Gig Workers and Employment Classification — Unraveling the Controversy

By Juliet Schor, Boston College*

Should gig workers be classified as independent contractors, entrepreneurs on their own, or should they have employee status and the rights that go with employment in most countries? This flashpoint of controversy has created a lot of activity in Europe and the United States. For example, the US Secretary of Labor in the Biden administration, just a few months ago, said that gig workers should be employees and that they deserve work benefits. This represents a huge shift from our previous presidential administration, which put forward several rules saying they should not. However, it has been very controversial. It is not clear what is going to happen in Massachusetts, where the gig companies have just pledged $100 million to overturn the law in the state, which says that gig workers are employees, as they did successfully in California. Companies claim that the application of employment in the gig economy will cause the disappearance of flexibility. In a widely cited op-ed, Uber’s CEO said that the employment system is outdated and unfair: it forces every worker to choose between being an employee with benefits but less flexibility, or to be a contractor with flexibility but no safety net.

In particular, these are the questions that we need to ask: do gig workers need more protections? Does employment status preclude flexibility? Will prices rise and reduce work? What does the evidence show? How can we reframe this debate? It has been a problematic debate so far. One of the big problems with it is that there have been lots of claims without evidence, particularly from the companies but not only from them. New research from the research team that I work with, using data from a nationwide delivery platform is the first study of its kind to address these questions. This research is carried out through a National Science Foundation project called “The Algorithmic Workplace” and funded by the Initiative on the Future of Work at the Human-Technology Frontier.

But before the research results discussion, let me make a few remarks about on-demand platform work. My first point is that it’s characterized by what I call heterogeneity up and down the line. This point has been under emphasized in the literature. In almost every way the platform economy is heterogeneous, whether it is the workforce, the kinds of platforms, heterogeneity over time, heterogeneity over place, heterogeneity in any individual app. Another key thing which relates to flexibility is the need for an on-demand supply of labour to address on-demand, customer demand. In other words, a very important point here is that a subset of platform economy is work that needs to be done at the moment of demand. A third, important issue is that there is not a profitable model so far in some of the biggest sectors such as ride-hailing and food delivery, and that’s not just in the United States. That’s pretty much across the board.

My team and I studied what social scientists call a natural experiment at a company that I’m going to call Bring Your Package (BYP). It is a national business-to-consumer package delivery platform. It works with many of the big-box stores and delivers packages on demand to residential customers. As California Assembly Bill 5 was going through the legislature, Bring Your Package decided to transition its drivers in California to become employees. Our team had access to data on drivers, schedules, hours, routes, etc. We also conducted interviews with key informants, with workers, frontline operational managers, support teams, and with management. We studied seven cities in California and beyond, from 2018 to 2020. This design exploits variation over time — from the pre-employee to the employee status condition in California, and then across the country, providing locational variation. The research focused on three issues: flexibility, working hours, and efficiency. We used multiple regression analysis to analyze the impacts of employment status. A major finding is that the scheduling system did not change after the shift to employment status. Workers retained the same flexibility that they had beforehand, as they were able to choose shifts that they preferred. The basic structures of the three worker groups in terms of hours worked remained. So the first big takeaway is that the company was able to shift these workers to employees and in terms of flexibility, things stayed mainly the same. There were a few changes which I will get to, but managers at the company expressed satisfaction with the new arrangement. In particular, it is very important that productivity and labor control were better as well as that predictability increased; the company had benefits from this new system as well.

Back to heterogeneity, which is related to the working hours in the platform work. We identified three group of workers. The first is long-hours workers (we might want to call them full time, although many of them are working less than 40 hours a week) who tend to do the majority of the jobs. They tend to be a small proportion of the BYP workforce. The second group is stable part timers and represent the largest group in all the cities. The third group, who we call intermittent drivers, are part of how the company gets flexibility, because they can rely on this group when demand is particularly high. Intermittent workers do not work very many hours. After doing the analysis for the California and non-California cases, we find that the California situation does not change much in terms of the proportions among the three categories of drivers.

On the question of working hours, we expected that the shift to employment would raise working hours. This is because the firm will has an interest lengthening hours because it is paying benefits to many of these workers and each employee has a higher per person cost than under the independent contractor condition. That is, we expected that employment status would raise working hours. Based on the multiple regression analyses, this turns out not to be the case for the full-time drivers who were already working the long hours that they wanted to work. However, hours do rise for both the stable drivers and the intermittent drivers. This may well be something that workers preferred, however, we do not have data about their preferences. All we know is that they worked more, controlling for demand. So employment status itself does have an impact. When workers transitioned into employees, they worked longer hours, unless they were already full-timers.

