As concerns about driver pay and employment status have taken center stage, many cities and states are perplexed by a simple question: “How much do ride-hail drivers earn?”
Finding an answer should be simple.
You start by understanding how much drivers earn, then estimate how much money they spend on their driving business. From there, a few math calculations should give you a straightforward answer.
Simple enough, right?
Wrong.
As the City of Seattle is learning, estimating just how much ride-hail drivers make can be a complicated matter.
Just like cities and states across the country, the City of Seattle is investigating whether Seattle ride-hail drivers need a minimum wage standard, and if so, what that standard should be.
They turned to economists James A. Parrott, from the Center for New York City Affairs at the New School; and Michael Reich, of the University of California, Berkeley, and commissioned them to conduct a study of ride-hail wages. The study took place during the entire month of October 2019, and its goal was to determine just how much ride-hail drivers in Seattle were making.
Meanwhile, Uber and Lyft decided to conduct their own study. They hired a team from Cornell University that was led by economic historian Louis Hyman. The Cornell study was conducted during one week in October 2019, and was delivered to the City of Seattle about a week before the Parrott-Reich study was published.
As it turns out, the two studies produced vastly different results.
In their report, Parrott and Reich demonstrated that drivers were making much less than the minimum hourly wage for workers in Seattle, which is $16.39 per hour. Based on data they collected, Parrott and Reich calculated that after subtracting expenses, drivers earned approximately $9.73 per hour.
In stark contrast, the Cornell University study concluded that drivers in the Seattle market were netting $23.25 per hour — significantly more than the minimum hourly wage.
So who is correct here? And how could academic studies from two reputable institutions arrive at such different conclusions?
The disparities between these two studies boil down to data, and data-backed assumptions. In this report we will review the methodologies and findings of both studies, and do the following:
- Explain how differences in earnings data used by the two teams contributed to the vastly different results;
- Examine the differences between the two teams’ calculations of driver expenses;
- Provide a recalculation of driver earnings over the course of a few months, incorporating ground truth data obtained by Gridwise directly from drivers through our app.
Different earnings data leads to different results
One of the reasons Uber and Lyft claimed their study was the more fact-based of the two was the quality of their information.
The Cornell researchers who conducted the Uber/Lyft study used ground truth data obtained from 14,000 ride-hail drivers. The data wereharvested directly from the records of drivers operating on the Uber and Lyft platforms.
Unfortunately for Parrott and Reich, they did not have access to this ground truth data. According to media reports, they requested it, but were told by Uber and Lyft that due to reasons of confidentiality, the companies could not share the information.
With no access to ground truth data, Parrott and Reich had to rely on survey data from some 7,400 drivers in the Seattle market, and from that data they determined how much money drivers were making.
Although surveys can offer good insight, they are inherently less accurate than data obtained directly from the source. So, the survey data collected by Parrott and Reich provided an opening for Uber and Lyft to criticize the study. When it was released, Uber’s spokesperson was quick to offer a harsh, pointed critique. As a July 12, 2020 New York Times article states:
“Uber said the Parrott-Reich study “is based on incomplete data and flawed assumptions about drivers’ experiences that are unsupported by facts, evidence or reality.”
Different perspectives on driver expenses create wide variations
The two studies made very different assumptions about driver expenses, which led to radical differences in driver earnings calculations. We’ll explore these differences, and leverage our data to help shed light on what assumptions should be made.
Fuel
The Cornell and Parrott-Reich studies used a similar per-mile estimate for fuel expenses.
Cornell Study: Median price $0.089 per mile, based on the average price for gas in Seattle ($3.255) during one week in October 2019.
Parrott-Reich Study: $0.094 per mile, based on a variety of vehicles (including SUVs), an average MPG rate for cities over three years, and an average annual mileage of 35,000.
The two studies used different methodologies, but came to a similar result.
Maintenance
The Cornell and Parrott-Reich studies used a similar per-mile estimate for fuel expenses. In fact, unlike many other expenses, the Cornell study estimated per-mile maintenance expenses to be higher than the Parrott-Reich study did.
Cornell Study: $0.08 per mile, based on standards set by the American Automobile Association (AAA), using a car owned for five years averaging approximately 75,000 miles of use.
