Company
Data is power in the gig economy
Transparency is the key to a functioning marketplace
Feb 28, 2024
✦
Alfie
What is the gig economy?
The gig economy is essentially a giant labour market where companies buy the services of workers. But unlike most labour markets they don’t employ the workers but instead engage them as independent contractors to perform individual tasks (’gigs’).
Let’s take the food delivery market in the UK as an example. On one side there are three main delivery companies: Just Eat, Uber Eats and Deliveroo. On the other side are about 150,000 food delivery drivers (it is a mix of two and four wheels but for simplicity I’ll call everyone a driver).
Drivers register on the apps (almost always plural) and can (pretty much) login whenever they want. While online they might be offered a job, which might include one, two or three deliveries. Drivers are free to accept or reject any job.
This model has proven to be very popular on both sides. Companies like the fact that drivers only get paid for actual deliveries so they don’t need to spend lots of time managing a workforce. Drivers like the flexibility to be able to set their own schedules. So far, so good.
How are prices (driver fees) set?
As in any market, the price (i.e. the fee a driver is paid for a delivery) is set by supply and demand.
Companies state how much they are prepared to pay drivers (this has nothing to do with the fees a customer ordering food might be charged). Drivers compare offers and try to sell their services to the highest bidder. If there are more drivers than jobs fees will fall, if there are more jobs than drivers fees will rise. Again, it sounds pretty efficient.
Unfortunately, the results suggest there might be a problem. While most sectors have seen wages increasing over the past few years amidst worker shortages and high inflation, the gig economy has seen the opposite. Data from Rodeo users show that fees on all three platforms meanwhile have fallen or remained flat since 2021, despite more than 20% inflation.
So, why isn’t it working?
There are two main reasons why we believe the market is failing to price drivers’ time properly.
Firstly, all three platforms allow drivers to share or rent their accounts to other people with no checks from the companies (’substitutions’). This makes it easily accessible work for people who do not have the right to work.
The result is a sorry state: vulnerable people with few other options often get exploited by people controlling lots of accounts (gangmasters) and legitimate drivers who have the right to work get undercut by people who will work for any wage. In most sectors the minimum wage helps to defend against this but it doesn’t apply in the gig economy.
The situation got so bad last year that the Home Office wrote to Uber, Deliveroo and Just Eat demanding they stop the practice. They promised to comply but have not set any dates (I submitted an FOI request to check).
Secondly, there is a massive asymmetry of information between and companies and drivers. Companies have lots of data. Drivers have very little. This means that the pay negotiation is heavily skewed in the companies’ favour.
Why does data matter so much?
In the early days most companies had transparent fee formulas: £x per mile, £y per minute, £z for bonus etc. It was therefore possible for drivers to compare pay across platforms. These have now been replaced by opaque algorithms - this job is £3.21 because the computer says it is £3.21.
The actual reason this is £3.21 is because an algorithm used millions of data points to decide that £3.21 is the lowest price at which that job is likely to get completed. It considers all the past jobs that have been accepted or rejected, how many drivers are online in the area and any other data that might help to identify the optimum fee (from the company’s perspective).
If it doesn’t get it right first time it can always have another go: jobs that get rejected multiple times are usually re-offered at a slightly higher rate - effectively a reverse (or Dutch) auction.
Now it’s time for the driver to make their decision - should you accept or reject the order? To do this you need information.
Firstly, how much is this job going to pay me for my time?
Already we hit a snag. Deliveroo proudly states that they (usually) tell drivers the destination in advance (jeez, thanks). They don’t, however, tell the driver how long the order is expected to take, despite the fact that they have an estimate (we know this because they sometimes terminate drivers’ accounts for failing to deliver within the unspecified time).
Drivers are being intentionally deprived of key information when trying to work out how good a job offer is.
Secondly, how does this compare to what other companies might be offer me?
Airlines use dynamic opaque algorithms to price their flights. This doesn’t seem especially unfair. The reason is that I can see what price every airline is offering before making a decision.
As a driver, you need to decide whether to take the job or wait for a better one to come along. You have a minute to decide. This means that you can’t compare different offers at the same time. And because there is no transparency or consistency in how fees are calculated you have no reference points.
So, there you are. The ping comes in. You have a choice to earn £3.21, a price set by a team of data scientists using millions of data points (some of which are your personal data incidentally). You don’t know how long it will take. You don’t know exactly how far you will have to travel. You don’t know whether the algorithm has been changed since yesterday. You don’t know whether it is better or worse than other offers that might be coming. You have 60 seconds.
The combination of opaque fee algorithms and time-pressured decisions severely undermine drivers' ability to exercise their market power in seeking out the highest offer for their time.
Why did this happen?
This imbalance of information isn’t accidental.
Deliveroo used to show drivers how long each job was predicted to take.
Uber and Deliveroo used to show drivers how many hours they had spent online.
All the platforms used to explain how their pay formulas worked.
They decided to stop because they realised that the more data the company has and the less data the driver has, the stronger their position.
So they started to collect more and more data about driver behaviour and share less and less data with drivers.
How can drivers get better access to their data?
In 2021 we decided to build an app where drivers could automatically import their data. We can only do this once every 24 hours and it requires drivers to login via a third party payroll API service but it works most of the time and it has enabled more than 15,000 drivers to track and analyse their income.
Importantly it also has enabled greater transparency about the entire market. 150,000 workers effectively have to reverse engineer an algorithm to know whether they’ve been given a pay cut and no individual driver has enough data to prove that there has been a change. By aggregating anonymised data of thousands of drivers (with their consent), we can look at average fees per order or fees per minute this week and compare to last week to see what has changed (spoiler alert - it’s generally gone down).
How did the platforms respond?
To their credit most companies have accepted that, while it might occasionally be inconvenient, drivers have the right to access and control of their personal data. But there’s always one laggard and in this case it was Deliveroo.
They responded by rewriting their driver contracts and then alleging that driver’s login details and earnings are confidential to Deliveroo (as, apparently were their legal threats against us). They also spent lots of money making it harder for drivers to export their data.
What is the solution?
Technology has made a new form of flexible labour markets possible. This can be a good thing for both companies and workers. But some companies have taken a shortsighted view and sought to build technical and legal barriers to skew this market in their favour.
For the gig economy to function properly, workers need access to a competitive, transparent market. Drivers need to be able to see how much each job will pay them for their time and have a reasonable assessment of how that compares with what other companies are offering them.
Workers need access and portability of their data. Unfortunately it seems that not all platforms can be trusted to provide this voluntarily. Fortunately, Open Banking provides a ready solution: all workers should have a mandatory real-time smart data API for all of their earnings and performance data.
More posts