Unlimited employee autonomy over working time? The potentially empty notion of autonomous workers in EU law

2.1. The CJEU's narrow interpretation

The CJEU interprets the derogation strictly, and excludes from its scope workers who enjoy only partial autonomy over their working time. With a very laconic reasoning, it pointed out that ‘it is apparent from the express wording of that provision’ that it applies only to workers whose working time as a whole is not measured or predetermined or can be determined by the workers themselves. ⁷ As this conclusion was reached in an infringement procedure about certain national working time rules that were generally applicable to adult workers, nothing in the facts of the case that the application of this principle should be limited only to specific cases. Consequently, if the factual circumstances show that the employer has even a very limited control over the employee's working time, the derogation shall not apply.

In one case, for example, university experts had full-time contracts (40 hours per week) and were also involved in a post-doctoral research project. The mere fact that the measure of working time was set out in the contract led the Court to conclude that, even in the case of university lecturers, at least part of their working time was determined by their employer, so the derogation for autonomous workers did not apply to them. However, the CJEU noted that it was for the national court to ascertain whether that was the case. ⁸ It is a good illustration of the Court's rigorous approach that, as I have indicated above, even the Commission had previously taken the view that academic staff could fall within the scope of the exception. Similarly, teachers carrying out casual and seasonal activities in holiday and leisure centres, for an annual maximum of 80 working days, were deemed not to qualify as autonomous workers as the Court found no evidence that the workers were free to choose the measure and the schedule of their working time. ⁹

The stringency of this approach can be well illustrated by the Hälvä case. ¹⁰ The case involved a child protection association which provided accommodation for children as close as possible to a family environment in seven ‘children's villages’, each with several houses for children. The case was about ‘relief parents’ working for the association who substituted for the foster parents while the latter were absent (justified by days off, annual leave or sick leave). During the absence of the foster parents, they were responsible for the daily running of a children's home and the care and upbringing of the children. The relief parents were employed as autonomous workers and their working time was organised accordingly in a very flexible manner: they worked for continuous 24-hour periods which could last several days, with the right to one day off per week and, on average, two weekends off per month. This schedule infringed the workers’ right to daily and weekly rest periods as prescribed by the WTD. The referring Finnish court had doubts as to whether these employees enjoyed whole working time autonomy and whether the derogation for autonomous workers could be applied.

The facts showed that the special character of their job indeed guaranteed a certain level of autonomy to the relief parents. The employer's representatives did not control their day-to-day work and the employer did not issue orders in respect of working time and rest periods during working days. Within the limits imposed by the needs of the children, a relief parent could him or herself decide on the organisation and content of his or her work. However, there were also some constraints on that freedom. First, a care and education programme was prepared for each child, to be followed by the relief parent in caring for the child, and with respect to which they had to write a report. Furthermore, the relief parent consulted with the foster parent regarding the running of the children's home for which he or she was responsible, together with practical matters related to it. Importantly, their contracts of employment stipulated that they were to work 170 or 190 periods of 24 hours annually from which 30 to 33 days of annual leave had to be deducted. The employer stipulated, in advance, both the beginning and the end of the working time. The director prepared, in advance, lists indicating day by day the house in which the relief parent was to work. The latter agreed with the foster parent the time at which the relief period begun.

From these facts, the CJEU concluded that the foster parents did not enjoy autonomy over their working time as a whole. It was largely determined by the contract of employment and by the employer. Although it was for the national court, which had a better knowledge of the facts, to have the final word, the Court left little room for the conclusion that the relief parents were actually autonomous workers. ¹¹

To summarise the case law, the strict interpretation of the exception for autonomous workers means that it is virtually impossible to find a worker to whom it could apply. For example, for an autonomous worker, the employer cannot even assign tasks like attending a meeting, being present during visits, or meeting the client at a predetermined time, however these should not exclude the application of the derogation. ¹² Employee control over working time is in essence a relative term: an employee generally has more or less control over working time, relative to the extremes. ¹³ However, there are probably no employees in the EU – not even top executive managers or highly trained experts –who are at least occasionally not given some of the tasks listed above, and consequently who enjoy autonomy over their working time as a whole.

