Over the past three years we have reported on a series of decisions where the courts have grappled with the question of ‘who employs the agency worker?’. Are they employees of the agency or the end user or neither?

Remember the landmark Court of Appeal decision of Dacas v Brook Street Bureau (2004) in which the concept of an implied contract of employment existing between an agency worker and an end user was explored? In 2005 and 2006 decisions in Cable and Wireless plc v Muscat swung the pendulum further in favour of finding an implied contract of employment with the end user when an individual had worked under the day-to-day control of the same end user for a significant period of time.

This year has already brought us two Employment Appeal Tribunal decisions on this issue - James v Greenwich Council and Cairns v Visteon UK Limited. While the possibility of an implied contract of employment still looms, these cases do swing the pendulum back slightly away from the end user.

So, what has the EAT now said and where does this leave us?


In Dacas, the worker had been supplied via an employment agency to work as a cleaner for the council for four years. At first instance the Tribunal found that neither the agency nor the end user (the council) employed the worker. The Court of Appeal said this was “simply not credible”. It held that just because the contracts between the various parties stated that the worker was not the employee of the agency or the end user, that was not the end of the matter. A contract of employment could still be implied and this was something tribunals in future cases should consider.

Unfortunately for Mrs Dacas, as her appeal was limited to claiming she was employed by the agency, her appeal failed. The Court of Appeal’s comments on the possibility of an implied contract with the end user were therefore not strictly binding on future tribunals.

We then had Muscat, in which Mr Muscat had worked for the predecessor of the end user. In readiness for a potential buy-out he was told he must become self-employed, providing his services via a service company. The Court of Appeal held that the Tribunal had rightly looked for and found an implied contract with the end user. Importantly, the Court of Appeal in Muscat held that tribunals and the EAT are bound by the Dacas guidance regarding the possibility of an implied contract of employment between the agency worker and the end user.

What happened in James?

Mrs James was employed directly by the council until 1997. She then had a career break and once again began to work for the council through an employment agency in 2001. In 2003 she decided to change agencies to receive a better hourly rate of pay. When absent for a month due to illness, another agency worker was provided to the council. When Mrs James wished to return to the position, she was told she was no longer required.

In this case, the EAT upheld the Tribunal’s finding that Mrs James (the agency worker) although supplied to the council (the end user) for five years was not employed by the council. The EAT confirmed that it was correct to look for mutuality of obligation. In this case, the end user did not have to accept the agency worker and the agency could at any time cease to provide the worker. In particular, significant weight was placed on the fact that Mrs James had changed agencies in order to obtain a higher rate of pay and that she was simply replaced with a different agency worker when ill. In particular, the EAT noted that the mere passage of time was not sufficient to require a contract of employment to be implied between the agency worker and the end user.

The EAT also gave the following general guidance on the circumstances in which tribunals might properly infer an implied contract of employment between an agency worker and the end user applying the Dacas' principle:

1. The issue is whether the way in which the contract is in fact performed is consistent with the agency arrangement or whether it is only consistent with an implied contract between the worker and the end user.

2. In a genuine agency relationship, the end user is not paying directly for work done by the worker, but for the services supplied by the agency in accordance with its specification. The money paid by the end user to the agency does not only represent wages but includes other elements such as expenses and profit.

3. The key feature is that the end user cannot insist on the agency providing a particular worker.

4. A genuine agency relationship may subsequently change, so that the agency arrangements no longer dictate or adequately reflect how the work is actually being performed and the reality of the relationship is only consistent with the implication of the contract. In such cases, it will be necessary to show that the worker is no longer working pursuant to the agency arrangements but, as a result of mutual obligations binding the worker and end user.

5. The mere passage of time does not justify an implied contract being made as a matter of necessity.

6. It would be more readily open to a tribunal to imply a contract in a case like Muscat where the agency arrangements were superimposed upon an existing contractual relationship, the arrangements in effect being a sham.

In essence, the EAT is saying that where the arrangements between the agency, worker and end users are genuine and representative of the actual working relationship, there is unlikely to be any evidence which would justify the tribunal implying a contract. What would be needed would be evidence of mutual obligation existing between the worker and end user which was incompatible with an agency arrangement.

