United Kingdom Employment Appeal Tribunal

APPEARANCES

 

For the Appellant	MR JOHN CAVANAGH QC (One of Her Majesty's Counsel) and MR PAUL CAPE (of Counsel) Instructed by: Messrs Eversheds LLP Solicitors Central Square South Orchard Street Newcastle upon Tyne NE1 3XX
For the first Respondents	MR STEFAN CROSS (Solicitor) Instructed by: Messrs Stefan Cross Solicitors St Mary's Business Centre Oystershell Lane Newcastle upon Tyne NE4 5QS
For the second Respondent	MS JANE WOODWARK (of Counsel) Instructed by: Messrs Thompsons Solicitors St. Nicholas' Building St. Nicholas Street Newcastle upon Tyne NE1 1TH


 


SUMMARY

Sex Discrimination

Equal Pay – Like work

1. Female train drivers made a claim under S1 of the Equal Pay Act 1970. The Claimants were in a group known as 'Metro Operators' and claimed parity of pay for like work with another group of train drivers. The majority of both groups were overwhelmingly male. The claims were based on indirect discrimination. It was said that because the proportion of those women in both groups together who were in the disadvantaged group was higher than the corresponding proportion of men, the pay differential had a disparate adverse effect on women and was thus 'tainted by sex'. Accordingly the Respondent was bound to provide objective justification for the disparity under S1(3). The Employment Tribunal accepted this case, although it could find no provision criterion or practice that led to the disparity. It went on to reject the Respondent's defence of objective justification.
2. The EAT allowed the appeal on several grounds. The principal ground was that the ET had misdirected itself as a matter of law in finding that there was a prima facie case that the pay disparity was tainted by sex regardless of the fact that the overwhelming majority of those in the disadvantaged group were male. The EAT held that in an equal pay claim by women based upon indirect discrimination, in the absence of some provision criterion or practice that might lead to a disparate impact on women, it was necessary for there to be at least a bare majority of women in the disadvantaged group. Even if a bare majority was not required the proportion of women in the disadvantaged group had to be substantial and approaching a majority; a percentage as in the present case of 15% or 8%, depending on how one constructed the pool, was quite insufficient.
3. Secondly the EAT held that the ET fell into error in constructing a pool of the disadvantaged that included women who were not found to be doing equal work with the comparators in the advantaged pool. The disadvantaged pool should only comprise those employees who were found to be doing like work with the better paid comparators.
4. Thirdly the EAT found that the ET had failed to have regard to material evidence which pointed conclusively in favour of objective justification for the pay disparity.

HIS HONOUR JUDGE SEROTA QC

Introduction

5. This case concerns an equal pay claim by female Metro Operators (known as "MOPs") who worked for the Respondents on the Tyne and Wear Metro. This is an appeal by the Respondent from the decision of the Employment Tribunal at Newcastle, entered onto the Register on 5 September 2005 (N.W. Garside Esq., Chairman). The hearing lasted 11 days. We shall refer to the Appellants as the Respondents, and the Respondents to this appeal as "the Claimants".
6. The decision was deferred until the Court of Appeal had delivered judgment in the case of Bailey v Home Office [2005] IRLR 369.
7. The Employment Tribunal concluded that the six Respondents were entitled to a decision that their contracts of employment should be treated as modified so as to be not less favourable than male comparators employed as drivers. They were entitled to compensation for unauthorised deduction from their wages.
8. The Employment Tribunal dismissed the claims of six other Claimants (there is no appeal against this decision) and found there had been no unauthorised deduction from their wages. The Employment Tribunal dismissed the claims of five of the Claimants, who are the First Respondents to this appeal, that they had been discriminated against on the grounds of their sex.
9. On 4 November 2005 the then President, Burton J, referred this appeal to a full hearing.

Factual background

10. We do not need to consider the factual background insofar as it related to the discrimination claims. In relation to the equal pay claims, there was little factual dispute. The dispute between the parties related to the correct application of the law and inferences that might properly be drawn from facts that were not in themselves controversial.
11. The Respondent is a statutory corporation that provides and promotes public transport in the Tyne and Wear area, which comprises five local authorities. The Respondent operates a light railway known as the Tyne and Wear Metro. Historically in the 1990's the metro was funded from fares and additionally by subsidies from the five local authorities which of course, in part, were derived from council tax.
12. The metro was governed by a Passenger Transport Authority ("PTA") that comprised councillors from the five local authorities and which set the budgets for the metro.
13. The Claimants in the proceedings, including the unsuccessful Claimants, were all Metro Operators. Their duties initially comprised not only train driving but revenue control and customer services. At the time the claims were presented in September 2004, the successful Claimants were exclusively working as drivers on the roster known as the "main link". The unsuccessful Claimants worked on the roster known as the "metro link". Between the presentation of the applications and the date of hearing, all Metro Operators, whether employed on the main link or the metro link, worked exclusively as train drivers. However, in September 2004 when the claims were presented, only those employed on the main link worked exclusively as drivers, while those on the metro link were also responsible for revenue collection.
14. The successful Claimants complained that their pay was substantially less than their comparators, those persons employed as train drivers. We shall shortly explain how it came about that there were different groups of drivers and Metro Operators. It was conceded by the Respondent that those Metro Operators employed on the main link as at the date of presentation of their applications were engaged on like work to the drivers. The concession did not extend to those Claimants employed on the metro link. The Respondent's defence included the case that the pay differential was justified by reason of a genuine material factor within the meaning of s.1 (of the Equal Pay Act 1970 (as amended)). We shall in due course refer to the section. For the sake of brevity we shall on occasion adopt the familiar abbreviation "GMF".
15. The relevant chronological background starts in 1996. At this time, the activities of the metro were not covered by revenue from fares and a significant subsidy was received from the constituent local authorities. The PTA resolved that this subsidy should be phased out over a period of four years. At this point in time, the drivers of the metro trains worked only as drivers. There were separate revenue control inspectors who formed two sub-groups, a revenue protection team and customer service agents. There was no flexibility between train drivers and other categories of staff and only train drivers were allowed to drive trains.
16. The Employment Tribunal found that train drivers had been recruited from train drivers and bus drivers which were both traditionally male-dominated professions. There was, however, no restriction on female applicants seeking train driver posts, and as the date of commencement of the proceedings, there were in fact five female drivers.
17. The Respondent wished to extend the metro system to Sunderland and had indeed received promises of funding from Central Government, the European Commission and from Railtrack. It was, however, unlikely that the extension to Sunderland would proceed unless operating costs were reduced so as to remove the need for the subsidy.
18. From 1996 until 1998 there were what the Employment Tribunal described as "difficult and protracted negotiations" between the Respondent and trades unions representing the drivers. These negotiations concluded with agreements entered into on 11 August 1998 so far as concerned the train drivers and on 8 November 1998 so far as concerned the new category of Metro Operators, known familiarly as "MOPs". These agreements were approved by ballots. A major plank of the agreements was that no further drivers would be recruited and the new grade of Metro Operators would be established. The Metro Operators would both drive and collect fares. As a result of the agreement, the drivers as a category would disappear over time.
19. The drivers gave up their restrictive practices and received a significant increase in pay of some 30% which was staged. Their pay was to be protected and would increase over time as a result of the annual pay round. Metro Operators' pay was fixed at a sum greater than the drivers' previous earnings before the 30% uplift, but was less than the drivers' enhanced rate of pay.
20. The drivers were thus, in a sense, "red circled". We recognise that the Employment Tribunal at paragraph 112 did not regard the position of the drivers as having been "red circled" in the "true" sense. The Employment Tribunal was minded to define red circling as confined to those cases where a group of employees' pay was subject to temporary protection until others caught up. In the case of the metro drivers, the pay differential and indeed pay advantage to the drivers would be perpetuated. It seems to us that whether or not this is "red circling" is a sterile discussion. The term is not a defined term of art in any event.
21. As a result of the agreement, the metro extension to Sunderland went ahead; no new drivers have been recruited and persons recruited to drive trains since the agreements came into force have been the lower paid Metro Operators.
22. It was anticipated that over a 25-year period, the agreements would result in a cost saving of £5.5 million. This was apparently never achieved and it became apparent by the middle of 2000 that costs savings would not be achieved. However, for other reasons, by the beginning of 2001 a "zero operating subsidy" had been achieved; i.e. the five local authorities were no longer obliged to subsidise the metro.
23. The Employment Tribunal conducted the case against a background of a "pool" in accordance, by way of example, with what was described by Lord Walker in Rutherford v Secretary of State for Trade and Industry [2006] IRLR 551:

"It is usual to regard the pool as consisting of the aggregate of the advantaged and disadvantaged and to express the proportions of the advantaged and the disadvantaged as percentages of the whole pool."

24. It is helpful to set out the following statistics. As at September 1999 there were five female drivers and 106 male drivers. The female drivers thus constituted 4.5% of the total. Thereafter the number of female drivers has remained constant but the number of males has reduced to 93. Thus in September 2004, the percentage of female drivers was approximately 5.1%.
25. Metro Operators as a category did not, of course, exist in 1999 but in September 2003 there were 13 female Metro Operators representing 14.29% of the total number of Metro Operators and 78 male Metro Operators. In September 2004 there were 13 female Metro Operators, representing 15.48% of the total Metro Operators and 71 male Metro Operators. The evidence was that between 2002 and 2004 the number of female Metro Operators had remained constant at 13, while the number of male Metro Operators had reduced from a highest figure of 80, to 71. There was always a large male majority. Apart from the first year, when Metro Operators were introduced as a grade and the percentage of female Metro Operators was 25%, the percentage of female Metro Operators had never exceeded 16.67% and, as we have mentioned, at the time the complaints were presented they formed approximately 15% of the total group.
26. There was uncontroverted evidence (that was overlooked by the Employment Tribunal) that 57 males and 5 females were employed as Metro Operators on the main link, as opposed to the metro link. If one, therefore, were to aggregate the groups of persons employed as Metro Operators on the main link with those in the category of drivers (99 male and 5 female) 50% of the women would be disadvantaged, so as to speak. Sixty-three percent of men would be advantaged and 50% of the women. If, however, one took the figures for the disadvantaged group as including those employed both on the metro link as well as the main link, 72% of all females would be disadvantaged as opposed to 43% of all males, and 27.8% of women would be in the advantaged group and 56.7% of males.
27. At some point in time, in early 2004, after a pay dispute that had involved a decision to take strike action, it was decided to establish a revenue collection squad to be responsible for revenue collection. Revenue collection was to cease to be the responsibility of Metro Operators who, as a result, would come to do the same job as drivers. As at the date of the hearing Metro Operators employed on both metro and main link were simply engaged in driving duties, but at the time of the presentation of the applications, as we have already noted, Metro Operators employed on the metro link were still responsible for revenue collection.
28. As at the date the applications were presented, Metro Operators' pay was approximately £5,000 per annum less than that of drivers'.
29. We do not propose to recite in any detail the allegations of discrimination on the grounds of sex that failed, but we note that these did include the allegation that the widely used description of Metro Operators as "MOPs" was derogatory.
30. We also note that underlying the Claimants' (First Respondents here) case is a sense of strong dissatisfaction with the relevant trades unions who are regarded by the Claimants as being male-dominated and resistant to equal opportunities.

The decision of the Employment Tribunal

31. We have already set out the background facts, the bulk of which are from the findings of the Employment Tribunal. We note that the Employment Tribunal read the evidence of Mr I Clayton, the Respondent's head of operations. Mr Clayton was not called to give evidence and consequently was not cross-examined. Nonetheless, it is apparent that the Employment Tribunal had regard to his evidence; see paragraphs 1 and 111. We mention this because Mr Cross (who appeared for five of the successful Claimants) appeared to object to the Respondent relying upon Mr Clayton's witness statement before us.
32. We also need to note that the case for the Claimants before the Employment Tribunal was put on the basis that they were disadvantaged by reason of a provision, criterion, or practice (sometimes abbreviated to "PCP") which it was said had a disproportionate (and unfair) effect on women. We shall come on to explain the difference between this kind of indirect discrimination and that the discrimination was in fact found by the Employment Tribunal to have existed. The Employment Tribunal concluded that the disparity between the two relevant groups was tainted by sex. The two groups were the advantaged group of drivers and the disadvantaged group of Metro Operators.
33. As we have mentioned, the advantaged group and the relevant comparators were the drivers. As a result of a concession by the Respondent, it was common ground that these Claimants, who were all Metro Operators on the main link, were engaged in like work as the drivers.
34. At the hearing, no evidence was advanced on behalf of the Metro Operators employed on the metro link, that they were doing either like work or work of equal value with the drivers. It is important to note that the parties did not agree that the drivers and Metro Operators on the metro link were doing work of equal value. This point was never determined by the Employment Tribunal and is, we believe, now subject to an application for review.
35. The Respondent says that the reason the Claimant did not adduce evidence to support its case as to the work carried out by Metro Operators on the metro link and the drivers being like work, was because Mr Cross had misinterpreted the nature of the concession. Our attention was drawn to paragraphs three and four of Mr Cross's preliminary skeleton argument for his clients in which he firstly sought to argue that as at the date of the hearing, Claimants in both groups only drove trains; their jobs were identical to the drivers and that it was admitted that "all the Claimants are employed on the like work". That was not in fact the case. In any event, no evidence was led as to why it was said Metro Operators on the metro link were doing the same work as the drivers; see Employment Tribunal decision paragraphs five and 54 for the limitation on the Respondent's concession.
36. Further, in his final submissions to the Employment Tribunal, Mr Cross submitted that this was a "like work" case, falling within s.1(2)(a) of the Equal Pay Act, rather than an "equal value" case falling within s.1(2)(c).
37. We have already mentioned that Mr Cross relied upon the argument that as at the date of the hearing all Metro Operators had become exclusively drivers. He sought to argue, therefore, that he could retrospectively so as to speak, rely upon that position for the purposes of his case to establish that all Metro Operators, whether employed on the main link or the metro link, were engaged in like work as the drivers. This was of obvious relevance to the construction of the relevant pool. This submission, however, was clearly rejected by the Employment Tribunal at paragraph 105:

"We have to examine the situation as we find it at the date of the presentation of the claim forms. We have to consider those metro operators who were working on the main link. Not all of the claimants were working on the main link at the time of the presentation of their claim forms. All metro operators are now, in fact, working on the main link. This is an entirely different situation than previously when the metro operators had mixed duties; driving trains and revenue collection. We have not been presented with any evidence whether the work of the metro operators, when they worked their mixed duties, was work of equal value. Reliance has been put on the concession by the respondent that those who work on the main link are, in effect, exclusively driving trains."