2.2. Reasons for a more permissive interpretation

In my view, it would be reasonable to interpret this exception less strictly or to amend the provision to make it more flexible. This may be justified on several grounds.

Although the Court has never elaborated any reasoning underpinning the assertion that only workers with full autonomy qualify as autonomous workers (apart from a broad reference to the express wording), Advocate General Kokott developed a reasoning for this conclusion, building not only on the Directive's wording, but also on the provision's context and objectives. ¹⁴ Below, when I argue for a more permissive interpretation, I also outline why I do not share the Advocate General's views.

First, the grammatical interpretation that forms the basis of the Court's jurisprudence is not clearly underpinned by the normative text. The Court took it as self-evident that autonomy must apply to the ‘whole’ of the working time, even though this word is not used in the text of the Article and the provision remains reasonably interpretable even if it is seen as a less stringent requirement.

In her reasoning, the Advocate General pointed out that the expression ‘activity concerned’, the specific characteristics of which justify the derogation, is used in the singular. She concluded from this that it is a question of a general consideration of the activity of the workers concerned, not of the particular nature of some of the duties of which that activity consists. ¹⁵ In my view, this claim justifies the opposite interpretation: the merely marginal control of time by the employer (e.g. a manager has to attend an hour-long weekly meeting every Monday morning) would not change the overall autonomous nature of the activity.

In the same vein, the Opinion also cited the definition of the term ‘working time’, which refers to the activity of the worker as a whole, rather than making a distinction between separate parts of that activity. ¹⁶ However, such a general approach to the definition of ‘working time’ is not reflected in case law, since the Court has classified periods of a very different nature as working time on the basis of this definition. For example, periods of vocational training required by the employer, ¹⁷ on-call duty ¹⁸ or stand-by time performed outside the workplace, if associated with constraints that objectively and very significantly limit the employee's right to control his or her own time, are all considered working time. ¹⁹ Consequently, the activities carried out by the worker during working time may differ significantly, ranging from the actual performance of certain tasks and participation in training to merely being on stand-by for orders. It seems similarly reasonable to hold that the employer's control over working time might differ depending on the specific activity in question. For instance, the employer may exercise strong control over compulsory training, but allow for more flexibility in other contexts.

The Court's strict interpretation can also be challenged by the teleological interpretation. The legal basis of the WTD is the right to health and safety at work, and all possible national exceptions, derogations and opt-outs should be subordinated to this principle. ²⁰ In her reasoning, the Advocate General found that the narrow interpretation of Article 17(1) is also necessary for the objective of worker protection. She pointed out that in cases of partial autonomy, the maximum permissible weekly and night working time could already be completely or largely used up by that part of working time which is defined by the employer. ²¹ It is evident that this would represent a clear circumvention of the WTD, even if the derogation were to explicitly apply also to partially autonomous workers.

Nevertheless, despite the marginal influence of the employer, the worker would still enjoy a considerable degree of autonomy over their own working time, and the guarantees listed in Article 17(1) could be set aside without endangering the aims of such provisions. For example, if only about 20% of the worker's weekly working time is affected by the employer's instructions, he or she can still have a decisive say over working time and the necessary rest. It also must be pointed out that reference to autonomous work does not mean that the employee is totally left without protection, as the exception sets aside only certain provisions of the WTD. The provisions relating to annual leave and the guarantees in respect of night working (except its maximum duration) still apply. Furthermore, even in the absence of the aforementioned minimal rest periods or the limit on the maximum level of weekly working time, all the requirements of the OSH Framework Directive must still be complied with, and due regard shall be paid to ‘the general principles of the protection of the safety and health of workers’.

A systematic interpretation may cast even more serious doubt as to whether the concept of full autonomy can be upheld. In short, it is questionable whether, with such a high degree of autonomy, autonomous workers still qualify as workers under EU law. This question will be addressed in the next section.

