- Unpaid labour should be recognised as a systemic feature of platform work, inherent in its current model of work organisation. Its prevalence and magnitude render it a pressing regulatory issue.
- Introducing minimum standards regarding working time and wages is a key step towards limiting unpaid labour and establishing fairer working conditions.
- Limiting unpaid labour requires recognition of the subordinate status of large segments of workers, improvements in employment stability and a floor of contractually guaranteed working hours. Platform work within the framework of hourly-paid employment is less frequently linked to unpaid work than piece-rate and self-employment models. But without predictability of hours, hourly-paid employment can also lead to unpaid labour.
- Data sharing and transparency in compliance with digital protection law, as well as some forms of data portability (of ratings, portfolios) for freelancers working remotely on platforms is necessary to promote career progression and upskilling.
- Freelancers and independent workers on platforms should be granted access to effective voice mechanisms through the creation of representation structures and inclusion in collective bargaining, affording them some presence in policy and regulatory processes.
- Protection should be extended to different categories of labour by revising EU competition law’s scope of application, thereby making it possible to improve working conditions through collective agreements.
The phenomenon of the platform economy is increasing in size and importance, with its business model infiltrating various branches of the economy. The Covid-19 crisis has been an exacerbating factor in this. While employers and some policymakers insist that platforms create jobs and new opportunities for workers, especially those who are hard to employ, booming academic research in this area has pointed to many challenges to work and employment conditions within platforms, notably unpredictable pay and working time (Pulignano et al. 2021; Berg 2016). One important additional challenge has received only scant attention, however, namely the amount of unpaid labour performed by platform workers (ILO 2021). This policy brief fills this gap, by providing a systematic overview of the various forms of unpaid labour across different types of platform work, analysing and quantifying its prevalence, identifying drivers, and formulating policy recommendations on how this challenge can be tackled.
Current studies report on myriad projects and tasks undertaken within the digital economy by ‘unpaid’ people and workers, who contribute to the creation of economic value without their input being remunerated or even recognised as labour. On one hand, this includes input provided outside any labour arrangements, such as using self-service technologies (such as shop checkouts), generating big data or player-produced modifications for video-games (Ekbia and Nardi 2017). On the other hand, digital labour platforms organise paid work in certain ways – such as piece rates, on-call working, self-employment – that allow them to remove portions of working time and many work-related activities from the scope of paid labour (see also Piasna 2019). This results in unpaid labour, which we define as ‘a worker’s time or effort outside the fixed hours and hourly rates of an employment relationship’ (Pulignano and Morgan 2021: 3), which according to Moore and Newsome (2018) should be considered a dimension of precarious work. In this policy brief, we examine activities experienced by platform workers as unpaid labour and undertaken across a variety of on-location delivery and online (remote) freelancing platforms,
across different countries and regulatory contexts in Europe.
We show that gig work in both on-location delivery and remote freelancing platforms includes unpaid labour. It involves the extraction of economic (‘surplus’) value from the workforce without compensation and usually consists of unremunerated, yet ‘productive’ activities (such as labour time spent waiting or searching for tasks/orders, travelling between orders, building a reputation) performed by the worker and/or freelancer beside their paid tasks. Although we recognise that some forms of unpaid labour may not be specific to platform work, and may characterise jobs outside and inside the platform economy (for example, creative freelancers), it is our contention that unpaid labour within platforms is distinctive and results from how platforms govern labour to match clients with workers through digital intermediation, such as algorithmic-based
optimisation mechanisms, performance ratings (including metrics and reviews) and processing of data on workers and clients. Among other things, we found platform-based reputation systems based on ratings to be unidirectional, with only workers held accountable, allowing clients’ and platforms’ unfair practices to go unchecked and increasing the burden on unpaid labour. Based on our findings, we distinguish two forms of unpaid labour (see Table 1). Both are present under different contractual arrangements and pay systems, in on-location (Deliveroo, Takeaway), as well as remote freelancing platforms (Upwork, Malt):
- time-based, including unpaid overtime, waiting time, time spent searching for tasks, travelling to work and between jobs, unpaid breaks at work;
- not time-based, including work intensification, pay-to-labour, including platform fees, and purchasing equipment.