38. The Employment Tribunal noted that there were proportionally more women among the Metro Operators than among the comparator drivers. The Employment Tribunal set out its findings as to the numbers and percentages of men and women in the respective groups; there may have been some dispute about these but the findings of the Employment Tribunal are set out at paragraphs 17 and 18. The Employment Tribunal was satisfied that the figure put forward by the Respondents were extracted from their records.
39. The Employment Tribunal went on to consider allegations relating to sex discrimination, and as we have noted, concluded that the use of the term "MOPs" was not discriminatory. We say nothing further about issues of discrimination.
40. The Employment Tribunal, having set out the parties' submissions, directed itself by reference to the law and in particular article 141 of the EC Treaty and the burden of proof Directive 97/80/EC to which we will refer later. It also made reference to s.1 of the Equal Pay Act, to which again, we shall refer shortly.
41. It then went on to consider a number of authorities and cited the judgment of Cox J in Ministry of Defence v Armstrong [2004] IRLR that:

"The complex and labyrinthine nature of equal pay law is well known. The existence of authorities which cannot always be easily reconciled, is, regrettably, confusing and unhelpful. In equal pay cases there is a need not to lose sight of the essential purpose of the legislation. In considering s.1(3) the fundamental question is whether the material cause of the pay difference between the applicant and her comparator is tainted by sex related factors. If it is, then the defence fails."

The Employment Tribunal also cited Cox J at paragraph 33:

"…In Strathclyde Regional Council & others v Wallace and others [1998] IRLR 146 it was held that the object of s.1 of the 1970 Act was to eliminate sex discrimination in pay, not to achieve fair wages. Accordingly, if a difference in pay between a female worker and her male comparator was explained by genuine factors which were not tainted by any sex discrimination that was sufficient to raise a valid defence under s.1(3). There was, in those circumstances, no further burden on the employer to justify anything."

42. The Employment Tribunal went on to direct itself as to the distinction between direct and indirect discrimination and that the concept of indirect discrimination under the Equal Pay Act when read together with European equal pay legislation and the case law was broader than that applicable under the Sex Discrimination Act 1975.
43. The Employment Tribunal went on to cite a passage from the speech of Lord Nicholls in Glasgow City Council v Marshall [2000] IRLR 272. We shall refer shortly to Lord Nicholls' speech. The Employment Tribunal then directed itself by reference to the decision of the European Court of Justice in Enderby v Frenchay Health Authority [1994] ICR 192 as to the approach to be adopted under s.1(3) of the Equal Pay Act to determine the question whether the rationale for a disparity in pay was or was not caused by a difference in sex. The explanation for the disparity might be directly or indirectly sexually discriminatory. The type of discriminatory pay differential being caused in such cases by some PCP had a disproportionate effect on the disadvantaged sex. The Employment Tribunal went on to consider a different approach to this issue, to be derived from the decision of the European Court of Justice, in R v Secretary of State for Employment ex parte v Seymour-Smith [1999] IRLR 253 in which a somewhat different approach had been taken as to the establishing whether the reason for the difference in pay was tainted by sex. We quote what the Employment Tribunal had to say at paragraph 84:

"84 . The second approach for establishing the third factor is that approved by the Court of Justice in Seymour Smith. The starting point is to look at the percentage of those who do qualify and to ask, do those figures cover enough individuals, are they purely fortuitous or significant or are the figures "relevant and sufficient" for the purposes of resolving the case before the tribunal? If these figures are not sufficient or significant, it maybe necessary to look elsewhere, including the figures of those who do not qualify. The relevant statistical comparison involves:

(1) taking as the pool 'the workforce' to whom the requirement applies.

(2) ascertaining the proportion of men in the workforce who are advantaged by being able to meet the requirement, and the proportion of men who are excluded from the right who are disadvantaged by being unable to meet the requirement and are disadvantaged.

(3) ascertained [sic] the proportion of women in the workforce who are advantaged by being able to meet the requirement, and the proportion of women who are excluded by not being able to meet the requirement and disadvantaged.

(4) Compare the result for men with those for women in order to see the percentage (not the numbers) of men in the workforce who are disadvantaged is considerably smaller than the percentage of women who are disadvantaged. The primary focus is on the proportions of men and women who can comply with the requirement, of the dispute requirement. If there is a considerable disparity of impact there is a need for objective justification."

44. The Employment Tribunal went on to consider the Court of Appeal decision in Bailey [2005] IRLR 369. At paragraph 94 it directed itself as to the questions it needed to pose and answer:

"94. After consideration of the Act and the cases it appears to us that we have to consider various stages:-

1 Are the complainants doing like work, work rated as equivalent or work of equal value to that of their male comparators.

2 If it is proved, or conceded, that the complainants are doing like work, work rated as equivalent or work of equal value to that of their comparators, and there is a variation in pay between the complainants' contracts and the comparators' contracts, it is presumed to be due to the difference of sex.

3 The respondent can rebut the presumption by showing that there is a genuine material factor which explains or is the reason for the variation between the complainants' contract and the comparators' contract and which is not tainted by sex. To do so the respondent must satisfy the Tribunal that the factor he relies on, as an explanation or reason for the difference:-

(a) is not a sham or pretence, but is genuine

(b) the variation between the women's contract and the men's contract is due to that factor; it must be causally relevant to the variation in pay. The factor must be material, that is, a significant and relevant factor.

(c) The factor put forward is not a difference of sex; the difference in sex can be direct or indirect sex discrimination.

(d) The factor relied on is, or in the case of equal value (s.1(2)(c)) may be, a material difference, that is, a significant and relevant difference between the complainants' contract and the comparators' contract.

4 The respondent who satisfies the Tribunal that the factor relied on is not due to a difference in sex does not have to show that the difference in pay between the complainants' contract and the comparators' contract is for a good reason and is not required to justify the differential in pay.

5 If there is evidence of sex discrimination the respondent is required to objectively justify the differential in pay.

6 Indirect discrimination can arise, but is not limited to, situations where the rate of pay is determined.

(a) by characteristics which are gender based e.g., part-time work.

(b) By factors, not being gender based, where the disadvantaged group is exclusively or significantly female or that a significant majority of the disadvantaged group are women.

(c) by the application of a provision, criterion or practice applied equally to men and women but is such that a substantially large number of women than men suffer a detriment.

7 To objectively justify the difference in pay the respondent must show that the means chosen to achieve that objective serve a real need on the part of the employer, are appropriate with a view to achieving the objective in question and are necessary to that end."

45. It is apparent that the formulation in paragraph 94 is significantly taken from the speech of Lord Nicholls in Glasgow City Council v Marshall.
46. The Employment Tribunal then went on to find that Metro Operators on the main link were doing like work with the driver comparators and that there was a variation of pay between them of some £5,000 per annum in favour of the drivers. This of course was common ground. It then went on to say that accordingly:

"the presumption applies that the difference in pay is due to the difference of sex."

Accordingly, the presumption could be rebutted by the Respondent bringing itself within s.1(3)(a) of the Act by proving a genuine material factor. The Employment Tribunal then set out the Respondent's general material factor defence, which related to the circumstances in which the two 1998 agreements were entered into. The Employment Tribunal then had this to say:

"97 The factor relied on is, so it seems to us, that for financial and economic reasons the respondent entered into two agreements. The agreements created a two category workforce capable of driving trains, the existing train drivers and the newly formed metro operators. The train drivers' agreement was to remove restrictive practices and introduce more flexibility. The metro operators agreement was to create a flexible workforce capable of train driving and revenue collection. The metro operators were to be paid less than train drivers.

98 The train driver category was closed, there were to be no more train drivers appointed and the number of train drivers would, by effluxion of time, cease to exist.

99 The factor, which Mr Parker put forward in his witness statement at paragraph 23, is that operational staff were not being efficiently utilised and the business needed more flexibility and to achieve that a new flexible category of staff was needed. We find that factor, at the date the agreements were entered into, was a material and significant factor and, caused the pay differential."

47. The Employment Tribunal found that there was no sham or pretence in the creation of the two groups; see paragraphs 101 and 102.
48. It went on to consider "red circling" of the drivers and found that the drivers' group was effectively closed.
49. We have already referred to paragraph 105 as to the relevant date to determine whether the Claimants were doing like work with the drivers. At paragraph 106 the Employment Tribunal stated that it did not have the exact figure for Metro Operators, at the time that the claim forms were presented, who worked on the main link. It also did not consider that the train driver group was, properly speaking, "red circled" because the pay of the drivers' group was protected over an extended period. The Employment Tribunal considered that "red circling" was confined to those cases where pay was protected until the disadvantaged group caught up. We have already expressed our views as to the relevance of the term "red circling".
50. The Employment Tribunal then went on to make this significant finding at paragraph 107:

"107 The exclusion of the metro operators from the train drivers' group is not based on the characteristics of gender. It applies equally to men and women. The metro operators in the disadvantaged group (the metro operators) is not significantly female, it is significantly male. Entry to the advantaged group is dependent upon an historical fact of being a train driver prior to the train drivers' agreement. It applies equally to men as it does to women within the metro operator groups, that group being substantially comprised of men."

51. The Employment Tribunal then went on to consider the relevant percentages of men and women within a pool that consisted of all Metro Operators (whether employed on the main or metro link) and the drivers' group. Those statistics showed that overall a greater proportion of women than men was in the disadvantaged group.
52. The crucial paragraph is at paragraph 109:

"109 We must consider whether the rate of pay is determined by sex, if it is the respondent must justify the disparity. In this case the pay difference is not based on characteristics which are gender based. There is no part time working. Both male and female can be and are metro operators. The difference in pay is equally applicable to male and female metro operators. There being no gender based factors we must then look to see if the disadvantaged group has a significant majority of female employees. The metro operators, as a group, have a significant majority who are male employees, in September 2003, 78 males and in September 2004 71 males compared with, in both years 13 female employees. Is a provision criteria or practice (PCP) applied equally to men and women, but is such that a substantially larger number of women than men face a detriment? The PCP has to be, that to be a member of the advantaged group of train drivers the employee must have been a train driver before that group became protected. That is the particular argument in Home Office v Bailey. But, the Court of Appeal in Bailey v Home Office decided that it was not necessary for there to be a PCP if the statistical evidence shows that a sufficient prima facie case of pay discrimination arises requiring the employer to objectively justify any material difference. It is necessary to look at both the proportions of each gender in the disadvantaged group. The proportion in this case is 2:1. That is sufficient for objective justification to be required."

53. Paragraph 109 of the decision is not altogether easy to follow but it is clear that the Employment Tribunal concluded that because a higher proportion of women than men were in the disadvantaged group, a prima facie case of taint by sex had been made out so that the Respondent was obliged to provide objective justification under s.1(3) of the Act if its defence of the claim was to succeed. This was so, even though the Employment Tribunal had concluded that:

"the pay difference is not based on characteristics which are gender based."

54. There was, therefore, no PCP. The Respondent's defence failed thus far, solely because of the relatively higher proportion of women then men in the disadvantaged group, based upon the total number of men and women in both groups combined.
55. The Employment Tribunal then went on to reject the Respondent's case on objective justification. The Employment Tribunal noted that projected savings had not been achieved and that the concessions as to flexibility as described by Mr Clayton "did not appear to be major". Further, certain concessions in relation to collection of revenue had been abandoned. The pay disparity favoured mostly men. Although there had been a need to eliminate subsidy that had long been reduced by other means, and it was "clear that the savings made by the two agreements was limited if not illusory." In the circumstances, the Employment Tribunal concluded that as the need to reduce subsidy was not there, the continuation of pay disparity, including sex discrimination, could no longer be justified. Flexibility had been eroded over time, and had not produced significant savings. It accordingly found in favour of the Claimants because the Respondent had failed to objectively justify the pay disparity as at the date of the presentation of the originating applications. The successful Claimants were therefore entitled to an equality clause.
56. However, those not on the main link were not entitled to the equality clause. The Employment Tribunal continued, at paragraph 117:

"117 We should add that this case has been presented in a particular way. There was a partial concession, by the respondent, that those on the main link at the time of the presentation of their claims were, subject to the genuine material factor defence, doing like work and entitled to the equality clause. No evidence as to those not on the main link either doing like work or work of equal value has been presented to us. No evidence on what the male metro operators were doing has been presented to us. The statistics put forward by the respondent deal with the train drivers as one group and the metro operators as another group. There has been no breakdown of metro operators group who were just train driving and those doing train driving duties and revenue collection throughout the period that we have considered."

As we shall explain shortly, the Employment Tribunal appears to have overlooked the evidence in this regard.

The notice of appeal

57. We now turn to the notice of appeal. In essence, the notice of appeal raises three principal issues. These may be summarised as follows by reference to Mr Cavanagh's skeleton argument:

"6 (1) If a Claimant shows that there has been prima facie indirect discrimination between the Claimants and the comparators, it is not sufficient that the respondent employer shows that there is a difference between the claimant's and comparator's cases which explains the differential in pay: it is necessary for the respondent to go further and to prove that the difference is objectively justified."

The Employment Tribunal is criticised for finding that objective justification was required for a number of reasons, principally, that the only relevant consideration was the proportion of each gender in the disadvantaged group among all of those in the advantaged and disadvantaged groups forming a single pool. This issue is referred to as "the disparate impact issue":

"6 (2) …the Tribunal erred in law in the identification of the relevant pools for the purposes of the disparate impact issue ("the correct pool issue"). In particular, the ET included Metro Operators within the disadvantaged pool who did not carry out exclusively driving duties and in respect of whom there was no finding or concession that they were carrying out like work with the Driver comparators; and

(3) Alternatively, if the ET was correct in law to look to [the Respondent] for objective justification, it erred in law, or reached a conclusion that no reasonable employment tribunal, properly directing itself in law, could have reached, in concluding that the variation in pay was not objectively justified ("the objective justification issue")."

