Abstract
This contribution explores a lesser discussed provision in the new Platform Work Directive (‘PWD’): Article 3 on ‘intermediaries’ (originally titled ‘subcontracting liability’ by the EP). Article 3 is possibly the most ambiguous provision in the Directive, for three reasons: the unusual drafting history of Article 3, the wording of Article 3 itself and, the amount of flexibility left to Member States to interpret Article 3 as they choose. Ultimately this means that Article 3 will likely have very divergent impact in practice. Article 3 provides that individuals providing work for an intermediary should enjoy the same level of protection ‘pursuant to the Directive’ as those contracting directly with digital labour platforms. There is also reference to joint and several liability, ‘where appropriate’. At the very least under Article 3, the employment presumption in Article 5 should also apply to intermediaries and the provisions on algorithmic management should apply directly to those in a subcontracting chain. The contribution details the unusual drafting history of Article 3, then contrasts Article 3's fit with other EU provisions on ‘subcontracting’, before discussing whether it was intended as, or might be interpreted as, alternatively a provision on joint liability or on joint employment. The final section provides a brief reflection on the concept of the employer, in light of the PWD.
Introduction
While correctly most attention regarding the new Platform Work Directive (PWD) is focused on the employment presumption in Article 5 or on the provisions about algorithmic management, this piece focuses on a less-discussed provision: Article 3 on intermediaries. 1 Rather curiously, Article 3 has been described in initial commentary on the new Directive both as important and, simultaneously, as a limited provision. 2
Article 3 was a late addition to the Directive. It was not included in the Commission's original proposal from December 2021 and was added initially by the European Parliament (EP), although the final version of Article 3 is different from the EP's proposal (and different again from the Council's starting position). In sum, Article 3 governs the situation when a digital labour platform (DLP) subcontracts for labour, so that individuals providing platform work or platform workers will be in a contractual relationship with a subcontractor, or ‘intermediary’, rather than with the platform. 3 The provision states that individuals providing work for an intermediary should enjoy the same level of protection ‘pursuant to the Directive’ as individuals contracting directly with DLPs.
The first observation in this piece is to note the ambiguity of Article 3. Article 3 is ambiguous, as will be discussed below, for three reasons. First, the wording of Article 3 itself is ambiguous; second, relatedly, Article 3 provides massive amounts of flexibility or freedom to Member States when implementing the provision; and third, uniquely, the drafting history of Article 3 contributes to its ambiguity. The purpose of this contribution is to discuss the intention behind Article 3 and to contemplate its possible effects in the different Member States. 4 The structure will be as follows.
The opening section, next, will discuss the unusual drafting history of Article 3. Then, the second section will discuss Article 3's ‘fit’ as a provision on subcontracting liability with other such EU provisions. The third section will discuss whether Article 3 was intended as, or might be interpreted as, alternatively a provision on joint liability or on joint employment. The fourth section uses the PWD to reflect briefly on the concept of the employer. The conclusion is that while Article 3 provides leeway for Member States to implement the provision widely (subject to provisions on freedom of establishment), it would have been a stronger provision if it was more limited yet mandatory. 5
Drafting history of Article 3
There was no version of Article 3 in the Commission's original proposal for a Platform Work Directive from December 2021. 6
The first version of Article 3 originated in a draft report, dated May 2022, by the European Parliament's rapporteur for the PWD, Italian MEP (and then S&D vice-president) Elisabetta Gualmini.
7
While Gualmini's report is more well-known for changes suggested to other parts of the Directive proposal (notably such as to strengthen the employment presumption and suggesting the provisions on algorithmic management should apply to all workers subject to algorithmic management whether or not platform work),
8
additionally, Gualmini proposed the addition of a new Article 12b on ‘subcontracting liability’, initially worded as follows: ‘(1) In order to tackle fraud and abuse with regard to the provisions of this Directive, Member States may, after consulting the relevant social partners in accordance with national law or practice, take additional measures on a non-discriminatory and proportionate basis in order to ensure that, in the event of a subcontracting chain, platform workers have an effective remedy whereby the digital labour platform of which the employer (service provider) is a direct subcontractor can be held liable, in addition to or in the place of the employer, with respect to any outstanding net remuneration corresponding to the minimum rates of pay and contributions due to the common funds or institutions of social partners’
9
According to reports of proceedings in the EP, Gualmini's proposals were strenuously debated in the EP's Committee, even before getting to the ‘tortuous’ interinstitutional negotiations between the Commission, EP and Council (so-called ‘trilogues’). 11
The final version of the report, which formed the EP's starting position for trilogues, was subtly different with regards to the proposed Article 12b. 12 For instance, the original inclusion of a possible due diligence defence was removed. In addition, the reference to ‘tackling fraud and abuse’ was moved to an amended subsection 4, and the substantive part of subsection 1 was widened to read: ‘including with’ respect to any outstanding remuneration and contributions due to the common funds or institutions of the social partners.
