Uber BV and others v Aslam and others

New ways of working organised through digital platforms pose pressing questions about the employment status of the people who do the work involved. The central question on this appeal is whether an employment tribunal was entitled to find that drivers whose work is arranged through Uber’s smartphone application (“the Uber app”) work for Uber under workers’ contracts and so qualify for the national minimum wage, paid annual leave and other workers’ rights; or whether, as Uber contends, the drivers do not have these rights because they work for themselves as independent contractors, performing services under contracts made with passengers through Uber as their booking agent. If drivers work for Uber under workers’ contracts, a secondary question arises as to whether the employment tribunal was also entitled to find that the drivers who have brought the present claims were working under such contracts whenever they were logged into the Uber app within the territory in which they were licensed to operate and ready and willing to accept trips; or whether, as Uber argues, they were working only when driving passengers to their destinations.

For the reasons given in this judgment, I would affirm the conclusion of the Employment Appeal Tribunal and the majority of the Court of Appeal that the employment tribunal was entitled to decide both questions in the claimants’ favour.

The first appellant, Uber BV, is a Dutch company which owns the rights in the Uber app. The second appellant, Uber London Ltd (“Uber London”), is a UK subsidiary of Uber BV which, since May 2012, has been licensed to operate private hire vehicles in London. The third appellant, Uber Britannia Ltd, is another UK subsidiary of Uber BV which holds licences to operate such vehicles outside London. In this judgment I will use the name “Uber” to refer to the appellants collectively when it is not necessary to differentiate between them.

The claimants, and respondents to this appeal, are individuals who work or used to work as private hire vehicle drivers, performing driving services booked through the Uber app. For the purpose of the decision which has given rise to this appeal, the employment tribunal limited its consideration to two test claimants, Mr Yaseen Aslam and Mr James Farrar, both of whom were licensed to drive private hire vehicles in London. Like the employment tribunal, I will use masculine pronouns for brevity when referring to Uber drivers in this judgment in circumstances where all the claimant drivers are male.

At the time of the employment tribunal hearing in 2016, there were about 30,000 Uber drivers operating in the London area and 40,000 in the UK as a whole. Some two million people were registered to use the Uber app as passengers in London.

As described in more detail in the decision of the employment tribunal, Uber’s business model is simple. Prospective customers download the Uber app (for free) to their smartphone and create an account by providing personal information including a method of payment. They are then able to request rides. To do so, they open the Uber app on their phone and make a request. In the period considered by the employment tribunal, users did not have to enter their destination when booking a ride through the app, but they generally did so. The Uber app identifies the passenger’s location through the smartphone’s geolocation system. Using the same technology, the app identifies the nearest available driver who is logged into the app and informs him (via his smartphone) of the request. At this stage the driver is told the passenger’s first name and Uber rating (as to which, see below) and has ten seconds in which to decide whether to accept the request. If the driver does not respond within that time, the next closest driver is located and offered the trip. Once a driver accepts, the trip is assigned to that driver and the booking confirmed to the passenger, who is sent the driver’s name and car details.

At this point the driver and passenger are put into direct contact with each other through the Uber app, but this is done in such a way that neither has access to the other’s mobile telephone number. The purpose is to enable them to communicate with each other in relation to the pick-up, for example to identify the passenger’s precise location or to advise of problems such as traffic delay. The passenger can also track the driver’s progress on a map on their smartphone.

The driver is not informed of the passenger’s destination until the passenger is collected. At that point the driver learns the destination either directly from the passenger or through the app (if the destination was entered when the ride was requested) when the driver presses “start trip” on his phone. The Uber app incorporates route planning software and provides the driver with detailed directions to the destination. The driver is not bound to follow those directions but departure from the recommended route may result in a reduction in payment if the passenger complains about the route taken.

On arrival at the destination, the driver presses “complete trip” on his smartphone. The fare is then calculated automatically by the Uber app, based on time spent and distance covered. At times and places of high demand, a multiplier is applied resulting in a higher fare. Drivers are permitted to accept payment in a lower, but not a higher, sum than the fare calculated by the app (although, in the unlikely event that a driver accepts a lower sum, the “service fee” retained by Uber BV is still based on the fare calculated by the app). Drivers are at liberty to accept tips but are discouraged by Uber from soliciting them.

The fare is debited to the passenger’s credit or debit card registered on the Uber app and the passenger is sent a receipt for the payment by email. Separately, the Uber app generates a document described as an “invoice” addressed on behalf of the driver to the passenger (showing the passenger’s first name but not their surname or contact details). However, the passenger never sees this document, which is not sent to the passenger but is accessible to the driver on the Uber app and serves as a record of the trip and the fare charged.

Uber BV makes a weekly payment to the driver of sums paid by passengers for trips driven by the driver less a “service fee” retained by Uber BV. In the cases of Mr Aslam and Mr Farrar, the service fee was 20% of the fares.

Drivers are prohibited by Uber from exchanging contact details with a passenger or contacting a passenger after the trip ends other than to return lost property.

Uber operates a ratings system whereby, after the trip, the passenger and driver are each sent a message asking them to rate the other anonymously on a scale of 1 to 5.

To become an Uber driver, a person can sign up online. They must then attend and present certain documents at the offices of the local Uber company (which, for the London area, is Uber London). The required documents comprise a national insurance certificate, driving licence, licence to drive a private hire vehicle, vehicle logbook, MOT certificate and certificate of motor insurance. The applicant must also take part in what the employment tribunal described as “an interview, albeit not a searching one”, and watch a video presentation about the Uber app and certain Uber procedures. This process is referred to by Uber as “onboarding”.

Individuals accepted as drivers are given free access to the Uber app through their own smartphone or may hire a smartphone for £5 a month from Uber BV configured so that it can only be used to operate the Uber app. The driver has to provide and pay for his own vehicle, which must be on a list of accepted makes and models, in good condition, no older than a specified age and preferably silver or black. Drivers must also bear all the costs of running their vehicles, including fuel, insurance, road tax and the cost of obtaining a private hire vehicle licence.

Individuals approved to work as drivers are free to make themselves available for work, by logging onto the Uber app, as much or as little as they want and at times of their own choosing. They are not prohibited from providing services for or through other organisations, including any direct competitor of Uber operating through another digital platform. Drivers can also choose where within the territory covered by their private hire vehicle licence they make themselves available for work. They are not provided with any insignia or uniform and in London are discouraged from displaying Uber branding of any kind on their vehicle.

The employment tribunal made a number of findings about standards of performance which drivers are expected to meet and actions taken where drivers fail to meet these standards. For example, the tribunal found that a “Welcome Packet” of material issued by Uber London to new drivers included numerous instructions as to how drivers should conduct themselves, such as “Polite and professional at all times”, “Avoid inappropriate topics of conversation” and “Do not contact the rider after the trip has ended”. Other material in the Welcome Packet, under the heading “What Uber looks for”, stated:

Taking these three metrics in reverse order, drivers whose acceptance rate for trip requests falls below a set level - which according to evidence before the tribunal was 80% - receive warning messages reminding the driver that being logged into the Uber app is an indication that the driver is willing and able to accept trip requests. If the driver’s acceptance rate does not improve, the warnings escalate and culminate in the driver being automatically logged off the Uber app for ten minutes if the driver declines three trips in a row. A similar system of warnings, culminating in a ten-minute log-off “penalty”, applies to cancellations by drivers after a trip has been accepted. The driver’s ratings from passengers are also monitored and the employment tribunal found that drivers who have undertaken 200 trips or more and whose average rating is below 4.4 become subject to a graduated series of “quality interventions” aimed at assisting them to improve. If their ratings do not improve to an average of 4.4 or better, they are “removed from the platform” and their accounts “deactivated”.

Uber also operates a “driver offence process” to address misconduct by drivers. This again involves a graduated series of measures, beginning with a “warning” message and potentially leading to “deactivation”.

In addition, Uber London handles passenger complaints, including complaints about a driver, and decides whether to make any refund to the passenger (sometimes without even referring the matter to the driver concerned). Such a refund will generally result in a correspondingly reduced payment to the driver, though the tribunal found that on occasions, when Uber London considers it necessary or politic to make a refund but there is no proper ground for holding the driver to be at fault, Uber London will bear the cost of the refund itself.

Uber will in some circumstances pay drivers the cost, or a contribution towards the cost, of cleaning vehicles soiled by passengers. The employment tribunal noted that it was not suggested that such payments were conditional upon Uber recovering this sum from the passenger.

Before using the Uber app as drivers for the first time, the claimants were required to sign a “partner registration form” stating that they agreed to be bound by and comply with terms and conditions described as “Partner Terms” dated 1 July 2013. In October 2015 a new “Services Agreement” was introduced to which drivers were required to signify their agreement electronically before they could again log into the Uber app and accept trip requests. The differences between the old and new terms are not material for present purposes and it is sufficient to refer to the new terms contained in the Services Agreement.

