Recognition of Qualifications in the Gig Economy
Regulating for Globalization
Please refer to this post as:, ‘Recognition of Qualifications in the Gig Economy’, Regulating for Globalization, 04/09/2019, http://regulatingforglobalization.com/2019/09/04/recognition-of-qualifications-in-the-gig-economy/
In a series of earlier entries on this blog, I have argued that all human activity is skilled; work involves the productive use of one’s skills; the most fundamental right in relation to work must be the right to work; an important derivative right thereof is the right to have one’s skills recognised; and the right to education and training merits special attention by virtue of a focus on skills. In this post, I continue my focus on skills by considering the extent to which the skills of platform workers in the EU can be recognised. This post builds in particular on the recent entry by Valerio de Stefano, Antonio Aloisi and Six Silberman, ‘A manifesto to reform the Gig Economy’. It aims to provide practical steps to implement one of the reforms they envisage, namely, a portable rating system for platform workers.
Platform workers provide their services in the EU through the medium of digital work intermediaries or so-called ‘platform employers’. These platform employers are considered innovative because they enable people or organisations to effectively outsource tasks to a large pool of on-demand workers at a very low cost. The ‘platform’—the technology enabling people to seek workers for a specific task or set of tasks—makes it easy to match supply and demand in the labour market because the platform facilitates access to a large and varied workforce. Essentially, the ability to increase the speed at which supply meets demand in the labour market facilitates a reduction in what economists call ‘transaction costs’, namely, the costs associated with locating and matching supply and demand. The development of digital work intermediaries or platforms has been branded the ‘gig economy’ because it enables almost anyone to perform work tasks, often from the comfort of their own home. An example will help to flesh out this point.
An Example: Amazon Mechanical Turk
Amazon Mechanical Turk (‘MTurk’) is Amazon’s online labour platform, which provides potential workers with the opportunity to earn extra money in their spare time, and potential employers with easy access to a large pool of on-demand workers who can perform basic, non-expert tasks. Why, then, should we focus on MTurk? One reason is that MTurk, unlike some other platforms, facilitates access to a global network of workers. Other platforms, by contrast, facilitate access to local workers only. Uber or Deliveroo, for example, only provide access to workers within one’s own jurisdiction, and most often only those within a certain distance of one’s actual location. A second reason is the availability of research data. Amazon MTurk has attracted the attention of numerous labour law scholars in recent years. A final reason is the role of technology in MTurk’s qualification system, a matter to which I now turn.
MTurk emerged as a response to an inability to automate certain low-skilled tasks in Amazon’s own business. Initially, Amazon required a system to clean data coming from its huge product catalogue. Amazon was unable to achieve this outcome computationally and so it decided to outsource this function through ‘crowdsourcing’. Crowdsourcing involves the outsourcing of functions which would otherwise be completed within a business to a large, undefined group of people in the form of an open call, generally through a specific platform or series of platforms on the internet. Initially, Amazon crowdsourced these functions to its own workers by setting up its own, internal online platform; Amazon’s workers could complete tasks available on the platform in their spare time. Over time, Amazon then permitted workers from outside the company to join the platform and also expanded the range of tasks available to be completed. It then included a mechanism to facilitate payment to workers.
How, then, does MTurk work on a day-to-day basis? In accessing the platform, the user is requested to sign up as a requester or worker. Requesters, or putative employers, post ‘human intelligence tasks’ or ‘HITs’—generally, low-skilled tasks involving, eg, identifying objects in photographs, completing surveys and transcribing audio or video clips. Requesters and workers are located anywhere in the world, although most are currently based in the USA and India. MTurk thus enables rapid access to a global on-demand workforce which is generally well-educated; according to Six Silberman and Lilli Irani, the most active MTurk workers hold a bachelor’s degree or higher. When a worker completes a task, it is then sent to the requester for approval; requesters can approve or reject completed tasks at will. Requesters can also set certain qualification levels for workers, eg, that they be located in a specific country; that they have a masters degree; etc. Amazon also has its own qualifications system, describing certain workers as ‘Masters’, an elite pool of workers who have attained a high degree of work experience on the platform, according to Amazon’s own technology, which identifies worker performance and monitors their performance over time. Once attained, the Masters qualification must be maintained; MTurk Masters must continue to meet MTurk’s statistical monitoring requirements to maintain their qualification.