2.3. A systematic perspective: are employees more independent than self-employed workers?

The WTD does not contain any explicit rule on the definition of the term ‘worker’, nor does it refer to national law. Consequently, the Court could easily conclude that it is an autonomous concept of EU law, and has built up a broad interpretation of the term. To decide on the personal scope of the WTD, the Court follows its Lawrie-Blum formula, which consists three conjunctive elements: for a certain period of time a person performs services, for and under the direction of another person, in return for which he receives remuneration. ²² This test was developed in the context of free movement of workers, yet the subordination element gained importance only in the jurisprudence concerning labour law Directives. ²³ In the application of the test in respect of the WTD, among others, staff employed in holiday and leisure centres, ²⁴ a foster parent, ²⁵ a civil servant in a social security public institution, ²⁶ a volunteer firefighter, ²⁷ an honorary magistrate ²⁸ and a non-commissioned officer in the army ²⁹ were all held to be workers.

With working time autonomy, the subordination element may be called into question, which can be interpreted in two ways. First, if a person enjoys full autonomy over his or her working time, it excludes the hierarchical relationship vis-á-vis the employer, or, second, other working conditions may still establish that there is subordination between the employer and the autonomous worker. If the first interpretation were accurate, then Article 17(1) would turn out to be an empty shell, as any person fulfilling the criteria for its application would not be considered a worker and would not come under the scope of the WTD. This interpretation would therefore render one of the Articles of the Directive meaningless. However, on the basis of the case law, this possibility cannot be clearly excluded.

Although the Court has not yet specifically examined what factors concerning working time can establish subordination between the parties, case law offers some valuable guidance in this regard. For a hierarchical relationship between the parties, it is necessary to examine in each particular case all of the factors and circumstances characterising the relationship. ³⁰ These include, in particular, the circumstances of recruitment, the nature of the duties entrusted to the person, the context in which those duties are performed, the scope of the person's powers and the extent to which he or she is supervised within the company, and the circumstances under which he or she could be removed. ³¹ Clearly, many of these elements do not relate to working time. For example, a person who otherwise enjoys considerable autonomy to decide when and how much to work may nevertheless be obliged to strictly follow the employer's instructions on how and which tasks shall be carried out, and may be subject to disciplinary action taken by the supervisor, upon whose decision the employment may be unilaterally terminated.

The Court has developed a broad understanding of subordination in the context of the labour law Directives. ³² For instance, even company board members afforded ‘a margin of discretion in the performance of their duties’ have been found to be workers, based on a reporting and cooperation obligation with a supervisory board which could also remove them any time. ³³ In another case, the Court acknowledged that foster parents have broad discretion as to the daily performance of their duties but permanent supervision and assessment of their activity by the public service place them in a hierarchical relationship. ³⁴ Similarly, honorary judges (magistrates) are independent when adjudicating, but are subject to administrative obligations which are sufficient for the establishment of a legal relationship of subordination. ³⁵ These cases show that a person might enjoy independence in some aspects of the employment, yet remain subordinate in other dimensions.

It is therefore not entirely impossible for a worker to have full freedom in terms of working time, while other working conditions put him or her in a position of subordination vis-à-vis the employer. Nonetheless, the Hälvä case demonstrates that these other conditions shall not result in any interference with the worker's autonomy over working time. Yet it is difficult to imagine that instructions on how the work should be carried out or obligations to report on performance, coupled with adequate sanctions for non-compliance, have no impact on the use of working time at all. It seems to me that working time cannot be separated from other working conditions, and if there is an element of subordination in the employment relationship, it will necessarily also encroach on the temporal dimension of work. Similarly, minimal subordination in some respects does not mean autonomy in working time. For example, the Commission has expressed its doubts as to the compliance with the WTD of Member States that exempt workers who either earn three times the minimum wage (Netherlands) or occupy positions of considerable importance or trust and receive a salary seven times the mandatory minimum wage (Hungary). ³⁶ These solutions are probably not in line with the WTD, as the measure of the wage has no direct connection with the parties’ control over the working hours.

The Court has also examined the relationship between working time and subordination in the context of self-employed workers. Accordingly, an independent service provider enjoys ‘more leeway’ ³⁷ or ‘independence and flexibility’ ³⁸ than employees who perform the same activity, as regards the determination of the working hours. It should be noted that this means only a certain degree of autonomy, which may be limited by some measures taken by the employer (or by the customer or the client).