Discussion of the relevant legal framework

58. General introduction.

We think it is helpful if we set out our views of the relevant legal framework before going on to consider the grounds of appeal and the arguments in support and in opposition. The starting point has to be Article 141 of the European Community Treaty:

"Article 141 (ex Article 119)

1. Each Member State shall ensure that the principle of equal pay for male and female workers for equal work or work of equal value is applied.

2. For the purpose of this Article, 'pay' means the ordinary basic or minimum wage or salary and any other consideration, whether in cash or in kind, which the worker receives directly or indirectly, in respect of his employment, from his employer.

Equal pay without discrimination based on sex means—

(a) that pay for the same work at piece rates shall be calculated on the basis of the same unit of measurement;

(b) that pay for work at time rates shall be the same for the same job.

3. The Council, acting in accordance with the procedure referred to in Article 251, and after consulting the Economic and Social Committee, shall adopt measures to ensure the application of the principle of equal opportunities and equal treatment of men and women in matters of employment and occupation, including the principle of equal pay for equal work or work of equal value.

4. With a view to ensuring full equality in practice between men and women in working life the principle of equal treatment shall not prevent any Member State from maintaining or adopting measures providing for specific advantages• in order to make it easier for the under-represented sex to pursue a vocational activity or to prevent or compensate for disadvantages in professional careers."

59. We also set out the relevant parts of Council Directive 9 February 1976 76/207/EEC on the implementation of the principle of equal treatment for men and women as regards access to employment vocational training and promotion and working conditions:

"Article 1

1. The purpose of this Directive is to put into effect in the Member States the principle of equal treatment for men and women as regards access to employment, including promotion, and to vocational training and as regards working conditions and, on the conditions referred to paragraph 2, social security. This principle is hereinafter referred to as "the principle of equal treatment".

[1a Member States shall actively take into account the objective of equality between men and women when formulating and implementing laws regulations administrative provisions, policies and activities in the areas referred to in paragraph 1.]

2. With a view to ensuring the progressive implementation of the principle of equal treatment in matters of social security, the Council, acting on a proposal from the Commission, will adopt provisions defining its substance, its scope and the arrangements for its application. -

Article 2

1. For the purposes of the following provisions, the principle of equal treatment shall mean that there shall be no discrimination whatsoever on grounds of sex either directly or indirectly by reference in particular to marital or family status.

2. For the purposes of this Directive, the following definitions shall apply—

— direct discrimination: where one person is treated less favourably on grounds of sex than another is, has been or would be treated in a comparable situation;

— indirect discrimination: where an apparently neutral provision, criterion or practice would put persons of one sex at a particular disadvantage compared with persons of the other sex, unless that provision, criterion or practice is objectively justified by a legitimate aim, and the means of achieving that aim are appropriate and necessary…

…

A person's rejection of, or submission to, such conduct may not be used as a basis for a decision affecting that person.

4. An instruction to discriminate, against persons on grounds of sex shall be deemed to be discrimination within the meaning of this Directive."

60. Article 141 was transposed into our domestic law by s.1 of the Equal Pay Act 1970. The preamble to the Act is in these terms:

"An Act to prevent discrimination, as regards terms and conditions of employment between men and women."

The relevant provisions of s.1 (as amended) are as follows:

"1 Requirement of equal treatment for men and women in same employment

[(1) If the terms of a contract under which a woman is employed at an establishment in Great Britain do not include (directly or by reference to a collective agreement or otherwise) an equality clause they shall be deemed to include one.

(2) An equality clause is a provision which relates to terms (whether concerned with pay or not) of a contract under which a woman is employed (the "woman's contract"), and has the effect that—

(a) where the woman is employed on like work with a man in the same employment—

(i) if (apart from the equality clause) any term of the woman's contract is or becomes less favourable to the woman than a term of a similar kind in the contract under which that man is employed, that term of the woman's contract shall be treated as so modified as not to be less favourable, and

(ii) if (apart from the equality clause) at any time the woman's contract does not include a term corresponding to a term benefiting that man included in the contract under which he is employed, the woman's contract shall be treated as including such a term;

…

[(c) where a woman is employed on work which, not being work in relation to which paragraph (a) or (b) above applies, is, in terms of the demands made on her (for instance under such headings as effort, skill and decision), of equal value to that of a man in the same employment—

(i) if (apart from the equality clause) any term of the woman's contract is or becomes less favourable to the woman than a term of a similar kind in the contract under which that man is employed, that term of the woman's contract shall be treated as so modified as not to be less favourable, and(ii) if (apart from the equality clause) at any time the woman's contract does not include a term corresponding to a term benefiting that man included in the contract under which he is employed, the woman's contract shall be treated as including such a term];

…

[(3) [An equality clause falling within subsection (2)(a), (b) or (c) above shall not] operate in relation to a variation between the woman's contract and the man's contract if the employer proves that the variation is genuinely due to a material factor which is not the difference of sex and that factor—

in the case of an equality clause falling within subsection (2)(a) or (b) above, must be a material difference between the woman's case and the man's; and

in the case of an equality clause falling within subsection (2)(c) above, may be such a material difference.]"

61. It is not controversial that the laws relating to equal pay and sex discrimination should, so far as possible, be treated as a single code; see Bridge L J in Shields v Coomes Holding [1978] IRLR 263 and 269, Browne-Wilkinson, J in Jenkins v Kingsgate Clothing [1981] IRLR 228 and Elias, J in Villalba v Merrill Lynch [2006] IRLR 437 at paragraphs 103 and 105.
62. It is also uncontroversial that the object of equal pay legislation is to prevent a difference of pay between men and women on the grounds of sex. It is not intended to achieve fair pay. We refer to the preamble to the Act itself and need not consider all the authorities, because they are extensively referred to in the recent decision of Elias J, in Villalba (supra):

"137 Third, and in our view very importantly, as Ms Rose fairly accepted, if her argument is correct then it means that a woman may recover under the equal pay legislation, read with Article 141, simply because she happens to be a woman. That fortuitous and chance fact will give her the right to a remedy in circumstances where a man would be denied one. Take a case where, for example, 100 men are employed in job A at £x and 100 men in job B at £x+1O% for historical reasons which could not be objectively justified. The men in the lower paid job have no grounds for complaint under discrimination law. Put a woman into a lower grade job, however, and, says Ms Rose, she can require the employer objectively to justify the difference in pay. Since he cannot do that, she must receive the higher pay. It cannot conceivably be said that her lower pay has anything whatsoever to do with discrimination on grounds of sex, yet she will be able to bring a claim because she happens to be a woman, in circumstances where a man could not do so. Once her claim has succeeded, of course, the men in job A can thereafter compare themselves with her so as to lead to an improvement in their pay also. The overall result may be a fairer wage, but the root by which it is achieved is the fact that a woman can rely upon the fortuitous and arbitrary fact of her sex as a basis for making the claim. At the heart of anti-discrimination law is the notion that it is wholly obnoxious to permit one sex to obtain benefits or advantages by the mere fact of their sex. Yet on this analysis the law has done just that; it is permitting the chance fact that the new recruit is a woman to determine legal rights. Of course the principle could directly benefit the men. If in the example we have given a woman is employed in job B then all the men in job A can immediately claim equal pay with her since there is no objective justification. Yet the notion that there has been sex discrimination is absurd. We think it would be surprising if any court were willingly to adopt an approach so lacking in principle."

We also refer to Elias J's summary at paragraph 183

"183 Summary

We entirely agree with Mr Linden that Ms Rose's argument on this point changes totally the concept of discrimination. The elimination of discrimination on grounds of sex cannot in our view begin to require objective justification for differences in pay in circumstances where the employer has satisfactorily rebutted direct sex discrimination and there is no independent evidence of any kind to show that sex has had any influence on the difference in pay. It is only in the language of Lewis Carroll that such a pay differential not tainted in any way by sex could be rendered unlawful under provisions which outlaw sex discrimination. It could of course be rendered unlawful under a wider principle of fair wages which entitled the tribunals to become wage setting bodies, but that is not the law."

Categories of equal pay cases

63. There are three different circumstances in which pay arrangements may be "tainted by sex". We refer again to the judgment of Elias J, in Villalba:

"114 Following Enderby, therefore, there are three different circumstances in which pay arrangements may, to use an expression frequently found in the cases, be 'tainted by sex'.

115. First, there may be a difference in treatment which is specifically on sex grounds. A woman is paid less simply because she is a woman. That is the classic form of direct discrimination.

116 Second, there may be a difference in treatment which, whilst not specifically on grounds of sex, results from the adoption of a criterion or practice which adversely impacts on women because they are women. Typically this may be because the social role which women habitually perform makes it more difficult for them to place themselves in the category of the worker attracting the higher pay. Treating part timers less favourably is the classic example.

117 Third, where cogent, relevant and sufficiently compelling statistics demonstrate that women suffer a disparate impact when compared with men, there is an irrebuttable presumption that sex has indirectly tainted the arrangements even though it may not be possible to identify how that has occurred, and the differential needs to be objectively justified."

64. The second category originates from the decision of the European Court of Justice in Bilka-Kaufhaus GmbH v Weber von Hartz [1986] IRLR 317 and the third category, from the decisions of the European Court of Justice in Enderby and Seymour-Smith.
65. The classic way of proving indirect discrimination had been on the basis of some PCP that was in effect discriminatory to one sex or the other. For example, in the case of women a PCP that required women could only be admitted to the advantaged group on the basis for example, of height requirements, or requirements of full-time working. Enderby extended the concept of indirect discrimination to what has become referred to as the "two groups" kind of case. The different ways in which indirect discrimination may be proved are comprehensively set out by Elias J in Villalba at paragraphs 110 to 113:

"110 The classic way in which indirect discrimination is established is to identify a practice, criterion or provision which adversely impacts upon women. Jenkins, Bilka and Rinner-Kühn all fall into that category. However, the concept of indirect discrimination was extended by the decision of the European Court of Justice in Enderby v Frenchay Health Authority [1993] IRLR 591. In that case Mrs Enderby was a speech therapist employed in the National Health Service. Speech therapists were overwhelmingly female. She claimed that she was paid less well than clinical psychologists and pharmacists, who were, in different degrees, predominantly male. It was assumed that the work performed was of equal value. The industrial tribunal had found in terms that in the fixing of pay there had been no discrimination, either direct or indirect. The employers therefore contended that they had rebutted the presumption imposed upon them by Article 141. Since there was no scintilla of sex discrimination, there could be no recovery under the principle of equal treatment. The applicant was not claiming that there were any specific hurdles which created an impediment to her gaining access to either of the other professions. The pay structures of the different professions were governed by different processes of collective bargaining and the tribunal had found that they were conducted without discrimination.

111 The ECJ rejected this submission. It recognised that the case was quite unlike the standard indirect discrimination case, such as Bilka, where there has been some criterion or practice resulting in the adverse treatment of women, and said this (paragraphs 16-19):

'However, if the pay of speech therapists is significantly lower than that of pharmacists and if the former are almost exclusively women while the latter are predominantly men, there is a prima facie case of sex discrimination, at least where the two jobs in question are of equal value and the statistics describing the situation are valid.

It is for the national court to assess whether it may take into account those statistics, that is to say, whether they cover enough individuals, whether they illustrate purely fortuitous or short-term phenomena, and whether, in general, they appear to be significant.

Where there is a prima facie case of discrimination, it is for the employer to show that there are objective reasons for the difference in pay. Workers would be unable to enforce the principle of equal pay before national courts if evidence of a prima facie case of discrimination did not shift to the employer the onus of showing that the pay differential is not in fact discriminatory: see, by analogy, Handels-og Kontorfunktionaerernes Forbund I Danmark v Dansk Arbejdsgiverforening (case 109/88) [1989] IRLR 532, paragraph 13.

In these circumstances, the answer to the first question is that, where significant statistics disclose an appreciable difference in pay between two jobs of equal value, one of which is carried out almost exclusively by women and the other predominantly by men, Article 119 of the Treaty requires the employer to show that the difference is based on objectively justified factors unrelated to any discrimination on grounds of sex'

112 In subsequent cases there have been attempts to limit the effect of this decision to cases where the disparate impact is such that the disadvantaged group are 'almost exclusively' women. That was the factual situation in the Enderby case and indeed the decision of the Court did not extend wider than that, as the extract reproduced above indicates. However, that approach has been rejected and the position appears to be that a presumption of unintentional indirect sex discrimination arises even where the statistics are less stark. Precisely how less stark is still a matter of some uncertainty, as we indicate, at paragraph 123 below.

113 In effect, therefore, Enderby establishes that statistics alone may in a sufficiently powerful case create an irrebuttable presumption of prima facie indirect sex discrimination. Once there is statistically relevant and material evidence to demonstrate that a group is in fact being adversely affected on sex grounds, which will oblige the employer to justify the pay arrangements. It is not enough in those circumstances for the employer to demonstrate that the arrangements have resulted without any direct sex discrimination being practised of any kind. In effect the court is holding that there must somewhere have been some element of indirect discrimination - in other words there is an assumed indirect discrimination. The statistics demonstrate a sufficiently marked adverse impact to constitute a prima facie case requiring not merely an explanation that the difference is not caused directly by sex, but in addition an objective justification. The statistics must at least show that it is reasonable to infer that the treatment of the disadvantaged group must have resulted from some factor or combination of factors which impinge adversely on women because of their sex even though no obvious feature causing this disparate treatment can be identified, and indeed even though the employer has apparently demonstrated to the contrary."

66. The present case, on the basis of the findings by the Employment Tribunal is the third class of case referred to by Elias J, the "two groups" case. Cases involving a PCP and "two groups" cases require proof of the disparate impact caused or assumed to be caused by the employer's pay arrangements; see Villalba at paragraphs 121 and 123.
67. Reference can also be made to the decision of the European Court of Justice in Seymour-Smith in relation to proof of indirect discrimination by the aggregation of pools of advantaged and disadvantaged workers.