The observations outlined in this policy brief are based on ongoing large-scale empirical research on gig work, the ResPecTMe Project funded by the European Research Council (ERC) under the European Union’s Horizon 2020 programme. Empirical evidence presented here comprises 62 in-depth interviews with platform workers and 16 working time diaries, collected during fieldwork in Belgium, France and the Netherlands between May 2020 and July 2021. Working time diaries were audio-recorded by respondents within a time span of 10 working days. The investigation includes on-location delivery platform work (Deliveroo, Takeaway) and online freelancing platforms where work is provided remotely (Upwork, Malt). It also offers comparisons across different models of organising work, employing and remunerating workers: hourly pay versus piece rates and task-based self-employment, and freelancing versus dependent employment.
Time-based forms of unpaid labour
Deliveroo is an entirely app-based food delivery platform, enabling the use of a large flexible workforce to offer fast and fully-tracked services. Since March 2020, Deliveroo has been using a ‘free-login’ system whereby workers are assigned orders based on their speed and spatial efficiency of delivery. These criteria are assessed algorithmically with the use of data collected from restaurants, riders and clients. We found that the provision of unpaid labour on Deliveroo results mainly from the ‘free login’ system and a ‘piece-rate’ payment structure, with workers paid per delivery. Indeed, because ‘everyone [with a bike and a smartphone] can connect to the app at any time’ (FRCM14) and compete for orders, workers experience long and unpaid waiting times. Over half (52 per cent) of working days for Deliveroo involved waiting time because the platform does not assign enough orders (see Figure 1). The uncertainty about the number of orders is increased by algorithmic management that allocates work based on workers’ availability, speed and order acceptance/ rejection: ‘Deliveroo introduced lots of insecurities because now anyone can log in any time (..) so you don’t have to book in advance but everyone can start to work on a Saturday evening, which means that you do maybe one order per hour’ (NLMR02). When after logging in workers see that it is a quiet time, they can wait without compensation or log out, forgoing potential earnings: ‘At 8am I logged in but it wasn’t that busy so I just went home’ (BEMF38); ‘today I worked even less than normally because there were no orders’ (BEMF39).
Waiting times at restaurants and clients’ doors are also unpaid and cannot be used to search for new orders: ‘I had to wait in a restaurant for 15 minutes, I’m not paid for that’ (FRCM21); ‘today I had to wait 20 minutes to get an order’ (NLMR02). Moreover, the time it takes a rider to fix any errors, also those made by restaurants or clients, is also unpaid: ‘I drove back to a restaurant because they forgot the drinks, for that extra effort and distance I don’t get paid’ (BEMF08).
Deliveroo adjusts per-delivery pay according to fluctuations in demand and supply, which results in highly volatile earnings: ‘We never know how much we will earn. We depend on the number of people logged-in, the rates that platform offers and on the clients’ (FRCM21). Because of long and frequent waiting times that are unpaid, aggravated by a limited availability of paid work, actual hourly earnings are much lower than the advertised rates and often insufficient to make ends meet: ‘You’re paid peanuts considering how long you had to wait before receiving an order and in between orders or how long you had to bike to reach the point you need to be (…) I earned €12.89 today but I had to cycle a lot. That’s not enough to cover my basic needs’ (BEMF39); ‘it is impossible to use Deliveroo as the main source of income’ (NLMF01). Because the workplace is not defined, travelling time to and from work, as well as between orders represents another time-based form of unpaid labour.
This is more commonly reported in relation to Deliveroo’s piece rate system than concerning the hourly-paid Takeaway workers (Figure 1). Riders often live in the suburbs, and it may take them ‘an hour of cycling to reach the city centre, where they can check in and start working’ (BEMF38). Importantly, in a piece rate system there is no guarantee of obtaining any paid activity after investing time in travelling to work.
Takeaway pays workers by the hour. In France, workers are guaranteed working hours in their contracts as Takeaway employs them directly. In Belgium and the Netherlands, they are often employed via intermediate employment agencies and are guaranteed at least two shifts of 2–3 hours per week. Takeaway uses performance ratings to assign work within a shift-based system. Workers indicate their availability and, depending on their algorithmically calculated performance (based on platform metrics and client reviews) are allocated working hours. Good performance is rewarded with extra hours of work (sometimes even stable employment) and pay raises. While this reduces the financial risk of unpaid labour associated with piece-based pay models (such as Deliveroo) by providing some employment stability, it does not fully liberate workers from unpaid labour, in the form of either compulsory unpaid breaks – ‘If you do two shifts in a row, there has to be a 2 hour break in between’ (FRCM25) – or unilateral shortening and prolonging of shifts by the platform – ‘My shift was ending at 8pm and I was sent an order at 7:56, which I’d give to the client around 8:30. But I was late and [finished at] 9.00, so I worked one extra hour. And I got paid only for 30 minutes extra’ (NLMR05).