In terms of the number of workers available, in 2011 Amazon declared that there were over 500,000 unique workers operating in 190 countries on MTurk. However, some researchers suggest that there are approximately 50,000 active or regular workers, of which there are even fewer—in the range of 7,500-10,000—full-time workers. These numbers are, surprisingly, quite small relative to other platforms. Upwork, for example, a platform hosting freelance workers from across the globe, provides access to more than 9 million freelance professionals and workers. Freelancer, a similar platform, by contrast, hosts approximately 31 million freelance professionals. Each of these platforms also provide internal qualification systems. Upwork provides online skills tests and a ratings system for its freelancers. Freelance workers can become ‘Top Rated’ by having a high job success score, being available for tasks and having 12-month earnings of at least $1,000. Similarly, Freelancer provides for a ‘Preferred Freelancer’ programme, which attracts higher paying tasks. Besides these specific qualifications, in general, qualifications or ratings systems involve a degree of automatic, automated or technological oversight and supervision of workers. Indeed, in some circumstances,
“algorithms are absorbing many organizational functions that managers traditionally would perform. Computer code may perform a variety of supervisory tasks from the mundane to the sophisticated: assigning tasks to workers, speeding up work processes, determining the timing and length of breaks, monitoring quality, ranking employee, and more. Code makes crucial on-the-spot decisions about individualized employees and what they need to be doing in real time.”
MTurk is itself an example of this phenomenon. As noted briefly above, MTurk uses technology of some kind—it does not disclose what sort—to assign their own ‘Masters’ qualification, in addition to assigning work in general. In this way, workers’ skill utilisation and development are mediated through technology in a largely unregulated—and undisclosed—manner. Another, less insidious method of skills recognition has also been deployed in the platform economy. That is the model of customer ratings whereby each worker receives ratings based on customer experiences. By leaving a rating on their experience, customers affect the ability of workers to receive work in the future. Good ratings translates to more work, bad ratings to less; this much is clear from empirical studies. The effect is that the customer determines how qualified the worker is and how skilled they are at the tasks which they are charged to perform.
A number of lessons can be drawn from the analysis of MTurk for the regulation of the platform economy generally, and for the utilisation, recognition and development of platform workers’ skills specifically. The first stems from the last-mentioned point, that of ratings and qualifications, and it has two aspects. First, the use of algorithms and customer ratings tends to tie workers to specific platforms, which seems contrary to the spirit of freedom and entrepreneurialism embedded in the platform economy. This has led to one commentator suggesting that ratings should be portable, allowing workers to move between platforms and deploy their skills as and when necessary, receiving adequate compensation and recognition for past work. This seems like an appropriate regulatory response to an otherwise precarious position. In terms of regulatory design, Jeremias Prassl recommends the General Data Protection Regulation as a starting point, which requires personal ‘data portability’; according to Prassl, ‘[p]ortable ratings could operate along similar lines, with standardised metrics accounting for experience, customer friendliness and work quality’. Obtaining such a degree of control and ownership of one’s own skills and qualifications is important.
This could operate in a similar manner to the European Skills Passport currently available in the EU, an electronic portfolio which gives a comprehensive picture of one’s skills and qualifications. The Passport can be easily formatted, translated and transmitted for the purposes of advertising or communicating one’s skills. It allows a person to upload documentation validating their skills and qualifications, such as degree certificates, attestations of employment, and so on. Adapting such a tool to incorporate the skills, experience and qualifications acquired and developed by platform workers in general, and those working in third-countries in particular, would be beneficial. This is something which the European Commission could develop as it has already developed a similar tool in respect of third-country nationals, namely, the Skills Profile Tool for Third-Country Nationals. Making available such an online tool to all platform workers would be easy as internet access is generally global. A European Skills Passport could provide a greater degree of transparency of skills and qualifications. In terms of how this would work in the platform economy, all workers could use the skills passport as the device for communicating their skills. Thus conceived, the skills passport would be similar to the European Qualifications Framework, discussed in a previous blog post which aims to be the standard by which all qualifications can be assessed. On this view, a European Skills Passport for would facilitate direct comparison of skills and qualifications by potential customers or employers, rather than relying solely on platform-specific reviews.
Relatedly, the European Commission has proposed a regulation concerning online intermediaries. According to the proposed regulation, providers of online intermediary services must set out, in their terms and conditions of use, the main parameters determining ranking and the reasons for the relative importance of each such criterion. This is subject to the caveat that any such disclosure may exclude ‘trade secrets’, within the meaning of the Trade Secrets Directive. This obligation may require platforms such as MTurk to disclose on what basis they rank and qualify platform workers.
