This is the 18th in our series of posts by job market candidates.
The gig economy represents between 200 and 400 million workers worldwide. In places where unemployment and informal work are widespread, such as in Latin America, jobs in app-based delivery services offer new and flexible employment opportunities. For many, these apps create much-needed jobs with minimal entry barriers, making gig work a lifeline for people without stable job options. In the context of food deliveries, traditional restaurant employees—like in-house waiters and drivers—may see their positions replaced by lower-paid, less secure gig roles without benefits.
Policymakers are increasingly exploring how to regulate the gig economy to maximize its benefits while safeguarding vulnerable workers. Yet, designing effective policies has been challenging without precise estimates of who benefits and who loses. Are these services a primarily positive force, creating fresh opportunities for workers? Or do they replace secure, well-paying jobs with ones that offer less stability and lower wages?
I examine these questions in my job market paper by investigating how delivery apps affect in-house restaurant employees and gig workers in Brazil. Studying these effects is challenging because it requires combining data on restaurant adoption of these services, restaurant workers, and platform-based workers. I address this challenge by linking data from a major delivery app with administrative employer-employee records in Brazil. This dataset enables a comprehensive look at formal employment nationwide and tracks restaurants’ technology adoption and the work histories and performance of app-based drivers. To study the effects on restaurants and workers, I exploit the staggered rollout of the platform across the country and use a matched difference-in-difference design that compares restaurants (and their workers) that adopt the online delivery platform to observationally similar restaurants and workers located in regions where no platform was available at the time.
Restaurants replace in-house waiters with delivery workers when they adopt delivery platforms
The first step is understanding how these apps impact restaurants. Are they helping restaurants expand their reach, or are they mainly a way to replace in-house workers with contract-based drivers? I find that restaurants reduce their in-house workforce by 6% one year after signing up on the delivery service. This downsizing affects waiters rather than kitchen staff (essentially unaffected). Waiters who are displaced are replaced one-to-one by gig delivery workers, suggesting that restaurants may not necessarily be expanding their output but are instead shifting towards delivery sales (Figure 1).
Figure 1. Replacement of in-house labor for outsourced delivery drivers
While some waiters see themselves replaced by gig workers, this displacement does not map one-to-one to employment impact. My data allows me to follow workers over time and evaluate how many displaced employees secure new positions—and the quality of these new roles. Comparing affected workers to similar restaurant workers in areas without app adoption, I find that while 6% of waiters lose their jobs, approximately 75% of them manage to secure new roles in the formal sector within a year. This reallocation limits the earnings impact to a modest 1.5% loss compared to their counterfactual scenario in which their employer would not have adopted the platform.
Spillover effects on non-adopting restaurants
Not all restaurants adopt delivery services, yet they too can feel the impact when neighboring businesses adopt. To study these dynamics, I geolocated every restaurant in Brazil and identified potential “spillover events.” Restaurants were exposed to these events if, at a particular moment in time, they had not adopted a delivery platform and were exposed to a large share of neighbours who suddenly adopted the service. Consistent with some crowd-out in this market, I find that these non-adopting restaurants, compared to their counterfactual, downsize by approximately 3% and are 5% more likely to close one year after their neighbours adopt the platform. Before their neighbors adopted, spillover restaurants were performing similarly. 8% of spillover restaurants do adopt within a year, similar to the unconditional probability of adoption during the same period (10%), mitigating the concern of selection into non-adoption.
The effect on employees at these non-adopting establishments is more pronounced than that at adopting establishments. Because their employers are more likely to shut down, these workers suffer a larger probability of displacement. Workers at non-adopting restaurants experience an average 6.6% decline in earnings. This underscores how the influence of delivery apps extends beyond adopting restaurants to impact workers across the industry.
Weighing the overall impact on worker earnings
So far, I have described what happens to formal workers in the restaurant sector when employers start offering delivery services through online platforms. However, to understand the overall impact of these platforms on workers' earnings, we need a couple more ingredients. First, Brazil has a significant informal sector that must be considered. A limitation of administrative datasets is that they do not allow us to observe informality directly. To consider these workers, I impute their earnings effect using a combination of the Brazilian census and the household survey.
In addition, we need to examine the potential gains for gig workers. Although these technologies decrease the number of in-house restaurant workers, they also create new opportunities for workers at a low entry cost. In my setting, approximately 80% of the workers in the delivery platform did not hold a formal job the quarter before working on the app. This implies that there are potentially large gains for these workers, but are these large enough to overcome the losses of restaurant workers?
Assessing these gains requires factoring in the opportunity costs of these workers (i.e. how much they could have made in another job), operating expenses (such as gas and vehicle maintenance), and differences in tax schemes between gig workers and formal employees. When all these elements are considered, I find that the net earnings increase for app-based drivers outweighs the income losses experienced by restaurant employees. On average, each adopting restaurant generates earnings equivalent to 4.6% of their pre-platform wage bill, driven mainly by the gains of informal and unemployed workers with limited outside options (Figure 2).
Figure 2. Overall impact of delivery platforms on workers' earnings
Beyond earnings: Assessing job quality
While earnings are a crucial part of the analysis, non-wage aspects like job security, flexibility, and health risks also affect worker welfare. If gig work offers valued flexibility, the benefits to workers may go beyond just wages. Alternatively, if working conditions lack security or involve higher health risks, these jobs could be less favorable. To account for these non-wage aspects, I use job-to-job transitions to rank how workers value gig work compared to other restaurant jobs. The intuition behind this exercise is that good employers will receive workers from other good employers and lose few employees to other firms. The advantage of this exercise is that good employers are not only defined based on the observable earnings they pay workers but also on any other non-pecuniary feature the job may offer. Interestingly, my findings indicate that app-based jobs rank within the 90th percentile of restaurant jobs in Brazil, suggesting that many workers value them highly, even beyond wage considerations.
Policy implications: Navigating the complex effects of gig work
The findings highlight both the benefits and challenges posed by delivery apps. While these services create new earnings opportunities, they also lead to job displacement and intensify competition, affecting non-adopting businesses. Additionally, the low entry barriers to gig jobs offer large gains for workers in vulnerable situations, but are often linked to limited protections for these workers. For policymakers, this presents a complex landscape. Regulatory efforts, such as the EU directive on platform work or Spain’s Ley de Riders, which focus on improving protections for app-based drivers, should safeguard job flexibility. In countries like Brazil, where informal work is common and unemployment remains high, balancing these factors could help ensure that gig economy growth supports, rather than undermines, long-term economic stability and worker welfare.
Pascuel Plotkin is a PhD Candidate at UBC’s Vancouver School of Economics. His research interests are in Labor, Development, and Applied Econometrics.
Join the Conversation