Parrott-Reich Study: $0.0646 per mile, based on actual driver surveys and standard numbers produced by Seattle area car dealers and service centers.
From this point forward, the two studies get further and further apart.
Insurance
While Parrott and Reich believe that insurance costs add up to $.07 per mile, the Cornell researchers did not include any per-mile expenses for insurance.
Cornell Study: Not included, because Seattle requires transportation network companies (TNCs) to provide it. This study maintains that supplemental insurance is “not a necessity.”
Parrott-Reich Study: $0.07 per mile, based on their driver survey.
Uber and Lyft assert that buying additional insurance is not necessary; however, many drivers do indeed purchase additional insurance for their ride-hailing businesses.
To be clear, drivers purchase outside insurance for good reason. If those who haven’t told their insurance companies about their ride-hail driving are involved in an accident, they can face stiff penalties. Drivers are also left uninsured in Periods 0 and 1 if they don’t buy additional insurance.
As flawed as survey data may be, we at Gridwise think understanding how many drivers purchase ride-hail insurance via a survey can be helpful.
We interviewed 734 ride-hail drivers and found that 40% do indeed purchase outside insurance.
This is to ensure they are protected during the periods of time when they don’t have passengers in the vehicle with them, as well as when they are driving outside the platform. Almost all insurance companies charge additional premium amounts for ride-hail drivers.
Here, it seems that having some per-mile expense for additional car insurance would be called for.
Vehicle Acquisition/Depreciation
The biggest single per-mile expense difference between the Cornell and Parrott-Reich studies is the cost of vehicle acquisition/depreciation.
While the Cornell study used a figure of just $0.02 per mile, the Parrott-Reich study used $0.23 per mile.
Cornell Study: $0.02 per mile. The researchers maintain that the ride-hail fleet does not reflect the same types of vehicles normally taken into consideration by the IRS when estimating depreciation. The vehicles most commonly used for ride-hail in Seattle tend to be less expensive, and thus, carry a lower loan or lease payment, and depreciate less. It is also assumed that the driver would have the car even if s/he were not driving for a TNC, so the expense is not due directly to driving for a TNC.
Parrott-Reich Study: $0.23 per mile. These economists used the price of acquisition rather than the price of depreciation. They acknowledge that drivers normally use up the value of a four-year car loan by driving more than they would without working for a TNC.
The key difference here is that the Cornell study assumes that drivers are likely to have their cars already, and spend most of their time driving recreationally, while the Parrott-Reich study assumes many drivers truly invest in their ride-hailing vehicle.
To get a better idea of how ride-hail drivers actually think about their vehicles, we can look to our latest Gridwise study.
We found that 47% of drivers purchased their vehicle strictly for their gig work, meaning most, if not all expenses for their vehicle should be allocated to business use, not personal use.
Our survey also found that of the time drivers are on the road, 71% of that time is spent driving for gig work. This again shows that much of the expense for their vehicle should be allocated to their business.
Cell Phone
The Cornell study does not take cell phone usage into account.
Cornell Study: This study does not take cell phone expenses into consideration when calculating driver expenses.
Parrott-Reich Study: $0.045 per mile, included because drivers require a recent vintage, large-format cell phone with an unlimited plan in order to efficiently perform their duties.
The Cornell study assumes that drivers will have the same cell phones and cell phone plans whether they were driving or not. Understanding how drivers adjust their cell phone expenses because they’re driving is difficult, but should be considered in future studies.
Vehicle Cleaning
The Cornell study does not take vehicle cleaning into account.
Cornell Study: This study does not take cleaning the driver’s car into consideration when calculating driver expenses.
Parrott-Reich Study: $0.04 per mile, included because drivers must keep their vehicles clean in order to ensure high ratings from riders.
While vehicle cleaning isn’t a huge expense, it is a real one for drivers.
While Gridwise is unable to leverage actual vehicle cleaning data for the purposes of this article, we can leverage data from a recent study. Of the 734 drivers we interviewed, more than 53% said they clean their cars every day, and another 27% clean their cars every few days.
Drivers clean their vehicles often because they understand how important having a clean vehicle is to keeping their ratings and tips high. As Ray, a driver in Boston, told Business Insider during an interview: “I want to make passengers feel good, because when they’re feeling good, I get more tips.”