We also used regression analyses to examine efficiency. In particular, we measured scheduled hours against the hours that were actually worked. One of the things about this platform and many platforms is that they schedule workers and they may or may not have enough work for them during those shifts. In the case of Bring Your Package, if a worker was scheduled for a shift and there were no packages to deliver, they would get sent home and receive half the pay for that shift under the independent contractor condition, and they would also get paid also under the employee condition. (That payment structure is a little more complicated.) An interesting point here is that shifting to employment status reduced the gap between hours worked and scheduled hours. In other words, delivery work became more predictable. People worked the hours that they were scheduled for, possibly because things were more predictable for the company. Interviews showed that the workers were more willing to work those hours and that the company may have had more control over the workers. You can think of this as an efficiency or productivity impact. That is a sort of positive thing, certainly for the firm. Whether it’s also positive for the workers, we cannot really tell here, but it likely is as they were more likely to get work when they’re signed up for shifts and were less likely to be sent home without work.

Of course, employee status also raises certain costs. We don’t know by how much because we did not have access to cost data. We suspect that costs rose between 10 and 30%, but these higher costs were partially offset by increased productivity and efficiency.

If the assumption about the cost-rise is correct, then one of the things that is likely to happen with a shift to employee status is that prices will also rise. But one thing we have to understand about the platform economy is that prices are going to rise in any case. A New Work Times piece published as prices started to rise in the ride-hailing industry in the US a few months ago made the obvious but not widely-noted point that prices have been subsidized by the platforms as they tried to dominate their markets. The companies gave consumers prices that were below the true costs of providing these services. It has been estimated that there was a 40% subsidy in ride-hailing sector.

Keeping the US picture in mind, I argue that workers need wage protections. In the US, there has been a collapse in earnings, particularly in ride-hailing and food delivery. One of the best studies of this kind found that between 2013 and 2017 earnings in transportation, including all kinds of drivers, fell by more than half. Proposition 22 took away minimum wage protection for gig workers. Supporters of Prop 22 promised a wage guarantee, but that actually turned out to be below minimum wage at about $5.64/an hour because it did not include the considerable time that drivers sit around waiting for jobs, rides or food. After expenses, many gig workers are making less than the minimum wage. During one of our interviews, a delivery courier tried to figure out what he had made in the past few weeks and found out it was $6 an hour. These workers need wage protections at the very least.

Can employment work for ride-hailing where there are no shifts? Our study was of a package delivery firm, which is a little bit different as it has shifts. There are examples where ride-hail is moving in that direction. In 2018, New York City put in regulations that some way toward an employment system where with a minimum wage guaranteed both before and after expenses, a driver cap and some restrictions from the platforms about who can sign on the app. Ride-hail could also institute shifts and retain most functional flexibility; shifts are necessary in order to pay for idle time. Food delivery has long worked with shifts, for instance. Companies can do a lot of prediction of demand now, which I think makes a shift model work. I do not think ride-hail is ruled out because it has a different labor model in terms of how people sign on.

But ride-hail also raises the issue of whether Uber and Lyft will ever be profitable? This is a big question. Uber is the most losing company in human history; it has lost more than 25 billion and some estimates are up to 28 billion. Every year the Uber CEO tells us this is the year they’ll achieve profitability, but both ride hailing and food delivery are losing money. Food delivery is really not a profitable business and these companies went into it thinking that their technology could make it profitable. I think it’s clear that’s not the case. Furthermore, in the United States ride-hail is in a kind of death spiral. Drivers are not showing up because of COVID. Uber is now more expensive than taxis — the industry that it destroyed — and it’s less reliable. There are many reasons to think that profitability is going to be elusive for quite a while.

My last point is a theoretical one. This perspective comes from the introduction to a special issue of the Socio-Economic Reviewon the platform economy that I have co-edited with a number of colleagues. The point is to understand that platforms are contested social structures. If we think about the actors on platforms — platform managers and users, where users are both consumers and earners, there are three social forces that are operating on platforms — domination, autonomy, and mutuality. The platforms differ in how much one of these three dominates. There is the striving for domination that platform owners and managers have. They want to control what is going on with the customers and the workers. There is autonomy, which is a key factor that drives workers to platforms — the ability to work when they want, that flexibility, as it’s called, be your own boss (the desire for freedom, independence and autonomy that that platforms offered to workers). And then there is mutuality, which has been less of an issue for ride-hailing and food delivery sectors; but is important on other sharing platforms including those like Airbnb. The state of the markets, regulations, civil society, all of these are affecting these contested social relations. This is one of the reasons we have so much heterogeneity up and down the line because the relative power of these three forces — domination, autonomy, and mutuality — varies a lot across different kinds of platforms, across time, and it varies across space.

*This blog is based on the keynote address delivered at the Future of Work Conference on September 10.

The opinions and views expressed in this publication are those of the authors. They do not purport to reflect the opinions or views of Reshaping Work.