This can be illustrated by the Court's approach to the issue of the subordination of platform workers as regards working time. In cases concerning Uber drivers, the Court broadly acknowledged that the platform exercised ‘decisive influence’ over the conditions under which the drivers provided the service, without going into a detailed assessment of the working conditions. ³⁹ Advocate General Szpunar highlighted the following elements of control: for most drivers, Uber was their only or main professional activity; drivers completing a large number of trips received a financial reward; Uber informed drivers of where and when they could rely on there being a high volume of trips and/or preferential fares; and client ratings could lead to the exclusion from the platform. ⁴⁰ Importantly, these cases were not about the labour law status of Uber drivers, yet it is apparent that these workers did not enjoy full autonomy in working time.

In the Yodel case, the Court completed a more thorough, yet somewhat formalistic analysis, which did not look beyond the formal letter of the parties’ contractual arrangements to detect other ways in which the platform might exercise control. ⁴¹ The case concerned a self-employed courier who worked according to his own schedule but could only carry out the tasks between 07:30 and 21:00 from Monday to Saturday. According to the Court, that requirement did not indicate the existence of an employment relationship, as it ‘is inherent to the very nature of that service, since compliance with those times slots appears essential in order to ensure the proper performance of that service’. ⁴² It is not clear to what extent such a purely organisational issue can be read as natural or inevitable by the Court. ⁴³ Nonetheless, while this ‘natural’ constraint not deemed incompatible with the higher level of freedom typical of the self-employed, it seems also unambiguous from the above that, with a similar restriction, it would not be possible to speak of an autonomous worker in the case of an employment relationship.

In summary, autonomous workers occupy a special place in the binary system of employees and the self-employed. While they enjoy full autonomy in terms of the measure and schedule of their working hours, they fall under the first category and consequently are subordinate to their employer as regards the other aspects of their employment. Compared to self-employed workers, they enjoy less overall independence and flexibility in their work, but are fully autonomous in terms of working hours, which shall not be limited by any employer control otherwise compatible with self-employment. This systematic analysis confirms the doubt as to whether

anyone falls into the autonomous worker category.

2.4. A suggested solution

The CJEU's narrow interpretation does not preclude the use of arrangements involving ‘partial’ working time autonomy. It merely means that in such cases, the basic protections of the WTD shall still be respected. For example, the working schedule of the above-mentioned relief parents is not against EU law, but needs to be designed in accordance with all rules on rest periods and maximum working time. Alternatively, parties may also conclude collective agreements for ‘semi-flexible’ schedules to deviate from some provisions of the WTD, but equivalent compensating rest periods shall be granted or, in exceptional cases, other appropriate protection. ⁴⁴ This allows the parties to adapt the regulatory framework to their individual needs, but Court jurisprudence is rather strict as to the appropriate compensation. ⁴⁵

As is apparent from the above analyses, I do not find the Court's current interpretation convincing. In fact, the Court has never really outlined why full autonomy is necessary to apply Article 17(1). Consequently, a more permissive interpretation may be possible without having to amend the WTD. Nonetheless, for the sake of legal clarity, I suggest a new wording that would enable the derogation for autonomous workers to occupy the position it should in the system of EU working time rules. Accordingly, the derogation would be possible if, on account of the specific characteristics of the activity concerned, the duration of the working time is ‘predominantly’ not measured and/or predetermined or can be determined ‘predominantly’ by the workers themselves. This solution would significantly extend the personal scope of the derogation, albeit it may also increase the risk that some workers facing some level of employer control are left without adequate protection. To address that, the derogation could be supplemented with some back-stop provisions, which would set more flexible but binding limits on the maximum weekly working time or minimum rest periods. ⁴⁶

3.1. Temporal freedom vs financial dependency

Even if it is primarily the worker who decides when and how much to work, the employer might strongly influence his or her decisions through a well-designed remuneration system.

As for the schedule of working time, while all hours are the same for machines, this is not the case for people. ⁵⁰ If the employer needs to cope with peak periods ⁵¹ and/or a certain level of cover is also required during unsocial hours (like night time, weekends, holidays), workers can be incentivised to be available during these periods with higher rates of pay.

