Who Are the Gig-Economy Workers?

Read transcript

With Professors Matthew Corritore and John-Paul Ferguson
J.P. Ferguson: Academics, policymakers and other thought leader types get very excited when we talk about the gig economy because we know enough about how employment systems work that we let our imaginations run wild. Our imaginations have run pretty far ahead of what we actually know about this type of employment.
Host: Welcome to Season 1, Episode 4 of Delve, a podcast from McGill University’s Desautels Faculty of Management where we’ll hear from management researchers and practitioners as they explore the latest ecological, social and economic challenges that we face as a society. I’m your host Mo Akif and today we’re talking about the gig economy and the extent to which it’s changing traditional work. Flexible, liberating, transformative. These are words used in popular culture to describe the gig economy, but how much of this is true? The reality is we don’t know a lot about gig workers, their backgrounds, their social networks, and how they fare at work. But professors, Matthew Corritore and John-Paul Ferguson will soon fill this void. They’re setting out to get a fuller picture of today’s gig workers separating fact from fiction and providing data for better regulation. Let’s listen in as they discuss what they’re looking to explore in their new research and how they’re going to accomplish this.
M. Corritore: My name is Matthew Corritore. I’m an assistant professor at the Desautels Faculty of Management, and I study nonstandard work arrangements, things like contract workers, gay workers like we’ll talk about today, as well as corporate culture and organizational performance.
J.P. Ferguson: I’m John-Paul Ferguson. I’m an assistant professor of organizational behavior at the Desautels Faculty of Management. I study careers, employment relationships, employment segregation and the like. This project got its start actually when neither of us was at McGill just as a bit of background. We both joined Desautels Faculty of Management in 2018 having come from Stanford in California and it’s very hard not to notice the gig economy when you’re in the Bay area. Virtually everywhere.
J.P. Ferguson: I mean Ubers are more common than non Ubers on the streets in some neighborhoods in San Francisco where both of us lived before we were thinking about this project and there were a lot of questions that people were already starting to ask about workers in the gig economy, but basic information was lacking in the sense of even how many people there are and the like. The best data that we could find was usually people who had said I managed to get access to records from a company like Uber or Lyft or Task Rabbit or something like that where okay, they have information from the company, but the worker themselves and what they do kind of gets lost in that kind of data.
M. Corritore: Yeah. I think the broad motivation of the study is thinking about how to define the gig economy and the extent to which it’s reshaping employment and the world of work. I think those are some of the big questions that that motivated this to the extent is gig work kind of replacing traditional work for some workers such that workers are doing a series of gigs and constructing their work from a series of gigs across multiple platforms. Or should we think of gig work as more of a supplements that workers who are still more engaged with the traditional workforce are kind of doing on the side as a supplementary part of their income?
J.P. Ferguson: One of the funny things when people talk about gig work as though it’s a brand new thing, this has always been a running question because part of my own work as a PhD student was studying the history of labor and employment and really short term informal employment is nothing new. The idea, for example, that you’d get a job that might last for a day or even a couple of hours and then there’s no guarantee of further employment, but also that you’re not necessarily going to show up again.
J.P. Ferguson: There’s plenty of models of that. Like Montreal itself famously has… The old Port of Montreal used the same system to hire dockworkers as most other ports in North America, which is called the ShapeUp system. This is the idea that you have a bunch of people who show up at the docks in the morning and the head of that particular shift comes out and picks people that he wants to work and they’re there for the day. There’s no guarantee that they’re going to be hired tomorrow. Now they don’t have to show up tomorrow either. So there’s a sense in which those old dockworkers chose their own hours, but we didn’t think of them as independent contractors. Right. We thought of that is just sort of a potentially crummy employment relationship.
M. Corritore: Likewise, if you go to a Home Depot or another big box hardware store early in the morning, you’ll find day laborers that are looking for work for that day only.
J.P. Ferguson: Yeah. Those of us that know a lot about the past of employment systems know that there is potentially a real dark side to such systems. So you really want to get some data to get a sense of well in aggregate do we think this is better or worse?
M. Corritore: Part of the motivation of this study in particular is that we really lack worker level data. A lot of the existing work on the gig economy has either used large scale government surveys that have a rather imprecise definition of gig work and provide a limited window into a worker’s experiences or involve, say a researcher partnering with only one gig platform, say a researcher partnering with Uber to study only Uber drivers.
