Grundy v British Airways Plc

Neutral Citation Number: [2007] EWCA Civ 1020 Case No: A2/2005/1949 IN THE SUPREME COURT OF JUDICATURE COURT OF APPEAL (CIVIL DIVISION) ON APPEAL FROM THE EMPLOYMENT APPEAL TRIBUNAL EAT/0676/04/RN Royal Courts of Justice Strand, London, WC2A 2LL Date: 23/10/2007 Before : LORD JUSTICE WALLER LORD JUSTICE SEDLEY and LORD JUSTICE CARNWATH - - - - - - - - - - - - - - - - - - - - - Between : MRS V GRUNDY - and - BRITISH AIRWAYS PLC - - - - - - - - - - - - - - - - - - - - - (Transcript of the Handed Down Judgment of WordWave International Limited A Merrill Communications Company 190 Fleet Street, London EC4A 2AG Tel No: 020 7404 1400, Fax No: 020 7831 8838 Official Shorthand Writers to the Court) - - - - - - - - - - - - - - - - - - - - - Appellant Respondent Ms J McNeill QC and Mr M Ford (instructed by Messrs Simpson Millar) for the Appellant Mr C Jeans QC and Mr A Short (instructed by British Airways Legal Department) for the Respondent Hearing date: Monday 26 February 2007 - - - - - - - - - - - - - - - - - - - - - Judgment Lord Justice Sedley : 1. 2. 3. Mrs Grundy is a long-serving member of British Airways’ cabin crew (CC). She was initially employed full-time, but in 1987 she took employment as support cabin crew (SCC). This was a new class of contract, set up under a collective agreement known as the Burford Bridge Agreement, between British Airways (BA) and what was then the single staff union. It provided for those cabin crew who elected to become SCC to nominate between 15 and 20 days in each 28-day period on which they wished to work. They were paid a daily rate for those of the nominated days on which BA offered them work. The arrangement thus had benefits of flexibility for both parties, but particularly for women like Mrs Grundy who, in the course of time, had childcare responsibilities. Until 1994 this was the only way in which cabin crew could work part-time. From 1994 recruitment of SCC was halted and crew were offered the option of fixed percentage hours equating to a third, a half or three quarters of full- time work. When employment as SCC was abolished in 2002, Mrs Grundy opted for a 75% CC contract. While full-time and post-1994 part-time cabin crew were on an incremental pay scale which gave them annual increases, support cabin crew had no such increments (and initially no pension, paid holiday or sick pay). As a result, by the time these proceedings under the Equal Pay Act 1970 were heard, Mrs Grundy’s pay was £13,589 as against the £17,499 paid to a man – Mr Wynne – who it was accepted was doing like work (also at 75% of full time) but had benefited as cabin crew from annual increments to which Mrs Grundy had not been entitled. Hence the claim under the Equal Pay Act 1970. A concomitant claim under the Sex Discrimination Act 1975 was withdrawn at first instance. The Equal Pay Act as amended makes the following material provisions: 1(1) If the terms of a contract under which a woman is employed at an establishment in Great Britain do not include …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 …………….. (3) An equality clause 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 – (a) …… must be a material difference between the woman’s case and the man’s; ………………. 4. 5. 6. The effect of these provisions can be summarised as follows: every woman's contract of employment includes an equality clause which entitles her to terms no less favourable than those of a man who does like work for the same employer, unless she has less favourable terms which are genuinely due to a material difference, other than the difference of sex, between her case and the man's. The self-evident purpose is to invite a case-specific inquiry which may become general only at the point where less favourable contractual terms for like work have been established and the employer is seeking to establish reasons unrelated to gender for the difference. Because it required radical adjustments to pay schemes, the Equal Pay Act 1970 did not come into effect until 29 December 1975. On the same date the principal provisions of the Sex Discrimination Act 1975 were also brought into effect, dealing with the non-contractual treatment of women and doing so by means of structured concepts of direct and indirect discrimination quite different from the mechanism of the Equal Pay Act. As the EAT in the present case pointed out, the two statutes are mutually exclusive. The Equal Pay Act nowhere calls in terms for a collective assessment of disparate impact: it requires only a comparison between the woman’s case and the man’s. How then does it come about that this case has been argued at each instance entirely in terms of disparate impact and justification? The explanation is to be found in the decision of the House of Lords in Glasgow City Council v Marshall [2000] ICR 196. This followed the House’s decision in Strathclyde Regional Council v Wallace [1998] ICR 205, in which Lord Browne-Wilkinson reviewed the cases and concluded: “…. [T]he 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 … It follows that the words “not the difference of sex” where they appear in s.