While Gualmini provided an explanatory statement for her report overall, this did not discuss the proposed new Article 12b. 13 From the recitals, it is almost certain that the proposed Article 12b was prompted by developments following the introduction of the so-called ‘Rider Law’ (RDL) in Spain (Stop Fake Autonomous Law), which had come into effect shortly beforehand, in August 2021. 14 The recitals referred to the ‘experience’ in Spain, noting that after the introduction of a new legal presumption of employment for delivery-based platform workers, there had been a significant increase in subcontracting in Spain in platform delivery work. 15 To avoid that same fate, the two most obvious options would be, first, more radically, to attempt to regulate when subcontracting by DLPs could occur or, second, to apply the employment presumption also directly to DLPs in relation to those working for or under subcontractors. While the second option is hypothetically a possible reading of eventual Article 3 PWD, it is not interpreted in that way in initial commentary on the PWD. 16 By contrast, with regard to the first option, Hießl makes the observation that the proposed Article 12b could indeed have been intended as a deterrent to subcontracting, with reference to its trio of mandatory inclusion of joint and several liability, covering substantive elements of the subcontractor's employment relationships, and no due diligence defence. 17 As will be seen below, the eventual Article 3 is far more ambiguous. It does not appear that deterrence of subcontracting was now an aim, if it ever was. Joint and several liability seems, on the one hand, possibly required (‘shall’), however on the other hand simultaneously optional (‘where appropriate’); it is not necessarily clear what joint liability is intended to cover.
Indeed, the Spanish RDL is widely regarded as one of the main inspirations for the PWD generally, although there are also significant differences between the PWD and RDL.
The two key features of the RDL are as follows: first, a presumption of employment and second, provisions on algorithmic transparency. 18 In comparison to the PWD, the RDL's employment presumption, in scope, is far more limited, applying only to individuals providing delivery services for digital platforms (although the proposal originally was for the RDL also to be wider): in other words, if referring to the most well-known DLPs, Uber Eats is covered (delivery service), however not Uber (ride hailing). 19 It is reported that during tripartite collective bargaining negotiations in Spain leading to the agreement of the RDL, the so-called ‘big’ DLPs asked for two concessions when agreement was close to being reached. 20 The first concession requested was for a longer timeframe before the RDL would become effective and the second concession requested was for the text of the RDL to allow subcontracting. Only the latter was granted.
However, a discussion of Spain would not be complete without further reference to employment-related regulation of subcontracting more generally. Rodríguez Cardo and Alvarez Alonso explain that subcontracting is otherwise ‘extensively’ regulated in Spain, with reference to Articles 42 and 43 of the Spanish Labour Code, which (unlike the PWD) are not sector specific. 21 Article 42 applies when ‘subcontracting affects the client undertaking's own activity’, which means that the definition of the latter in case law becomes evidently important. 22 Rodríguez Cardo and Alvarez Alonso explain that the effect of Article 42.2 is that the contractor is obliged ‘to check if the contractor or subcontractor is complying with its duties to pay social security contributions’, plus joint and several liability for debts in respect of wages owed by the subcontractors to their employees and for debts related to social security contributions during the term of the subcontracting. 23 These two features of Article 42 might be one possible explanation for the inclusion of wages and unpaid social security contributions in the wording of the proposed Article 12b PWD, if the Spanish approach more broadly was a template for the amendment. On the other hand, the proposed Article 12b PWD only referred to an ‘effective remedy’, whereas in Spain, there is additionally Article 43 on the illegal assignment of workers, which can also impact the identity of the employer. This provision means that if a subcontractor is deemed to be a ‘shell’ intermediary and not a true employer, its behaviour will be deemed to be a ‘very serious infringement’, with consequently stronger financial penalties attached, and the worker in that instance can choose which of the two entities is to be regarded as their permanent employer. 24
While initially it was predicted that it would not be difficult to show an illegal assignment under Article 43 if DLPs subcontracted with a view to avoiding the RDL, 25 Hießl points out that in practice, a number of subsequent cases have failed, for example against Cabify and Auro, and highlights a judgment from the Valencia Labour Court in 2022 where it was suggested that a claim under Article 43 would only rarely succeed where the subcontractor's role is ‘entirely negligible’. 26
The EP's Committee on Employment and Social Affairs (EMPL) voted to adopt Gualmini's amended report in December 2022, with a subsequent vote by the full EP in February 2023.
27
During this time, the Council agreed its own mandate for trilogue negotiations; this also included a version of what is now Article 3.