J.P. Ferguson: Yeah, I think that’s an easy thing to overlook when you say, “Oh I have data on Uber drivers.” Well you don’t even really have that, right? You have data on Uber drivers when they’re working for Uber and what they do in the other hours of their day is just not going to be in that kind of data. So really simple questions that you might want to ask like how many platforms do gig workers work on? What population would you even start to?
M. Corritore: In other words, we want to figure out whether gig work is completely substituting for more traditional forms of employment or it’s merely supplementing or complementing those other forms which are very important. We also want to know some details about the actual tasks and skills that these workers are using. In other words, is most of our sample going to be focusing mostly on ride sharing. They’re basically just driving for Uber, driving for Lyft or another service, or are they doing a wide variety of tasks as part of a kind of a gig portfolio? Are they driving for a couple of hours and then also going to hang someone’s drapes right through a platform like Task Rabbit, which would be an atypical collection of tasks and skills to think about in terms of one’s job.
M. Corritore: As part of this respondent driven sampling methodology, we also want to learn something about the social networks of these gig workers themselves. We want to try to understand, are kind of gig workers working in isolation where they don’t know a lot of other gig workers? Or do they have a robust social network of of other people in this type of employment who they’re relying upon for advice, for guidance, for insight into how to kind of maximize the welfare that they derive from these types of jobs?
J.P. Ferguson: Yeah. We often find whenever we start to study a line of work out in the world that seems really isolating, especially because we have an internet, that almost always I start looking at an occupation like the people who study the scans that radiologist send somewhere for analysis. These are people who sit in a dark room looking at a computer screen all day. Then you find out that they have equivalent of a Slack channel with like 300 other people who look at radiology scans and this very active social life. Or the nurses who draw blood at dialysis clinics and then it turns out that they have forums that they post on and talk to each other all the time.
J.P. Ferguson: That’s important, right? Because the assumption with this kind of research is that this population is hidden to a researcher, whether that be us, whether it be a census taker, or someone else, but they aren’t hidden to each other. It’s just that we don’t have access to the same social world.
M. Corritore: We think the results of the survey will have important implications. For example, as scholars of work and employment and sociologists, we’re very worried to some extent about these types of gig arrangements as alternative work arrangements that don’t have a lot of the securities of traditional employment and avenue to save for retirement, for example, access for traditional benefits. We really want to try to understand by going in detail with these workers, are they satisfied with this type of work or do they actually want to transition more into traditional employment?
M. Corritore: What aspects of the work are particularly troublesome? Are there some advantages of the work that we need to highlight and what type of worker is working for what type of platforms are experiencing those benefits? Which are very important questions. Not only for scholars and academics, but have huge implications for policy makers in thinking about regulations of this type of work. And the general public, people who might be thinking about moving into this type of gig work. We want to learn from this sample of workers in order to give them accurate information about the types of things they might expect from this line of work.
J.P. Ferguson: I mean at a personal level, when you think about taking on gig work, one of the things that’s always seemed really neat about it to me is you end up un-bundling tasks from jobs. By that I mean that there are things that someone who has a particular job does that may or may not seem appealing to you. But the other parts of the job, whether it be it’s social status, the other things you have to do, the employment qualifications that you need in order to take it on have never appealed. When we start to separate out those tasks, you might start to see different types of people who are willing to do different types of work.
J.P. Ferguson: I bring this up, if you think about like an Uber driver or otherwise, I know lots of people who have driven for Uber who never would have driven a cab. There was nothing about that that appealed to them. There was the sense that the only way you could drive around in the middle of the night and pick up passengers for pay and the other things that are the tasks involved in this work was to work for a cab company. That didn’t appeal for any number of reasons, the cost of entry, but also the social status of a cab driver was seen as something very different. It’s not the work itself because we all drive, right? But something about the way it was tied together with that whole job and occupation was very different.
J.P. Ferguson: I’ve known lots of people who take on jobs on Task Rabbit, like maybe folding, unfolding people’s laundry is a classic example where people weren’t going to work in the laundry, they weren’t going to become a maid or a house cleaner in a lot of cases. But they’re perfectly happy to do individual loads of laundry for people for pay. Notice again, you’re un-bundling this type of work, this task that people can do from the job as we’ve usually defined it, and a lot of the types of work get bound up with all these sorts of issues of social status or identity that we have. An interesting thing about gig work is it throws a lot of those ideas into question.