Linked to this is the role of differentiation in ranking. According to art 6 of the proposed regulation, online intermediaries must include, in a statement of its terms and conditions of use, any differential treatment that they may or shall give to certain products or services, in particular to such products or services which are distinguished by different rankings. Article 7 concerns access to personal data. According to the terms thereof, online intermediaries are obliged to include, in their terms and conditions of use, a description of any technical or contractual access, or lack thereof, of users to any data which they or their customers provide in participating in the online intermediary. This resonates strongly with the point noted above, made by Prassl, that online intermediaries should facilitate access to a worker’s personal data for the purposes of communicating to others their skills and qualifications to date. Finally, art 13 requires the Commission to encourage the development and proliferation of codes of conduct for the platform economy. The Commission has, however, acknowledged the difficulty with such codes of practice, observing
“A prerequisite for the effectiveness of this measure is the alignment of all interest involved and the existence of sufficient incentives for both business users and platforms to participate. Achieving this, particularly regarding ranking, is not realistic. In addition, stakeholders need to be represented equally, a criterion that leaves the much less organized business stakeholders at a disadvantage.”
It therefore seems that a significant degree of regulatory uncertainty remains. Nonetheless, eight German labour platforms have signed up to the ‘Crowdsourcing Code of Conduct’, according to which local wage standards are guaranteed. The code of conduct also provides that measures will be taken as regards training and retraining opportunities, prizes and awards, and experience points and badges in respect of work achieved. The code is officially supported by the German Crowdsourcing Association.
The primary problem with the above-mentioned approach is its reliance on private codes of conduct. As the Commission rightly observed, ‘stakeholders need to be represented equally’ and this is something which needs to be achieved so that a fair balance may be struck between convenience for consumers and fair standards for workers. Moreover, private codes of conduct can be messy. There is normally not just one private code of conduct; there are several. Furthermore, such codes of conduct are rarely judicially enforceable and, even where codes are taken seriously by consumers, suppliers and workers, it is not always clear that they are effective. It is perhaps the case that more time is needed to fully explore and develop the culture of utilisation, development and recognition of codes of conduct in respect of the platform economy and labour supply chains more generally. Certainly, at this stage it is simply too difficult to assess how successful these will be. Indeed, in respect of much of the platform economy, the primary concern at present is whether, and to what extent, the labour market intermediary or platform employer is an employer at all. Much remains to be worked out in this respect.
 Valerio De Stefano, ‘The Rise of the Just-in-Time Workforce: On-Demand Work, Crowdwork and Labor Protection in the Gig Economy’ (2016) 37 Comp Lab L & Pol’y J 471; M Six Silberman & Lilly Irani, ‘Operating an Employer Reputation System: Lessons from Turkopticon, 2008-2015’ (2016) 37 Comp Lab L & Pol’y J 505; Miriam Cherry, ‘Beyond Misclassification: The Digital Transformation of Work’ (2016) 37 Comp Lab L & Pol’y J 577; Jeremias Prassl and Martik Risak, ‘Uber, Taskrabbit and Co: Platforms as Employers – Rethinking the Legal Analysis of Crowdwork’ (2016) 37 Comp Lab L & Pol’y J 619; Antonio Aloisi, ‘Commoditised Workers – Case Study Research on Labour Law Issues Arising from a Set of On-Demand/Gig Economy Platforms’ (2016) 37 Comp Lab L & Pol’y J 653; Jeremias Prassl, Humans as a Service: The Promise and Perils of Work in the Gig Economy (OUP 2018); and Sarah Kessler, Gigged: The Gig Economy, the End of the Job and the Future of Work (Random House 2018).
 Silberman and Irani (n 1) 510.
 Cherry (n 1) 596-597.
 Rossana Ducato, Miriam Kullmann and Marco Rocca, ‘Customer ratings as a vector for discrimination in employment relations? Pathways and pitfalls for legal remedies’ Marco Biagi Conference, Modena, 19-21 March 2018.
 Humans as a Service: The Promise and Perils of Work in the Gig Economy (n 1) 111-113.
 ibid 112.
 According to the Commission, ‘Commission Staff Working Document Impact Assessment’ SWD(2018) 138 final 109, however, it is likely that the algorithm used to rank users may be a trade secret.
 ibid 115.
 In the supply chain context generally, see Jette Steen Knudsen, ‘The Growth of Private Regulation of Labour Standards in Global Supply Chains: Mission Impossible for Western Small- and Medium-Sized Firms?’ (2013) 117 J Bus Ethics 387; Axel Marx and Jan Wouters, ‘Redesigning enforcement in private labour regulation: Will it work?’ (2016) 155 Int’l Lab Rev 435; and Francis Churchill, ‘Certification ‘making no difference’ to supply chain labour abuses’ Supply Management (5 June 2018) <https://www.cips.org/en/supply-management/news/2018/june/certification-makes-absolutely-no-difference-preventing-labour-exploitation/> (accessed 7 February 2019).