28
The Council proposed a new Article 2a, titled ‘intermediaries’, which tersely stated: ‘Member States shall ensure that the use of intermediaries does not lead to a reduction in the protection afforded by this Directive’
The next stages in the story of the adoption of the PWD are well-known: trilogue negotiations led to a first provisional agreement between the different EU institutions being reached in December 2023, 30 which almost immediately failed when reportedly as many as 12 Member States expressed discontent with the agreement in the Council's Committee of Permanent Representatives (Coreper). 31 Reportedly, from the point of view of Member States in the Council, the first provisional agreement had strayed too far from the Council's agreed mandate. 32 However, it should be emphasised that Article 3 was not the source of disagreement within the Council: rather disagreement was linked to the mechanics of the employment presumption. 33 Council documentation confirms a continued commitment in the Council to a provision on ‘intermediaries’. 34
The first failed provisional agreement was followed after further trilogue negotiations by a second, different, provisional agreement, reached in February 2024.
35
While the second provisional agreement also almost failed in the Council, it subsequently formed the basis for the final Directive.
36
In its final version, titled ‘Intermediaries’, Article 3 reads as follows: ‘Member States shall take appropriate measures to ensure that, where a digital labour platform makes use of intermediaries, persons performing platform work who have a contractual relationship with an intermediary enjoy the same level of protection pursuant to this Directive as those who have a direct contractual relationship with a digital labour platform. To that end, Member States shall take measures, in accordance with national law and practice, to establish appropriate mechanisms, which shall include, where appropriate, joint and several liability systems’
A provision on subcontracting liability?
The EP's proposed original version of Article 3 was titled ‘Subcontracting Liability’, in comparison to the final version's title of ‘Intermediaries’, if the change in title alone is significant. 38
As well as the title in the EP's proposed Article 12b, the substantive wording of the proposed Article 12b also seemed heavily to draw on two provisions specifically on ‘subcontracting liability’ in earlier Directives: Article 8 in the Employers’ Sanctions Directive (ESD) 2009 and Article 12 in the Posted Workers Enforcement Directive (PWED) 2014. 39 This is the alternate and stronger explanation as to why the proposed Article 12b included reference to wages and social security contributions, when otherwise neither were included elsewhere in the PWD. 40
In terms of whether Article 3 PWD and its predecessor versions constitute a good ‘fit’ with other Union provisions on subcontracting liability, Article 8 ESD and Article 12 PWED also include a permissive or mandatory due diligence defence (in this instance, for the DLP). 41 As mentioned above, in the EP's final version of the proposed Article 12b, in comparison, there was no due diligence defence. 42 This was a potentially significant – or in the context of previous EU practice, unusual – omission. While the final version of Article 3 does not track the wording of Article 8 ESD and Article 12 PWED in the same way that the EP's proposed Article 12b did; similarly to the proposed Article 12b, the final version of Article 3 also does not include a due diligence defence for DLPs. On the other hand, as Article 3 is conditioned by multiple references to national law and practice, if by reference solely to the wording, there would not seem to be a hurdle if a Member State were to choose to introduce a due diligence defence for DLPs as, in the wording of Article 3, ‘appropriate to its legal system’. Whether this flexibility in Article 3 would offset a potential challenge, if the principle of effectiveness of EU law were to be raised, is a different matter. 43
In addition, while Article 8 ESD and Article 12 PWED are explicitly provisions on subcontracting liability, they are both factually highly specific: applying only to the situation of employing illegally staying third-country nationals (ESD) and to posted workers in the construction sector (Article 12 PWED). By contrast, more recently, there is another new Directive which might be described as dealing more generally with the area of business and human rights, including subcontracting. This is the new Directive on Corporate Sustainability Due Diligence 2024 (CSDDD or CS3D). 44 Also taking into account the Forced Labour Regulation 2024, these three new instruments might be described as a trio of new aligned EU measures, each contributing in their own way to the broader field of business and human rights - although the CSDDD would be the dominant instrument for these purposes. 45 Nevertheless, it must be noted that in February 2025 the Commission introduced an ‘Omnibus simplification package’, which included a proposal for a Directive to amend significantly, inter alia, the CSDDD. 46 As part of this package, another Directive was speedily passed in April 2025, to postpone the implementation of the CSDDD and other corporate sustainability reporting. 47 With regards to proposed substantive changes to the CSDDD, both the Commission's proposal and the Council's negotiating mandate include a provision to repeal the planned EU-wide harmonised civil liability regime in the CSDDD. 48
However, there is a relevant difference between Article 3 PWD and the original CSDDD in the provision on civil liability, even if there is only a small chance that the latter will now survive in its original form. 49 While an ‘in-scope’ company under the 2024 version of the CSDDD can potentially be liable for the ‘adverse impacts’ of their subsidiaries as well as ‘business partners in the chain of activities of the company’ at Article 29, 50 this is only where a company has intentionally or negligently failed to put in place a prevention or corrective action plan, and that failure is what has caused damage to a natural or legal person. 51 In other words, there is a very clear fault requirement for a parent company or company subcontracting before civil liability can attach. 52 It will also presumably be difficult to show that the failure to enact or implement a prevention or corrective action plan is what has caused the relevant damage. 53
In comparison, if looking at the wording of Article 3 PWD alone, a DLP under Article 3 could potentially be held jointly and severally liable for a subcontractor's wrongdoing without any requirement for either fault or causation on the part of the DLP. In this way, if only on the face of it, Article 3 PWD seems far more generous to claimants than the counterpart Article 29 CSDDD, as initially enacted. The question then becomes why platform work would receive potentially more claimant-friendly regulation. With regards to Article 12 PWED, the Commission's explanatory memorandum at the time explained that construction workers were singled out for special treatment because this is an area where the ‘protection of workers’ rights is a matter of particular concern’. 54 However, it is not clear that persons providing work for platforms or platform workers are necessarily in a worse position compared to the various other individuals affected by the CSDDD alleging adverse human rights or environmental impacts.