Burden of proof in disparate impact cases

68. Most of this subject (save with reference to more recent authorities) has been comprehensively considered by Elias, J in Villalba and we cannot improve on his analysis, so we take the matter relatively shortly.
69. The European Court of Justice in Enderby made clear that in principle, the burden of proof rested on a Claimant to establish sex discrimination. However, it recognised that in certain circumstances it was appropriate to reverse the burden of proof when a measure had an adverse impact on substantially more members of one or other sex.

13. It is normally for the person alleging facts in support of a claim to adduce proof of such facts. Thus, in principle, the burden of proving the existence of sex discrimination as to pay lies with the worker who, believing himself to be the victim of such discrimination, brings legal proceedings against his employer with a view to removing the discrimination.

14. However, it is clear from the case law of the court that the onus may shift when that is necessary to avoid depriving workers who appear to be the victims of discrimination of any effective means of enforcing the principle of equal pay. Accordingly, when a measure distinguishing between employees on the basis of their hours of work has in practice an adverse impact on substantially more members of one or other sex, that measure must be regarded as contrary to the objective pursued by Article 119 of the Treaty, unless the employer shows that it is based on objectively justified factors unrelated to any discrimination on grounds of sex…"

70. We would also refer to what Elias J said in Villalba.

"90 The equal pay appeal

The issue can be put simply. Ms Rose submits that once a woman is held to be working on work of equal value with a male comparator, the employer must show that there is an objective justification for the difference in pay, or else the claim must succeed. It is not enough simply to demonstrate that there is no discrimination on grounds of sex. Moreover, the need to establish this defence arises whether or not the employer has adopted arrangements which have an adverse or disparate impact on women as a group."

Later in his judgment, Elias J rejected the argument and had this to say at paragraph 124:

"124 What is striking about this whole jurisprudence is that prior to any uncertainties that may have been created by Brunnhofer, no one has ever thought that the mere fact that a woman is paid less than a man for work of equal value is enough to trigger the obligation objectively to justify the difference in pay. If it were enough then the legislation would be concerned with fairness rather than with sex discrimination. All the case law is premised on the basis that whilst that basic comparison is enough to raise a prima facie case of direct sex discrimination which the employer will have to rebut, if the claim is one of indirect discrimination then the claimant will have to establish a prima facie case. This may be done either from the relevant statistics or in some other way (perhaps, for example, because the criterion involved plainly discriminates against women, such as treating part timers less favourably, in which case formal statistics may not be necessary). But it is only once indirect discrimination is established that the employer has to go on to show objective justification."

We would again refer to what Elias J further had to say at para 137, which we have already set out. Elias J made clear that if the law permitted a woman to claim an equality clause with a male comparator simply because she was a woman anti-discrimination law would be stood on its head. A male doing like work would not be able to make such a claim if the comparator was male. The case would have nothing to do with sex discrimination yet, if the argument were correct the woman would succeed but not the man. Yet once the woman succeeded the man would also succeed by using the successful woman as his comparator. This might achieve a fairer wage but would do so by the route of permitting the woman to obtain a benefit simply by virtue of her gender. Elias J suggested that no court would be prepared to willingly adopt an approach so lacking in principle. We have that very much in mind.

71. The question has been considered on a number of occasions by courts in the United Kingdom. We would refer to the decision of the House of Lords in Strathclyde Regional Council & Others v Wallace [1998] IRLR 146. That was a case in which it was conceded that the pay arrangements did not manifest any discrimination on grounds of sex. There was no statistical basis for alleging that women had been adversely affected. The Employment Tribunal found that although the employer had relied upon factors which had explained the pay differentials and which were not at all related to sex, nonetheless the equal pay claim succeeded because the employer had not objectively justified the difference in treatment between the applicants and the comparators. That is the argument that was addressed to Elias J in Villalba and was in effect addressed to the Employment Tribunal in the instant case.
72. Elias J put it in this way at paragraph 127 in Villalba:

"127 It is perhaps hardly surprising that the House of Lords held that where the allegation is of unlawful treatment contravening laws which discriminate on grounds of sex, an applicant cannot expect to succeed if there is no discriminatory treatment on grounds of sex."

73. The House of Lords returned to consider the point in Glasgow City Council. The leading speech was given by Lord Nicholls. Lord Nicholls had this to say:

"I can well understand that an instructor in a special school, whether a woman or a man, may feel aggrieved that a teacher in the same school is being paid more for doing the same or broadly similar work. I have more difficulty in understanding how, in the absence of sex discrimination, this perceived unfairness is said to be caught and cured by a statute whose object, according to its preamble, is to prevent discrimination between men and women as regards terms and conditions of employment. The instructors' contention is that this conclusion follows from the clear wording of section 1. Further, they contend that this conclusion is not surprising. Proof that women are being paid less than men for like work is prima facie evidence of sex discrimination. Part of the purpose of the Act of 1970 was to ensure that discrimination does not arise through accident or inertia. If an employer fails to rebut the presumption of sex discrimination because he is unable to show a proper reason for the disparity in pay, the case falls within the mischief the Act was intended to remedy. This conclusion may go further than the provision regarding equal pay for equal work in Article 119 (now renumbered 141) of the E.C. Treaty. But there is no reason why the equality of pay legislation in a member state should be confined in its scope to that of article 119.

I am unable to agree with the main thrust of this submission or with the approach adopted by the industrial tribunal. This approach would mean that, in a case where there is no suggestion of sex discrimination, the equality clause would still operate. That would be difficult to reconcile with the gender-related elements of the statutory equality clause. The equality clause is concerned with variations in pay or conditions between a woman doing like work with a man and vice versa. But if the equality clause were to operate where no sex discrimination is involved, the statutory starting-point of a gender-based comparison would become largely meaningless. On this interpretation of the Act, what matters is not sex discrimination. What matters is whether, within one establishment, there is a variation in pay or conditions between one employee doing like work with another employee. The sex of the employees would be neither here nor there, save that to get the claim off the ground the chosen comparator must be of the opposite sex. On this interpretation the Act could be called into operation whenever mixed groups of workers are paid differently but are engaged on work of equal value. In such a case the statutory equality clause would operate even when the pay differences are demonstratively free from any taint of sex discrimination. Indeed, a notable feature of the industrial tribunal's decision in the present case is that a male instructor succeeded as well as seven female instructors. It is a curious result in a sex discrimination case that, on the same facts, claims by women and a claim by a man all succeed.

I do not believe the Act of 1970 was intended to have this effect. Nor does the statutory language compel this result. The scheme of the Act is that a rebuttable presumption of sex discrimination arises once the gender-based comparison shows that a woman, doing like work or work rated as equivalent or work of equal value to that of a man, is being paid or treated less favourably than the man. The variation between her contract and the man's contract is presumed to be due to the difference of sex. The burden passes to the employer to show that the explanation for the variation is not tainted with sex. In order to discharge this burden the employer must satisfy the tribunal on several matters. First, that the proffered explanation, or reason, is genuine, and not a sham or pretence. Second, that the less favourable treatment is due to this reason. The factors relied upon must be the cause of the disparity. In this regard, and in this sense, the factor must be a "material" factor, that is, a significant and relevant factor. Third, that the reason is not "the difference of sex." This phrase is apt to embrace any form of sex discrimination, whether direct or indirect. Fourth, that the factor relied upon is or, in a case within section l(2)(c), may be a "material" difference, that is, a significant and relevant difference, between the woman's case and the man's case.

When section 1 is thus analysed, it is apparent that an employer who satisfies the third of these requirements is under no obligation to prove a "good" reason for the pay disparity. In order to fulfil the third requirement he must prove the absence of sex discrimination, direct or indirect. If there is any evidence of sex discrimination, such as evidence that the difference in pay has a disparately adverse impact on women, the employer will be called upon to satisfy the tribunal that the difference in pay is objectively justifiable. But if the employer proves the absence of sex discrimination he is not obliged to justify the pay disparity."

This passage is also cited by Elias J in Villalba. It can be seen that Lord Nicholls' speech is the basis for the approach taken by the Employment Tribunal at paragraph 96, as his speech was explained by HHJ Clark in Kings' College v Clark [2003] EAT/1049/02. However, as Elias J noted in the Villalba, Lord Nicholls' speech was in some sense ambiguous:

"129 The House of Lords therefore held that when read in context, the material factor defence did not require the employer to demonstrate a justifiable reason for the differential in the absence of any evidence at all of sex discrimination.

130 We pause respectfully to note that there is arguably some ambiguity in the judgment of Lord Nicholls. His third requirement is that the employer should show that the reason for the difference in pay is not the difference of sex. If that is satisfied, then he does not have to prove a 'good' reason for the pay disparity (ie objective justification.) The question is, how does he prove that the reason is not a difference of sex? In our view Lord Nicholls was clearly assuming that where the claimant shows that there is the necessary relevant adverse disparate impact, that of itself makes the decision tainted by sex and so requires objective justification. We think that is plain from the passage which we have emphasised in his judgment. It seems to us that this is necessary to comply with Enderby. In other words we consider that he was intending to treat a decision as tainted by sex if it falls into any of the three kinds of discrimination which we have outlined above. In effect an employer whose pay arrangements result in an adverse disparate impact cannot demonstrate that his pay arrangements are not tainted by sex; the presumption of indirect discrimination arises and he will have to show that the arrangements are objectively justified.

131 However, the Court of Appeal has apparently taken a different view in Armstrong v The Newcastle upon Tyne NHS Trust Hospital [2006] IRLR 124. The Court in that case has read Lord Nicholls as establishing that even where there is an adverse disparate impact the employer can avoid liability simply by showing that the difference is genuinely caused by a reason which is other than sex. If he proves that, then objective justification does not arise: (see in particular the judgments of Arden LJ at paragraphs 32-33 and Buxton LJ at paragraph 110 and the questions he poses for the tribunal to answer on remission at paragraph 129.). In our respectful view this is to ignore the argument which was run and lost in Enderby which does not appear to have been referred to by the court on this point. Once the relevant disparate impact is shown, the employer must effectively satisfy the tribunal of both aspects ie that there is an objective justification which is not itself the difference of sex.

132 Mr Linden suggested that we were bound by Armstrong, which of course post dates Brunnhofer, and which we should assume had been considered by it. Plainly it is wholly at odds with Brunnhofer, but that case was not cited to it, or if it was, it was not referred to in the judgments (save indirectly in a passage reproduced in the employment tribunal's decision), and therefore it seems to us that the Armstrong decision does not relieve us of the duty to determine whether Brunnhofer did achieve the sea change which the appellant maintains that it did. We have respectfully indicated our difficulty with the judgment in Armstrong because Enderby is an important step in the appellant's argument, and we think that Lord Nicholls' observations in Marshall both can and should be read as consistent with it. Suffice it to say that if we are wrong and the law is indeed as the Court of Appeal has construed Lord Nicholls' remarks, then the change brought about by Brunnhofer would be even more profound than is otherwise the case.

74. The issue also came before the Court of Appeal in Nelson v Carillion Services [2003] ICR 1257. The leading judgment was given by Simon Brown LJ, with whom Dyson and Scott Baker LJJ agreed. Simon Brown LJ considered Lord Nicholls' speech, which we have quoted and went on to consider the effect of s.63A of the Sex Discrimination Act 1975 which provides for a reverse burden of proof, where a complainant had proved facts from which a Tribunal could conclude, in the absence of an adequate explanation, that the Respondent had committed an act of discrimination. He had this to say:

26. It seems to me tolerably clear that the effect of section 63A was to codify rather than alter the pre-existing position established by the case law. The burden of proving indirect discrimination under the 1975 Act was, as Mr Langstaff accepts, always on the complainant, and there pursuant to section 63A it remains, the complainant still having to prove facts from which the tribunal could conclude that he or she has been unlawfully discriminated against "in the absence of an adequate explanation" from the employer. Unless and until the complainant establishes that the condition in question has had a disproportionate adverse impact upon his/her sex the tribunal could not in my judgment, even without explanation from the employer, conclude that he or she has been unlawfully discriminated against.

27. This to my mind accurately reflects the position laid down by the Court of Justice in Enderby [1994] ICR 112 and that, indeed, is hardly surprising. True it is, as Mr Langstaff points out, that the Burden of Proof Directive expressly provides that 'this Directive shall not prevent member states from introducing rules of evidence which are more favourable to plaintiffs'. It is difficult to see, however, why the United Kingdom should have wished to introduce a rule more favourable to complainants (a) than had earlier been established by the domestic authorities, or (b) than the Court of Justice thought appropriate for a claim directly brought under Article 119.

28. All these considerations notwithstanding, it is the applicant's submission that when one comes to a claim brought under the 1970 Act the burden lies throughout on the respondent employer, the complainant in an indirect discrimination case having to advance no more than 'a credible suggestion' of disproportionate adverse impact.

29. This somewhat surprising conclusion, Mr Langstaff submits, is justified, indeed required, by the plain language of section 1(3) itself, a provision differently framed from anything in the 1975 Act. Section 1(3) in terms puts the burden of proof on the employer. And support for this approach, he argues, is to be found in Lord Nicholls's speech in Glasgow City Council v Marshall [2000] ICR 196 (see paragraph 9 above). He relies in particular upon Lord Nicholls's statement, at p 203, in the second of the two paragraphs I have cited, that:

'In order to fulfil the third requirement"—that the reason for the less favourable treatment is not "the difference of sex", a phrase "apt to embrace any form of sex discrimination, whether direct or indirect"—"he must prove the absence of sex discrimination, direct or indirect.'

30. The argument is in my judgment unsustainable. In the first place, it overlooks the very next sentence in Lord Nicholls's speech:

'If there is any evidence of sex discrimination, such as evidence that the difference in pay has a disparately adverse impact on women, the employer will be called upon to satisfy the tribunal that the difference in pay is objectively justifiable.'

31. This seems to me to recognise that it is for the complainant initially to establish that the matter complained of has indeed had 'a disparately adverse impact on women'. Secondly, however, it should be appreciated that Glasgow City Council v Marshall was not concerned with the burden of proof at all; that simply was not an issue in the case.