M. Corritore: It’s really suggesting that these new technology platforms that are connecting buyers and sellers, that are connecting gig workers with clients who demand these services might be kind of changing work in that way and making certain types of work more appealing to people in ways that that type of work previously wouldn’t be appealing. So that’s a really exciting part of this study to get some insight into that.
J.P. Ferguson: Another piece that sometimes we get concerned about is how you get a job and who you get a job from on these different platforms. Because for example, think about something like, I don’t know Airbnb where you used to… I don’t want to use Airbnb because Airbnb doesn’t really work as a gig economy that you’re renting out your apartment, but let’s go with something like Task Rabbit instead. Historically, if you wanted to get your laundry done and you went to a laundromat and they refused to serve you, for example, because of the color of your skin. Well there’s a regulation against that. There’s someone that you can complain to.
J.P. Ferguson: Now let’s just imagine you’re posting jobs and there’s a profile of you on the site and people can bid on these jobs or not and you find that people are less likely to agree to do work for you or indeed you’re looking for work and people are less likely to hire you. In aggregate, some of the same problems in the labor market in terms of discrimination can still exist, but who records that and if it is a widespread problem, how do we deal with that? How much of a problem that is depends on what the total experiences are that people have in the gig economy and in terms of data, it’s the wild west. We just don’t know how big such potential problems are.
M. Corritore: You have cases where individual buyers on these platforms are making decisions that may be biased in some way and they’re doing it with a lot of frequency. The traditional workforce, traditional employment, a lot of those decisions are being made by organizations that have regulations and policies in place that are designed at least to some extent to limit that type of bias. We’re concerned and we’re interested in learning the extent to which you put a responsibility for those decisions in the hands of the individual buyers. Do we see some of those biases magnified?.
J.P. Ferguson: The question is what are the benefits of this type of employment relationship, and are they worth some of the risks of changing from the system that we’ve already had in place? If you think that there are some, for example, lowered public safety in using one particular gig economy service, but then you find out also that the workers who provide that service think it’s a really bad job, that changes the calculation in your head about should we be regulating this, should we just be banning this particular type of practice?
M. Corritore: A lot of the rhetoric around gig work I think has this idealized vision of what it is in mind or you hear a lot of people talking about the flexibility benefits of this type of work. I think what they have in their mind is that you can flexibly change between different types of jobs working at different times in the day constructing this gig portfolio. That may be the case. There might be some benefits of flexibility, but there also might be some real hidden costs or unforeseen costs from that. What is the cost of switching between different types of jobs during the day, different types of tasks? What type of training or skills might a worker have to need and incur the cost of acquiring in order to to work across those diverse set of tasks? So I think policy makers and then the public to some extent have an idealized vision or a typical vision in their mind of what the gig economy is.
J.P. Ferguson: I think something that’s really struck me since coming to Quebec from California is the flexibility that we associate with gig work. Its effects exist independently of the rest of public policy and how society is set up. Having short term employment contracts in California where health insurance is vastly more expensive and prior to a few years ago didn’t necessarily exist at all for people in these positions versus flexible short term work in an economy where it is reasonable to imagine that people are going to be able to take care of things like medical emergencies independent of their exact employment status from day to day. I think you’re going to have very different reactions to something like gig work. So we won’t necessarily expect that everyone’s experience is the same across different economies. But I think there’s no… Because all of our economies look slightly different, there’s no one size fits all response probably that this type of work.
M. Corritore: Something we’ve talked about is if it’s successful in Montreal, successful in Toronto, what are the different Metro areas, the different locales that we might expand to? So we naturally talked about San Francisco as a place that one day if we find this survey is successful, it works as anticipated and the insights are interesting, we might expand to different cities in the US and try to see the extent to which the findings that we reveal in one city hold in another.
J.P. Ferguson: That’s essentially… I mean we talk big dreams of expansion. The main survey we’re hoping to have launched here in Montreal in the spring and summer to start here and then expand in a couple of months time into Toronto as well to start gathering results.
M. Corritore: We’re starting out as seeds for our survey. Those initial respondents that we’re going to study, we’re going to look at Uber drivers in Montreal and they’re going to recruit other respondents who are in their network that we can later interview.
J.P. Ferguson: If you’ve ever done a survey or even read about survey design, you know that the best types of surveys have random samples. We go out and we pull a random selection of people from the population and the reason why random sampling is really good is random samples tend to be representative samples. So if 30% of your survey works in the gig economy and it’s really a random representative sample, you feel okay saying that 30% of the workforce works in the gig economy.
