Regulating labour platforms, the data deficit

It is widely reported that there is a data deficit regarding working conditions in the gig economy. It is known, however, that workers are disadvantaged because they are not classed as employees with the result that they lack work-related entitlements and may not be protected by the social welfare safety net. Nor is this compatible with the social market economy enshrined in the European Union treaties. Two obstacles are that labour law and social policy are mainly a national competence and that platforms are reluctant to share data with regulators. In this paper I take the specific case of offline labour platforms intermediated by app and smart phone such as driving and delivering and look for new pathways between access to data and the shaping of public policy in member states with potentially legal certainty..


Introduction
'A paradox in the digital and Internet economy is that never before has so much data been collected, and never before has it been so difficult to access. The value of this data is likely to be much higher for social and public policy purposes than it is for private purposes of the platform operators' (Codagnone, Biagi and Abadie 2016:60,61) The labour platform model of the gig economy offers popular services but depends for its competitive advantage over established service providers on low wages, precariousness and lack of in-or -out of work entitlements and benefits. This raises the question of its compatibility with the concept of the social market enshrined in the European Union (EU) treaties, specifically with regard to employment practices. There are different models of the social market throughout the EU but it has been broadly defined as a 'fundamental social model that ensured people's rights, inclusion to social protection, a good wage model that shared productivity through collective bargaining and social dialogue' (Burrow 2018) and, specifically, 'the creation of a world of work that is humane, socially balanced and sustainable' (Jürgens, Hoffmann and  A.J. Hawley / European Journal of Government and Economics 7(1),[5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] 6 Schildmann 2017:222). This is reflected in the concept of 'good work' in the gig economy by Taylor (2017) who measures it against a number of criteria such as, 'employment quality, working conditions, consultative participation and collective representation in addition to wages' and also 'good gigs' by Balaram, Warden and Wallace-Stephens (2017) and as 'Gute Arbeit' in Germany (German Federal Ministry of Labour and Social Affairs 2017). With the rise of the gig economy is there a need therefore for new regulation or are existing rules being breached or evaded? How do we know? This is a question of data, its availability and quality. Two important considerations are the role of the EU, since labour law is mainly a national competence, and the perennial cleavage of opinion about when and whether to regulate any economic activity. It is, however, the subject of data in relation to working conditions in the gig economy, and specifically to labour platforms, which is principally addressed in this paper.
Labour platforms do not provide the entitlements generally expected for their workers because they do not accept that they are employees. There is little if any corroborated, independent data, however, on how much they are paid, the hours they work, discrimination of any kind, the transparency of rating systems and opportunities for worker training and representation. We do not know much about their conditions except anecdotally and what platforms are willing to tell us through the medium of a few privileged researchers. By contrast, independent survey data presents a per country aggregate of demographics, people's motivations and total hours worked on platforms. Except in one case that I have found (Balaram, Warden and Wallace-Stephens 2017) they do not reveal data on conditions on specific platforms. Nor, why should the latter reveal them for competitive or any other reasons?
Since their workers are generally treated as self-employed, even though they may be largely or even wholly dependent upon a single platform, their conditions presently do not fall under national labour law regulations, for example, the minimum national wage (or minimum living wage in the United Kingdom) or transposed EU legislation in the case, for example, of the In the first half of this paper I start with a definition of what is meant by the catch-all term of the 'gig' economy and the part described as 'labour platforms' with which this paper is concerned. There follows an overview of the literatures of organisation and management, paradigms of political economy and cultural responses towards rapid technological change that may have some explanatory value for the rise of the phenomenon and also differing attitudes towards it. These differences, I show, are being played out in an EU context as well as a national one.
The second half of the paper is devoted to how data might be acquired and put to work systemically to improve the conditions of gig workers. I start with a brief review of the data from surveys and studies carried out in Europe and the United States. If there is a change in public policy, the "elephant in the room" however, will be data compliance by platforms, an issue which looks as though it will present formidable obstacles and one on which little research has been done with the notable exception of Arun Sundararajan (2017). I sketch out his approach generally known as 'shared regulation' and the few others with the similar objective. Since labour law and social policy are mainly national competences I consider the EU's use of soft power to achieve its objectives in this field. Specifically I look for a pathway between access to data on working conditions and legal certainty in establishing minimum standards for workers.
Finally, I conclude by suggesting further areas of research for a deeper evaluation of opportunities for change in this field following the recommendations and principles of the recently proclaimed European Pillar of Social Rights (EC 2017).