The Services Agreement is formulated as a legal agreement between Uber BV and “an independent company in the business of providing transportation services”, referred to as “Customer”. It contains an undertaking by “Customer” to enter into a contract with each driver in the form of an accompanying “Driver Addendum”. This arrangement is inapposite for the vast majority of drivers who sign up as individuals and not on behalf of any “independent company” which in turn engages drivers.

In the typical case where “Customer” is an individual driver, the nature of the relevant services and relationships as characterised by the Services Agreement is that Uber BV agrees to provide electronic services (referred to as the “Uber Services”) to the driver, which include access to the Uber app and payment services, and the driver agrees to provide transportation services to passengers (referred to as “Users”). The agreement states that Customer acknowledges and agrees that Uber BV does not provide transportation services and that, where Customer accepts a User’s request for transportation services made through the Uber app, Customer is responsible for providing those transportation services and, by doing so, “creates a legal and direct business relationship between Customer and the User, to which neither Uber [BV] nor any of its Affiliates in the Territory is a party” (see clause 2.3).

Clause 4.1 of the Services Agreement states that:

Clause 4.2 gives Uber BV the right to change the fare calculation at any time in its discretion “based upon local market factors”; and clause 4.3 provides that Uber BV and/or its Affiliates reserve the right to adjust the fare for a particular instance of transportation services (eg where the driver took an inefficient route) or to cancel the fare (eg in response to a User complaint).

In addition to the written agreements between drivers and Uber BV, Uber also relies in these proceedings on written terms and conditions (the “Rider Terms”) which passengers are required to accept before they can use the Uber app. The version of the Rider Terms current at the time of the tribunal hearing was last updated on 16 June 2016. These Rider Terms state that they constitute an agreement between the rider, Uber BV and the local Uber company operating in the relevant part of the UK. As mentioned, in the case of London, the relevant company is Uber London.

Clause 2 of Part 1 of the Rider Terms states that, as set out in clause 3, “Uber UK” (a term defined to include Uber London) accepts private hire vehicle bookings (“PHV Bookings”) made using the Uber app. Clause 3 states:

Under Part 2 of the Rider Terms, riders are granted a licence by Uber BV to use the Uber app, described in clause 2 as “a technology platform that enables users ... to pre-book and schedule transportation, logistics, delivery, and/or vendor services with independent third party providers of such services, including independent third party transportation providers.”

The operation of private hire vehicles in London is regulated by the Private Hire Vehicles (London) Act 1998 and regulations made under it. Under that Act a vehicle may only be used for private hire if both vehicle and driver are licensed by the licensing authority, which is Transport for London. A licence is also required to accept bookings (referred to in the Act as “private hire bookings”) for the hire of a private hire vehicle to carry one or more passengers. Thus, section 2(1) provides:

At all relevant times Uber London has held a private hire vehicle (“PHV”) operator’s licence for London. Section 4(2) of the Act places an obligation on the holder of such a licence to secure that:

Pursuant to regulation 9(3) of the Private Hire Vehicles (London) (Operators’ Licences) Regulations 2000, as originally formulated, it was a condition of the grant of a London PHV operator’s licence that:

The obligations of the operator under the Act and regulations also include keeping records of all bookings accepted and of all private hire vehicles and drivers available to the operator for carrying out bookings accepted by him.

The rights claimed by the claimants in these proceedings are: rights under the National Minimum Wage Act 1998 and associated regulations to be paid at least the national minimum wage for work done; rights under the Working Time Regulations 1998 which include the right to receive paid annual leave; and in the case of two claimants, one of whom is Mr Aslam, a right under the Employment Rights Act 1996 not to suffer detrimental treatment on the grounds of having made a protected disclosure (“whistleblowing”).

All these rights are conferred by the legislation on “workers”. The term “worker” is defined by section 230(3) of the Employment Rights Act 1996 to mean:

A “contract of employment” is defined in section 230(2) of the Act to mean “a contract of service or apprenticeship, whether express or implied, and (if it is express) whether oral or in writing.” An “employee” means an individual who has entered into or works under a contract of employment: see section 230(1). However, the terms “employer” and “employed” are defined more broadly to refer to the person by whom an employee or worker is (or was) employed under a worker’s contract: see section 230(4) and (5).

Similar definitions of all these terms are contained in section 54 of the National Minimum Wage Act 1998 and regulation 2(1) of the Working Time Regulations 1998.

The effect of these definitions, as Baroness Hale of Richmond observed in Bates van Winkelhof v Clyde & Co LLP[2014] UKSC 32; [2014] 1 WLR 2047, paras 25 and 31, is that employment law distinguishes between three types of people: those employed under a contract of employment; those self-employed people who are in business on their own account and undertake work for their clients or customers; and an intermediate class of workers who are self-employed but who provide their services as part of a profession or business undertaking carried on by someone else. Some statutory rights, such as the right not to be unfairly dismissed, are limited to those employed under a contract of employment; but other rights, including those claimed in these proceedings, apply to all “workers”.

Following a preliminary hearing, the employment tribunal decided that the claimants were “workers” who, although not employed under contracts of employment, worked for Uber London under “workers’ contracts” within the meaning of limb (b) of the statutory definition quoted at para 35 above. The tribunal further found that, for the purposes of the relevant legislation, the claimants were working for Uber London during any period when a claimant (a) had the Uber app switched on, (b) was within the territory in which he was authorised to work, and (c) was able and willing to accept assignments.

An appeal by Uber from this decision to the Employment Appeal Tribunal was dismissed, as was a further appeal to the Court of Appeal (Sir Terence Etherton MR and Bean LJ, with Underhill LJ dissenting). The Court of Appeal granted Uber permission to appeal to this court.

Limb (b) of the statutory definition of a “worker’s contract” has three elements: (1) a contract whereby an individual undertakes to perform work or services for the other party; (2) an undertaking to do the work or perform the services personally; and (3) a requirement that the other party to the contract is not a client or customer of any profession or business undertaking carried on by the individual.

This case is concerned with the first of these requirements. It is not in dispute that the claimant drivers worked under contracts whereby they undertook to perform driving services personally; and it is not suggested that any Uber company was a client or customer of the claimants. The critical issue is whether, for the purposes of the statutory definition, the claimants are to be regarded as working under contracts with Uber London whereby they undertook to perform services for Uber London; or whether, as Uber contends, they are to be regarded as performing services solely for and under contracts made with passengers through the agency of Uber London.

It is Uber’s case that, in answering this question, the correct starting point is to interpret the terms of the written agreements between Uber BV and drivers and between the Uber companies and passengers. Uber relies on the terms of these written agreements quoted above which state that, when a request to book a private hire vehicle made through the Uber app is accepted, a contract is thereby created between passenger and driver, to which no Uber entity is a party and under which the driver is solely responsible for providing transportation services to the passenger. Uber also relies on terms of the written agreements which state that the only role of Uber BV is to provide technology services and to act as a payment collection agent for the driver and that the only role of Uber London (and other Uber UK companies) is to act as a booking agent for drivers.

Uber maintains that the approach adopted by the employment tribunal, the Employment Appeal Tribunal and the majority of the Court of Appeal was wrong in law because it involved disregarding, without any legal justification, the clear and unambiguous terms of the written agreements.

There is a difficulty which, in my view, would be fatal for Uber’s case even if the correct approach to deciding whether the claimants were working under workers’ contracts with Uber London were simply to apply ordinary principles of the law of contract and agency. This difficulty stems from the fact that there is no written agreement between Uber London and drivers. In these circumstances the nature of their relationship has to be inferred from the parties’ conduct, considered in its relevant factual and legal context.

It is an important feature of the context in which, as the employment tribunal found, Uber London recruits and communicates on a day to day basis with drivers that, as mentioned earlier: (1) it is unlawful for anyone in London to accept a private hire booking unless that person is the holder of a private hire vehicle operator’s licence for London; and (2) the only natural or legal person involved in the acceptance of bookings and provision of private hire vehicles booked through the Uber app which holds such a licence is Uber London. It is reasonable to assume, at least unless the contrary is demonstrated, that the parties intended to comply with the law in the way they dealt with each other.