The Cairns decision

Shortly after the James decision was handed down, we then had Cairns. In this case, Mrs Cairns worked for Visteon directly between 1998 and 2001. However, in 2001 she became employed by an employment agency, MSX, which provided her services to Visteon. Following a dispute in relation to timesheets, Visteon told MSX that it no longer required Mrs Cairns’ services. As MSX were unable to find alternative work for her, she was made redundant by MSX. In this case, although Mrs Cairns had an employment contract with MSX, she tried to claim that she had a parallel implied employment contract with Visteon.

The Tribunal and EAT rejected her claim stating that to imply a contract by conduct, it was necessary to show that the conduct of the parties was consistent only with there being a contract of service between them. On the facts of this case, the conduct was equally consistent with Mrs Cairns’ services being supplied via an agency. This case considered the possibility of an agency worker having parallel contracts, one being with the agency and the other with the end user in respect of the same work. Although the EAT’s decision did not categorically rule out the possibility of parallel contracts in all circumstances, it did suggest that it would be hard for employees to overturn the established principle that a servant cannot have two masters in respect of the same job. Essentially, where an agency worker had an employment contract with the agency, it is unlikely that an implied contract with the end user will be established.

So where does this leave us?

The basic principle of Dacas and Muscat, still stands. So tribunals should always investigate whether an implied contract of employment has arisen between the agency, worker and the end user. However, James appears to limit the potential impact of Dacas and Muscat in stating that the mere passage of time would not of itself be sufficient to establish an implied contract where the arrangement is genuine and accurately represents the actual relationships between the parties.

James stresses the importance of a consideration of mutuality of obligation. The importance of mutuality of obligation will often leave an agency worker in a difficult position as often an agency is not obliged to provide a particular worker nor is the end user obliged to accept a particular worker put forward. Furthermore, simply because a worker does accept work for a period of time does not mean they are obliged to accept it in future nor is the end user obliged to provide the work indefinitely. Following Cairns, where the agency worker has a contract of employment with the agency, it is highly unlikely that an implied contract of employment will be found with the end user.

Both James and Cairns, send out a signal that the question of whether a contract of employment will be implied between an agency worker and the end user will ultimately come down to whether the implication of such a contract is necessary to give effect to the business reality of the arrangement. This may come as a relief to businesses worried they may have unwittingly increased their headcount following Dacas. However, there still remains a risk that an agency worker who works for the same end user for a significant period of time, with day to day control being exercised by the end user, may be able to establish that they have an implied contract of employment with the end user.

What can end users do to limit risk?

  • Keep assignments as short as possible. Restricting the length of an assignment will not in itself avoid a finding that an agency worker is employed by the end user. However, it will at the very least limit the extent to which the individual will accrue employment rights based on length of service.
  • Exert minimum control - to the extent possible. Day to day control and control in relation to matters such as disciplinary and grievance issues, should rest with the employment agency.
  • Minimise integration into the end-user’s workforce. Obvious indicators of integration (such as uniforms, entitlement to similar benefits, inclusion in organisational charts, business cards and identity badges) should be avoided. More generally, companies should aim to create a clear distinction in the workforce between employees and agency workers.
  • • Robust documentation - All contracts and other documentation should support the intentions of the parties as regards employment status (which, in the end user’s case, will usually involve, as a minimum, ensuring that the agency employs the worker in question and that there are express statements to the effect that the worker is not an employee of the end user). The Conduct of Employment Agencies and Employment Businesses Regulations 2003 are intended to make it clearer on what basis the agency is contracting. Case law has, however, made it clear that the labels that parties attach to the relationship are by no means conclusive.
  • • Indemnification - An end user could seek indemnities in the contract with the employment agency against any claims by the agency workers that they have become employed by and accrued employment rights against the end user. Commercially, this may be difficult to negotiate, particularly given the current state of the law, so would need to be considered on a case by case basis. Also, any agreement to indemnify an end user would no doubt be reflected in the agency’s charges.

End of the saga?

Unlikely. The EAT in James repeated observations made by courts on previous occasions that many agency workers are left vulnerable and in need of protection. However, like other tribunals before them, the EAT sees this as an issue to be resolved by legislation review not case law. The legislative answer is unlikely to come from the current Private Members Bill which has little prospect of proceeding through Parliament. However, legislation may be forced from Europe: the draft Temporary Workers Directive currently placed on the back burner in Brussels may yet be revived