On the other hand, it is highly unlikely that Article 3 PWD would be implemented in this more generous way. If anything, the reality is that Article 3 PWD is weaker in comparison to other EU provisions on subcontracting: a Member State, under Article 3, might choose not to institute any joint liability or could limit the scope of joint liability in accordance with its legal system. On the former, to explain, it is not even clear that joint and several liability is mandated by Article 3: the wording of ‘shall’ is potentially undercut by ‘where appropriate’. As for the latter, there are various modifications to joint and several liability systems possible, for instance, in common law countries short of proportionate liability, such as the so-called ‘major/minor’ hybrid or the ‘contributory negligence’ hybrid. 55 It is for these reasons that civil society organisations have critiqued Article 3 as a weak provision on subcontracting liability. There tend to be two critiques in this respect: first, that joint and several liability should have been mandatory or more mandatory, to avoid any ambiguity in this regard; and second, possibly that there should have been – more in line with the CSDDD more generally – additionally a requirement for DLPs to undertake due diligence monitoring of their subcontractors. 56 (Due diligence checks, as referred to here, are separate from a due diligence defence as discussed earlier: this would be more similar to a requirement to audit the company's value chain). Interestingly, when Gualmini's draft report was being debated in the EP, one proposed amendment was to add a new subclause to proposed Article 12b to require exactly such due diligence checks by platforms, however this was not included in the EP's eventual starting position. 57
A final comparison might be drawn between Article 3 PWD as a provision on subcontracting liability and, beyond the EU, the lengthy process at the United Nations (UN) towards a possible new treaty on business and human rights. 58 As currently drafted, the new UN treaty would not include a due diligence defence for companies. 59 Rather, in the more modern way, it would require signatory State Parties to establish a ‘comprehensive and adequate system of legal liability’ for corporate human rights abuses, 60 alongside requiring States to make human rights due diligence legislation mandatory in their national legislation. 61 With potential relevance for subcontracting or third party liability generally, the new treaty would, however, go beyond existing leading case law in two ways: first, by mandating aiding and abetting claims to be allowed and, 62 second, possibly by requiring States to institute a new corporate duty to prevent human rights abuses by third parties ‘where the enterprise controls, manages or supervises the third party.’ 63 On the other hand, treaty negotiations have been ongoing for over a decade, and it is unclear if a final version will ever be agreed. 64
Joint liability or joint employment?
The prevalent interpretation of Article 3 PWD is that it represents, in circumstances where appropriate, joint liability between a DLP and intermediary, however not joint employment. 65 On the one hand, this does not matter much in practice. For example, when discussing joint and several employer liability in Spain, Rodríguez Cardo and Alvarez Alonso state that joint liability in Spain achieves a similar effect to joint employment in the USA. 66 On the other hand, that conclusion depends on two key interpretations: first, in the context of Article 3, what joint liability covers and; second, additionally, the meaning of joint employment. The next part of this section discusses first the possible meanings of joint liability under Article 3: another example of ambiguity in the provision.