On Upwork (a global remote labour platform) and Malt (a remote labour platform operating in France and in the French-speaking part of Belgium), workers are paid per task, for which they compete with other freelancers or are selected by clients directly. Workers on both platforms are presumed to be self-employed. Key to their employability and income is reputation, which they build based on reviews submitted by clients after task completion, with performance rating systems awarding them badges, such as ‘Top Rated’ and ‘Rising Star’ on Upwork and ‘Super Malter’ on Malt. Employability is not governed autonomously by freelancers, however, but determined by algorithm-based performance ratings systems. Hence, freelancers work long (and often unsocial) hours without any guarantee they will receive a paid task. Time and effort spent searching for available work and applying for tasks represents one of the main forms of unpaid labour: ‘I am always there checking what is available’ (FRCM13) (see Figure 2). Remote platform workers report that access to tasks and the ability to set pay levels depend on an algorithmically calculated reputation based on reviews from previous clients, but the algorithm’s decision-making is opaque: ‘It was never explained how the algorithm works, but they would hint that if the contract is for a larger sum of money or for a longer time, it’s better for your job success rate’ (BECM02). A failure to achieve the highest score represents a serious threat, impacting employability and future pay: ‘I was contacted by a Malt advisor, who explained a client didn’t want to pay in full for the translation because it’s not good. I kept demanding the full payment but I started reflecting because a bad review on Malt is thousand times worse than not getting my money. So I said: I accept these conditions only if the client doesn’t give me a bad review.’ (FRCM03). It is not uncommon for workers to vastly undervalue their own work by setting lower rates simply to receive first reviews and improve
employability: ‘Clients are like: ‘Oh, but this was also a part of the project, and this, and that. So at the end you keep doing things for free or you reduce your rates just for the sake of getting good reviews’ (BECM02). Freelancers commonly over-deliver to receive the highest possible scores, which results in ‘working unpaid extra time and efforts because it’s outside the hours and outside the job description’ (NLLV01). Workers log in frequently and accept assignments under the threat of having their profile deactivated. In this context, freelancers are in an asymmetrical position in relation to their clients, who often ‘ask the impossible to freelancers within a minimum space of negotiation from the side of the freelancer’ (FRCM02). Consequently, freelancers spend long hours communicating with clients, often to resolve inconsistencies between task descriptions and clients’ expectations (reported in 88 per cent of the working days analysed). Freelancers also complete extra tasks to keep clients satisfied and get good reviews (in 30 per cent of the working days analysed), and extra tasks needed for project maintenance, planning or work-related tools (that is, work intensification and extensification) (in 57 per cent of working days) (see Figure 2).
Forms of unpaid labour that are not time-based
Workers on Deliveroo also reported a variety of non-working time–related tasks that they tend not to be paid or reimbursed for, such as purchasing and maintaining work equipment, including an outfit, bike or phone with an internet plan: ‘Before work I repaired my bike and did a bit of maintenance myself; what I earned did not cover these costs’ (BEMF39). Under the piece rate system, pay can be further reduced by incidents, road accidents and weather conditions, which may constrain the number of orders workers can undertake and require them to engage in various actions to remedy their situation: ‘I worked less today because I got a flat tire’ (FRCM21). On Takeaway, as workers are paid per hour and the platform removes slack time from paid hours by dispatching as many orders as possible to be delivered within the paid working time: ‘At the very moment you’ve delivered an order, bam, you get a new one’ (BEMF18); ‘I need to be fast with picking up the order and delivering it. If I’m not fast enough, I might face some trouble. I might even lose my job. And you don’t know how many orders you might have. You might do in a 4-hour shift more than 30 or 35km, so in the whole day you might do 70km’ (NLMR05). Intensifying work by squeezing shifts creates unpaid labour for workers, particularly in the absence of stable guaranteed working hours, as workers can be deactivated at any time if they cancel already accepted orders. Therefore, workers tend not to refuse extra, usually unpaid, work within their regular paid hours in order to be able to receive work in the future: ‘I need good scores to guarantee work’ (FRCM22). Workers who ride more slowly or who reject orders are sanctioned: ‘If a rider keeps getting less than two deliveries per hour, he’ll receive a warning, and then another one and eventually will be kicked out’ (BECM10).
