New study reveals how gig platforms “wait out” workers to slash wages

TÜBINGEN – Researchers from the Max Planck Institute for Intelligent Systems (MPI-IS), the Tübingen AI Center and Ellis Institute Tübingen have conducted a study which uncovers a mathematical strategy that explains how digital labor platforms can systematically suppress wages. The research, titled “Stochastic Wage Suppression on Gig Platforms and How to Organize Against It,” develops a mathematical model motivated by digital labor markets, including crowd-work, ride-hailing, and food delivery. It shows how a buyer that posts prices and can wait for workers to accept them, may keep payments very low when some workers are willing to accept tasks at very low prices.

“Our starting point was the large empirical literature showing that many digital workers earn very little, once unpaid time and other hidden costs are taken into account,” says Ana-Andreea Stoica, who is a research group leader in the Social Foundations of Computation Department at MPI-IS and first author of the publication. “These inequalities are particularly visible in global data-work supply chains, where workers in lower-income countries perform essential labor for companies and users elsewhere. We wanted to understand one structural mechanism that can sustain these outcomes, and when collective action can change them.”

The “Waiting Game” Strategy

Ana-Andreea Stoica, Celestine Mendler-Dünner, and Moritz Hardt, who is the Director of the Social Foundations of Computation Department, investigated a phenomenon that is referred to as “stochastic wage suppression” (SWS). This strategy allows platforms to have a massive number of tasks completed by digital workers from all across the world, while the platforms’ total spending on wages is remarkably low – sometimes paying a total cost that remains nearly the same even as the number of tasks grows.

How is this possible? Some digital platforms take advantage of workers' uncertainty about their labor costs. Workers accept a task when their estimated costs are lower than the pay offered. However, in markets with many new workers signing up to a platform, people often struggle to accurately estimate their labor costs. They often realize later that they have spent a lot of time searching for relevant information, new tasks, and learning how to navigate the platform. Ultimately, newcomers can often miscalculate their costs and earn less per hour than they anticipated. This can incentivize platforms to 'wait out' higher-priced labor: platforms set low prices and wait until someone accepts the task. Food delivery platforms are one example. Such platforms can be very unregulated, creating vulnerable conditions for drivers.

“One reason why platforms pay low wages is algorithmic,” explains Stoica. She and her team want to understand the structural reasons for low wages. Secondly, they are investigating possible ways to organize people in order to overcome price dumping.

“If everyone agrees on a minimum price for labor, like a minimum market rate, then as the market grows and more tasks become available, employers have to pay at least that. Hence, the market should develop linearly. However, if there is no such minimum market rate and there is always someone willing to work for very low pay, the average cost of labor can remain low, even as the market grows,” Stoica continues.

Why broad, untargeted organizing may fall short

While digital platforms are making huge profits, one might think that low wages would trigger protests. Although this does sometimes happen, for example with Uber or Lyft drivers taking to the streets to demand higher pay, more transparency and better working conditions, Stoica’s study provides a sobering look at why many current attempts to organize gig workers struggle to make a dent in platform profits. The researchers found that 'horizontal' collective action, whereby a random group of workers from different categories (such as market segments) come together to demand a minimum price, is largely ineffective. This is because the platform can simply wait for a worker who is not part of the collective to accept the low price. Such untargeted efforts merely reduce the pool of available workers without forcing the platform to increase its wages. For a protest movement to be successful, it would require almost 100% participation, which is obviously hard to achieve.

The way forward in collective action: a "vertical" strategy

The research suggests a powerful alternative: 'vertical' or targeted collective action. Rather than recruiting random workers from everywhere, the study suggests that efforts should be concentrated on specific market segments. By reaching the segment of the market on which the low-price strategy depends, even a comparatively small collective can change the market outcome. In the researchers’ model, targeted coordination can cause total payments to grow proportionally with the amount of work being procured.

This is an area in which platforms themselves can make a difference: If a platform is willing to collaborate with workers, it could contact new workers and say, “Hey, we saw your account was created a few days ago. Here's some basic information and some links.” It is harder to pay dumping prices if a minimum number of workers are well informed. The platform can act as a mediator between employers and workers. They can also schedule workers to avoid over-supply at certain times, which would push prices down. Existing worker-led initiatives, such as Turkopticon, illustrate how information-sharing tools can help crowd-workers exchange knowledge about requesters and working conditions.

Thirdly, although this is not done in all countries, digital platforms can be regulated through government intervention, with laws setting minimum wages and ensuring that gig work is classified as wage work rather than freelance.

Impact for the Future of Work

“Our results highlight a structural imbalance in digital labor markets,” Stoica concludes. Although platforms currently have the upper hand, workers can regain control by shifting from broad, decentralized campaigns to coordinated, segment-specific bargaining.