Whereas the previous section contrasted Article 3 with other provisions on subcontracting liability, this section unpacks the meaning of Article 3. Article 3 requires the same level of protection for those subcontracted as the directly contracted, however: ‘pursuant to this Directive’. The latter is potentially a significant qualification. On the dangers of subcontracting, the recitals once again refer to two risks: first, employment misclassification and, second, automated decision-making and automated monitoring. 67 At the very least then, Article 3 would appear to require two minimum elements from Member States: first, the employment presumption under Article 5 must also apply to intermediaries, and second, in line with the wording in Article 2(1)(b), the provisions on algorithmic management must apply directly to DLPs in relation to individuals working for or under intermediaries even though there is not a contractual relationship with the DLP. Another argument, in theory, would be that the ‘same level of protection’ for those subcontracted should further mean that the employment presumption ought to apply also to DLPs in relation additionally to those subcontracted (where there is control and direction by the DLP), but this is a minority view. 68
On the narrowest view of joint liability under Article 3, joint liability might only cover those two minimum elements. 69 For instance, the DLP might be deemed jointly liable if an intermediary misclassifies the employment status of an individual working for it, where this is an offence in itself. To take the UK as an example: unless this is linked to the breach of another statutory employment right (such as a failure to pay the national minimum wage if the individual is a ‘limb b’ worker), there is no penalty per se for misclassifying a worker or employee. 70 The closest analogy would be where there is a failure to give a written statement of employment particulars, and here the penalty is only two to four weeks of pay (and again linked to the breach of another statutory employment right). 71 If the position is at all similar to the UK, and if that is the extent of joint liability under Article 3, this really may not be very extensive – a potentially massive difference from joint employment. In comparison, the assumption seems to be in initial analysis of the PWD that, possibly, joint liability under Article 3 might be interpreted more broadly by Member States, to cover the subcontractor's employment relationships more generally. 72 This would be perhaps surprising as Article 3 no longer states this (in comparison to the EP's proposal) and the joint liability could potentially be very wide, going beyond, for example, Article 12 PWED, which is limited to joint liability for wages. In addition, an intermediary under Article 3 extends potentially beyond the first layer of subcontracting, in comparison again to Article 12 PWED. 73
If a Member State were to choose to implement joint liability more widely, it is highly foreseeable that they might try to introduce one or other types of limitation to accompany that wider interpretation. There is here the same tension as noted above between the multiple references to ‘appropriateness’ and national law and practice in Article 3 (plus, additionally, co-ordination with other Directives as well as individual national laws), versus the principle of effectiveness of EU law. 74 Whether or not the following would survive a challenge from the principle of effectiveness, a Member State might attempt to limit joint liability to wages and social security contributions (akin to Article 12 PWED and, for example, as discussed earlier, Article 42 of the Spanish Labour Code), or they might choose to introduce a fault requirement (similar to the CSDDD) or attempt to introduce a due diligence defence (following Article 12 PWED or Article 8 ESD). 75 Even the aforementioned Article 42 of the Spanish Labour Code does not institute joint liability for employment rights unless the subcontractor is undertaking the same activity as the lead company. There would also possibly be freedom of establishment considerations for Member States if interpreting joint liability more broadly. 76
However, even with a wider view of joint liability, Hießl helpfully notes that the identity of the employer does still matter in other respects. 77 For example, the identity of the employer will determine which collective agreement is applicable and against which entity or entities a potential unfair dismissal claim could be brought, amongst others. In comparison to joint liability, joint employment would affect also the identity of the employer, but how significant this is will depend on the second key interpretation noted above: the meaning of joint employment.
Rodríguez Cardo and Alvarez Alonso have referred to joint employment in the US. 78 Joint employment in the US is normally regarded as the most developed - and also most significant - example in practice of joint employment. However, there is potentially an important difference between joint liability in the US as an example, and if joint employment were to be recommended or encouraged under the PWD. As obvious as it sounds, a finding of joint employment under the Fair Labour Standards Act 1938 (FLSA) only applies to that particular statute, with no consequences for employer status beyond the FLSA. Similarly, where there is a finding of joint employment under the National Labor Relations Act (NLRA), being declared a joint employer will only apply for the purposes of the NLRA (and, in case law, a joint employer will only be required to collectively bargain over terms and conditions of employment which it has the power to determine). 79 There are various reasons why the PWD does not encourage or recommend joint employment at Article 3, but in addition to the more obvious ones (to be discussed below), it is suggested here that a further possible reason is the difference between what is termed here a ‘domain specific’ application of joint employment and a ‘non-domain specific’ application of joint employment.
With regards to identifying employees, the employment presumption in Article 5 PWD is not limited to the enjoyment of particular rights under the Directive but is intended to affect the status of the platform worker for all purposes. Where the presumption is satisfied, the platform worker will be entitled to claim statutory employment rights available to those with an employment relationship in that jurisdiction. First, symmetry with Article 5 but, second, if joint liability under Article 3 is interpreted more widely as discussed above, might suggest that if the Directive were to encourage joint employment, this could in turn mean recommending a finding of joint employment as a new starting point to apply across domains in the context of platform work – a move substantially beyond joint employment under the FLSA or the NLRA in the US. There is a separate debate to be had here as to whether the definitions of both employee and employer should be universal to a jurisdiction or change according to the ‘function’ or regulatory domain. 80 If Article 3 were to suggest joint employment, another broader question is why this should be limited to platform work.