Uber maintains that the acceptance of private hire bookings by a licensed London PHV operator acting as agent for drivers would comply with the regulatory regime. I am not convinced by this. References in the Private Hire Vehicles (London) Act 1998 to “acceptance” of a private hire booking are reasonably understood to connote acceptance (personally and not merely for someone else) of a contractual obligation to carry out the booking and provide a vehicle for that purpose. This is implicit, for example, in section 4(2) of the Act quoted at para 31 above. It would in principle be possible for Uber London both to accept such an obligation itself and also to contract on behalf of the driver of the vehicle. However, if this were the arrangement made, it would seem hard to avoid the conclusion that the driver, as well as Uber London, would be a person who accepts the booking by undertaking a contractual obligation owed directly to the passenger to carry it out. If so, the driver would be in contravention of section 2(1) of the Private Hire Vehicles (London) Act 1998 by accepting a private hire booking without holding a private hire vehicle operator’s licence for London. This suggests that the only contractual arrangement compatible with the licensing regime is one whereby Uber London as the licensed operator accepts private hire bookings as a principal (only) and, to fulfil its obligation to the passenger, enters into a contract with a transportation provider (be that an individual driver or a firm which in turn provides a driver) who agrees to carry out the booking for Uber London.

Counsel for Uber sought to resist this interpretation of the legislation on the basis that the legislation was enacted in the context of “a long-established industry practice” under which PHV operators may merely act as agents for drivers who contract directly with passengers. Uber has adduced no evidence, however, of any such established practice which the Private Hire Vehicles (London) Act 1998 may be taken to have been intended to preserve. I will consider later two cases involving minicab firms which were said by counsel for Uber to show that the courts have endorsed such an agency model. But it is sufficient to say now that in neither case was any consideration given to whether such an arrangement would comply with the licensing regime. The same is true of cases also relied on by Uber (along with a notice published by HMRC in 2002) which are concerned with how VAT applies to the supply of private hire vehicles. That material in my view has no bearing on the issues raised in these proceedings.

It is unnecessary, however, to express any concluded view on whether an agency model of operation would be compatible with the PHV licensing regime because there appears to be no factual basis for Uber’s contention that Uber London acts as an agent for drivers when accepting private hire bookings.

It is true that the Rider Terms on which Uber contracts with passengers include a term (in clause 3 of Part 1, quoted at para 28 above) which states that Uber London (or other local Uber company) accepts private hire bookings “acting as disclosed agent for the Transportation Provider (as principal)” and that such acceptance “gives rise to a contract for the provision to [the rider] of transportation services between [the rider] and the Transportation Provider”. It is, however, trite law that a person (A) cannot create a contract between another person (B) and a third party merely by claiming or purporting to do so but only if A is (actually or ostensibly) authorised by B to act as B’s agent.

Authority may be conferred by a contract between principal and agent. It cannot be said, however, that the Rider Terms establish a contract between drivers and Uber London. There is no evidence that drivers were ever sent the Rider Terms let alone consented to them. In any case the Rider Terms state that they constitute an agreement between the rider, Uber BV and the relevant local Uber company: they do not purport to record an agreement to which any driver is a party. In accordance with basic principles of contract and agency law, therefore, nothing stated in the Rider Terms is capable of conferring authority on Uber London to act as agent for any driver (or other “Transportation Provider”) nor of giving rise to a contract between a rider and a driver for the provision to the rider of transportation services by the driver.

The only written agreements to which drivers were parties were agreements with Uber BV, the Dutch parent company. No other Uber company was a party to those agreements. In any case, although clause 2.2 of the Services Agreement describes what is to happen if a driver accepts a trip request “either directly or through an Uber Affiliate in the Territory acting as agent”, there is no provision which purports to confer the driver’s authority on any Uber Affiliate to accept such requests on his behalf.

An agency relationship need not be contractual. What is required is an overt act by the principal conferring authority on the agent to act on the principal’s behalf. Even if lacking such actual authority, a person (A) who purports to act as agent for another (B) may still affect B’s legal relations with a third party under the principle of ostensible or apparent authority, but only if B has represented to the third party that A is authorised to act as B’s agent and the third party has relied on that representation.

The employment tribunal made no finding that drivers did any overt act that conferred authority on Uber London to act as the driver’s agent in accepting bookings so as to create a contract between the driver and the passenger, nor that drivers did or said anything that represented to passengers that Uber London was authorised to act as their agent. Uber’s case that Uber London acted as a booking agent for drivers has been based solely on the written agreements referred to at paras 22-28 above - which, for the reasons given, do not support it. When pressed during oral argument on how the alleged agency relationship between drivers and Uber London was created, leading counsel for Uber, Ms Dinah Rose QC, suggested that it was created when a driver attended Uber London’s offices in person and presented the documents required in order to be authorised to use the Uber app. So far as I am aware, this was the first time that such a suggestion had been made in these proceedings. Not only is it unsupported by any finding of the employment tribunal but, so far as this court has been shown, there was no evidence capable of founding such an inference.

In order to found such an inference, it would be necessary to point, at the least, to a prior communication from Uber London to the individual concerned or other background facts known to both parties which would lead reasonable people in their position to understand that, by producing the documents required by Uber London, an individual who did so was thereby authorising Uber London to contract with passengers as his agent, rather than - as seems to me the natural inference - merely applying for a job as one of Uber’s drivers. There is no finding of the employment tribunal that any such communication was made nor that anything occurred during the “onboarding” process which could, even arguably, be construed as an act by the prospective driver appointing Uber London to act as his booking agent.

Once the assertion that Uber London contracts as a booking agent for drivers is rejected, the inevitable conclusion is that, by accepting a booking, Uber London contracts as principal with the passenger to carry out the booking. In these circumstances Uber London would have no means of performing its contractual obligations to passengers, nor of securing compliance with its regulatory obligations as a licensed operator, without either employees or subcontractors to perform driving services for it. Considered against that background, it is difficult to see how Uber’s business could operate without Uber London entering into contracts with drivers (even if only on a per trip basis) under which drivers undertake to provide services to carry out the private hire bookings accepted by Uber London.

Given the importance of the wider issue, however, I do not think it would be right to decide this appeal on this basis alone and without addressing Uber’s argument that the question whether an individual is a “worker” for the purpose of the relevant legislation ought in principle to be approached, as the starting point, by interpreting the terms of any applicable written agreements.

In advancing this argument, Uber has to confront the decision of this court in Autoclenz Ltd v Belcher[2011] UKSC 41; [2011] ICR 1157.

In the Autoclenz case the claimants worked as “valeters” performing car cleaning services which the company (Autoclenz) had contracted to provide to third parties. In order to obtain the work, the claimants were required to sign written contracts which stated that they were subcontractors and not employees of Autoclenz; that they were not obliged to provide services to the company, nor was the company obliged to offer work to them; and that they could provide suitably qualified substitutes to carry out the work on their behalf. As in the present case, the claimants brought proceedings claiming that they were “workers” for the purposes of the legislation conferring the rights to be paid the national minimum wage and to receive statutory paid leave. The employment tribunal held that the claimants came within both limbs of the definition of a “worker” and appeals by Autoclenz were dismissed at every level including the Supreme Court.

In the Supreme Court the sole judgment was given by Lord Clarke of Stone-cum-Ebony, with whom the other Justices agreed. In his discussion of the legal principles, Lord Clarke drew a distinction between certain principles “which apply to ordinary contracts and, in particular, to commercial contracts”, and “a body of case law in the context of employment contracts in which a different approach has been taken” (see para 21). It can be seen from a passage quoted by Lord Clarke (at para 20) from the judgment of Aikens LJ in the Court of Appeal [2010] IRLR 70, paras 87-89, that the principles applicable to ordinary contracts to which he was here referring were: (i) the “parol evidence rule”, whereby a contractual document is treated, at least presumptively, as containing the whole of the parties’ agreement; (ii) the signature rule, whereby a person who signs a contractual document is treated in law as bound by its terms irrespective of whether he or she has in fact read or understood them; and (iii) the principle that, generally, the only way in which a party to a written contract can argue that its terms do not accurately reflect the true agreement of the parties is by alleging that a mistake was made in drawing up the contract which the court can correct by ordering rectification.

Whilst stating that nothing in his judgment was intended to alter these principles as they apply to ordinary contracts, Lord Clarke endorsed the view of Aikens LJ that, in the employment context, rectification principles are not in point and there may be reasons other than a mistake in setting out the contract terms why the written terms do not accurately reflect what the parties actually agreed.

Beginning at para 22 of the judgment, Lord Clarke considered three cases in which “the courts have held that the employment tribunal should adopt a test that focuses on the reality of the situation where written documentation may not reflect the reality of the relationship”. From these cases he drew the conclusion (at para 28) that, in the employment context, it is too narrow an approach to say that a court or tribunal may only disregard a written term as not part of the true agreement between the parties if the term is shown to be a “sham”, in the sense that the parties had a common intention that the term should not create the legal rights and obligations which it gives the appearance of creating: see Snook v London and West Riding Investments Ltd[1967] 2 QB 786, 802 (Diplock LJ). Rather, the court or tribunal should consider what was actually agreed between the parties, “either as set out in the written terms or, if it is alleged those terms are not accurate, what is proved to be their actual agreement at the time the contract was concluded”: see para 32, again agreeing with observations of Aikens LJ in the Court of Appeal.