Defining the 'gig economy'
There is no precise definition in the literature which contributes to the difficulty in measuring it.
The EU still refers to it by the loose, catch-all term 'collaborative economy' which could include both for-profit and not for-profit models and those where only the expenses of providing the service are recovered plus a small percentage (e.g. Blablacar, a car-sharing service in France). A similar development is true of Airbnb, an accommodation-sharing service, where a large proportion of properties have been bought solely for renting out. 'Gigs', however, is more widely understood as referring to activities where labour is performed rather than the exploitation of A.J. Hawley / European Journal of Government and Economics 7(1),[5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] 8 assets such as accommodation. 'Gigs or 'gig work' therefore best describes services performed by individual workers dependent upon platform intermediaries, sometimes several, to which they are connected by a smartphone or computer app, typically poorly paid (sometimes less than the national minimum wage), capable of being 'deactivated' at any time, without a collective voice, and lacking in either any of the work-related entitlements enjoyed by workers classified as 'employees' or full access to statutory state benefits when they need them. By contrast, surveys show that the flexibility to work when and where they want is liked by gig workers. Equally, however, many say have little other choice. Work in the gig economy can be online 'mind' services ranging from simple, repetitive tasks (as in Clickworker) to those requiring higher skill levels (as in Upwork) or offline physical services (such as ride-hailing or delivering). It is with the offline labour platform model with which this paper is concerned and how it fits into the interlocking circles of the overall 'collaborative economy' is shown in Figure1. 'Gigs' are very often neither 'peer-to-peer' as has been said nor in the case of Uber occasional. According to the European Court (ECJ 2017: §47) 'it has become apparent that most trips are carried out by drivers for whom Uber is their only or main professional activity'.
Although I am chiefly concerned with the physical rather than mind services, the social dimension of poorly paid, precarious work without entitlements or benefits can equally apply to both.

Literature and theory
The gig economy is a recent development brought to popular attention under the title 'collaborative consumption' by Botsman and Roo (2010) followed by Rifkin (2014) whose theme was 'the collaborative commons'. It has since become the object of wider scholarly research. Bardhi and Eckhardt (2012), for example, found that a car-sharing model (Zipcar) set up primarily for access rather than profit (costs of operation covered) with benefits for sustainability failed to gain brand recognition loyalty. Martin (2016) has drawn attention to the co-option of the 'sharing' ideal by 'the market', Stark (2015, 2016)  surveyed ways in which its drivers are obliged to 'earn' their ratings under a reputation system to which most of them object, including, having also to perform 'emotional labour' for example, tolerating rude behaviour and providing (at their own expense) small comforts for passengers or having them sit in the front passenger seat. A plethora of reports have emerged from the EU institutions, national governments and independent research centres.

Organisational aspects -bad for workers, good for consumers (and potentially investors)
The rise of the gig economy can be situated within the context of the information society described by Nye (2014) and Nagirnaya (2014) while its salient organisational characteristics are identified as information asymmetry Stark 2015:2016), ruthless, flexible and evasive entrepreneurialism (Mejia 2016;Sennett 2006;Elert and Henrekson 2016). Traits such as disruptiveness and lack of corporate social responsibility are pre-figured in earlier texts by Christensen (1997), Christensen, Raynor and McDonald (2015) and Crouch (2006), while it remains unclear whether Schumpeter's (1934,1939,1942) theory of creative destruction is applicable as I have found no data which confirms that there has been a net increase in new jobs. In terms of wages, evidence is of the opposite. Consumers, however, may benefit from lower costs and greater convenience.

Management aspects -back to the 19th Century
Management is about how workers are treated and the organisational model described above example, shares the principle of scientific management with its algorithmic control and panoptic surveillance of every task undertaken by its workers. Its aim is the same -to achieve the greatest possible return for the minimum outlay using new developments in technology and in this respect and others it goes well beyond Taylor. In his theory maximum output by workers equated with highest wages and self-satisfaction, for which management was responsible for providing all the resources and skills training required. Labour platforms such as Uber or Deliveroo, a goods-delivery service, treat their workers as instrumental, merely a commodity, a factor of production to be acquired (and disposed of) at the lowest possible cost in terms of wages, entitlements and representation, exemplifying Marx's conception of 'the reserve army of unemployed labour' (Braverman 1974(Braverman [1998:265-276) which is made available for low paid, unskilled service jobs first by mechanisation, then automation, and now (latterly) digitalisation.
By contrast, the peer-to-peer nature of gig work has been portrayed by platforms as a democratisation of labour by eliminating the traditional hierarchies of both Taylorism and the HR school and fostering flexible work, wherever and whenever desired. Offline labour platforms, however, have substituted an often mythic peer-to-peer claim for bogus self-employment.