There are also other more basic reasons why the PWD does not encourage or recommend joint employment. If the definition of a worker is politically sensitive, the definition of employer is likely to be sensitive too. Van Schadewijk notes that the Commission's original draft of the Directive on Transparent and Predictable Working Conditions (TPWCD) 2019 included a definition of the employer, 81 however this was removed on the advice of the Committee of Regions to the effect that defining the employer is a matter for Member States. 82 The PWD only just passed as is without adding any further controversial elements. A further obvious factor is the continued adherence to a single employer model in some or many continental legal systems, even if this is fraying and where there are increasing exceptions. 83 Indeed, it might be suggested there is no need to develop a joint employment model in continental legal systems where there is a prohibition between formal and ‘actual’ employer, if the subcontractor were not to play any meaningful role and assuming there is control and direction by the DLP. 84 On the other hand, as powerful as that prohibition is, the underlying assumption is that only one party will be exercising control and direction, when factually this may not always be the case. To refer back to Hießl, when reviewing recent claims against platforms, she adds that subcontractors do sometimes ‘add rules’: notably such as an obligation to work, or a maximum aggregate weekly hour limit, or restrictions as to the use of vehicles or IT. 85 Multiple parties might be exercising control and direction.
All of the above explains why there is only tacit recognition at best of joint employment in the PWD, and nothing more. Article 4(3) refers to the ‘party or parties’ responsible for the obligations of the employer, but that is it. Even the EP's wider proposed Article 12(b) seemed limited to joint liability over joint employment, with its reference to ‘effective remedies’ and the reference in the text to the original ‘employer’.
While the Directive does not recommend or encourage joint employment, there is still good reason to suggest further exploration of the doctrine. Weil has written extensively about the difficulties which the ‘fissured’ workplace poses for conventional legal analysis: ‘fissuring’ in this respect might refer to ‘vertical disintegration’ within corporate groups or outside corporate groups by way of subcontracting or outsourcing. 86 When Weil previously was the Administrator of the US Department of Labour's (DOL) Wage and Hour Division (WHD), he produced an Administrator's Interpretation (AI) on Joint Employment in 2016, since rescinded. 87 In that AI, Weil noted the increasing practical importance of the doctrine, given that multiple entities may be involved in or determine an individual's working conditions. 88 The DOL was stated to be involved in ‘hundreds’ of joint employer enquiries under the FLSA alone each year. 89 Indeed, claims have also been recently made in case law in the UK, albeit unsuccessfully, to advocate for joint employment for the purposes of particular statutory employment rights, 90 beyond existing limited recognition of dual employers under the law on vicarious liability. 91 Additionally, the Institute of Employment Rights in the UK has advocated for a new, more broadly applicable, joint employer standard. 92
This section concludes by briefly reflecting on the standard for joint employment in the US under the FLSA and NLRA, as the most developed examples of this doctrine, if joint employment were to be explored further. On the other hand, the US is possibly not the best advert in this respect: under both the FLSA and NLRA, joint employment standards in recent years have been extremely volatile and have changed backwards and forwards. 93 The practical significance of a finding of joint employment under the FLSA might also be rare beyond the most exploitative work: the rate of the federal minimum wage is notoriously comparatively low. 94 In addition, outside of the FLSA, with regards to at least some joint employment case law in the US under Title VII Civil Rights Act 1964, a joint employer may only be liable to the extent of its own wrongdoing. 95
At least traditionally, joint employment is not applied in the same way under the FLSA and NLRA, although this may change if a rule similar to the previously rescinded 2020 Final Rule on joint employment under the FLSA is re-introduced under the new Presidency. 96 Federal courts and the DOL historically have taken a wider economic dependency approach towards identifying joint employers under the FLSA, 97 whereas the National Labor Relations Board (NLRB) and federal courts take a control-based approach to identifying joint employers under the NLRA. 98 This stems from the instruction under the NLRA to apply the common law of agency to questions of personal scope, 99 whereas the broader ‘to suffer or to permit’ standard is expressed in the statute of the FLSA. 100 However, and importantly, the volatility with regards to joint employment in the US is likely not a reflection of the doctrine itself, but more a reflection of the wider political environment and the accompanying changing compositions of the DOL and NLRB. Indeed, the doctrine of joint employment is both well-established and long-established in the US: under the FLSA, recognition of joint employment is normally attributed to the US Supreme Court's decision as long ago as 1947 in Rutherford Food Corp v McComb, 101 and, with regards to its modern articulation, the ‘first’ Browning-Ferris Industries decision in 1982 with regards to the NLRA. 102 It might be suggested that the joint employer standard overall was stable under the FLSA until 2020, albeit with some differences between the Circuits. 103 Moreover, even with the volatility in this area, none of the relevant entities have recommended rejecting joint employment, although admittedly the effect may not be very different practically with regards to the narrower direct, substantial and immediate control test recommended at different times by the DOL and by the NLRB. The current position in mid-2025 is that the US Code of Federal Regulations (CFR), with regards to joint employment under the FLSA has been vacated and has been blank since 2021, 104 whereas the previous 2020 Final Rule on joint employment by the NLRB is effective by default since the replacement 2023 NLRB Final Rule was vacated by a District Court in March 2024. 105 From late January 2025 onwards, the NLRB has been unable to take action as it lacks its minimum quorum of three. 106