J.P. Ferguson: But at the very least, if we think that gig work is fairly rare, then you’re going to have to sample an incredibly large number of people just with a straight random sample, if you ever want to go out and gather information on a large enough population of gig economy workers to tell you something general about the population. So that is what has tended to push us toward a what’s referred to as respondent driven sampling.
M. Corritore: This is a really exciting technique that’s been used in prior work in sociology to discover what researchers call hidden populations, populations that we can’t define a priority. There’s no directory that we can look at in order to sample. So it uses a variant of snowball sampling, which essentially interviews a particular number of seeds that we’ve identified, respondents that we know are in the gig workforce. After we’ve gathered information from them, ask them to provide referrals for other gig workers they know, other people that are in their social network. We then try to collect surveys from those. Typically snowball sampling can produce unrepresentative samples. This methodology collects characteristics of the social network of these respondents that we collect, certain parameters that we can use to, after we gathered the data, adjust the data to become more representative.
J.P. Ferguson: I’m going to be a tremendous nerd for a moment and see if I can give an intuitive explanation of how respondent driven sampling works. Because anyone who’s interested enough to listen to a research podcast might want to know some of the details. At the root of respondent driven sampling is what is usually referred to as a Markov chain. Markov chains are famous in statistics and artificial intelligence. Markov chain, and particularly what are referred to as first order Markov chains, are situations where your probability of the next event only depends on your probability of the present event. So snowball sampling, which Matt referred to a minute ago, snowball sampling is where I as a researcher contact some people and I survey them. Then I ask them to give me some suggestions about whom I should follow up with. Then I try to hunt down those people and I talk to them and I asked them for recommendations and so on. So there’s that idea that my list of contacts is snowballing over time as I follow up with all these people.
J.P. Ferguson: But notice in every step of that process, it matters who I follow up with and who chooses to talk to me. So me as the person who’s doing that surveying ends up biasing what that sample looks like because it’s going to be the people of this group that I’m most likely to talk to and who are most likely to trust me. That shows up in a lot of situations where it could cause a problem. Respondent driven sampling was originally developed to look at intravenous drug users. If you’re a user of heroin and you’re answering a survey and someone comes up to you at the end and says, “Could you write down the names of nine other heroin users so I can contact them.” I mean you’re supposed to stop snitching if you’re involved in intravenous drug use. So you’re unlikely to volunteer names and there’s no reason these people are necessarily going to respond to me when I come knocking on their door.
J.P. Ferguson: On the other hand, and this is the key innovation with respondent driven sampling, is instead of me following up with my respondents’ contacts, I ask them to recruit other intravenous drug users, other contacts to come talk to me. It’s often the case that even if someone’s not going to trust me, they might trust their friends. The first part of this technique that’s important is to notice that the people who respond to the survey recruit the other people who are going to show up in the survey. Now in and of itself, that just helps with a lot of those trust issues and starts to reduce the bias.
J.P. Ferguson: But formally in statistics when we talk about Markov chains, because a Markov process only relies on the previous period, there’s a thing called the Law of Large Numbers in Markov chains where if you have a sufficiently large number of events, after a while the distribution of states in that Markov process is independent of the prior recommenders and so on. What that means is whereas in the snowball sampling, I the researcher am present at every stage potentially biasing that process. If you engage in a respondent driven sampling process after you’ve gathered enough people, essentially the identities of earlier recommenders and so on don’t matter as much anymore. As long as you can keep track of who recommended whom, and that’s built in in terms of how you recruit the subjects for these kinds of surveys, then you have a way empirically to take that bias out and build up a representative population even though you didn’t know who the group was that you wanted to sample from in the first place.
M. Corritore: A lot of the work right now is focused on finalizing our survey design, making sure we’re asking the right questions to be able to get insight into the answers we want. Then as John Paul mentioned, we expect to launch the survey in Montreal this summer. Start to recruit those initial seeds and then watch as our sample grows over time, which will be very exciting.
Host: That was Professor Matthew Corritore and Professor John-Paul Ferguson talking about their new research on gig economy workers, a project made possible through the support of the Social Sciences and Humanities Research Council of Canada. We hope to catch up with them once they’ve gathered and analyzed their data. It promises to be eyeopening, so stay tuned for that in the near future. If you enjoyed this podcast and want more insights, you can subscribe on your podcast app of choice or visit us at mcgill.ca/delve.