Wages may also be useful in controlling the number of hours worked. If the compensation plan offers higher rates or better paid assignments after a certain number of hours or tasks (rides, deliveries, gigs etc.) have been completed within the given period of time, the worker might not want to stop working until the more lucrative level is reached. For this purpose, a system where compensation increases exponentially with time worked can be particularly effective. Empirical evidence shows that workers on zero-hours or on-call contracts have no specific working hours set in their contracts with the – at least theoretical – right to refuse the employer's call to work. In practice, however, these contracts result in a broad fluctuation in working hours, unreliable rest periods and little or no input for workers into their schedules. ⁵² In Advocate General Szpunar's words, ‘without exerting any formal constraints over drivers, Uber is able to tailor its supply to fluctuations in demand’. ⁵³

Flexibility in working time might be a priority for the workers themselves, but this can be easily undone if the job offers decent income only after a higher level of commitment. For example, empirical studies have revealed that for platform workers, ‘flexibility is just a kind of solace: to earn a significant sum of money, workers might also have to work more hours every day than a “standard” worker. Since they have to be available “around the clock”, this kind of flexibility does not entail a greater freedom for the worker.’ ⁵⁴ Similarly, underemployment may lead those who choose formally flexible or part-time working to be available as much as possible and to give up their working time preferences.

Flexible working time arrangements often mean that employers pay wages based on performance rather than hours worked. At first glance, this is also beneficial to the employee, as the more tasks completed, the higher the salary. However, this leaves administrative tasks or time spent waiting or in search of the next task without additional compensation. Thus, what may at first sight appear to be an attractive per-unit fee for a ride, customer etc., can obscure significant time-consuming, unremunerated tasks and inactive periods, while the amount of such time remains the worker's risk. ⁵⁵

3.2. Competition among workers

Working time autonomy can be significantly constrained if many, or in some cases numerous, workers occupy the same position at the company. If this is the case, the employer can increase their availability by creating strong competition between them. For example, it is sufficient to allocate tasks to workers on a first come, first served basis, with short deadlines for acceptance. ⁵⁶ Thus, those who want to get the highest-paying or the easiest ‘jackpot’ tasks should be on call for as long as possible, even constantly. ⁵⁷

This can be further exacerbated by overstaffing, when the employer deliberately hires more workers than can be assigned with sufficient paid work. ⁵⁸ The highly competitive work environment is especially apparent in the platform economy, where workers around the globe can compete for online tasks, which also has an adverse effect on wages. ⁵⁹ From the platforms’ perspective, ‘the worst case…is an order that is not delivered. A driver who does not receive an order does not matter’. ⁶⁰ For workers, however, temporal flexibility has less value if sufficient work is not guaranteed for everyone. ⁶¹

Workers’ performance can be even more influenced if the employer assigns the most favourable tasks or shifts on the basis of past performance. In this context, the employer may rely on systematic reports and evaluations of the worker performance, taking into account how many hours and when the worker has worked in the past, or whether the worker has previously refused to perform any assigned tasks or shifts. ⁶² In this competition, only those who are ready to be constantly available or to work long hours can get the most lucrative jobs. ⁶³ Nonetheless, even the best performers lack any real bargaining power to set the pace of work, as it is defined by the employer (or its clients or the platform). ⁶⁴

What is in fact fierce competition for jobs can, with the right measures, appear as a game for workers. For instance, statistics on the hours performed, clients served and wages earned are available or sent directly to the worker on a daily, weekly or monthly basis in a comparative way, so that the worker can check how he or she ranks among the employer's current workforce. Above-average performers can be awarded well-sounding titles (e.g. Rider of the Week) and/or some extra money, while less productive colleagues can receive ‘you can do it better’ messages. The logistics sector in the United States, for example, uses apps that rank workers in video game-like leaderboards, based on their performance tracked in real time. ⁶⁵ Such gamification can replace older forms of surveillance with seemingly ‘playful’ forms. ⁶⁶ Nonetheless, workers can be put under considerable pressure if they have to constantly compete with both their colleagues and their own past performance. ⁶⁷