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 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 …” 7. 8. 9. It is not necessary here to set out the facts of Marshall, save that they had produced a decision at first instance in the applicants’ favour by a literal application of s.1(2) and 1(3) of the Equal Pay Act. Lord Nicholls, in the single reasoned speech, held that this was not the right approach. In order to advance a defence that the difference in pay was not due to the difference in sex, the employer “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.” At least three material points emerge from Lord Browne-Wilkinson’s and Lord Nicholls’ reasoning. First, sex discrimination enters the s.1(3) defence as an aspect of the onus resting on the employer: it has to be disproved, not proved. Secondly, if he can disprove discrimination, the argument is concluded in his favour; if not, he must go on to show objective justification. Thirdly, and importantly, Lord Nicholls holds: “Whether a pay disparity has a disparately adverse impact on women is primarily a question of fact.” The Watford employment tribunal (Mr Adamson, Mr Pepper and Mrs Scott) found in Mrs Grundy’s favour and directed that her pay, adjusted to 75%, be brought with retrospective effect to the top of the cabin crew pay scale. There is no issue about the appropriateness of this relief. In arriving at it, however, the tribunal first considered whether Mrs Grundy’s employment history, “with the failure of the Respondent to provide her with increments each year, was indirectly discriminatory on the ground of sex and unlawful”. They held the pay differential to be a “policy, criterion or practice of not paying increments to support cabin crew”, which “was to the detriment of a considerably larger proportion of women than men”. They went on to hold that it was not justified. The first of these findings must have been intended to reflect s.1(2)(b)(i) of the Sex Discrimination Act 1975 (as amended with effect from 12 October 2001), which, however, speaks of a “provision, criterion or practice”; but whichever of these elements the tribunal took to be at issue, what was in fact at issue was a contractual term, namely the provision for Mrs Grundy’s pay, which was less favourable than Mr Wynne’s notwithstanding that they were doing the same work. The second finding, on justification, reflected s.1(2)(b)(ii) of the Sex Discrimination Act. 10. On appeal to the EAT (Judge McMullen QC, Dr Fitzgerald and Ms Switzer), the same approach was adopted by the parties and (with an introductory explanation reflecting an understandable unease at the assimilation of the two statutes) by the appeal tribunal. The EAT declined to interfere with the ET’s decision on justification but overset it on the issue of disparate impact. They held that the ET had erred in law in focusing on the small, disadvantaged, SCC group rather than on the larger, advantaged, CC group. They went on to hold that only the latter was permissible in law and, since it was common ground that this produced a nugatory difference, substituted a finding that there was no disparate impact. 11. In this court Jane McNeill QC for Mrs Grundy accepts that the larger group displays no appreciable adverse impact, but submits (a) that there was nothing wrong in law with the ET’s focus on the smaller group and (b) that even if there was, it does not follow that the larger group was determinative, so that the most the EAT could properly have done was remit the case for rehearing. Christopher Jeans QC for BA argues that the EAT was right in relation to disparate impact but wrong in relation to justification, so that even if Mrs Grundy succeeds on the former issue, she loses on the latter. 