Within remote labour platforms such as Upwork and Malt, freelancers are required to invest their own money in purchasing the platforms’ virtual currency to be able to apply for jobs (for example, connects on Upwork), as well as to pay platform fees and commissions (reported in 70 per cent of the working days analysed). Sometimes freelancers are also asked by potential clients to prove themselves by sending samples of their work or conducting small tasks for free (in 5 per cent of the working days analysed). Moreover, freelancers tend to cover the costs of training to improve their profiles and promote themselves on the platform, while such skills may not be transferrable to other contexts. Many respondents see it as an investment in the form of ‘paying for the chance to earn’ (NLLV01) and a form of unpaid labour: ‘If it’s a one-time job, you pay two connects. And if it’s a longer period, it’s six. Sometimes I use six connects because I want something long term but it’s not and therefore I hardly get my investment in connects back’ (FRCM06). Overall, one outcome of lowering rates to get good reviews and investing one’s own money (27 per cent of working days analysed) is that freelancers may struggle financially: ‘Today I had to work for four hours without a break because the client was in a great hurry. 3000 words in 4 hours, it was extremely tiring and stressful. I’ll get only $67 for this and when I deduct my own costs, it doesn’t leave me enough to live on. but I cannot afford to leave the client disappointed’ (FRCM16).
In its Work Programme the European Commission announced that a legislative initiative on improving platform workers’ working conditions should be tabled by the end of 2021. The findings presented in this policy brief illustrate that it is vital that the policy debate on possible solutions in this area recognises that the issue of unpaid labour is built into the way the platform economy currently operates. Our research indicates that unpaid labour leads to an overestimation of the value produced by platform work and an underestimation of its costs, as completion of tasks that are paid does not represent workers’ entire temporal and financial investment, effectively lowering their earnings. We contend that the extent of unpaid labour inherent in platform work sheds new, and critical, light on the claims made by platforms about their positive employment effects. Labour platforms provide access to paid work, but they also rely heavily on unpaid labour for their profits. In our view, limiting unpaid labour within platform work can generate further (paid) employment possibilities. Unpaid labour impacts not only income, but also working conditions more generally, including working time and work intensity. This is because, as findings indicate, providing unpaid labour is contingent upon the platform’s use of new digital intermediation technologies to govern labour. Algorithmic control systems optimise the process by analysing and using workers’ performance
ratings, metrics and data collected from clients and users to make decisions about allocation of future tasks and worker retention. This results in different forms of unpaid labour, with algorithmic management in general increasing their burden. While unpaid labour may include non-time-based forms, such as payment of platform’s fees and commissions, time-based forms of unpaid labour are prevalent on the remote work platforms we analysed (Upwork and Malt). They are directly linked to lengthy and unsocial hours spent on the platform to augment reputation ratings, which are platform-specific and thus may in fact ‘lock-in’ freelancers to that particular platform. But reputation is essential for freelancers to develop their portfolios and establish economic transactions and relations with clients in the market more generally (Pulignano et al. 2021). This calls for policy interventions aimed at extending protection for different forms of labour under EU competition law. Within on-location food delivery platforms, different compensation systems are crucial to explaining the nature of unpaid labour. Unpaid labour in general, and its time-based forms in particular, is undertaken with a relatively higher frequency by piece rate workers (Deliveroo) than by hourly-paid ones (Takeaway). The latter, however, still experience a considerable burden of non-time-based forms of unpaid labour, such as shortening of shifts without compensation and squeezed shifts, resulting in work intensification. Unpaid labour in the different forms revealed in this study can have spillover effects on people’s working lives. We point to four main effects of this kind (see Figure 3). First, work intensification leads to poor health and well-being by impairing both physical health and job satisfaction. Second, unpredictability
related to working longer and unsocial hours disrupts work–family life balance and accentuates work–life conflicts. Third, investing one’s own money and equipment when there is a poor return on investment because of centralised algorithmic decision-making can lead to a lack of autonomy, income insecurity and, possibly, precarity. Fourth, unidirectional and non-portable ratings in a context of increasing competition for assignments may enhance clients’ authority over workers, ultimately increasing power asymmetries in favour of work providers and severely limiting participation and workers’ voice as important features of democracy at work.