3.3. Communication strategies, reporting obligations

A cleverly crafted communication strategy is a softer form of control, which is apparent especially in online jobs. ‘Autonomous’ workers receiving regular text messages (emails, push messages, other messaging applications) reminding them on awaiting clients, tasks or expiring deadlines can effectively be influenced in their decision-making on their availability. For instance, if the gig worker wants to quit the application it immediately sends a pop-up message on the number of clients waiting nearby or on the amount of money that could be earned in the next couple of hours. ⁶⁸

The worker might be subject to strict reporting obligations, which can take different forms. First, even if the worker is free to schedule his or her own working time, he or she must report regularly on the completed hours, tasks, or on the clients served. Such data will then be the input for the evaluations mentioned above. Second, the worker might be asked to report beforehand on his or her availability for the upcoming period. Once the worker has reported that he or she is free to work on a particular afternoon, he or she is likely to be reluctant to change the plan for the sake of some private life matter.

While algorithmic management builds on the collection and processing of a vast pool of data, coupled with the above-mentioned communication strategies, this does not mean that working time management is transparent to the worker at all. On the contrary, application-based shift systems often come with an informational gap on the validity and accuracy of the employer's decisions or announcements, which means that workers ‘shall play a game without having access to the rulebook’. ⁶⁹ If the employer ‘hides behind the algorithm’, workers are left without the ability to discuss their concerns with a human manager. ⁷⁰ Another example of information asymmetry is food delivery or personal transport applications that do not reveal the final destination until the worker has accepted the task. ⁷¹

3.4. Sanctions

A negative aspect of control is the use of sanctions or reprimands for workers who are absent or late, or who refuse or cancel tasks assigned to them. ⁷² For instance, the employer does not directly schedule the working time of its workers but those who are not available for a minimum duration of time or who reject a certain number of tasks will face sanctions. Such sanctions could include the ‘deactivation’ or suspension of their contract (meaning that they cannot apply for work for a certain period), ⁷³ levelling-down (only less advantageous tasks are available), or even the termination of the contract. For instance, UK workers in marginal part-time employment (on-call work or zero-hours contracts) often face the ‘fear of being zeroed down’, meaning that the employer might not offer the worker any further job opportunities once the he or she rejects a call to work. ⁷⁴ These techniques are also widely used in the gig economy. While there might be some scope for refusing work or deciding when to work, it is effectively curbed by a fear of possible sanctions, such as not receiving future work offers. ⁷⁵

3.5. The combination of control measures

The examples mentioned above – especially if more forms are used in combination – can allow the employer to have significant influence over working time. In the age of digitalisation, this can be easily done, for example, if an application offers the next tasks to the formally autonomous worker, evaluating the quantity, time, quality and any refusals of previous tasks. Technological developments can therefore create paradoxical situations where, although the worker is not subject to any direct supervision, there is an increasing real risk of intensified control and monitoring due to the technological connection between the parties. ⁷⁶

This is often referred as the ‘autonomy paradox’: the more freedom and autonomy that flexible work designs and technology offer, the more intensely employees work, and the more they feel controlled and constrained by their work. ⁷⁷ With algorithms and artificial intelligence, it is now possible for employers to track the working time data of up to thousands of workers in real time, to continuously analyse and evaluate (process) such datasets, and to use them to design the next schedules that are the most favourable for the company. ⁷⁸ Technical advances have made it possible to re-establish the monitoring and supervision that is characteristic of a bilateral employment relationship but without any of its associated rights and benefits. ⁷⁹ Moreover, the precarity caused by unstable schedules does not affect working time alone, as it permeates through the interconnected domains of workers’ lives, and can lead to work-life conflicts, reduced earnings and short job tenure. ⁸⁰

The solutions presented above are therefore factors to be evaluated when the ‘autonomous’ nature of the work is called into question, and also in misclassification cases. If a worker works under a remuneration system which indirectly obliges him or her to be available for a time equivalent to a full-time job and the schedule is strongly controlled by the employer, while he or she must regularly report on his or her performance, which leads to certain reprimands or awards, then these facts might suggest that there is an employment relationship between the parties and the derogation for autonomous workers cannot be relied upon.