City hired them.
Parrott and Reich’s report was made public on July 7, and in a shock to them, the Cornell research team published its report on the same day. Both reports, and both research teams, headlined their “bottom line” finding as to how much TNC drivers make in Seattle; and of course their numbers were very different; we’ll come back to that. But the more glaring contrast was the extent to which it showed how inaccurate the PRR survey data is; the underlying demographics don’t match, and in particular it highlights how over-represented full-time drivers are in the PRR/Parrott and Reich study. While it’s technically possible that Uber and Lyft fabricated the data that they gave to Cornell (the Cornell team looked for inconsistencies between the two independently-supplied data sets from Uber and Lyft, and found none), for all intents and purposes it is the “ground truth” in representing a week of TNC driving in Seattle. We can still argue with Cornell’s analysis, but its underlying data set is sound. On the other hand, if the Parrott/Reich underlying data set is heavily biased — and it clearly is — then their numerical analysis on top of it is practically worthless.
According to Parrott and Reich and the survey, 32% of Seattle TNC drivers are full-time, i.e. they drive 32 hours or more per week. According to Hyman’s team and the Uber/Lyft data, 15% are. Parrott and Reich say 21% drive 10 hours or less per week; Hyman says it’s more like 50%. The distributions are completely different across the two studies:
You can imagine how Parrott and Reich reacted to all this: they fired off an angry written response. Hyman and his team, with the help of Uber and Lyft, returned in kind. But academic cat-fighting aside, the truth is that neither research team, and their respective report, is neutral; both were hired to present a certain skew (regardless of whether they admit it), and both delivered that in their analysis. But even if we ignore the conclusions and recommendations in the reports — and we should — we can learn a lot about the policy issues from examining the way each team did their analysis.
The goal of both studies is the same: answer the question “how much do TNC drivers make?” Parrott and Reich, on behalf of Seattle, want to make the number as small as possible so they can make the case to raise it. Hyman, on behalf of Uber and Lyft, wants to make the number as large as possible so they can lobby to leave it alone. An objective point of comparison we can use is the minimum wage for regular employees, represented as dollars per hour.
The rules for gaming this system aren’t hard. If you want to make it look smaller, decrease the numerator (dollars) and increase the denominator (hours). If you want to make it look bigger, do the reverse. And that’s exactly what the two teams do.
If we want to calculate compensation, then Dollars = wages + tips – expenses. Wages is what Uber and Lyft pay their drivers directly; tips are paid by customers (but thankfully through the app so it’s captured in the app data that Cornell worked with). But tips are not included in regular employees’ minimum wage, so that’s a major point of disagreement: if we want to objectively see how much a driver is making, we include tips; if we want to compare it to standard minimum wages, we leave it out. Or to be more blunt: if you want compensation to look higher, you include it; if you want it to look lower, you leave it out. Unsurprisingly, Parrott and Reich left tips out, and Hyman included them.
And then there’s expenses, the really big can of worms. The problem with expenses is that they are inconsistent between part-time and full-time drivers. Full-time drivers often buy a separate car just for TNC driving; they have to pay for licensing it, maintaining and repairing it, etc. A full-time driver logs between 32,000 and 35,000 miles per year; their car will last about four years if well-maintained, so with an average cost of $28,000 to $32,000 for a TNC vehicle, amortized over four years that’s about $7,000 to $8,000 of vehicle expense just for purchasing the vehicle alone.
There are other expenses that all TNC drivers have: among them, gasoline and cleaning are both big-ticket items — though both also scale depending on the number of trips driven.
And then if you are really committed to thinking about full-time TNC drivers as the equivalent of regular employees, there’s health and dental insurance, and things like retirement account contributions.
Parrott and Reich’s approach is to maximize expenses, both because they are prioritizing full-time drivers, and because deducting all those expenses makes the wages look smaller; they even included the cost of a cell phone in their list of expenses. Hyman’s approach is much more parsimonious with the expenses, as one would expect.
As for hours, that is its own mess. The TNC industry divides up a driver’s time in the car into three buckets, known as P1, P2, and P3.
P1 is the time when a driver is logged into the app and is waiting to be dispatched.
P2 is the time a driver spends, after being dispatched, driving to a pickup location and waiting for the passenger(s) to arrive.
P3 is the time a driver spends actually driving from the pickup location to the drop-off location.
Both Uber and Lyft pay drivers for P3 time, using complicated formulas based on both time and distance. They also pay for a small amount of P2 time, if the distance to the pickup location takes longer than a minimum amount of time or the driver must wait more than a certain number of minutes for the passengers to show up. But most P2 time goes unpaid. And all P1 time is unpaid.
Obviously Uber and Lyft have the power to decide the payment structure, and have used that power to minimize drivers’ pay. And one could make a simple argument, as Parrott and Reich do, that Uber and Lyft should pay drivers for all time spent in P1, P2, and P3. But the ecosystem is so much more complicated than that, because both the companies and the drivers are trying to game the system to their advantage. Many drivers — and a large percentage of the full-time drivers — are logged into both the Uber and Lyft apps at the same time, and will take the first dispatch offered on either app. That minimizes their P1 time, but in a way that is opaque to the companies (Hyman’s team identified the overlapping P1 time for drivers logged into two apps and factored it out; the PRR survey also tried to capture the time drivers spent logged into multiple apps). Some drivers take care of other activities during P1 time that could hardly be considered “working” for Uber or Lyft; for example, if you visit the parking lot at Sea-Tac Airport where TNC drivers wait to be dispatched, you will find many students logged into the app while they sit in their car and study.