Paradigms of political economy and cultural attitudes
Opinion on the extent to which the gig economy should be regulated varies throughout the EU and within the EU institutions themselves. The broadest division which animates policy makers and is already evident in current regulation is between pro-laissez faire, pro-choice ('leaving it to the market') and pro-social justice ('pro-values'). There are also those who do not unreservedly accept the application of new digital technology seeing it as 'the tyranny of rational choice solutions, alienated from concrete social practices' (Strong and Sposito 1995: 268) (Weber [1904(Weber [ /5], 1930 in which individuals are trapped by teleological efficiency, rational calculation and controlthe bureaucratisation of social order, the 'disenchantment' of the world (Greisman and Ritzer 1981:35). Michael Sandel (2012) draws attention to the 'market society' in which we now live rather than simply the 'market economy', and in which non-market values are 'crowded out'.

A European dimension
These cleavages manifests themselves in several ways. Firstly, they align closely with a revealed that Members of the EP were divided almost equally between those who called for further regulation at EU level and those who either did not or whose view was that it was expressly a matter for member states. Thirdly they are demonstrated in the communications of the European Commission which show that priority in policy making with regard to the gig economy has been placed more on the economic aspects -namely growth and competitiveness -than with social aspects. It is noteworthy that the Commission has consistently declined to bring forward any new legislation on the gig economy since publication of its set of non-binding guidelines contained in its Agenda for the collaborative economy (EC2016).
The stance of the Commission is seen by some, principally by trade unionists and others on the Left, as consistent with its liberalising, deregulatory, pro-business tendency. By contrast, there is a push back by industry and market economists who hold it responsible for largely perverse effects on business of social measures such as the Working Time Directive (WTD) and others relating to agency, temporary and part time workers, the prevention of discrimination against minority groups and health and safety. None of these, however, materially affect the conditions of gig workers because they only apply to employees. The proclamation of the Time Directive, both of which were already 'in the pipeline'. This is where access to data on working conditions has a critical role to play in shaping public policy and how this might be approached is discussed in the second half of this paper.

The data deficit
The lack of empirical data is widely acknowledged ( however, to talk about a lack of access as never before has so much data been collected as it has by platforms (Codagnone, Biagi and Abadie 2016;60,61). A review I conducted of the relatively small number of independent surveys that have been carried out in Europe reveals a focus on overall numbers, motivations and demographics of both service providers and consumers using offline labour platforms but very little data on specific working conditions and employment status of the former. Further investigation is required into these, notably pay, leave, health and safety, working hours, tax, insurance, collective bargaining and quality of working life (Huws et al. 2016:51). More detail has come from surveys commissioned by, or facilitated in the United States and Europe by Uber, but they have not been independently corroborated. The most noteworthy of these are the analyses by Krueger (2015, 2016). These have attracted criticism in the press and among other researchers being described as 'mostly descriptive and inconclusive' and whose findings on earnings as 'an utter misrepresentation of the reality' (Codagnone and Martens 2016:17,18). The operative data that would make a difference in formulating public policy, however, is not from surveys, whatever their credibility.
The data, in the era of Big Data are already there. They are collected by platforms twenty four hours a day, justified in some cased for security reasons. Uber, for example, collects licence, insurance and background checks (medical and criminal records) on each of it drivers as well as credit card details and mobile phone numbers for every rider. In effect it 'holds the key to the digital identities of its millions of contractors and customers' (Lashinsky (2017:140). The task is making it available to those who legitimately need it (and not, by oversight, to those who do not as has recently been reported), namely policy makers, regulators and tax authorities. This raises questions of commercial sensitivity and data privacy protection. Several approaches will be considered in the next section.