32. There seems to me every reason for approaching the burden of proof in indirect discrimination cases in the same way irrespective of whether they are brought under Article 119 (now Article 141 EC), under the 1975 Act, or under the 1970 Act. As Lord Browne-Wilkinson said in Strathclyde Regional Council v Wallace [1998] ICR 205, 212:

'The cases establish that the Equal Pay Act 1970 has to be construed so far as possible to work harmoniously both with the Sex Discrimination Act 1975 and article 119. All three sources of law are part of a code dealing with unlawful sex discrimination: see Shields v E Coomes (Holdings) Ltd [1978] ICR 1159 and Garland v British Rail Engineering Ltd [1981] ICR 420. It follows that the words "not the difference of sex" where they appear in section 1(3) of the Equal Pay Act 1970 must be construed so as to accord with the Sex Discrimination Act 1975 and article 119 of the EC Treaty, i e an employer will not be able to demonstrate that a factor is 'not the difference of sex' if the factor relied upon is sexually discriminatory whether directly or indirectly. Further a sexually discriminatory practice will not be fatal to a subsection (3) defence if the employer can "justify" it applying the test in the BilkaKaufhaus case [1987] ICR 110.'"

75. We would also refer to paragraph 36:

"36. I have in short come to the clear conclusion that in an indirect discrimination case the burden of proving disproportionate adverse impact lies on the complainant and that merely to raise 'a credible suggestion' that, were the relevant (valid and significant) statistics provided, these might establish disproportionate impact is not sufficient for the complainant's purposes and imposes no further burden of explanation upon the employer."

We also note that Simon Brown LJ made it clear at paragraph 38 that it was for the complainant to provide the necessary statistics in order to demonstrate disproportionate adverse impact.

76. In Cheshire and Wirral Partnership NHS Trust v Abbott [2006] IRLR 546, Keene LJ noted at paragraph 17 that:

"It is well established that in a case such as these [an equal pay case] the burden of proving sex discrimination lies initially on the person asserting that such discrimination exists."

77. The Claimants have sought to rely upon the decision of the Court of Appeal in Bailey. In that case, Peter Gibson LJ made clear that although he had some doubts as to the reasoning of the Court of Appeal in Nelson, it was part of the reasoning of the court and binding upon the Court of Appeal. Waller LJ, in a passage relied upon by the Claimants had this to say at paragraph 37:

"37. I have to say that I do not understand why the applicants in this case should have to accept that there was some onus on them to show 'disparate impact'. I understand that the concession was made in reliance on Nelson v Carillion Services [2003] IRLR 428. We did not explore that decision during the hearing of the appeal and it may be that the concession in the light of that authority was rightly made. But, by an applicant being compelled to take on that burden, cases, in my view, get into an unnecessary evidential tangle, because one finds oneself considering (a) is there a prima facie case that the difference in pay is due to sexual discrimination; (b) if so, what is the employer's position – is it open to the employer to prove that it was not, and can he do that without establishing what his genuine reason was? Or is the establishment of a prima facie case sufficient to place on the employer the burden of establishing a genuine reason, which is not the difference in sex?"

Sir Martin Nourse agreed with both judgments.

78. The decision of the European Court of Justice in Brunnhofer v Bank der Osterriechischen Postparkasse [2001] IRLR 571 led the Claimant in Villalba to claim that once an employee had shown she performed work of equal value (and by parity of reasoning alternatively had performed like work) to a comparator of the opposite sex, the employer was then obliged to provide objective justification. This view was rejected by HHJ Clark in Parliamentary Commissioner for Administration v Fernandez [2005] UKEAT 0573/04, HHJ Clark's decision was not followed by the EAT presided over by HHJ Ansell in Sharp v Caledonia Group Services [2006] IRLR 4. The point was, as we have already mentioned, considered by Elias J in Villalba. We have already referred to paragraphs 90, 124 and 137 of his judgment, in which he emphatically rejected the suggestion that a Claimant did not have to establish a prima facie case of disparate impact, in cases of indirect discrimination. He considered that Sharp was wrongly decided, and followed HHJ Clark's decision in Fernandez. In our opinion, we must have regard to the decision in Colchester Estates v Carlton Industries [1984] 2 All ER 601, Nourse J considered the position where there were conflicting decisions of courts of co-ordinate jurisdiction. He considered that if a decision of a court had been reached after full consideration of an earlier decision which went the other way, it should not be open to review on a third occasion when the same point arose for decision. He said:

"There must come a time when a point is normally to be treated as having been settled at first instance. I think that should be when the earlier decision has been fully considered, but not followed, in a later one. Consistently with the modern approach of the judges of this court to an earlier decision of one of their number… I would make an exception only in the case, which must be rare, where a third judge is convinced that the second was wrong in not following the first."

Villalba was clearly one of the rare cases referred to by Nourse J. In our opinion, Elias J, having considered in the greatest detail and with the greatest care, all the relevant authorities, including Sharp and Fernandez, has preferred Fernandez. Regardless of other authorities, in our opinion we are obliged to follow Elias J's decision in Villalba and we do so. We do note in any event that the facts of Sharp are not comparable to those of this case. Sharp was concerned with a case of a single Claimant claiming equality with a single comparator. There was no question of her being a member of a disadvantaged group, as the Claimants in these proceedings claim.

79. The Employment Tribunal has found, in effect, that this was a "two group" case. It accepted that the exclusion of Metro Operators from the train drivers' group was not based on any characteristics of gender, nor was the pay difference between Metro Operators and drivers based upon characteristics which were gender-based. We accept that PCP cases and "two group" cases are not self-contained categories. The distinction between one case and another may well not be clear. Indeed, it may well be said, that in a "two group" case where the statistics, if available, prove adverse disparate effect, they simply reveal that there must be some PCP which cannot readily be discerned. We would refer to what Peter Gibson LJ had to say in Bailey at paragraph 29:

"29. I am left unpersuaded by the EAT's attempt in paragraph 26 of its judgment, cited in paragraph 15 above, to identify a clear and sensible difference between a condition or requirement case on the one hand and cases where no such condition or requirement is identified. In each case the ET is concerned to determine whether what on its face is a gender-neutral practice may be disguising the fact that female employees are being disadvantaged as compared with male employees to an extent that signifies that the disparity is prima facie attributable to a difference of sex. The distinction drawn by the EAT between true 'requirement or condition' cases and 'cases involving disparity of pay which has arisen between two work groups' may not always be capable of clear definition or consistent practical application, as Ms Gill pointed out. Usually the disparity of pay between two work groups will reflect the fact that they do different work, and there may well be features of the work of the advantaged group which could be elevated to a requirement or condition. For example, where the disadvantaged group works in ordinary office or shop hours and the advantaged group does shifts or unsocial hours, it might be said that there was a requirement or condition for entry into the advantaged group of availability to work shifts or unsocial hours. The same statistics would be available whichever categorisation was used, and yet, if the Home Office was right, if it could be labelled a 'requirement or condition' case the Seymour-Smith approach could lead to a conclusion of prima facie discrimination whereas, if it were categorised as a case involving disparity of pay between two work groups, prima facie discrimination could only be found if the disadvantaged group was predominantly female (or male) and the advantaged group predominantly of the other gender. As Ms Gill submitted, the difference between a formal requirement or condition for obtaining a benefit which divides two groups of workers and a division by reference to jobs for which different amounts are paid is one of form rather than substance, and a common approach to the two types of case has the merit of ensuring that the 1970 Act is applied consistently to all forms of indirect discrimination."

80. Nevertheless, it is convenient for the purposes of discussion to distinguish between those cases where a PCP can be discerned and those cases, of which the present is one, where no PCP can be discerned, but where it is said statistics point to a disparate adverse effect which can be inferred to be tainted with sex. For the purposes of convenient shorthand, we shall continue to refer to such cases as "two groups" cases.
81. It is clear that where one has a "two group" case and the only evidence that supports an inference of disparate adverse effect comprises statistics, one needs to take considerable care when examining the statistics. That is apparent from the passage from Enderby (paragraph 17) to which we have referred, and the passages in the judgment of Elias J in Villalba at paragraphs 113 and 117 where, in particular, he stated that "cogent, relevant and compelling statistics" are required to explain in the absence of a discerned PCP why women were impeded from joining the advantaged group because of their sex.
82. It is also necessary to bear in mind that small numbers can easily lead to distortions; see, for example, Waller LJ in Bailey at paragraph 34 and Lord Nicholls in Barry v Midland Bank [1999] IRLR 36.

The numbers within the groups

83. There is authority on how many persons in the disadvantaged group need to be of one sex, when the advantaged group is predominantly composed of the other sex. The first question posed by the Court of Appeal in Enderby was as follows:

"7. In its first question, the Court of Appeal wishes to know whether the principle of equal pay for men and women requires the employer to prove, by providing objective justification, that a difference in pay between two jobs assumed to be of equal value, of which one is carried out almost exclusively by women and the other predominantly by men, does not constitute sex discrimination."

84. The European Court of Justice continued:

"16. However, if the pay of speech therapists is significantly lower than that of pharmacists and if the former are almost exclusively women while the latter are predominantly men, there is a prima facie case of sex discrimination, at least where the two jobs in question are of equal value and the statistics describing that situation are valid."

17. It is for the national court to assess whether it may take into account those statistics, that is to say, whether they cover enough individuals, whether they illustrate purely fortuitous or short term phenomena, and whether, in general, they appear to be significant."

We draw attention to the use of the term "almost exclusively" for the proportion of women in the disadvantaged group and "predominantly" in the relation to the proportion of men in the advantaged group, which would be sufficient to raise the prima facie case of sex discrimination.

85. In the Northern Ireland case of British Road Services v Loughran [1997] IRLR 92, the Northern Ireland Court of Appeal was considering an equal pay claim in which the group said to be disadvantaged comprised nine female clerical officers and seven male clerical officers. It is helpful to have regard to what MacDermott L J at paragraph 63 (having referred to Enderby):

"63. As I have already stated, the mischief at which the legislation is aimed is women being paid less than men for performing work of equal value. Where the women are members of an exclusively female group it is fair to assume that there is discrimination – an assumption could not easily be made if that group were a mixed group. It seems to me that the more women there are in the group the easier it would be to draw an assumption in their favour – conversely, if there were more men in the group it is unlikely that such an assumption could be fairly drawn – indeed it probably could not be drawn at least without convincing evidence. I would also add that the composition of a group may lead to a presumption one way or the other and in the light of relevant evidence that presumption will be revealed as sound or unsound and a final determination will be reached having regard to all the evidence."

86. Kerr L J also referred to this point. He referred to the decision of the European Court of Justice in Bilka and continued:

"101 In its judgment, the Court of Justice was less explicit than the Advocate-General in explaining its reasoning. The Court did not express disagreement with the Advocate-General, however, and concluded that, based on the percentages of women and men in either group, a prima facie case of discrimination may arise. In paragraph 16 of the judgment the Court stated:

'(16) However, if the pay of speech therapists is significantly lower than that of pharmacists and if the former are almost exclusively women while the latter are predominantly men, there is a prima facie case of sex discrimination, at least if the two jobs in question are of equal value and the statistics describing that situation are valid.'

102 It was argued on behalf of the appellant in this court that the Court of Justice here confined the circumstances where prima facie evidence of discrimination might arise solely by reference to the numbers in each group to those cases where the female group was 'almost exclusively' female and the male group was at least mainly men. I cannot accept that argument. Firstly, as I have pointed out, it was the applicant in Enderby [1993] JRLR 591 who first used the expression 'almost exclusively' to refer to the female group. I consider that the Court of Justice was merely reflecting the factual situation which obtained in that case in so stating its conclusion. Secondly, and more importantly, the relevance of the number of females in the group is as an indicator of its being traditionally a less well paid group on account of its being composed mainly of women. Logically a group comprising 75% females and 25% males has the capacity to provide such an indication. Whether it does in fact is a matter for the tribunal to decide. This was recognised by the Court of Justice in paragraph 17 of its judgment which states:

'(17) It is for the national court to assess whether it may take into account those statistics, that is to say, whether they cover enough individuals, whether they illustrate purely fortuitous or short-term phenomena, and whether they appear to be significant.'

103 It would be wholly illogical, in my opinion, to conclude that where a group consisted of 98% females the existence of separate collective agreements would not be enough to prevent a prima facie case of discrimination but it would be where women represented 75% of the group. A group which was composed of 75% women may have been, until lately entirely female. Conversely a group which is currently almost exclusively female may have recently been evenly divided between the genders. There is no sensible reason that one should concentrate solely on the bare statistics current at the time of the complaint to provide the exclusive guide to whether the female-dominated group is less well paid because they perform what is traditionally regarded as women's work.

104 A significant majority of women within a group may be as reliable an indicator of a history of unequal treatment as is a group composed virtually exclusively of women. It cannot be right that the latter is automatically sufficient to raise a prima facie case of discrimination but the former is not. I am satisfied that this was not the intention of the Court of Justice in Enderby [1993] IRLR 591. As I have already said, it is important to recall the nature of the case made by each of the parties before the Court of Justice. The applicant had argued that there was indirect discrimination unless the employer could objectively justify the difference in pay. That followed from the mere fact that a group of workers consisting almost exclusively of women was paid less than a group of workers consisting mainly of men. The health authority had contended that prima facie indirect sex discrimination could only occur where a hurdle or requirement had been imposed by the employer which women found difficult to surmount. The Court of Justice rejected the health authority's argument and accepted the case made by the applicant. In doing so it adopted the phraseology which had been used by the applicant to describe the actual percentages of men and women in the two professions. I do not consider that the Court intended to propound a principle that unless the disadvantaged group could be described as being composed 'almost exclusively' of females that a prima facie case of discrimination could not arise. On the contrary, its enjoinder to national courts to assess the statistics to determine whether they 'cover[ed] enough individuals' is a clear indication of its recognition that the percentages and the significance to be attached to them could vary from case to case."

87. The decision of the Court of Appeal of Northern Ireland was thus prepared to extend protection to a disadvantaged group where the "majority" were of the disadvantaged sex as opposed to where the disadvantaged group was "almost exclusively" of the disadvantaged sex.
88. In Bailey v Home Office [2005] IRLR 369 the Court of Appeal considered a case in which there were 189 women in the disadvantaged group and 184 men. Peter Gibson L J had this to say at paragraph 30:

"30. I can see no justification for the imposition of a high threshold for satisfying the test of prima facie discrimination. Where, as here, there is one group of employees of an employer which contains a significant number, even though not a clear majority, of female workers whose work is evaluated as equal to that of another group of employees of the employer who are predominantly male and who receive greater pay, it would be very surprising if an ET were to be precluded by the presence in the disadvantaged group of a significant number of men from holding that that disparity in favour of men required justification by the employer. In the present case it may well be that, as the Home Office suggests, there is a genuine material factor which is not the difference of sex and which justifies that disparity. Whether there is such factor is for further determination."

