Mr Szilagyi worked for builders Protectacoat. He claimed unfair dismissal. Protectacoat, pointing to a written "partnership agreement", argued that he was a partner, not an employee, and so could not claim unfair dismissal. Mr Szilagyi argued that the practical arrangements (such as Protectacoat providing his van and tools, and their general degree of control over him) showed he was in reality an employee. Protectacoat's publicity material said that they did not employ sub-contractors. He said he was not free to do other work.
An employment tribunal held that the "partnership agreement" was a "sham" and that Mr Szilagyi was an employee.
The Court of Appeal has clarified the correct test for establishing a 'sham' contract. From the leading judgment by Lady Smith, the following principles emerge:
- the question is always what the true legal relationship is between the parties. If there is a contractual document, that is ordinarily where the answer is to be found. But, if it is asserted that the document does not represent or describe the true relationship, the court or tribunal has to decide, on all the evidence, what the true relationship is;
- the court or tribunal has to consider whether or not the words of the written contract represent the true intentions or expectations of the parties, not only at the inception of the contract but, if appropriate, as time goes by;
- commenting on Consistent Group Ltd v Kalwak&Ors  EWCA Civ 430 (29 April 2008), a document which can be shown to be a sham "designed to deceive others" will be wholly disregarded in deciding what is the true relationship between the parties, but it is not only in such a case that its contents cease to be definitive.
Sedley LJ summed up the position most concisely, saying "...in the field of employment at least, it is more helpful and relevant..to ask in a case like this not whether the written agreement is a sham but simply what the true legal relationship is. Although there will be in many cases (as there was in this one) an intention to conceal or misrepresent the actual relationship, there is no logical reason why this should be a universal requirement.