Deriving data -cooperation or coercion?
If the objective is to ensure that workers on offline labour platforms enjoy the statutory entitlements and social welfare benefits available to those classed as employees, as envisaged by the EPSR, then platforms need to be monitored. This suggests mandatory access to the data and hours over a period in an easily communicable way would help gig workers to determine the hourly rate for the work they have done. This is not necessarily clear given their irregular hours and may even be less than the national minimum or living wage.
These are examples of one-way transmission of data by the platform, for one specific purpose, and only the French system is designed to catch all platform workers, who may, moreover, be working on several platforms. A more ambitious development would be the twoway question and reply procedure proposed by Sundararajan (2017). Here the platform's database would be interrogated by any legitimate state authority for, presumably, any data which it is authorised to access. This would be achieved by the use of application programming interfaces (that is, computers talking to computers). In this decentralised system data remains inside platforms, no commercially sensitive data is released to third parties and only queried for the attainment of public policy objectives. The process is described as 'digital audit' which might suggest either an obligation on the part of the platform to comply with minimum employment entitlements or only, at least at first, a data gathering exercise for policy formulation within an ethos of best practice. Uber has on more than one occasion declared itself willing to make data on the earnings of its drivers available for tax purposes. Travis Kalanick, its co-founder and former CEO, offered to do so in France (Le Monde 2015) if an agreement was reached with government, and most recently in its White Paper on Work and Social Protection in Europe (Uber 2018:21) featured a screen shot on its app of pay out and hours worked over a period which could be shared directly with the tax authorities. In voluntarily making this data offer, Uber strenuously denies acting as an employer -this runs wholly counter to its business model -and its drivers can choose whether to use the option or not. The reception of such data might be useful if also made available to social policy makers, but this raises the issue of data protection under the EU's General Data Protection Regulation (GDPR) of 2018, which is supreme over national law and has immediate effect. Nor could they depend on it for a complete picture as Uber drivers might also be working on other platforms without a similar option.
In Sundararajan's model which he calls 'data-driven regulation', regulatory responsibility is 'delegated to the party that naturally has the data' (Sundararajan 2017:24). He compares this favourably with two other forms of self-regulation such as rating systems ('peer-regulation'), and transportation network companies ('TNCs', or 'Self-Regulatory Collectives') where government sets standards with which drivers working on a platform registered as TNC must comply. Rating systems, however, have been much criticised for being faked and lacking transparency while the standards with which TNCs such as Uber need to comply are primarily concerned with consumer protection. In both cases the aim is promoting trust. This is undeniably important but neither of these forms of regulation address the central issue of worker entitlements and social benefits. On the contrary, they are aimed at 'market' issues. Where the profit interests of platforms do not coincide with the wellbeing of their workers -and they evidently may not as they deny responsibility for them as employees -then a regulatory authority with the power of enforcement needs to step in.
As has been said, platforms already hold all the data that regulators might need to know and data analytics can present a picture of demographics (age, ethnicity, gender, family status), motivations and any other aspects that might be of interest for social cohesion in addition to pay, hours, leave and representation. Here, there are several operative issues. Firstly, each government, particularly those member states within the eurozone to which EPSR is primarily addressed, needs to decide on the objectives of its employment and social policies, secondly the data it needs to achieve them, thirdly whether platforms are to be mandated or consensual on the basis of best practice and finally, what are the privacy issues? On the first two, it is intended that member states will be guided by accompanying actions prescribed by the EPSR which will be discussed later in this paper. On the third, a dialogue between government and an industry association such as SEUK (Sharing Economy UK) might be constructive. Privacy is of concern to both platforms and to their workers. The former need to be assured that there is no possibility of commercially sensitive data being leaked to competitors and the latter that the privacy of their personal data is protected in conformity with the GDPR. Digital audit using APIs as proposed by Sundararajan would reassure platforms on the first score.
The concept of the national hourly minimum wage is difficult to apply in the case of labour platform workers who can log-on and log-off whenever they want and who may work on several platforms with different pay rates and ways of determining them (for example by time, journey, delivery). The more important objective for public policy in ensuring that platform workers earn a living wage, defined as 'as a wage above the legal minimum that reflects the cost of living families face . . the living wage should be based on an up-to-date basket of goods and services, reflecting social consensus as to what constitutes a decent standard of living' (Resolution Foundation 2016). In the UK, the national minimum wage (NMW) has been 're-branded' by government as the national living wage (NLW) but may not reflect what is required to meet this definition. Moreover, paying the NLW is voluntary for employers. The rise of the platform economy presents a further complication and that is why access to current data held by platforms in an era of non-standard employment is now required in setting a rate for a living wage.