89. Mr Cavanagh submitted that the reference by Peter Gibson L J to a "clear majority" must be understood in the context of the factual basis of that case, set out at paragraph 7 of the decision. The majority of women in the disadvantaged group was 191 to 189 and Mr Cavanagh submitted that Peter Gibson L J was speaking of a "bare majority" as opposed to a "substantial majority" which might be sufficient to raise a prima facie case of discrimination. We believe Mr Cavanagh to be correct in this submission but in any event, in the light of the authorities that he cited to us, it seems to us that a "significant number" must still be a substantial percentage, just falling short of a "bare majority". We repeat, however, that in our opinion at least a "bare majority" is necessary.
90. In Strathclyde Regional Council v Wallace [1998] ICR 205, the House of Lords considered a case in which the disadvantaged group comprised 81 male and 53 female employees. It was conceded in that case that there was no discrimination on those statistics but we agree with the submission of Mr Cavanagh that the House of Lords clearly agreed that those statistics would not support a case of discrimination. We draw attention to the speech of Lord Browne-Wilkinson at paragraphs E-H, page 210:

"Finally, I must state an agreed fact of the greatest importance. The disparity in pay between the appellants and principal teachers has nothing to do with gender. Of the 134 unpromoted teachers who claimed to be carrying out the duties of principal teachers, 81 were men and 53 women. The selection by the appellants in this case of male principal teachers as comparators was purely the result of a tactical selection by these appellants there are male and female principal teachers employed by the respondents without discrimination. Therefore the objective sought by the appellants is to achieve equal pay for like work regardless of sex, not to eliminate any inequalities due to sex discrimination. There is no such discrimination in the present case. To my mind it was be very surprising if a differential pay structure which had no disparate effect of impact as between the sexes should prove to be unlawful under the Equal Pay Act 1970. The long title to that Act describes its purpose as being "An Act to prevent discrimination, as regards terms and conditions of employment, between men and women."

91. Mr Cavanagh drew our attention to the fact that Lord Browne-Wilkinson (with whom the other Law Lords agreed) held that the Northern Ireland case of McPherson v Rathgael Centre for Children and Young People [1991] IRLR 206 was wrongly decided. In that case, the disadvantaged group had comprised either four or five men and one woman and the advantaged group comprised one man. It was self-evident, submitted Mr Cavanagh, that there could be no indirect discrimination where women formed 15% or 20% only of the disadvantaged group. The factual basis of that case had some similarity to the present.

Selection of the Pool

92. The choice of pools is a matter for the Employment Tribunal. It is, however, helpful to have in mind the guidance given by Sedley L J in Allonby v Acrington and Rossendale College [2001] ICR 1189:

"18 The appeal tribunal held that there was no error of law in the tribunal's choice of a pool. Lindsay J quoted the decision of the President, Waite J, in Kidd v DRG (UK) Ltd [1985] ICR 405,415:

'The choice of an appropriate section of the population is in our judgment an issue of fact (or perhaps strictly a matter for discretion to be exercised in the course of discharging an exclusively fact-finding function)…'

I would sound a strong note of caution about this. As the appeal tribunal's excellent analysis of the possible pools shows, once the impugned requirement or condition has been defined there is likely to be only one pool which serves to test its effect. I would prefer to characterise the identification of the pool as a matter neither of discretion nor of fact-finding but of logic. This was the approach adopted by this court in Barry v Midland Bank plc [1999] ICR 319 334, and endorsed by Lord Slynn of Hadley on further appeal [1999] ICR 859, 863. Logic may on occasion be capable of producing more than one outcome, especially if two or more conditions or requirements are in issue. But the choice of pool is not at large."

93. These dicta was followed in Cadman v Health and Safety Executive [2004] IRLR 971 and in Cheshire and Wirral Partnership NHS Trust v Abbott [2006] IRLR 546. In the Cheshire and Wirral case, Keene L J pointed out that statistics had to be "valid" and "significant" relying on Enderby covering sufficient individuals and not be purely fortuitous or short-term phenomena. Consequently, an employee is not entitled to identify an artificial comparator group (see paragraph 17). Further, he noted at paragraph 20:

"But even in non-piecework cases, one should at least strive to include all the advantaged workers, where work is held or assumed to be of equal value, in the comparator pool. In any event, to limit the comparator pool to the all male porters amongst the hospital ancillary workers employed by the appellant who were receiving a bonus and to omit the caterers should at the very least be regarded as arbitrary and contrary to the approach laid down in Enderby. It would artificially bolster the case for the respondents by leaving out the other ancillary workers who received the advantage of the bonus."

94. By the same token, in our opinion, it is necessary to pool all the disadvantaged group together. We can see no justification whatever for including in the disadvantaged pool persons who are not said to be disadvantaged or in respect of whom there is no evidence of disparate impact.

Objective justification

95. Again, this matter is not controversial. Cases such as BA v Starmer [2005] IRLR 862 and Allonby make clear that the test is objective and the employer's objectives must be legitimate and the means chosen to achieve them should be appropriate and reasonably necessary to achieve that end; in the judgment of Sedley L J in Allonby at paragraphs 20-30.
96. The authorities to which we have referred also make clear, as Mr Cavanagh submitted, that the decision of the employer and its business reasons should be respected but not uncritically accepted.
97. We would also refer to the judgment of Buxton L J in Armstrong v Newcastle upon Tyne NHS Hospital [2006] IRLR 124 at which Buxton L J had this to say at paragraph 127:

"127 Failure to adjust the system

The employment tribunal went on to say, in its paragraph 32, that even if the disparity had been justified when CCT was introduced, the continuation over time of a system whereby more men than women received bonuses, even though that system was in the process of being phased out, was indirectly discriminatory. This conclusion was not fully explained, and I do not think that it was justified. I do not see how it can be said that a failure to deprive the male comparators of part of their income was discriminatory, if the assumption is that their original receipt of that part of their income was not discriminatory. And the same is true of the decision to put the domestics out to tender. If that decision was discriminatory when made, it continued as such thereafter. If it was not discriminatory when made, it could not become such thereafter."

Reasons for decisions

98. The principle that an Employment Tribunal must give adequate reasons for its decisions is well known and need not be repeated; see for example Anya v University of Oxford [2001] ICR 847 and English v Emery Reimbold [2003] IRLR 710.
99. We remind ourselves also as to the circumstances in which an Employment Appeal Tribunal can substitute its findings for those of an Employment Tribunal and either allow or dismiss an appeal outright; see Hellyer Brothers v McCloud [1987] ICR 526. An appeal can be allowed where the decision of the Employment Tribunal is plainly and arguably wrong and there is no need to make further findings of fact or re-investigate those facts; see page 547 D-E.
100. Finally, we remind ourselves of what Mummery L J had to say in Yeboah v Crofton on perversity appeals. The high threshold that an Appellant has to cross

"such an appeal ought only to succeed where an overwhelming case is made out that the Employment Tribunal reached a decision which no reasonable Tribunal, on a proper application of the evidence and the law, would have reached. Even in cases where the Appeal Tribunal has 'grave doubts' about the decision of the Employment Tribunal."

Grounds of appeal and arguments in support

101. The Notice of Appeal raises three main issues. Mr Cavanagh's case was that the grounds of appeal need to be seen against the background that where a Claimant can show prima facie indirect discrimination between themselves and comparators, it is not sufficient for the employer to show a difference between the cases of the Claimant and the comparators that explains the differential in pay; the Respondent needs to go further and prove that the difference is objectively justified. In the present case, the Employment Tribunal found that objective justification was required and went on to reject the Respondent's case on objective justification. The key issue for the Respondents, and the first ground of appeal, has been characterised as "the disparate impact issue". The Respondent's first ground of appeal was to the effect that the Employment Tribunal concluded that there was a prima facie case of discrimination simply on the basis of a consideration of the proportion of each gender in the pool of both the advantaged (drivers) and disadvantaged (Metro Operators) groups who were Metro Operators. This is described as "the disparate impact issue".
102. The second principal ground of appeal has been characterised as "the correct pool issue". It is said the Employment Tribunal fell into error in its identification of the relevant pools for the disparate impact issue, in particular by including Metro Operators within the disadvantaged pool who did not carry out exclusively driving duties (those on the metro link). They did not carry out exclusively driving duties and no concession or finding was made that they carried out like work to drivers.
103. The third ground of appeal was characterised as "the objective justification issue". The Respondent has submitted that if the Employment Tribunal was correct in looking to the Respondent for an objective justification of the pay differential it reached a perverse conclusion that the variation in pay was not objectively justified.

The disparate impact issue

104. Mr Cavanagh makes two principal points. First, the Employment Tribunal misdirected itself in law in that in particular it failed to take account of the numbers of women in the disadvantaged group and the fact of the proportion and percentage of the disadvantaged group that was female.
105. Secondly, in the light of its findings, the only outcome open to a reasonable Employment Tribunal would be to find that there had been no disparate impact on women caused by the pay differential between drivers and Metro Operators. Accordingly, there was no need for the Respondent to provide objective justification for the pay differential. The Employment Tribunal found there was a material reason for the difference in pay, and accordingly equal pay challenges should have been dismissed.
106. Mr Cavanagh submitted that the Employment Tribunal had failed to direct itself that the burden of proving disparate impact, so as to trigger the Respondent's obligation to provide objective justification rested on the Claimants. There was a suggestion in Sharp that whenever a woman proved she did like work to that of a male comparator but received lower pay triggered an obligation on the part of the Respondent to provide objective justification, even where the Respondent was able to demonstrate there had been no direct discrimination on the grounds of sex and no adverse effect on women. However that approach had been disapproved in Villalba and was contrary to what Simon Brown L J had said in Nelson at the passage we have quoted earlier.
107. Mr Cavanagh submitted the failure of the Employment Tribunal to correctly direct itself as to the burden of proof was more than a mere technicality. The Employment Tribunal drew attention to what it perceived as the Respondent's failure to provide sufficient information about the composition of the material workforce (although the evidence was in fact before it) and this error of the Employment Tribunal led to a further error in relation to the identification of the relevant pools. Mr Cavanagh submitted that it could not be certain that had the Employment Tribunal correctly directed itself on the burden of proof it would have come to the same conclusion, and this, with other misdirections, gave the Respondents cause for concern as to whether the Employment Tribunal properly understood the issues. Some evidence of sex discrimination was not enough. The Employment Tribunal found that the Respondent's obligation to objectively justify the pay differential arose "if there was evidence of it". This is contrary to Nelson. He relied upon both Enderby and Nelson to support his submission that the duty on the part of the employer only arose when discrimination in the sense of disparate adverse effect of the pay differential on women was proved on the balance of probabilities, or at least by a prima facie case.
108. Mr Cavanagh's principal point, however, was that the Employment Tribunal was wrong in law to find that discrimination had been made out because the proportion of women in both pools who were part of the disadvantaged group was greater than that of the proportion of males. The Employment Tribunal should not have solely looked at this but in particular at the actual numbers and the proportion in the disadvantaged group that was female.
109. He reminded us that the purpose of the Equal Pay Act was to deal with sex discrimination in pay, not to achieve fair pay. The Employment Tribunal was not satisfied that there was discrimination by reason of a PCP (contrary to the submission made to it by Ms Woodwark for Mrs Fulton) but found there was discrimination simply on the basis of the relative proportions of women and men in the disadvantaged group, as compared to their numbers overall. The case was therefore, a "two groups" case as Mr Cavanagh has characterised it. At paragraphs 107 and 109, the Employment Tribunal twice drew attention to the fact that the disadvantaged group (Metro Operators) was significantly male. That was the correct starting point, having regard to Enderby, Loughran, Wallace, and Bailey. Once the Employment Tribunal had found that the majority in the disadvantaged group were not female, and that there was no sex taint, the authorities required the Employment Tribunal to find that there was no indirect sex discrimination. Accordingly, it should simply have dismissed the claims.
110. Mr Cavanagh took us through the authorities to which we have already referred as to the necessary proportion of the disadvantaged sex in the disadvantaged group to found a case of disparate impact. He recognised that the most favourable of these authorities to the Claimants was Bailey and he reserved the right to argue that that case was wrongly decided. However, as we have already noted, he submitted that Peter Gibson LJ did not intend to suggest that there could be an inference of indirect sex discrimination in the absence of there being a majority of the disadvantaged sex in the disadvantaged group. Further, he submitted Peter Gibson LJ was speaking about cases where there were two work groups but a PCP explained why there were more women in one group than the other and where there might be a PCP hidden below the surface which statistics could help to identify. In the present case, however, there is a clear finding that there was no PCP. Accordingly, this was a "pure" "two groups" case and it was wholly appropriate to focus or at least look with care at the number of women in the disadvantaged group. In Bailey, the statistics showed that on aggregate, women were six times more likely than men to be in the disadvantaged group. There was nothing in Bailey to justify a finding of discrimination where the proportion of women in the disadvantaged group was as low as 8%. Even applying Bailey at its highest, the number of women in the disadvantaged group was too small to justify a finding of indirect sex discrimination. The small number of women in both groups also suggested that the statistics needed to be treated with particular care; that there should be close scrutiny of all other available material.
111. Mr Cavanagh went on to submit that the reasons given by the Employment Tribunal for its findings on indirect discrimination were simply inadequate; there is simply no adequate explanation in paragraph 109 as to why a prima facie case requiring the employer to prove objective justification had been established.
112. In relation to Bailey, and in particular the judgment of Waller LJ, Mr Cavanagh submitted that his comments were obiter and that Nelson remained good law.
113. We do not need to consider submissions to the effect that the Equal Pay Act was intended to deal with discrimination on the grounds of sex in pay rather than achieve fair pay, because we do not consider this proposition to be controversial.
114. Further, Mr Cavanagh submitted that it is quite clear that to use the convenient shorthand, this was a "two group" case. Accordingly, it was necessary for there to be special care in considering statistical evidence. This was so because there was no PCP that made it difficult for women to enter the advantaged group. There was obviously a significant likelihood that any disparity in the proportions of men to women in the disadvantaged group as compared to the overall pool, comprising both groups, advantaged and disadvantaged, might result from chance rather than discrimination. Accordingly, the statistical evidence needed to be clear to support a conclusion that any disparity was tainted by sex rather than having an innocuous cause. He repeated what Elias J had said in the passage in Villalba at paragraph 117 to which we have already referred as to the need for cogent relevant and sufficiently compelling statistics in the absence of a PCP.
115. The absence of a PCP was itself a contra-indication of sex discrimination. That heightened the need for requiring the statistics to be compelling. The existence of a PCP is, of course, a strong indication of discrimination. Mr Cavanagh submitted that the statistics must be "so clear as to speak for themselves without any other sign of sex discrimination". There was no indication that the Employment Tribunal was aware of this.
116. He went on to make further submissions about the relevant proportion of women in the disadvantaged group and their numbers. There was, he said, no authority to suggest that when considering statistical evidence to determine whether there was indirect sex discrimination the Employment Tribunal should focus only on the proportions of each gender among all of those in the advantaged and disadvantaged groups who were in the disadvantaged groups. The authorities showed that Employment Tribunals should consider the actual numbers and the gender split within the disadvantaged group. Looking at the proportions as the Employment Tribunal did in this case, might give a false picture. The relatively small numbers involved rendered the statistics liable to distortion, but they would not in fact provide evidence of indirect sex discrimination. He reminded us of the authorities to which we have referred, such as Enderby, Loughran, and Villalba and suggested that regard must be had to the actual numbers. The factual background of this case, submitted Mr Cavanagh, was similar to that considered by the House of Lords in Strathclyde. In Strathclyde, the male majority in the disadvantaged group was smaller than in the present case, but the House of Lords evidently considered it unarguable that a "disadvantaged" female group forming part of a group with a significant male majority could come within the scope of the Equal Pay Act. In that case, again, an advantaged group had been retained on financial grounds because the employers could not afford to promote new teachers to that grade.
117. It was contrary to logic and common sense to say that there was adverse treatment of women by reason of the pay differential in a group that was 85% male (13 women out of 84, or 15.48% if one aggregated the Metro Operators on the metro link and the main link). The proportion is, of course, even more favourable to the Respondent's case if one simply takes those employed on the main link.
118. Mr Cavanagh submitted that we should allow the appeal. There were no further facts to be found and it was clear on the facts found by the Employment Tribunal and uncontroversial evidence before it that there was simply no evidence of indirect sex discrimination. There was, accordingly, no need for objective justification and we should simply allow the appeal.