After quoting (at para 34) a further statement of Aikens LJ contrasting the circumstances in which contracts relating to work or services are often concluded with “those in which commercial contracts between parties of equal bargaining power are agreed,” Lord Clarke ended his discussion of the law (at para 35) by saying:

Applying that approach to the facts of the Autoclenz case, Lord Clarke concluded that, on the basis of findings of fact not capable of challenge on appeal, the employment tribunal was entitled to hold that the contractual documents did not reflect the true agreement between the parties - in particular insofar as the documents stated that Autoclenz was under no obligation to offer work to the claimants, nor they to accept it, and that the claimants had a right to provide a substitute. The tribunal was entitled to find that the actual understanding of the parties was that the claimants would be available to work, and would be offered work, whenever there was work available, and that they were required to perform the work personally. It followed that the employment tribunal was entitled to hold that the claimants were “workers” working under contracts of employment.

Uber submits that what the Autoclenz case decided is that, for the purposes of applying a statutory classification, a court or tribunal may disregard terms of a written agreement if it is shown that the terms in question do not represent the “true agreement” or what was “actually agreed” between the parties, as ascertained by considering all the circumstances of the case including how the parties conducted themselves in practice. If, however, there is no inconsistency between the terms of the written agreement and how the relationship operated in reality, there is no basis for departing from the written agreement.

Uber further submits that there is no inconsistency in the present case between the written agreements between Uber, drivers and passengers and how that tripartite relationship actually operated in practice. In particular, Uber argues that the facts found by the employment tribunal (or alternatively, which the tribunal should have found) are consistent with the written terms stipulating that the drivers were performing their services under contracts made with passengers through the agency of Uber London and not for or under any contract with any Uber company. Uber submits that there is in these circumstances no legal basis for finding that the terms of the written agreements did not reflect the true agreements between the parties and hence for departing from the classification of the parties’ relationships set out in the contractual documentation.

This argument was accepted by Underhill LJ in his dissenting judgment in the Court of Appeal. In his view (stated at para 120):

The judgment of this court in the Autoclenz case made it clear that whether a contract is a “worker’s contract” within the meaning of the legislation designed to protect employees and other “workers” is not to be determined by applying ordinary principles of contract law such as the parol evidence rule, the signature rule and the principles that govern the rectification of contractual documents on grounds of mistake. Not only was this expressly stated by Lord Clarke but, had ordinary principles of contract law been applied, there would have been no warrant in the Autoclenz case for disregarding terms of the written documents which were inconsistent with an employment relationship, as the court held that the employment tribunal had been entitled to do. What was not, however, fully spelt out in the judgment was the theoretical justification for this approach. It was emphasised that in an employment context the parties are frequently of very unequal bargaining power. But the same may also be true in other contexts and inequality of bargaining power is not generally treated as a reason for disapplying or disregarding ordinary principles of contract law, except in so far as Parliament has made the relative bargaining power of the parties a relevant factor under legislation such as the Unfair Contract Terms Act 1977.

Critical to understanding the Autoclenz case, as I see it, is that the rights asserted by the claimants were not contractual rights but were created by legislation. Thus, the task for the tribunals and the courts was not, unless the legislation required it, to identify whether, under the terms of their contracts, Autoclenz had agreed that the claimants should be paid at least the national minimum wage or receive paid annual leave. It was to determine whether the claimants fell within the definition of a “worker” in the relevant statutory provisions so as to qualify for these rights irrespective of what had been contractually agreed. In short, the primary question was one of statutory interpretation, not contractual interpretation.

The modern approach to statutory interpretation is to have regard to the purpose of a particular provision and to interpret its language, so far as possible, in the way which best gives effect to that purpose. In UBS AG v Revenue and Customs Comrs[2016] UKSC 13; [2016] 1 WLR 1005, paras 61-68, Lord Reed (with whom the other Justices of the Supreme Court agreed) explained how this approach requires the facts to be analysed in the light of the statutory provision being applied so that if, for example, a fact is of no relevance to the application of the statute construed in the light of its purpose, it can be disregarded. Lord Reed cited the pithy statement of Ribeiro PJ in Collector of Stamp Revenue v Arrowtown Assets Ltd(2003) 6 ITLR 454, para 35:

The general purpose of the employment legislation invoked by the claimants in the Autoclenz case, and by the claimants in the present case, is not in doubt. It is to protect vulnerable workers from being paid too little for the work they do, required to work excessive hours or subjected to other forms of unfair treatment (such as being victimised for whistleblowing). The paradigm case of a worker whom the legislation is designed to protect is an employee, defined as an individual who works under a contract of employment. In addition, however, the statutory definition of a “worker” includes in limb (b) a further category of individuals who are not employees. The purpose of including such individuals within the scope of the legislation was clearly elucidated by Mr Recorder Underhill QC giving the judgment of the Employment Appeal Tribunal in Byrne Bros (Formwork) Ltd v Baird[2002] ICR 667, para 17(4):

The Regulations referred to in this passage are the Working Time Regulations 1998 which implemented Directive 93/104/EC (“the Working Time Directive”); and a similar explanation of the concept of a worker has been given in EU law. Although there is no single definition of the term “worker”, which appears in a number of different contexts in the Treaties and EU legislation, there has been a degree of convergence in the approach adopted. In Allonby v Accrington and Rossendale College (Case C-256/01) [2004] ICR 1328; [2004] ECR I-873 the European Court of Justice held, at para 67, that in the Treaty provision which guarantees male and female workers equal pay for equal work (at that time, article 141 of the EC Treaty):

In Hashwani v Jivraj[2011] UKSC 40; [2011] 1 WLR 1872 the Supreme Court followed this approach in holding that an arbitrator was not a person employed under “a contract personally to do any work” for the purpose of legislation prohibiting discrimination on the grounds of religion or belief. Lord Clarke, with whom the other members of the court agreed, identified (at para 34) the essential questions underlying the distinction between workers and independent contractors outside the scope of the legislation as being:

In the Bates van Winkelhof case at para 39, Baroness Hale cautioned that, while “subordination may sometimes be an aid to distinguishing workers from other self-employed people, it is not a freestanding and universal characteristic of being a worker.” In that case the Supreme Court held that a solicitor who was a member of a limited liability partnership was a worker essentially for the reasons that she could not market her services as a solicitor to anyone other than the LLP and was an integral part of their business. While not necessarily connoting subordination, integration into the business of the person to whom personal services are provided and the inability to market those services to anyone else give rise to dependency on a particular relationship which may also render an individual vulnerable to exploitation.

The correlative of the subordination and/or dependency of employees and workers in a similar position to employees is control exercised by the employer over their working conditions and remuneration. As the Supreme Court of Canada observed in McCormick v Fasken Martineau DuMoulin LLP2014 SCC 39; [2014] 2 SCR 108, para 23:

Once this is recognised, it can immediately be seen that it would be inconsistent with the purpose of this legislation to treat the terms of a written contract as the starting point in determining whether an individual falls within the definition of a “worker”. To do so would reinstate the mischief which the legislation was enacted to prevent. It is the very fact that an employer is often in a position to dictate such contract terms and that the individual performing the work has little or no ability to influence those terms that gives rise to the need for statutory protection in the first place. The efficacy of such protection would be seriously undermined if the putative employer could by the way in which the relationship is characterised in the written contract determine, even prima facie, whether or not the other party is to be classified as a worker. Laws such as the National Minimum Wage Act were manifestly enacted to protect those whom Parliament considers to be in need of protection and not just those who are designated by their employer as qualifying for it.

This point can be illustrated by the facts of the present case. The Services Agreement (like the Partner Terms before it) was drafted by Uber’s lawyers and presented to drivers as containing terms which they had to accept in order to use, or continue to use, the Uber app. It is unlikely that many drivers ever read these terms or, even if they did, understood their intended legal significance. In any case there was no practical possibility of negotiating any different terms. In these circumstances to treat the way in which the relationships between Uber, drivers and passengers are characterised by the terms of the Services Agreement as the starting point in classifying the parties’ relationship, and as conclusive if the facts are consistent with more than one possible legal classification, would in effect be to accord Uber power to determine for itself whether or not the legislation designed to protect workers will apply to its drivers.

This is, as I see it, the relevance of the emphasis placed in the Autoclenz case (at para 35) on the relative bargaining power of the parties in the employment context and the reason why Lord Clarke described the approach endorsed in that case of looking beyond the terms of any written agreement to the parties’ “true agreement” as “a purposive approach to the problem”.