Another driver, Al Castillo from New York City, is also serious about keeping his car clean. “When [passengers] see that your car is clean,” says Castillo, “they know that you know what you’re doing. When they see that, they just relax.”
Licensing Fees
The Cornell study does not take licensing fees into account..
Cornell Study: This study does not take licensing fees into consideration when calculating driver expenses.
Parrott-Reich Study: $0.16 per mile, included because of the regular and extra fees drivers must pay in Seattle.
The Parrott-Reich study lays out various vehicle expenses that only drivers would incur, including:
- A business license that is required by the City of Seattle ($110 per year)
- Washington State vehicle registration fees (estimated cost $148.25)
- TNC vehicle inspection fees ($55 per year)
The Parrott-Reich study also includes driver’s license and vehicle registration fees that drivers may incur whether they were ride-hail drivers or not.
Given that many of these fees are exclusive to ride-hail drivers, and many drivers would not have their vehicles if they were not ride-hail drivers, it seems appropriate to have some licensing fees accounted for in the per-mile expense calculation.
Health Insurance
The Cornell study does not take into account health insurance costs.
Cornell Study: This study does not take health insurance costs into consideration when calculating driver expenses.
Parrott-Reich Study: $0.1076 per mile, included because many drivers (27 percent) do not have health insurance, and an additional 37 percent have incomes low enough to qualify for Medicaid coverage. In the final report of the study, Parrott and Reich assert that drivers should be able to purchase life insurance, and the cost must be included as a driver expense.
In the City of Seattle, wages are calculated as follows:
Wages include salary, hourly pay, piece rate pay, commissions, and non-discretionary performance bonuses. Tips and employer payments toward medical benefits plans are not wages.
This definition makes it clear that Seattle’s $16.39 per hour minimum wage is to be calculated without medical benefits plans. Given this definition, it does appear that health insurance costs should be deducted as an expense as Parrott-Reich do in their study.
Independent Contractor Taxes
The Cornell study does not take into account independent contractor taxes.
Cornell Study: This study does not take independent contractor taxes into consideration when calculating driver expenses.
Parrott-Reich Study: $0.0835 per mile for payroll taxes, and $0.0112 per mile for the state business tax, included because TNCs view their drivers as independent contractors. Under this arrangement, taxes are a driver expense and must be paid, ultimately, out of driver gross earnings.
Ride-hail drivers would certainly not face independent contractor taxes if they were not drivers, so it does seem appropriate for these taxes to be included in driver expenses.
Recalculating the Parrott-Reich study using ground truth data
To address Uber’s and Lyft’s issue that the Parrott-Reich study does not use ground truth data, we’ve recalculated driver earnings using their formula and Gridwise’s ground truth data.
Additionally, we’ve calculated per-hour earnings not just for October 2019, but October 2019 through June 2020. This will better show the true earnings per hour, and frequent fluctuations of earnings per hour that drivers face.
See a Seattle week-by-week breakdown of ride-hail driver earnings below.
Please note that driver earnings plummeted in March 2020 because of COVID-19 — and then quickly rebounded to be even higher than pre-COVID-19 as the vast majority of drivers stopped driving even as demand somewhat rebounded.
How can cities and states gain a better understanding of ride-hail driver earnings?
Unfortunately for cities and states across the country, one-off surveys, studies, and interviews are inherently biased and can have holes easily poked in them. Without a significant amount of ground truth data across a wide timeframe, cities and states may find it difficult to convince policymakers they have the clean and accurate data.
As concerns about securing minimum hourly rates for ride-hail drivers continue to multiply, cities and states will need to be able to leverage the same ground truth data used by TNCs. Gridwise is uniquely positioned to provide ground truth data that will allow cities and states to build objective driver earnings models that stand up to scrutiny.
Gridwise is an independent data source that can be used to support labor studies related to gig-drivers, or even those related to understanding other ride-hailing and delivery insights such as utilization, origin and destination patterns, and much more.
Not only does Gridwise have insights for Uber and Lyft, but also for DoorDash, Uber Eats, Postmates, Instacart, and most other major gig service companies.
If you or your company would benefit from gaining access to ground truth ride-hailing and delivery analytics, reach out to us at data@gridwise.io.