The correct pool issue

119. The point raised here is that the Employment Tribunal misconstructed the disadvantaged pool. As we have noted, the Employment Tribunal found that the Metro Operators working on the main link were disadvantaged, but not those employed on the other roster, the metro link. However, for the purposes of constructing the disadvantaged pool, the Metro Operators from both metro link and main link were aggregated. The disadvantaged pool should only consist of those found to be engaged in equal work with their comparators. This is a submission which we accept. If there was to be a pool, it is logical that it should consist of all those who were advantaged (i.e. the drivers) and all those who were disadvantaged, i.e. Metro Operators employed on the main link. The Employment Tribunal was wrong when it said at paragraph 117 that there was no evidence of the number of Metro Operators who did not carry out exclusively driving duties. Mr Cavanagh pointed to the witness statement of Mrs Fulton (the second Respondent to this appeal) and a questionnaire provided by the Respondent. There was also a specific reference in the Respondent's skeleton argument for final submissions. These three pieces of evidence were before the Employment Tribunal and revealed that there were 57 male and five female Metro Operators on the main link. If one aggregated these with the 99 male and five female train drivers, 63% of the men were advantaged and 50% of the women. Women constituted 8.77% of the disadvantaged pool and 4.81% of the advantaged pool. There were 27 Metro Operators employed on the metro link, 20 men and seven women. In the light of those figures, Mr Cavanagh submitted the Employment Tribunal was bound to find there was no prima facie case of discrimination on the grounds of sex, so there was no need for the Respondent to objectively justify the pay differential. Accordingly, he submitted, the EAT should substitute a finding that there had been no indirect sex discrimination.
120. He made similar points as to the burden of proof and inadequate reasons, as he had in relation to his first ground of appeal. He had a further point on the burden of proof. The Employment Tribunal at paragraph 117 appears to have assumed that the burden was on the Respondent to show how many Metro Operators were engaged in equal work with drivers for the purposes of the genuine material factor issue. The burden, however, fell on the Claimants. If the Employment Tribunal lacked sufficient information, then the claim should have failed because the Claimants had failed to make out a case of disparate impact. In relation to inadequacy of reasons, he submitted that the Employment Tribunal had failed to explain why it included the Metro Operators employed on the metro link in the disadvantaged pool.

Objective justification issue

121. The Employment Tribunal found at paragraphs 110, 113, and 114 that as the objectives of the 1998 changes had not been achieved, and there was no longer a need to reduce subsidy, so therefore, there was no longer a justification for the disparity in pay. Mr Cavanagh submitted that there were two errors in this reasoning. First, he relied upon the passage in the judgment of Buxton L J in Armstrong at paragraph 127 to which we have already referred in relation to decision that were not discriminatory when made so could not later become discriminatory. The fact that the subsidy had been eliminated did not mean that the Respondent need no longer be concerned about savings to be made from flexibility and efficiency. There was a continuing need to avoid unnecessary costs and achieve efficiencies. These needs continued when the target savings had been achieved. Secondly, the Employment Tribunal found that when the changes were introduced, the pay differential was objectively justified. The Employment Tribunal considered that the pay differential ceased to be justified when the savings did not materialise. The Employment Tribunal, however, failed to take a further key consideration into account, which showed that the objective justification continued. Once it was clear that the anticipated savings had not materialised, the Respondent could either continue with the arrangements, or make the Metro Operators into drivers (with a consequent 30% pay increase, something which would be costly: the expense would be in the region of some £600,000 per annum, taking into account national insurance and pension contributions). Alternatively, the drivers would have to be demoted to Metro Operators with a reduction in their pay and fairly obvious adverse industrial relations consequences. This point was made to, but was not addressed by, the Employment Tribunal. Mr Cavanagh submitted there was no reason the Employment Tribunal could have failed to conclude that there was objective justification for the continuing pay arrangements, despite the lack of anticipated savings.
122. He went on to submit that the Employment Tribunal was wrong in its evaluation of the benefit of the changes. He criticised the Employment Tribunal for an inconsistency in its reasoning, that led it to find that the new arrangements did not lead to major benefits such as could amount to objective justification. The Employment Tribunal accepted that the Respondent had entered into protracted and difficult negotiations with the trades unions and that the increased wages paid to drivers were intended to be funded by increased productivity, as the new arrangements would reduce the number of drivers. Contrary to that finding, the Employment Tribunal found there could be no savings in cost because the drivers had received a pay increase and the Metro Operators were paid more than the drivers had previously been paid. This finding, at paragraph 110, ignored the earlier finding that the flexibility negotiated with the trades unions had had the effect of improving productivity.
123. The Employment Tribunal accepted that there had been concessions in negotiations, but said that they did not appear to be major. Nevertheless, such matters as "sign on times" (the times between signing on and taking out the trains) had been significantly reduced and the shifts had become "flexible" or variable in length, allowing for greater usage of available time. The Employment Tribunal suggested that this flexibility eroded over time without making any findings as to whether these two "major contributors" to increase productivity had been lost or eroded. The Employment Tribunal, therefore, had failed adequately to explain why it had reached the conclusion that drivers' concessions in relation to flexible working and removal of restricted practices were not "major", especially when the evidence was undisputed. The finding was that they were won only after difficult and protracted negotiations.

Claimants' submissions

Submissions of Mr Cross.

124. Mr Cross appeared on behalf of all the successful Claimants, except Ms Fulton. Mr Cross's submissions were that the Employment Tribunal had properly directed itself as to the law and was entitled to make the findings it did on the facts. There was accordingly no merit in the appeal.

Indirect discrimination.

125. Mr Cross spoke of a trend in equal pay cases for Respondents not to deal with the real issue, namely whether the genuine material factor defence was untainted by sex. He submitted that employers were seeking to avoid having to justify pay differentials by concentrating on creating new hurdles between s1.2 and s1.3 of the Act.
126. We considered that Mr Cross's submissions came close to the argument rejected in Strathclyde and Villalba and other cases that mere proof of a differential in pay involving like work between a man and a woman was sufficient to require the employer to provide objective justification. He relied on the judgment of Waller LJ in Bailey, paragraph 34:

"It is to be observed that in paragraph 14 the ECJ was considering the cases which had previously been determined by it and placing them in two categories. One comprised cases involving measures distinguishing between employees on the basis of their hours of work which in practice had an adverse impact on substantially more members of one or other sex. The other was where a system of pay, wholly lacking in transparency, was applied. In paragraph 15 it was pointed out that the Enderby case was different in that it did not fall into either of those categories. In paragraph 16 the ECJ turned to the situation where the facts were as set out in the question which it was answering, and gave its answer in those terms. The ECJ does not purport to be defining exhaustively the only circumstances in which a prima facie case of discrimination can be established. For example, no one would suggest that the first category should be limited to cases involving part-time workers. What the ECJ was doing was describing the circumstances which applied in the cases already decided and another different set of circumstances which applied to the case which it was then deciding."

127. Mr Cross submitted that there is no two-stage process in which a woman, pointing to a higher paid man has to show a prima facie case of discrimination, which the employer then has the evidential burden of justifying. In the present case the Claimants succeeded because the Respondent failed to discharge the evidential burden.
128. Even if one applied the "controversial" decision in Nelson, the Employment Tribunal have found that the Claimants crossed the appropriate threshold in establishing a prima facie case of indirect discrimination. The decision of the Court of Appeal in Bailey supported the view that once a prima facie case had been made out of indirect discrimination, the employer needed both to show that the pay differential was not tainted by sex but also that it was objectively justified.

The correct pool issue.

129. Mr Cross submitted that there was no requirement in equal pay cases that there should be a pool. The Respondent was also incorrect to submit that there was more than one test for indirect discrimination. In submissions, Mr Cross raised the issue as to why the advantaged group had been closed. Was it because it was tainted by discrimination? We note that this seems rather contrary to the Employment Tribunal finding that it was not so tainted. In any event, it seems to us that once the group was closed it was closed to both men and women.
130. If it was correct to have a pool, Mr Cross submitted, the appropriate course was to compare the advantaged group with the disadvantaged group as the Employment Tribunal did, and there was no doubt that the proportion of women in the disadvantaged group was significantly greater than the proportion of women both in the advantaged group and overall. That was all the Claimants needed to establish the prima facie case of indirect discrimination. The Employment Tribunal could either have regard to the proportion of women in the aggregated two groups who were in the disadvantaged group, as compared to the proportion of men, or alternatively, look at the disparate impact on the two groups.
131. Mr Cross relied upon a passage in Cheshire and Wirral partnership NHS Trust v Abbott [2006] IRLR 546. He took us a passage in the head note:

"Although in a case of indirect sex discrimination in pay it is for the employee to identify a comparator group and to produce statistical evidence to show an appreciable difference in pay for jobs of equal value, the employee is not entitled to identify an artificial or arbitrary group. In principle, the comparison should between the disadvantaged group and the advantaged group. As a matter of statistics, a more reliable result is likely to be forthcoming if one takes as large a group as possible, so long as that group shares the relevant characteristics and can be seen as doing work of equal value."

The content of the head note appears to be borne out by the judgment of Keene LJ at paragraphs 17-20. Mr Cross also submitted, basing himself on the decision in Kidd v D R G [1985] ICR 405 that the selection of an appropriate section of the community within which the comparison should be made was one for the Employment Tribunal and was a question of fact not law.

132. The approach of the Employment Tribunal was consistent with the approach of the House of Lords in Rutherford. Proportions and ratios were used in that case, rather than numbers. It was a question of fact for the Employment Tribunal to identify the pools and to determine whether there was a prima facie case that the threshold had been crossed.
133. Mr Cross submitted that what the Respondent was trying to do was what the Claimants in Nelson and Enderby had sought to do; that is, to take an arbitrary date for the pool. As at the date of hearing, multi-tasking had finished. The choice of the pool, taking into account all relevant groups, was consistent with the decision of the Court of Appeal in Barry v Midland Bank [1998] IRLR 138. We were taken to the head note:

"The appropriate pool of men and women for comparison was all those to whom the relevant provisions of the agreement applied at the time that the determination of the applicant's employment. It was necessary to look at all part-time workers at that time and the average of their hours of work throughout their service and to compare the men and the women in the advantaged and the disadvantaged groups. The disadvantaged group in that pool were those part-time workers whose hours of work at termination were less than the average of their hours of work throughout their service. As no statistics were available as to the composition of the advantaged and disadvantaged groups in the pool, the applicant had failed to prove her case of indirect discrimination."

134. It was necessary to look at the widest possible pool, as the Employment Tribunal has done.

Objective justification.

135. The Respondent called no witnesses on justification. It put all its eggs in the basket of the pool for comparison, and lost. It called no evidence, it was submitted, in relation to the general material factor defence as to the cost of removing the pay differential between drivers and Metro Operators. In any event, the Respondent saved no money and never in fact implemented the decision to replace drivers with multi-skilled workers. The only difference was the 25% pay advantage enjoyed by drivers. It seems to us that this argument overlooks the fact that a saving in cost, in terms of the amounts paid to drivers who were Metro Operators and drivers in the drivers' group was considerable and this was a saving achieved by the Respondent as a result of the 1998 agreements.

Submissions of Ms Woodwark for Mrs Fulton

136. Ms Woodwark adopted Mr Cross's submissions.

Finding on indirect discrimination.