12. The figures on which Ms McNeill relies are helpfully distilled in the following table, which takes three representation years (1994, 1998 and 2002) from the agreed annual statistics: Females Males Advantaged 6,273 (1994) 8,501 (1998) 8,592 (2002) 3,147 (1994) 4,028 (1998) 4,130 (2002) Disadvantaged 471 (1994) 99 (1998) 42 (2002) 24 (1994) 5 (1998) 3 (2002) Total Pool 6,744 (1994) 8,600 (1998) 8,634 (2002) 3,171 (1994) 4,033 (1998) 4,133 (2002) Disadvantaged as % of pool 6.98% (1994) 1.15% (1998) 0.49% (2002) 0.76% (1994) 0.12% (1998) 0.07% (2002) Gender Ratio (F/M) 1.99:1 2.11:1 2.08:1 19.6:1 19.8:1 14:1 2.13:1 2.13:1 2.09:1 9.2:1 9.6:1 7:1 Advantaged as % of pool 93.02% (1994) 98.85% (1998) 99.51% (2002 99.24% (1994) 99.88% (1998) 99.93% (2002 1:1.07 1:1.01 1:1 13. The table shows clearly enough that, while the ratio of women to men in CC was constant at about 2:1 throughout these years, the female:male ratio in SCC averaged almost 18:1 and at its lowest was 14:1. In other words, anything which impacted adversely on SCC was going to hurt a far larger proportion of women than if it were to impact on CC. But because of the relative size of the two groups, it was going to hurt far fewer women in absolute numbers: in 1994, 471 as against 6,273; in 1998, 99 as against 8,501; in 2002, 42 as against 8,592. The disadvantaged female:male ratio within the total cabin crew workforce moved accordingly in these years from above 9:1 to 7:1; but the advantaged ratio remained almost constant, at marginally (1:1.07) above and latterly at 1:1. Everything therefore depended on where the ET decided to focus its analysis. 14. The ET said: “23. ….[W]e consider it relevant to look at both the advantaged (full-time, part-time and job share cabin crew) and disadvantaged groups (support cabin crew) in order to establish whether there has been discrimination[,] but that our focus must be on the disadvantaged group. 24. The inability to receive an increment is a detriment as it results in the employee of more than five years standing receiving a lower rate of pay than she otherwise would. Having regard to Allonby v Rossendale College [2001] ICR 1189 we consider the pool for comparison to be all the cabin crew, whether full-time, part-time, job share or support.” There is no complaint about §24. The issue is the last proposition in §23: “our focus must be on the disadvantaged group”. 15. The employment tribunal made an unimpeachable analysis of the figures summarised above and concluded that the ratio of the global percentage of advantaged women to that of advantaged men in 2002 was so small (99.53%:99.93%) as to be insignificant. Next, however, they compared the ratios affecting the disadvantaged group and found, in summary, that a comparison of the overall female:male ratio of about 2:1 with the female:male ratio in SCC of 14:1 showed a relative disadvantage of 86.67%. They concluded: “27. The proportion of women in the disadvantaged group is considerably higher than the proportion in the advantaged group. In the context of the respondent’s cabin crew workforce which is comprised predominantly of women we consider that this reveals, and we find[,] that the policy[,] criterion or practice of not paying increments to support cabin crew was to the detriment of a considerably larger proportion of women than of men.” 16. The issue is not the tribunal’s logic, which is faultless, but its premise. Contrary to what has been suggested, the tribunal in this passage do not ignore the larger group: they correctly analyse the situation of women within it and find no significant disadvantage. They then carry out a comparative exercise between the advantaged and the disadvantaged group and reach the conclusion which I have quoted. It is important to set out the terms (italicising the key phrases) in which they do it: “26. We considered the proportion of women to men within the overall workforce of the cabin crew. In 2002 the total female cabin crew (including support cabin crew) was 8,994 and the total male cabin crew (including support cabin crew) was 4,133, a proportion of 2.18:1. Women in both groups [viz advantaged and disadvantaged] comprised a considerably larger proportion than men. In respect of the disadvantaged group, the women comprise 42 out of 13,127, the men comprise 3 out of 13,127, the proportion of women disadvantaged to the proportion of men disadvantaged amount to 14:1. In the disadvantaged group 86.67% more women than men are disadvantaged…” 17. The EAT set out the tribunal’s findings and said this: “59. ….. As can be seen, the Tribunal directed itself by reference to Rutherford (No.1) and of course did not have the judgment of the Court of Appeal in Rutherford (No. 2). It is now plain from our discussion of the legal principles that it is an error of law to have reversed the focus, as here, from the advantaged to the disadvantaged group.” 