Such an approach is further justified by the fact that all the relevant statutes or statutory regulations conferring rights on workers contain prohibitions against contracting out. Thus, section 203(1) of the Employment Rights Act 1996 provides:

These provisions, as I read them, apply to any provision in an agreement which can be seen, on an objective consideration of the facts, to have as its object excluding or limiting the operation of the legislation. It is just as inimical to the aims of the legislation to allow its protection to be limited or excluded indirectly by the terms of a contract as it is to allow that to be done in direct terms.

Take, for example, the following provisions contained, respectively, in clauses 2.3 and 2.4 of the Services Agreement:

If or in so far, however, as these contractual provisions purport to agree matters of fact rather than law, then (leaving aside the fact that the relevant “Affiliates” including Uber London were not parties to the agreement) Uber would no doubt seek to rely on case law which has recognised a principle of “contractual estoppel” - whereby parties can bind themselves by contract to accept a particular state of affairs even if they know that state of affairs to be untrue: see eg First Tower Trustees Ltd v CDS (Superstores International) Ltd[2018] EWCA Civ 1396; [2019] 1 WLR 637, para 47. This would preclude a driver from asserting in any legal proceedings that he is performing transportation services for or under a contract with any Uber company or that he is directed or controlled in connection with the provision of transportation services by any Uber company. The result - which was patently the drafter’s intention - would be to prevent a driver from claiming that he falls within the statutory definition of a “worker” so as to qualify for the rights conferred on workers by statutory provisions such as those contained in the National Minimum Wage Act 1998. As such, these provisions in the agreement are just as much provisions which purport to exclude or limit the operation of the legislation as would be a term stating that “Customer acknowledges and agrees that Customer is not and shall not be deemed to be a ‘worker’ for the purposes of the National Minimum Wage Act 1998” or a term stating that “Customer acknowledges and agrees that, notwithstanding the provisions of the National Minimum Wage Act 1998, Customer shall not be entitled to be paid the national minimum wage.” In each case the object of the provision is the same. Consequently, in determining whether drivers are entitled under the provisions of the 1998 Act to be paid the national minimum wage, section 49(1) of the Act renders the clauses quoted above void. The same applies to all other provisions in the Services Agreement which can be seen to have as their object precluding a driver from claiming rights conferred on workers by the applicable legislation.

If, as I conclude, the way in which the relevant relationships are characterised in the written agreements is not the appropriate starting point in applying the statutory definition of a “worker”, how is the definition to be applied?

In the Autoclenz case it was said (at para 35) that “the true agreement will often have to be gleaned from all the circumstances of the case, of which the written agreement is only a part.” More assistance is provided by the decision of the House of Lords in Carmichael v National Power plc[1999] 1 WLR 2042. That case concerned tour guides engaged to act “on a casual as required basis”. The guides later claimed to be employees and therefore entitled by statute to a written statement of their terms of employment. Their case was that an exchange of correspondence between the parties in March 1989 constituted a contract, which was to be classified as a contract of employment. The industrial tribunal rejected this case and found that, when not working as guides, the claimants were not in any contractual relationship with the respondent. The tribunal made this finding on the basis of: (a) the language of the correspondence; (b) the way in which the relationship had operated; and (c) evidence of the parties as to their understanding of it. The House of Lords held that this was the correct approach. Lord Irvine of Lairg LC said at p 2047C that:

In the Carmichael case there was no formal written agreement. The Autoclenz case shows that, in determining whether an individual is an employee or other worker for the purpose of the legislation, the approach endorsed in the Carmichael case is appropriate even where there is a formal written agreement (and even if the agreement contains a clause stating that the document is intended to record the entire agreement of the parties). This does not mean that the terms of any written agreement should be ignored. The conduct of the parties and other evidence may show that the written terms were in fact understood and agreed to be a record, possibly an exclusive record, of the parties’ rights and obligations towards each other. But there is no legal presumption that a contractual document contains the whole of the parties’ agreement and no absolute rule that terms set out in a contractual document represent the parties’ true agreement just because an individual has signed it. Furthermore, as discussed, any terms which purport to classify the parties’ legal relationship or to exclude or limit statutory protections by preventing the contract from being interpreted as a contract of employment or other worker’s contract are of no effect and must be disregarded.

This last point provides one rationale for the conclusion reached in the Autoclenz case itself. The findings of the employment tribunal justified the inference that the terms of the written agreements which stated that the claimants were subcontractors and not employees of Autoclenz, that they were not obliged to provide services to the company, nor was the company obliged to offer work to them, and that they could provide suitably qualified substitutes to carry out the work on their behalf, had all been inserted with the object of excluding the operation of employment legislation including the National Minimum Wage Act 1998 and the Working Time Regulations 1998. Those provisions in the agreements were therefore void.

In determining whether an individual is a “worker”, there can, as Baroness Hale said in the Bates van Winkelhof case at para 39, “be no substitute for applying the words of the statute to the facts of the individual case.” At the same time, in applying the statutory language, it is necessary both to view the facts realistically and to keep in mind the purpose of the legislation. As noted earlier, the vulnerabilities of workers which create the need for statutory protection are subordination to and dependence upon another person in relation to the work done. As also discussed, a touchstone of such subordination and dependence is (as has long been recognised in employment law) the degree of control exercised by the putative employer over the work or services performed by the individual concerned. The greater the extent of such control, the stronger the case for classifying the individual as a “worker” who is employed under a “worker’s contract”.

This approach is also consistent with the case law of the CJEU which, as noted at para 72 above, treats the essential feature of a contract between an employer and a worker as the existence of a hierarchical relationship. In a recent judgment the Grand Chamber of the CJEU has emphasised that, in determining whether such a relationship exists, it is necessary to take account of the objective situation of the individual concerned and all the circumstances of his or her work. The wording of the contractual documents, while relevant, is not conclusive. It is also necessary to have regard to how relevant obligations are performed in practice: see AFMB Ltd v Raad van bestuur van de Sociale verzekeringsbank (Case C-610/18) EU:C:2020:565; [2020] ICR 1432, paras 60-61.

Section 28(1) of the National Minimum Wage Act establishes a presumption that an individual qualifies for the national minimum wage unless the contrary is established. This is not a case, however, which turns on the burden of proof.

The claimant drivers in the present case had in some respects a substantial measure of autonomy and independence. In particular, they were free to choose when, how much and where (within the territory covered by their private hire vehicle licence) to work. In these circumstances it is not suggested on their behalf that they performed their services under what is sometimes called an “umbrella” or “overarching” contract with Uber London - in other words, a contract whereby they undertook a continuing obligation to work. The contractual arrangements between drivers and Uber London did subsist over an extended period of time. But they did not bind drivers during periods when drivers were not working: rather, they established the terms on which drivers would work for Uber London on each occasion when they chose to log on to the Uber app.

Equally, it is well established and not disputed by Uber that the fact that an individual is entirely free to work or not, and owes no contractual obligation to the person for whom the work is performed when not working, does not preclude a finding that the individual is a worker, or indeed an employee, at the times when he or she is working: see eg McMeechan v Secretary of State for Employment [1997] ICR 549; Cornwall County Council v Prater[2006] EWCA Civ 102; [2006] ICR 731. As Elias J (President) said in James v Redcats (Brands) Ltd[2007] ICR 1006, para 84:

In many cases it is not in dispute that the claimant is doing work or performing services personally for another person but there is an issue as to whether that person is to be classified as the claimant’s employer or as a client or customer of the claimant. The situation in the present case is different in that there are three parties involved: Uber, drivers and passengers. But the focus must still be on the nature of the relationship between drivers and Uber. The principal relevance of the involvement of third parties (ie passengers) is the need to consider the relative degree of control exercised by Uber and drivers respectively over the service provided to them. A particularly important consideration is who determines the price charged to the passenger. More generally, it is necessary to consider who is responsible for defining and delivering the service provided to passengers. A further and related factor is the extent to which the arrangements with passengers afford drivers the potential to market their own services and develop their own independent business.

In all these respects, the findings of the employment tribunal justified its conclusion that, although free to choose when and where they worked, at times when they are working drivers work for and under contracts with Uber (and, specifically, Uber London). Five aspects of the tribunal’s findings are worth emphasising.

First and of major importance, the remuneration paid to drivers for the work they do is fixed by Uber and the drivers have no say in it (other than by choosing when and how much to work). Unlike taxi fares, fares for private hire vehicles in London are not set by the regulator. However, for rides booked through the Uber app, it is Uber that sets the fares and drivers are not permitted to charge more than the fare calculated by the Uber app. The notional freedom to charge a passenger less than the fare set by Uber is of no possible benefit to drivers, as any discount offered would come entirely out of the driver’s pocket and the delivery of the service is organised so as to prevent a driver from establishing a relationship with a passenger that might generate future custom for the driver personally (see the fifth point, discussed below). Uber also fixes the amount of its own “service fee” which it deducts from the fares paid to drivers. Uber’s control over remuneration further extends to the right to decide in its sole discretion whether to make a full or partial refund of the fare to a passenger in response to a complaint by the passenger about the service provided by the driver (see para 20 above).