137. Ms Woodwark submitted that on the basis of the speech of Lord Nicholls in Glasgow v Marshall, once a woman showed her pay was less than that of a man engaged in like work, the employer was required to demonstrate that the pay differential was not tainted by sex. If this differed from Nelson, she submitted, the decision of the House of Lords should be preferred.
138. She also relied upon Bailey as the authority for the proposition that a high threshold was not necessary to establish a prima facie case of discrimination. It was open to the Employment Tribunal to find that the long-standing disparity in favour of men as between drivers and Metro Operators required objective justification.
139. The Employment Tribunal correctly addressed the issues of indirect discrimination and it is not legitimate to go through the Employment Tribunal reasoning with a fine toothcomb. In this regard, she relied upon Hollister v National Farmers' Union [1979] ICR 541. This principle is, of course, well-known.
140. On the figures, there was sufficient evidence of disparate impact. Ninety-nine men were advantaged (73.5% of the male population) and only five women (50% of the female population). The ratio of disadvantaged females to males (following Rutherford) showed 50% of women but only 36% of men. Further, the difference was long-standing. The fact that females work in a male-dominated occupation does not mean that the pay practice is not discriminatory against women. If the evidence shows that there is prima facie discrimination, the employer is bound to show that this is not by reason of sex.
141. The Employment Tribunal was well aware of the limited number of women in the pool, so it well had in mind the fact that statistics might be "skewed".
142. The reasons given by the Employment Tribunal were sufficient. She referred to English v Emery Reinbold; the Respondent knew why it had lost and the Claimant knew why it had won.

Pool for comparison.

143. The pool selected was sufficient and adequate reasons were given as to why it was selected.

Objective justification.

144. The Employment Tribunal was entitled to find that the Respondent had failed to objectively justify the pay differential. Ms Woodwark drew our attention to the very high threshold required for a party to succeed on a perversity appeal; see Yeboah v Crofton, the relevant passage of which we have already cited.

Conclusions

Introductory observations.

145. One matter stands out above all others. At the time of the presentation of the originating applications, there were 13 female Metro Operators and 71 male Metro Operators. The female operators represented 15.48% of the pool taken by aggregating Metro Operators on the main link and metro link. If one excludes those employed on the metro link and examines the position so far as concerns the main link only, there were five female Metro Operators and 57 male operators. Approximately 8%, therefore, of Metro Operators on the main link were female. Ninety two percent, therefore, of the disadvantaged group was male. The Employment Tribunal was thus faced on the evidence before it with a case where the vast majority of those in the disadvantaged group were male.
146. There is a need, one would have thought, for the most careful consideration as to how pay disparity in such circumstances could be said to be tainted by sex and disadvantageous to women, when depending on how one takes the pool, there are either six times or 12 times more men than women in the disadvantaged group. On these figures, and in the absence of any evidence of direct discrimination or a PCP unfavourable to women, the finding that the disparity in pay amounts to discrimination against female Metro Operators seems absurd.
147. Statistics are a tool which may used to prove there has been some discrimination, in this case, where a PCP cannot be discerned. However, statistics are only a tool and statistics taken in isolation cannot always be sufficient to prove a case for Claimants as regards disparate impact. They should never be followed mechanistically. Clearly, in a case where most of those affected are female, the case is different. However, although it may be possible, it seems very difficult to envisage a situation in which there can be said to be indirect discrimination against women in the absence of a PCP by reason of a pay differential between one group of workers doing like work and another when both the group said to be advantaged and the group said to be disadvantaged are both mainly male. This is all the more so in a case such as the present where the percentage of women in the disadvantaged group is so low. The same point holds good where, whatever the make up of the advantaged group, the disadvantaged group is predominantly male.
148. In our opinion, arguments as to burden of proof are of little determinative effect in this case. Wherever the burden lay of proving disparate impact, the figures speak for themselves and in our opinion those figures cannot possibly support a finding of indirect discrimination or a prima facie case of indirect discrimination. In a case such as this, comparing the ratios of men and women in both groups, advantaged and disadvantaged, is not of the greatest assistance. We ask what the position would be if there were no women in the advantaged group and one or two women in the disadvantaged group. Whatever number of women were in the disadvantaged group, 100% of all women would be in the disadvantaged group and thus, on the Claimants' case, there would be a prima facie case that the pay differential was tainted by sex. This would be so even though in terms of numbers the vast majority of the disadvantaged group were also male. On that premise, the great majority of the disadvantaged would be male as opposed to female employees. One needs, therefore, to consider carefully how the Employment Tribunal came to its decision, which is one which at first blush appears to be so startling.

The disparate impact point

Burden of proof.

149. We accept Mr Cavanagh's submissions in this regard. In our opinion, the burden of proving disparate impact rests on the Claimants, for the reasons that we gave in our discussion of the law. We are bound by the decision in Nelson and also by Villalba. It is clear to us that the Court of Appeal in Bailey was not seeking to depart from the decision in Nelson and indeed, Peter Gibson LJ, with whom Sir Martin Nourse agreed, accepted that he was bound by Nelson. Waller LJ also was not saying that the decision was wrong, although he expressed the belief that for the Claimant to have the burden of proof of disparate impact complicated the presentation of cases.
150. One cannot raise an inference of disparate impact in a case where the pay differential is said to be the result of indirect discrimination simply by having a female point to a male comparator doing like work with enhanced pay. It is necessary to show disparate impact on women as being part of the disadvantaged group. It is not a question of preferring Lord Nicholls' speech in Glasgow. The Court of Appeal in Nelson has explained his speech in a manner that is binding on us and the Employment Tribunal, for the reasons we have given, is bound by the judgment of Elias J in Villalba. Even if we are wrong as to our views on the burden of proof, in our opinion whoever bears the evidential burden, it is necessary for the Employment Tribunal to find, on all the evidence before it, there is a prima facie case of disparate impact tainted by sex before moving to the question of objective justification. Here again we accept Mr Cavanagh's submissions. The evidence, as we have already said, not only fails to show that the pay differential was tainted by sex but shows almost conclusively that it was not. The Employment Tribunal has, in our opinion, fallen into error in going straight to objective justification simply because it found that a higher percentage of women overall as compared to men were in the disadvantaged group.
151. As we have said, statistics are a tool only and the Employment Tribunal must look at all other available material. The authorities to which we have referred make it clear that the numbers involved in both groups are highly relevant, see for example Enderby, BRS v Loughran, and Bailey, Villalba and Rutherford. The Employment Tribunal should also have had regard to the fact that the advantaged group (the drivers) was found to be closed and at the time of closure the Employment Tribunal made no finding that there was any discrimination in entry to the group and indeed found to the contrary. The Employment Tribunal found there was no PCP and that in itself was a contra-indication of discrimination. This point is of course not decisive and neither is the fact that the advantaged group was closed. However, in the absence of a PCP where the only evidence of disparate impact that can be relied upon comprise statistics, those statistics must be so clear as to speak for themselves. We find it difficult to conceive of a situation in which it would be correct to make a finding that a pay disparity is tainted by sex and discriminatory against women, where there is no direct discrimination and no provision criterion or practice that might lead to a disparate impact on women, even if the overall proportion of women in the disadvantaged group is higher than that of men but where the overwhelming majority of the disadvantaged group is male. Even were it possible to conceive of a case where such a finding might be appropriate the finding in the present case, in our opinion, stands the law on its head. When taking the best case for the Claimants, 85% of those affected are male, and in the worse case (which, as appears shortly, is the correct case) 92%. The effect of the decision of the Employment Tribunal is to give effect to what might be described as "fair pay" rather than applying the law relating to equal pay.
152. It cannot be said that the statistics relied upon by the Employment Tribunal are "cogent relevant and sufficiently compelling" as to described by Elias J in Villalba, thus pointing towards disparate impact on women. As we have said on several occasions, they in fact prove the reverse.
153. We also consider there is force in Mr Cavanagh's submission that the Employment Tribunal in any event failed to have regard to the fact that small numbers can easily lead to distortions. There were five female operators only in the pool of main link Metro Operators, 13 women in all if one adds those employed on the metro link. A change of one or two only would have made a significant difference to the proportion.
154. It seems to us that this case is factually very similar to that in the Strathclyde case, save that the male majority in the disadvantaged group in that case was smaller than in the present (81 male – 53 female) and it is clear that the House of Lords considered the argument of disparate impact in that case to be untenable.
155. We agree that on the facts as found by the Employment Tribunal the decision cannot stand. There is no point in remitting this matter to the Employment Tribunal because the facts that were found in our opinion conclusively point to there being no evidence of disparate impact. Although it is not strictly necessary in the circumstances for us to do so, we would add that we do find Mr Cavanagh's submissions on the absence of satisfactory reasons to be made out. Paragraph 109 of the decision of the Employment Tribunal does not give any explanation as to why the Employment Tribunal ignored the fact that women were so small a minority in the disadvantaged group.

The correct pool

156. Even if we were wrong as to our views on disparate impact, we would also allow the appeal on the grounds that the Employment Tribunal has failed to identify the correct pool of those disadvantaged. The Employment Tribunal's findings were that the closing of the drivers' group was not discriminatory and further, at the time of the closure of the group, there was no finding of any discrimination in recruitment to this grade. The Employment Tribunal appears to have overlooked the evidence of numbers of female Metro Operators employed on the main link and metro link. Mr Cavanagh demonstrated to us that these figures were set out not only in the evidence of Mrs Fulton, but also in the written closing submissions of Mr Cape, junior counsel for the Respondent (see paragraph 39). They also appeared in a questionnaire completed by the Respondent that was before the Employment Tribunal.
157. In our opinion the disadvantaged pool should only consist of those persons who were found to be engaged in equal work with the comparators. The Employment Tribunal found that only those employed on the main link were disadvantaged. It should have omitted the Metro Operators employed on the metro link from the pool for the purposes of comparison. The error, it seems to us, is fundamental because it significantly reduced the proportion of women in the disadvantaged group by approximately half to some 8%. The decision of the Employment Tribunal is accordingly irretrievably flawed.
158. However, we stress that in our opinion, whatever pool is taken, as we have made clear, the evidence points to the fact that the pay differential between drivers and Metro Operators had no disparate impact on women.
159. Again, it is unnecessary for us to do so, but we believe that Mr Cavanagh's submissions relating to the burden of proof on this issue and the adequacy of reasons seem well-founded. At paragraph 117, the Employment Tribunal appears, for example, to have assumed that the burden was on the Respondent to show how many operators were engaged in like work with drivers for the purposes of the genuine material factor issue rather than the Claimant. There is also no explanation as to why Metro Operators in both the metro link and main link were aggregated together in one pool.

Objective justification

160. In the light of our findings on the first two grounds, it is unnecessary for us to determine the point. We shall shortly express our views on this ground of appeal also. We accept that it is for the Employment Tribunal to establish the facts and evaluate them and we should not lightly interfere with their findings of fact. Nonetheless, we are satisfied that the Employment Tribunal's finding that objective justification has not been made out cannot stand. This is in fact not a perversity appeal but is based upon the Employment Tribunal failing to address highly material evidence and failing to give adequate consideration to the fact that the agreements when negotiated in 1998, which gave rise to the pay discrepancy, were not discriminatory. The Employment Tribunal was satisfied that the 1998 agreements were hardly bargained between the Respondent and trades unions, negotiations lasted several years and were protracted and difficult (see Employment Tribunal paragraph 11). It is apparent, having regard to the influence of the trades unions, that it would be commercially impossible for the Respondent to re-negotiate the agreements to the detriment of the drivers so as to remove the pay differential. The only option would be to increase salaries of the Metro Operators to bring them into line with the drivers, which would cost £600,000 per year. We would again refer to what Buxton LJ said in Armstrong:

"I do not see how it can be said that a failure to deprive the male comparators of part of their income was discriminatory, if the assumption is that their original receipt of that part of their income was not discriminatory. The same is true of the decision to put the domestics out to tender. If that decision was discriminatory when made, it continued as such thereafter. If it was not discriminatory when made, it could not become such thereafter."

161. In our opinion, Mr Cavanagh is correct in submitting that the Employment Tribunal accepted that when the 1998 agreements were made, the changes effected by them were objectively justified. We find it difficult to envisage how, if the additional payments made to the drivers were not discriminatory at the time that categories were established how they become discriminatory later, bearing in mind that the drivers' group was a closed group. The Employment Tribunal based its decision on the fact that the pay differential became discriminatory when savings did not materialise and therefore the justification ceased. However, one asks forensically, why should the Employment Tribunal ignore the practical impossibility of reducing the drivers' pay or the significant expense of increasing that of Metro Operators without any economic justification? Why are these reasons immaterial? Why is the saving that was afforded by employing Metro Operators as drivers by virtue of the agreement also to be regarded as immaterial? There was also the benefit of the extension of the metro to Sunderland. The Employment Tribunal has not addressed the issue of what the Respondent should have done; should it have attempted to cut the pay of the drivers, with obvious industrial relations consequences, or should it significantly have increased the pay of Metro Operators, despite the expense, and that it does not appear to have been economically justified.
162. Had the Employment Tribunal considered these matters in our opinion it would have been bound to find, however generous an ambit one gives an Employment Tribunal for fact-finding and evaluation of facts, that the Respondent had objectively justified the pay differential.
163. Further, the Employment Tribunal findings that there had been no major concession in relation to flexibility and that there was no longer a need to reduce subsidy, ignored a number of important matters:

(a) the increased flexibility of drivers as a result of the agreement;

(b) at the time of the agreement it was considered that the 30% uplift in drivers' pay would be paid for by increased productivity that would reduce the number of train drivers, as it did; that saving clearly continued because many of the Metro Operators who carried out driving duties, but for the 1998 agreements, which had the effect of closing the drivers' category, would have had to be employed as drivers.

164. There is no mention of this in the Employment Tribunal judgment. Sign on times were reduced, shifts became flexible and again, a point ignored by the Employment Tribunal, the Sunderland extension was made possible. To these must be added the cumulative effect of the other matters referred to by Mr Clayton at paragraph 26 of this witness statement, and which we are satisfied was evidence properly considered by the Employment Tribunal.
165. On this ground also, therefore, the decision of the Employment Tribunal is flawed and cannot stand.
166. We would like to express our gratitude for the great assistance we received from all advocates in this case, Ms Woodwark, Mr Cavanagh QC, Mr Cape and Mr Cross.