Even, however, assuming the widest interpretation of joint employment under US jurisprudence, it is notable that there is typically a high threshold in practice before joint employment is found. For example, in cases under the FLSA where joint employment has been established, using the theoretically wider economic dependency test, the facts tend to include a combination of multiple visits per week by personnel from the end-user (or, alternatively, the subcontracted workers will be working on the end-user's premises), 107 plus also supervision or instructions (whether direct or indirect) by personnel from the end-user. 108 Whether this is a reflection of the structure of the ‘traditional economy’ rather than the test of economic dependency is a different debate. 109 Control and/or supervision is one of the six to eight factors typically applied under the economic dependency test under the FLSA. 110 If the gig economy equivalent is something akin to the five ‘forms of control’ listed by the UK Supreme Court in Uber, while these were deemed to be ‘potent’ and ‘significant’ in the UK, 111 the Court of Appeals for the Third Circuit (‘Third Circuit’) in Razak v Uber Technologies Inc in 2020, on very similar or the same facts, found that there were ‘genuine disputes’ as to whether this was sufficient control. 112 Razak was not, however, a case about alleged joint employment: the question was whether drivers were employees under the FLSA or independent contractors. On the facts of a case more similar to the situation of an intermediary under the PWD, for example, the Eleventh Circuit in Layton v DHL Express (USA) Inc found that DHL was not a joint employer under the FLSA. 113 In Layton, DHL had ‘hired third-party contractors who employ couriers to deliver DHL's packages’. 114 In deciding against joint employment, perhaps tellingly, Wilson CJ noted: ‘DHL engaged in a limited amount of monitoring at the warehouse, but Drivers were basically unsupervised while completing their most essential job function which took up the majority of the workday—making deliveries.’ 115
On the other hand, this is not to underplay the potential significance of the doctrine of joint employment: rather these are more observations about the interpretation of, in this instance, the economic dependency test. Very similar facts in Uber in the UK led to a finding of intermediate worker status. 116 The UK Supreme Court implied new contracts (to provide work personally) between the drivers and Uber London. 117 On the basis that the drivers were paid by Uber BV; if it had been possible, potentially a finding of joint employment might have been equally sensible. 118 However, the definition of joint employment again would be key: if the claims would be frustrated by an arid discussion as to whether one entity exercised ‘more’ control when trying to establish joint employment; 119 also, if claims substantively would need to be satisfied against each entity; if proportionate liability would apply or whether a simplified version of joint and several liability would apply. 120
Concept of the employer
While the concepts of control and direction in Article 5 PWD have been criticised as inappropriate to identify an individual as a worker or employee, this section makes the brief observation that control and direction might, conversely, be useful for identifying employers. 121 This is an observation that applies also beyond platform work, but is prompted by the ideas of control and direction in Article 5 in relation to identifying employment relationships. To impose an employer's obligations on an entity, tentatively it is suggested here that the entity should have the power to control, or rather substantially to determine, an individual's working conditions. 122
If an ‘employer’ does not have the power substantially to determine working conditions, it might be objected that it may be unfair, but more importantly may not be effective, to impose an employer's obligations on that entity. 123 This might also be clearer (and possibly broader) than an alternative such as the employer is the entity ‘for whom the employee works’. 124 It would be subsidiary also to observe that the CJEU, in its limited case law in this respect (for example, Albron Catering BV v FNV Bondgenoten but more notably AFMB), also uses a control test to identify the relevant employer. 125 Normally, the assumption is that the test for identifying an employer must be symmetrical to the test for identifying an employee or worker. For example, Advocate General Pikamäe in AFMB reasoned in this manner. 126 This section additionally makes the brief observation that perhaps the tests do not need to be symmetrical. 127 If, hypothetically, a different test were to be used to identify workers or employees, a worker, for example, could potentially be identified on the basis of their predominantly personal work or on the basis of their economic dependence, whereas their employer or employers could be identified on the basis of which entity or entities substantially determine(s) the individual's working conditions. 128 On the other hand, this argument strongly depends on how control and direction is interpreted in this context.