On the flip side, the companies have their own goals with P1 time. A defining metric of competition between the two companies (and taxis, and public transit) is the amount of time a passenger must wait before being picked up. Uber and Lyft want that number to be as low as possible, but to do that they need to have many TNC drivers spread across the city, in P1, waiting for a dispatch. If they have to pay for P1 time, then having a low wait-time for passengers gets very expensive — prohibitively expensive. Parrott and Reich rightly point out that Uber and Lyft are passing that cost on to their drivers by essentially forcing them to log a lot of unpaid P1 time.
Parrott and Reich believe the answer to this problem is that Uber and Lyft need to pay their drivers for P1 time, but they also need to “better manage their drivers.” By that they mean, “keep the drivers busier with more dispatches, so they spend less time in P1.” That works out great for full-time drivers, who are of course the policy priority for Parrott and Reich. But it’s terrible for part-time drivers, because the only way to keep drivers busier is to have fewer drivers.
Parrott and Reich convinced the New York City TLC to adopt a minimum wage formula that looks like this:
A certain amount per mile, plus a certain amount per minute, divided by the company’s utilization rate: the percentage of an average driver’s time that is spent in P3. The net effect of dividing by the utilization rate is that the wage goes down as drivers spend more of their time in P3. So they log more billable hours, but they get paid slightly less per hour.
Since Uber and Lyft have no control over how much time drivers spend in P1, the only way they can prevent this system from becoming explosively expensive is to cap the total number of drivers. And that is exactly what they did in New York City. The winners were the full-time drivers; the losers were the casual and part-time drivers, many of whom are now prevented from driving at all — or they must schedule their driving shifts in advance, undoing one of the greatest benefits of TNC driving for them: the schedule flexibility. For the past five years here in Seattle, as the debate has raged in Council Chambers on how to regulate TNC driving, the part-time drivers have warned that a system that prioritizes full-time drivers will eventually end up in part-time drivers being forced onto a schedule. And now we see that is exactly what happened in New York City.
At this point, there’s no real point in groping through the wage, expense or hours calculations for either study; the underlying data in the Parrott and Reich study is garbage, and both studies skew their analysis and recommendations in the direction that favors the one who commissioned their study. Instead, we should be focused on the key policy questions, and forcing city officials to be transparent about who they are prioritizing in the policy decisions they are making in setting a minimum wage. Once the policy questions are answered, there is enough raw data in the Cornell report to calculate out the proper minimum wage.
The key policy question, one that drives so many others, is whether the city should prioritize full-time drivers over part-time and casual drivers. As we’ve seen above, it affects almost everything: what expenses to include, how much the system needs to try to maximize paid hours for full-time drivers, the economic model for the companies. According to the Uber/Lyft data set that Cornell analyzed, only 15% of TNC drivers are full-time, but for them the job is their livelihood. Should they be prioritized over people driving as a second job to help make ends meet or deal with emergency expenses? Should the city try to increase the number of full-time TNC driver jobs? Should the city be picking winners and losers at all between full-time and part-time drivers?
But there are other policy considerations as well, such as the environmental and transportation-grid impacts of managing the number of TNC drivers on the road at any given time. However, that also ties in, at least in the short term, with the COVID-19 pandemic, and how long it will be before people see public transportation as safe. If it takes a while, Uber and Lyft may be in high demand and instituting a wage policy that incents minimizing the number of drivers to increase driver utilization may be the wrong short-term action.
There are equity issues as well. For example, driving a taxi or a TNC vehicle is often one of the few jobs available to immigrant men in the United States when they first arrive, and a minimum wage policy could either help or hurt them depending upon their circumstances: are they trying to earn a living wage to support a family, or are they driving in their spare time while they take classes by day?
And then there’s the policy question of whether Uber and Lyft have too much economic power over their drivers, and the related question of whether in practice TNC drivers are employees and should be treated that way with the full set of employment benefits that full-time employees receive. Federal law doesn’t yet recognize that they are employees of the company (though California now does), but the winds of change seem to be blowing in that direction. On the other hand, neither Uber nor Lyft is actually profitable; both continue to lose hundreds of millions of dollars each year. Parrott and Reich engage in some questionable math to argue that the companies’ commissions per ride are too high and ought to be brought down; but their calculations are based entirely on local expenses and ignore the corporate R&D, operations, marketing and other administrative overhead that the company carries. It’s those central costs that are keeping the two companies in the red. And it’s entirely possible that Seattle’s efforts to regulate the market to better support workers tips it to the point where the companies simply can’t make the business work here and decide to pull out. We saw that with bike-share and with car-share; it could happen to TNCs too. Or worse: only one company survives and has a local monopoly.
The word on the street is that the Office of Labor Standards and the City Council are likely to pick the issue of minimum wage for TNC drivers soon, possibly even in early August.
I hope you found this article valuable. If you did, please take a moment to make a contribution to support my ongoing work. Thanks!