Second, the contractual terms on which drivers perform their services are dictated by Uber. Not only are drivers required to accept Uber’s standard form of written agreement but the terms on which they transport passengers are also imposed by Uber and drivers have no say in them.

Third, although drivers have the freedom to choose when and where (within the area covered by their PHV licence) to work, once a driver has logged onto the Uber app, a driver’s choice about whether to accept requests for rides is constrained by Uber. Unlike taxi drivers, PHV operators and drivers are not under any regulatory obligation to accept such requests. Uber itself retains an absolute discretion to accept or decline any request for a ride. Where a ride is offered to a driver through the Uber app, however, Uber exercises control over the acceptance of the request by the driver in two ways. One is by controlling the information provided to the driver. The fact that the driver, when informed of a request, is told the passenger’s average rating (from previous trips) allows the driver to avoid low-rated passengers who may be problematic. Notably, however, the driver is not informed of the passenger’s destination until the passenger is picked up and therefore has no opportunity to decline a booking on the basis that the driver does not wish to travel to that particular destination.

The second form of control is exercised by monitoring the driver’s rate of acceptance (and cancellation) of trip requests. As described in para 18 above, a driver whose percentage rate of acceptances falls below a level set by Uber London (or whose cancellation rate exceeds a set level) receives an escalating series of warning messages which, if performance does not improve, leads to the driver being automatically logged off the Uber app and shut out from logging back on for ten minutes. This measure was described by Uber in an internal document quoted by the employment tribunal as a “penalty”, no doubt because it has a similar economic effect to docking pay from an employee by preventing the driver from earning during the period while he is logged out of the app. Uber argues that this practice is justified because refusals or cancellations of trip requests cause delay to passengers in finding a driver and lead to customer dissatisfaction. I do not doubt this. The question, however, is not whether the system of control operated by Uber is in its commercial interests, but whether it places drivers in a position of subordination to Uber. It plainly does.

Fourth, Uber exercises a significant degree of control over the way in which drivers deliver their services. The fact that drivers provide their own car means that they have more control than would most employees over the physical equipment used to perform their work. Nevertheless, Uber vets the types of car that may be used. Moreover, the technology which is integral to the service is wholly owned and controlled by Uber and is used as a means of exercising control over drivers. Thus, when a ride is accepted, the Uber app directs the driver to the pick-up location and from there to the passenger’s destination. Although, as mentioned, it is not compulsory for a driver to follow the route indicated by the Uber app, customers may complain if a different route is chosen and the driver bears the financial risk of any deviation from the route indicated by the app which the passenger has not approved (see para 8 above).

I have already mentioned the control exercised by monitoring a driver’s acceptance and cancellation rates for trips and excluding the driver temporarily from access to the Uber app if he fails to maintain the required rates of acceptance and non-cancellation. A further potent method of control is the use of the ratings system whereby passengers are asked to rate the driver after each trip and the failure of a driver to maintain a specified average rating will result in warnings and ultimately in termination of the driver’s relationship with Uber (see paras 13 and 18 above). It is of course commonplace for digital platforms to invite customers to rate products or services. Typically, however, such ratings are merely made available as information which may assist customers in choosing which product or service to buy. Under such a system the incentive for the supplier of the product or service to gain high ratings is simply the ordinary commercial incentive of satisfying customers in the hope of attracting future business. The way in which Uber makes use of customer ratings is materially different. The ratings are not disclosed to passengers to inform their choice of driver: passengers are not offered a choice of driver with, for example, a higher price charged for the services of a driver who is more highly rated. Rather, the ratings are used by Uber purely as an internal tool for managing performance and as a basis for making termination decisions where customer feedback shows that drivers are not meeting the performance levels set by Uber. This is a classic form of subordination that is characteristic of employment relationships.

A fifth significant factor is that Uber restricts communication between passenger and driver to the minimum necessary to perform the particular trip and takes active steps to prevent drivers from establishing any relationship with a passenger capable of extending beyond an individual ride. As mentioned, when booking a ride, a passenger is not offered a choice among different drivers and their request is simply directed to the nearest driver available. Once a request is accepted, communication between driver and passenger is restricted to information relating to the ride and is channelled through the Uber app in a way that prevents either from learning the other’s contact details. Likewise, collection of fares, payment of drivers and handling of complaints are all managed by Uber in a way that is designed to avoid any direct interaction between passenger and driver. A stark instance of this is the generation of an electronic document which, although styled as an “invoice” from the driver to the passenger, is never sent to the passenger and, though available to the driver, records only the passenger’s first name and not any further details (see para 10 above). Further, drivers are specifically prohibited by Uber from exchanging contact details with a passenger or contacting a passenger after the trip ends other than to return lost property (see para 12 above).

Taking these factors together, it can be seen that the transportation service performed by drivers and offered to passengers through the Uber app is very tightly defined and controlled by Uber. Furthermore, it is designed and organised in such a way as to provide a standardised service to passengers in which drivers are perceived as substantially interchangeable and from which Uber, rather than individual drivers, obtains the benefit of customer loyalty and goodwill. From the drivers’ point of view, the same factors - in particular, the inability to offer a distinctive service or to set their own prices and Uber’s control over all aspects of their interaction with passengers - mean that they have little or no ability to improve their economic position through professional or entrepreneurial skill. In practice the only way in which they can increase their earnings is by working longer hours while constantly meeting Uber’s measures of performance.

I would add that the fact that some aspects of the way in which Uber operates its business are required in order to comply with the regulatory regime - although many features are not - cannot logically be, as Uber has sought to argue, any reason to disregard or attach less weight to those matters in determining whether drivers are workers. To the extent that forms of control exercised by Uber London are necessary in order to comply with the law, that merely tends to show that an arrangement whereby drivers contract directly with passengers and Uber London acts solely as an agent is not one that is legally available.

It is instructive to compare Uber’s method of operation and relationship with drivers with digital platforms that operate as booking agents for suppliers of, for example, hotel or other accommodation. There are some similarities. For example, a platform through which customers can book accommodation is likely to have standard written contract terms that govern its relationships with suppliers and with customers. It will typically handle the collection of payment and deduct a service fee which it fixes. It may require suppliers to comply with certain rules and standards in relation to the accommodation offered. It may handle complaints and reserve the right to determine whether a customer or supplier should compensate the other if a complaint is upheld.

Nevertheless, such platforms differ from Uber in how they operate in several fundamental ways. Notably, the accommodation offered is not a standardised product defined by the platform. Customers are offered a choice among a variety of different hotels or other types of accommodation (as the case may be), each with its own distinctive characteristics and location. Suppliers are also responsible for defining and delivering whatever level of service in terms of comfort and facilities etc they choose to offer. Apart from the service fee, it is, crucially, the supplier and not the platform which sets the price. The platform may operate a ratings system but the ratings are published in order to assist customers in choosing among different suppliers; they are not used as a system of internal performance measurement and control by the platform over suppliers. Nor does the platform restrict communication between the supplier and the customer or seek to prevent them from dealing directly with each other on a future occasion. The result of these features is that suppliers of accommodation available for booking through the platform are in competition with each other to attract business through the price and quality of the service they supply. They are properly regarded as carrying on businesses which are independent of the platform and as performing their services for the customers who purchase those services and not for the platform.

A platform of this kind was the subject of Secret Hotels2 Ltd (formerly Med Hotels Ltd) v Revenue and Customs Comrs[2014] UKSC 16; [2014] 2 All ER 685, a case on which Uber has strongly relied. In that case a company (“Med”) marketed hotel rooms and holiday accommodation through a website. One difference from the typical model described above was that Med reserved many hotel rooms in its own name, for which it paid in advance. The issue was whether, for the purposes of assessing liability for VAT in accordance with Directive 2006/112/EC (“the Principal VAT Directive”), Med was purchasing accommodation from hoteliers and supplying it to customers as a principal or whether Med fell within a category of persons who “act solely as intermediaries” to whom more favourable tax treatment applied.