For example, the exact wording of ‘substantially determine’ is used in the UK to identify the employer in the context of whistleblowing legislation when there is an alleged extended employment situation; however, this can be interpreted narrowly or broadly. 129 In McTigue v University Hospital Bristol NHS Foundation Trust, the Employment Appeal Tribunal sensibly rejected a comparative test ‘based on who determined the “majority of the terms”’. 130 By contrast, more recently, when an argument was made in a case under the Equality Act 2010 in the UK that an entity was ‘effectively dictating’ rates of pay at a subcontractor's organisation, the Court of Appeal rejected the very concept of ‘effectively dictating’ pay, as well as finding that this was not met on the facts. This was the case of Royal Parks Ltd v Boohene in 2024. 131 As a third comparison, the same wording in Boohene was included in a hypothetical scenario posed in the Third Restatement of Employment Law in the US from 2015, where it was stated that sufficient control should be deemed to be present on those facts (although the suggestion is admittedly complicated by the reference in the reporters’ notes accompanying that hypothetical to a case under the FLSA when control is only one factor and the governing approach is one of economic dependency). 132 The experience of joint employment in the US was touched on above. 133
Under the NLRA, in recent years, the NLRB have put forward, at different times, two different versions of a control test: one (narrower) requiring direct, substantial and immediate control in order to constitute a joint employer, 134 and the other (wider) accepting indirect, reserved and potentially limited control can sometimes be sufficient to constitute a joint employer. 135 However, the same can be stated of any test to identify indeed either workers or employers: interpretation is key. For example, a test of economic dependence to identify workers could become – and sometimes is – distracted into a different test of ‘dependence for income’, 136 whereas the experience of Deliveroo in the UK Supreme Court demonstrates a narrow approach to personal performance. 137
As for whether control could be particularly useful in the context of platform work (whether to identify employees or employers), in the one direction, there is Uber by the Supreme Court in the UK and Glovo by the Supreme Court in Spain. 138 The Supreme Court in Uber memorably referred to Uber's ‘tight’ control over drivers at least on the facts at that time, 139 but with broader significance, the Supreme Court in Glovo emphasised the importance of the ‘app’ as representing the essential means of performance (with ownership of the latter as particularly important for identifying employees). 140 If transplanted to other legal systems, the latter in particular could be highly significant, as the DLP will presumably always own or manage the app. In a subcontracting situation, even if the app is modified for subcontractors, presumably the DLP will be the entity making the modifications to ‘its’ app. 141 In the other direction, there is potentially again Razak v Uber Technologies Inc in the US, where the question of control was deemed not suitable for summary determination, 142 but the stronger counter-example is the reasoned order by the CJEU in B v Yodel Delivery Network Ltd. 143 The CJEU in Yodel emphasised the wide flexibility to drivers, when applying a test of control. Key factors were deemed to be the ability of drivers themselves to subcontract, the freedom of drivers to choose when to work and how many deliveries to make, and the ability of drivers to deliver simultaneously for competitors: common features of platform work. 144 On the other hand, this is potentially another example of where interpretation is key.
For example, in comparison to Yodel, the DOL in the US issued a new Final Rule on determining independent contractor status in early 2024; 145 albeit subsequent to the change of Presidency in 2025, the DOL in May 2025 announced it would no longer be using this Final Rule and would be reverting to a ‘Fact Sheet’ from 2008. 146 In the ‘supplementary information’ accompanying the 2024 Final Rule, the DOL opined that freedom in scheduling is of minor importance in relation to other possible signs of control by the putative employer. 147 Freedom in scheduling was, however, contrasted with price-setting. 148 In comparison, after the Spanish Supreme Court's decision in Glovo, the company reportedly responded by giving drivers the ability to modify the default price set for a trip by up to 30%. 149 A response to this, in turn, might be that the ability only to ‘modify’ is not true price-setting, but also that the instruction should be to look at the nature and degree of the putative employer's control in the round rather than focusing only on one subfactor - whether under the ‘economic reality’ approach to the FLSA or under the ‘primary of facts’ approach recommended by the International Labour Organisation and in the PWD. 150 Also, interestingly under that same heading of the nature and degree of the putative employer's control, the DOL pertinently noted in the same supplementary information from 2024 that the ability to work for competitors may not be a sign of entrepreneurialism, but rather ‘can too often be necessary for financial survival in the modern economy’. 151
Conclusion
This contribution focused on a lesser discussed and lesser-known provision in the new PWD: Article 3 on intermediaries. Article 3 was intended to provide some protection to individuals providing work for platform companies or platform workers in a subcontracting chain. At the very least, the employment presumption in Article 5 will apply to ‘intermediaries’, and the provisions on algorithmic management will apply directly to those working in a subcontracting chain. Article 3 also includes reference to joint and several liability, although that is seemingly not mandated (‘where appropriate’), and it would appear to be up to Member States to determine what precisely joint liability will cover. It was discussed that Article 3 is potentially the most ambiguous provision in the Directive, both because of its unusual drafting history but also because of the excessive flexibility given to Member States in the wording of Article 3. Ultimately, Member States have significant freedom to interpret Article 3 as narrowly or as widely as they choose, which will probably mean that there will be very divergent impacts when translated into national law and practice. Member States which supported the Directive during its drafting phase will probably implement the Directive in a more ambitious manner, whereas Member States which were more hesitant will likely take the opportunity to interpret Article 3 in a more limited way. 152 The contribution also discussed whether there is any encouragement or recognition given to joint employment by the Directive, finding the answer to be in the negative, and at the end made some brief reflections as to the appropriateness of control as a test for identifying employers.
Footnotes
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this Article.
Funding
The author received no financial support for the research, authorship, and/or publication of this Article.