Making data work
It is, so far, not clear how a how a conflict of interests between labour platforms and the social market can be resolved either through voluntary or coercive means. A first step is access to all the data necessary for the attainment of the employment and social policy objectives outlined in the EPSR. In this section therefore, I consider the accompanying actions prescribed by the EPSR to visualise how the use of data might create a pathway from its rhetoric to making a difference to the lives of gig workers some of whom are 'just about managing' to make a living and some who are not.
The EPSR is a soft power instrument relying for eventual effect on a raft of other soft power procedures and measures. The most prominent of these are the Social Scoreboard and the Country Specific Recommendations (CSRs) whose progress towards implementation are to be to be monitored on the European Semester. Legal certainty may also arise from Treaty clauses What is clear is that little direct evidence of gig workers' conditions, and specifically on offline labour platforms, finds its way onto the Social Scoreboard and hence influences social policyat least so far. Both this type of work, often still referred to as 'a-typical' or 'non-standard' despite its growing prevalence, and a Union system intended for measuring its impact are, after all, recent phenomena. Few member states have yet adjusted their employment or social security systems. The lack of an adequate statistical base prevents a comprehensive quantitative picture of those affected. The Commission (EC 2018) has decided that a way forward is for another soft law instrument, namely a Recommendation by the Council, to be approved by the co-legislators, on access to social protection for workers and the self-employed. It asks member states to commit themselves 'to collect and publish reliable statistics on access to social protection broken down by labour market status (self-employed/employee), type of employment relationship (temporary/permanent, part-time/full-time, new forms of work/standard employment), gender, age and citizenship'. New forms of work are understood to mean ondemand and platform work which are currently excluded from social protection. In this way and cooperating with Eurostat, suitable indicators should become available. In the end it comes down to member states to decide on their policies towards the social protection of workers but they are likely to be influenced by the data that is exposed on the Scoreboard and which become the subject of Country Specific Recommendations.

Conclusions
The argument of this paper is that the timely and comprehensive provision of data is required in adjusting labour law and social welfare polices to protect workers on offline labour platforms which I started by situating within the overall terminology of the collaborative, gig, sharing or peer-to-peer economy. My argument is based on two assertions that may not be obvious.
Firstly, it is not so much about looking for data -they are already there in abundance -but about access to them as observed at the outset (Codagnone et al. 2016:60,61). This led to a number of unresolved questions that will require further research. One is about the conduct of platforms, access to whose data on working conditions is required. be of significance for platform workers but this was already planned before the proclamation of the EPSR and will only benefit them if it is applied to all workers regardless of the employment relationship.
Since the European Economic Community (the 'common market') became the EU at Maastricht, the main thrust has been to make free movement in its various forms a reality. This has met with varying degrees of success, of which services, except financial, have scarcely been one despite the Services Directive of 2006. A major exponent of the labour platform model, Uber, has recently been ruled by the ECJ to be outside the acquis. Trade is an exclusive EU competence; employment and social policy are not. Attempts to harmonise the latter have accordingly been limited but they remain a Union objective of closer integration within the social market model laid down in the Treaty.
We have also witnessed a push-back against extension of the social acquis since the Great Recession following the banking collapse of 2007/2008 and the ensuing sovereign debt and euro crises. Germany has been regarded as an epitome of the social market yet in 2015 the then federal finance minister Wolfgang Schäuble commented (Varoufakis 2017:211,212) that the 'over-generous' European social model was no longer sustainable and had to be ditched.
Comparing welfare states in Europe with India and China where no social safety net exists at all, he argued that 'Europe was losing competitiveness and would stagnate unless social benefits were curtailed en masse'. Moreover, some do not believe that controlling the gig economy is a matter for jurisdictions and statutory regulation. Beyond protecting consumer safety, this should be left to market forces. This position is expressed by Niemietz and Zuluaga (2016) in their discussion paper on the UK's taxi industry for the Institute of Economic Affairs, a neo-liberal think tank. For them competitive pressures and trial and error will eventually determine what is tolerated through 'private' regulation and the emergence of 'regulatory brands' related to reputation, services offered and the suitability of their regulatory mechanisms (2016:37). As platforms are still an evolving phenomenon, this is an area for further research as is the sustainability of their business model should greater restraints be imposed upon them such as treating their workers as employees.
There is no reference, however, in the 'market' view to the quality of employment. Hans Kundnani (2018), to whom I am indebted for drawing my attention to Varoufakis' book argues that a 'competitive Europe' has now become the model for the European Commission and of pro-Europeans led by Angela Merkel, which 'bears little resemblance to the one of the "pro-European" imagination with its emphasis on the "social market economy". In becoming more "competitive", suggests Kundnani, 'the EU may be hollowing out the model for which it once stood'. Hence, for the present, only the exercise of soft power in relation to bringing social responsibility to labour platforms seems to be available to it. It may not be enough. In a globalised labour market which is rapidly changing due to technological developments the overarching question is whether there is room for an economic model that is both 'competitive' and 'social,' to which the EU aspires and which bears heavily on the integration project.