The Supreme Court held that the correct legal analysis of the tripartite relationship between Med, hoteliers and customers was that Med marketed and sold hotel accommodation to customers as the agent of the hoteliers and was in these circumstances acting solely as an intermediary for VAT purposes. In analysing the relationship, Lord Neuberger of Abbotsbury (with whom the other members of the court agreed) identified the correct approach at para 34 as follows:

Secret Hotels2 was not an employment case: it concerned the classification of a relationship for VAT purposes. In applying the VAT legislation, the proper approach - established by binding decisions of the CJEU cited at paras 22-29 of the judgment in Secret Hotels2 - is informed by the policy that “taxable persons are generally free to choose the organisational structures and the form of transactions which they consider to be most appropriate for their economic activities and for the purposes of limiting their tax burdens”, albeit that this is subject to an exception for “abusive transactions”: see Revenue and Customs Comrs v RBS Deutschland Holdings GmbH (Case C-277/09) EU:C:2010:810; [2011] STC 345, para 53, cited by Lord Neuberger in Secret Hotels2 at para 24. I have already explained why a different approach is required in applying the employment legislation invoked in the present case, which is underpinned by different policy considerations.

That said, even if the relationships in the Secret Hotels2 case were viewed through the lens of employment law, I see no reason to question the analysis that Med was acting solely as a booking agent for hoteliers. For the reasons given at para 104 above, the manner in which such a platform operates is materially different from Uber’s business model. Even the practice of reserving hotel rooms in Med’s own name was, as Lord Neuberger pointed out at para 49 of the judgment, consistent with its status as an intermediary: a customer who subsequently booked one of the rooms would not contract with Med, but would contract through Med with the hotelier; and, if not all reserved rooms were booked by customers at the hotel for the season in question, the amount paid by Med was carried forward to the next season. In short, I do not consider that the decision of this court in Secret Hotels2 provides any support for Uber’s case in the present proceedings.

I mentioned earlier (at para 48 above) the reliance placed by Uber on two judicial decisions involving minicab firms which are said to have recognised in an employment context that such firms may act as booking agents for drivers who provide transportation services directly to passengers under contracts with passengers and do not work for the minicab firm. Uber contends that its own business model is similar to that of such firms, with the principal differences being only the scale of its operations and the fact that Uber uses software to take bookings rather than doing so by telephone.

The principal case relied on is Mingeley v Pennock (t/a Amber Cars)[2004] EWCA Civ 328; [2004] ICR 727, in which the claimant driver brought a claim against a private hire vehicle operator trading under the name of “Amber Cars” alleging discrimination on the ground of race. The claimant owned his own vehicle and was responsible for obtaining a PHV driver’s licence. He was one of some 225 drivers who paid a weekly fee to Amber Cars for access to what was initially a radio and later a computer system through which trip requests from customers were allocated to drivers. There was no obligation to work but, when he chose to work, the driver was obliged to wear a uniform and to apply a fixed scale of charges set by the operator. He collected and was entitled to keep the full fare paid by the customer. The operator had a procedure for dealing with complaints from passengers about the conduct of the driver and had the power to order a refund of the fare to the passenger.

The Court of Appeal affirmed the decision of the employment tribunal that it had no jurisdiction to hear the claim as the claimant was not “employed” by the operator within the meaning of section 78 of the Race Relations Act 1976. Maurice Kay LJ (with whom Sir Martin Nourse and Buxton LJ agreed) regarded it as fatal to the claim that the claimant was “free to work or not to work at his own whim or fancy” (para 14) and held that the absence of an obligation to work placed him beyond the reach of section 78. Buxton LJ gave as an additional reason that, even when working, a driver was not employed by Amber Cars “under … a contract personally to execute any work or labour” as his only such obligation was owed to the passenger.

The definition of “employment” and related expressions in section 78 of the Race Relations Act 1976 has been substantially replicated in section 83(2)(a) of the Equality Act 2010, which defines “employment” to include “employment under a contract of employment, a contract of apprenticeship or a contract personally to do work”. Although not the same as the definition of a “worker” and a “worker’s contract” in section 230(3) of the 1996 Act, the two definitions have been held to have substantially the same effect: see Pimlico Plumbers Ltd v Smith[2018] UKSC 29; [2018] ICR 1511, paras 13-15.

It is not necessary for present purposes to express any view on whether the Mingeley case was correctly decided. I do not accept, however, that the fact that the claimant in that case was free to work as and when he chose was a sufficient reason for holding that, at times when he was working, he was not employed under a contract to do work for the firm. If that conclusion was justified on the facts of the Mingeley case, it would have to be on the basis that the claimant was not to be regarded as working for the minicab firm when transporting passengers in circumstances where the firm did not receive any money in respect of any individual trip undertaken by him. This arrangement was materially different from Uber’s business model.

The other employment case involving a minicab firm on which Uber relies is Khan v Checkers Cars Ltd, an unreported decision of the Employment Appeal Tribunal handed down on 16 December 2005. The facts were that the British Airports Authority had granted the respondent firm (“Checkers”) an exclusive licence to provide a taxi service at Gatwick airport. The firm had a fleet of over 200 drivers, one of whom was the claimant, who plied for hire at the airport taxi-rank. Checkers took a commission and imposed numerous conditions on its drivers, including requiring them to charge set fares, use fixed routes and wear a uniform. Drivers were entirely free to choose whether and when they worked but they were not permitted to drive for anyone else. The issue was whether the claimant had a right not to be unfairly dismissed, which depended on showing that he had been continuously employed by Checkers under a contract of employment for a period of not less than two years: see section 108 of the Employment Rights Act 1996.

The Employment Appeal Tribunal upheld the finding of the employment tribunal that the claimant did not satisfy this requirement because the absence of any obligation to work other than when he chose was inconsistent with the conclusion that there was any contract of employment which subsisted when the claimant was not working. Langstaff J, however, also observed, obiter, at para 32 of the judgment:

In making the observation quoted above, Langstaff J was drawing an analogy with the facts of Cheng Yuen v Royal Hong Kong Golf Club[1998] ICR 131. In that case the claimant worked as a caddie for individual members of the respondent golf club. He was issued by the club with a number, a uniform and a locker. Caddying work was allocated to available caddies in strict rotation. They were not obliged to make themselves available for work and received no guarantee of work. The club was not obliged to give them work or to pay anything other than the amount of the fee per round owed by the individual golfer for whom they had caddied. On an appeal to the Privy Council the majority of the Board held that the only reasonable view of the facts found was that the claimant had not been employed under a contract of employment by the club either on a continuing basis or separately each time he agreed to act as a caddie, and that the club did no more than grant him a licence to enter into contracts with individual golfers on terms dictated by the administrative convenience of the club and its members. Lord Hoffmann, who dissented, thought that it was more realistic to regard the claimant as a casual employee of the club, particularly when (as he observed at p 139) “the claimant had to work for the person to whom he was allocated according to the club’s system at a rate of pay fixed by the club and in the manner determined by the club”. Without expressing any view on the correctness of the decision, I do not consider that this case provides a relevant analogy in determining whether the employment tribunal was entitled to find that the claimants in the present case are workers.

In Quashie v Stringfellow Restaurants Ltd[2012] EWCA Civ 1735; [2013] IRLR 99 the claimant was a lap dancer who performed for the entertainment of guests at the respondent’s clubs. An important factual finding was that the respondent was not obliged to pay the claimant any money at all. Rather, the claimant paid the respondent a fee for each night that she worked. Doing so enabled her to earn payments from the guests for whom she danced. She negotiated those payments with the guests and took the risk that on any particular night she might be out of pocket. The Court of Appeal held that on these facts the employment tribunal had been entitled to find that the claimant was not employed under a contract of employment (either for each engagement or on a continuous basis). Again, the facts were very different from those of the present case and I do not find this decision of any assistance.

It is firmly established that, where the relationship has to be determined by an investigation and evaluation of the factual circumstances in which the work is performed, the question of whether work is performed by an individual as an employee (or a worker in the extended sense) or as an independent contractor is to be regarded as a question of fact to be determined by the first level tribunal. Absent a misdirection of law, the tribunal’s finding on this question can only be impugned by an appellate court (or appeal tribunal) if it is shown that the tribunal could not reasonably have reached the conclusion under appeal: see Lee Ting Sang v Chung Chi-Keung[1990] 2 AC 374, 384-385; Clark v Oxfordshire Health Authority[1998] IRLR 125, paras 38-39; the Quashie case, para 9.

On the facts found in the present case, and in particular those which I have emphasised at paras 94-101 above, I think it clear that the employment tribunal was entitled to find that the claimant drivers were “workers” who worked for Uber London under “worker’s contracts” within the meaning of the statutory definition. Indeed, that was, in my opinion, the only conclusion which the tribunal could reasonably have reached.

It does not matter in these circumstances that certain points made by the employment tribunal in the reasons given for its decision are open to criticism, nor is it necessary to discuss such particular criticisms, since none of the errors or alleged errors affects the correctness of the tribunal’s decision. I agree with the majority of the Court of Appeal that there are some points made by the employment tribunal which are misplaced (see in particular para 93 of the Court of Appeal’s judgment). I also agree with the analysis set out at paras 96 and 97 of that judgment of the 13 considerations on which the tribunal principally based its finding that drivers work for Uber. I agree with the majority of the Court of Appeal that those considerations, viewed in the round, provided an ample basis for the tribunal’s finding.

The secondary question which arises in the light of this conclusion is: during what periods of time were the claimants working?

The Working Time Regulations 1998 and the National Minimum Wage Regulations 2015 each contain provisions for measuring working time. But before those provisions are applied, it is necessary to identify the periods during which the individual concerned is a “worker” employed under a “worker’s contract” so as to fall within the scope of the legislation.

As mentioned earlier, the employment tribunal found that a driver was “working” under such a contract during any period when he (a) had the Uber app switched on, (b) was within the territory in which he was authorised to use the app, and (c) was ready and willing to accept trips. Uber contends that it was not open to the tribunal in law to make this finding and that the tribunal should have found that the claimants (on the assumption that they were “workers” at all) were only working under workers’ contracts during periods when they were actually driving passengers to their destinations. Alternatively, Uber contends that the tribunal should have found that a claimant was only working under such a contract from the moment when (and not until) he accepted a trip request.

I think it clear - as did all the members of the Court of Appeal, including the dissenting judge, Underhill LJ, if he was wrong on the main issue - that a driver enters into, and is working under, a contract with Uber London whereby the driver undertakes to perform services for Uber London, if not before, then at the latest when he accepts a trip. If the driver afterwards cancels the trip, that signifies only that the obligation undertaken to pick up the passenger and carry the passenger to his or her destination is then terminated. It does not mean that no obligation was ever undertaken. The more difficult question is whether the employment tribunal was entitled to find - as by implication it did - that a worker’s contract came into existence at an earlier stage when a claimant driver logged onto the Uber app.

Uber argues that it is clear from the tribunal’s own findings that drivers when logged onto the Uber app are under no obligation to accept trips. They are free to ignore or decline trip requests as often as they like, subject only to the consequence that, if they repeatedly decline requests, they will be automatically logged off the Uber app and required to wait for ten minutes before they can log back on again. Furthermore, when logged onto the Uber app, drivers are at liberty to accept other work, including driving work offered through another digital platform (see para 16 above). Counsel for Uber submitted that, on these facts, a finding that a driver who switches on the Uber app undertakes a contractual obligation to work for Uber London is not rationally sustainable. Nor can the fact that the driver is ready and willing to accept trips logically alter the position so as to give rise to such an obligation.

The fact, however, that an individual has the right to turn down work is not fatal to a finding that the individual is an employee or a worker and, by the same token, does not preclude a finding that the individual is employed under a worker’s contract. What is necessary for such a finding is that there should be what has been described as “an irreducible minimum of obligation”: see Nethermere (St Neots) Ltd v Gardiner [1984] ICR 612, 623 (Stephenson LJ), approved by the House of Lords in Carmichael v National Power plc[1999] 1 WLR 2042, 2047. In other words, the existence and exercise of a right to refuse work is not critical, provided there is at least an obligation to do some amount of work.

In the present case Uber London in the Welcome Packet of material issued to new drivers referred to logging onto the Uber app as “going on duty” and instructed drivers that: “Going on duty means you are willing and able to accept trip requests” (see para 17 above). Logging onto the Uber app was thus presented by Uber London itself to drivers as undertaking an obligation to accept work if offered. The employment tribunal also found that the third in the graduated series of messages sent to a driver whose acceptance rate of trip requests fell below a prescribed level included a statement that “being online with the Uber app is an indication that you are available to take trips, in accordance with your Services Agreement.” The reference in this message to the Services Agreement must have been to clause 2.6.2, which stated:

Counsel for the third respondent suggested that this clause is inconsistent with clause 2.4 of the Services Agreement, which provided that:

Whilst the irreducible minimum of obligation on drivers to accept work was not precisely defined in the Services Agreement, the employment tribunal was entitled to conclude that it was in practice delineated by Uber’s criteria for logging drivers off the Uber app if they failed to maintain a prescribed rate of acceptances. Uber seeks to characterise this system as merely a way of seeking to ensure that drivers do not remain logged onto the app, perhaps through inadvertence whilst away from their vehicle, at times when they are not in fact available to work. However, if that were the only purpose of automatically logging off a driver, it is hard to see why the driver should then be shut out from logging back onto the Uber app for a ten-minute period. It was open to the tribunal on the evidence, including Uber’s internal documents, to find that this exclusion from access to the app was designed to operate coercively and that it was reasonably perceived by drivers, and was intended by Uber to be perceived, as a penalty for failing to comply with an obligation to accept a minimum amount of work.

It follows that the employment tribunal was, in my view, entitled to conclude that, by logging onto the Uber app in London, a claimant driver came within the definition of a “worker” by entering into a contract with Uber London whereby he undertook to perform driving services for Uber London. I do not consider that the third condition identified by the tribunal that the driver was in fact ready and willing to accept trips can properly be regarded as essential to the existence of a worker’s contract; nor indeed did the tribunal assert that it was. But it is reasonable to treat it, as the tribunal did, as a further condition which must be satisfied in order to find that a driver is “working” under such a contract.

This brings me to the question of what periods during which a driver is employed under a worker’s contract count as working time.

For the purpose of the Working Time Regulations 1998, “working time” is defined in regulation 2(1), in relation to a worker, as “any period during which he is working, at his employer’s disposal and carrying out his activity or duties”.

There is no difficulty in principle in a finding that time when a driver is “on call” falls within this definition. A number of decisions of the CJEU establish that, for the purpose of the Working Time Directive, to which the UK Regulations aim to give effect and which defines “working time” in the same way, time spent on call counts as “working time” if the worker is required to be at or near his or her place of work. For example, in Ville de Nivelles v Matzak (Case C-518/15) EU:C:2018:82; [2018] ICR 869 the CJEU held that time spent by firefighters on stand-by at their homes, which were required to be within eight minutes travelling distance of the fire station, was working time.

On the facts of the present case, a driver’s place of work is wherever his vehicle is currently located. Subject to the point I consider next, in the light of this case law the tribunal was justified in finding that all time spent by a driver working under a worker’s contract with Uber London, including time spent “on duty” logged onto the Uber app in London available to accept a trip request, is “working time” within the meaning of the Working Time Directive and Regulations.

The point that - like the majority of the Court of Appeal and Judge Eady QC in the Employment Appeal Tribunal - I have found more difficult is whether a driver logged onto the Uber app in the area in which he is licensed to work can be said to be “working, at his employer’s disposal and carrying out his activity or duties” if during such times the driver is equally ready and willing to accept a trip request received from another PHV operator. It was argued with force by counsel for Uber that a driver cannot reasonably be said to be working for and at the disposal of Uber London if, while logged onto the Uber app, he is also at the same time logged onto another app provided by a competitor of Uber which operates a similar service.

I have concluded that this question cannot be answered in the abstract. I agree with Judge Eady QC when she said in her judgment dismissing Uber’s appeal to the Employment Appeal Tribunal (at para 126) that it is a matter of fact and degree. Like the majority of the Court of Appeal, I also agree with her that:

So far as this court has been shown, no evidence was adduced at the hearing in the employment tribunal in 2016 that there was at that time any other app-based PHV transportation service operating in London or that drivers logged into the Uber app were as a matter of practical reality also able to hold themselves out as at the disposal of other PHV operators when waiting for a trip. No finding was made by the tribunal on this subject. In these circumstances I do not consider that the tribunal was wrong to find that periods during which its three conditions were met constituted “working time” for the purpose of the Working Time Regulations 1998.

What counts as working time for the purposes of the right to be paid the national minimum wage is defined by the National Minimum Wage Regulations 2015. These regulations contain complex provisions for measuring hours worked depending on how the work is classified. Before the employment tribunal there was an issue as to whether a driver’s working hours should be classified as “time work”, as Uber argued, or as “unmeasured work”, as the tribunal held. It was accepted by Uber that “time work” could only be the appropriate category if the driver is working only when carrying a passenger and not otherwise. As I have concluded that the tribunal was entitled to reject that contention, it follows that the tribunal was also entitled to find that the claimants’ working hours were not “time work”. As it was common ground that those hours did not fall within the definitions of “salaried work” or “output work”, it further follows that the tribunal was entitled to find that they constituted “unmeasured work”, which is a residual or default category. On this point too, therefore, there is no basis for interfering with the employment tribunal’s decision.

For all the reasons given, I would dismiss this appeal.

After the hearing of the appeal but before this judgment was handed down, Lord Kitchin fell ill and it was uncertain when he would return to work. With the agreement of the parties, the presiding judge, Lord Reed, gave a direction under section 43(3) of the Constitutional Reform Act 2005 that the court was still duly constituted by the remaining six Justices, all of whom are permanent judges.