18. The EAT went on to hold – as the employment tribunal had also held - that the disparities within the larger group were not significant. They concluded: “Although the impact of course on the individual claimants in the disadvantaged group is serious, the infection is contained.” This passing remark highlights the real difference between the Equal Pay Act criteria and the wider approach now taken. The Equal Pay Act, at least on the face of it, is concerned with the individual treatment of such “infections” (and to a limited extent with their aetiology), but not with their epidemiology. 19. Since I shall have to go through some of the case-law, it is no disrespect to the EAT to cite only their conclusion on it, which was, as they themselves make clear, the basis of their oversetting of the employment tribunal: “…[T]he correct approach is to focus on the advantaged group and not the disadvantaged group. It is not incorrect to look at other proportions and other numbers before finally focusing on the advantaged group.” 20. 21. If this is the law, first-instance tribunals, not least the one which decided this case, as well as employers seeking to get their pay structures right and unions trying to ensure that they do so, must be in despair. If the principle is that the sole relevant group is the advantaged one, why look at “other proportions and other numbers” at all? If it is right to look at them, what legitimate bearing do they have? To understand what has driven the EAT to this unsatisfactory position it is necessary to consider the majority decision of the House of Lords in Secretary of State for Trade and Industry v Rutherford (No.2) [2006] ICR 785, albeit that decision postdates the EAT’s. Another division of this court – see Chaudhary v BMA [2007] EWCA Civ 789 - has been considering Rutherford (No.2) concomitantly with us, and our 22. 23. judgment in the present case has been held back to take account of the outcome, to which I will come. But before doing so, it is important to note the differing contexts in which the assessment of disparate impact has arisen and can arise. Both the Strathclyde and the Marshall cases, cited in paragraph 6 above, were equal pay cases. Neither of them, however, dealt with the constitution of the pool where a s.1(3) issue arose. Rutherford, which does deal with the constitution of a pool, does so in relation to the impact on men and women respectively of an upper age limit of 65 for redundancy and unfair dismissal rights in the light of article 141 of the EC Treaty, which requires member states to apply “the principle of equal pay for male and female workers for equal work or work of equal value”. It does not necessarily follow that such principles as can be derived from Rutherford about the constitution of the pool for the purposes of article 141 can simply be read across into the s.1(3) exercise; but the starting point is undoubtedly that the Equal Pay Act has to be read and applied so as to give effect, where required, to article 141, with the result that any principles of law governing the latter will govern the former too. First, then, what are the principles to be derived from Rutherford? I respectfully share the difficulty of the other division of this court (Mummery, Smith and Maurice Kay LJJ) in finding a common ratio among the speeches in their Lordships’ House. I would respectfully express my sympathy with Carnwath LJ’s remarks on this question. Like the division which decided Chaudhary, I think it is safe to take the reporter’s headnote as representing the essence of the decision: “The applicants were male employees dismissed by their respective employers when they were over 65. On their claims for compensation for unfair dismissal and redundancy pay, to which, the employers being insolvent, the Secretary of State was respondent, the employment tribunal disapplied sections 109(1)(b) and 156(1)(b) of the Employment Rights Act 1996 in so far as they precluded claims by employees aged over 65 on the ground that the upper age limit provisions had a disparate impact on men that could not be objectively justified and were, accordingly, contrary to the principle of equality of pay between the sexes in article 141 EC. The Employment Appeal Tribunal allowed an appeal by the Secretary of State, holding that, in considering disparate impact, the employment tribunal had selected the wrong pool and should have taken the entire workforce to which the age limit provisions applied, and that on consideration of that pool those provisions did not have a disparate impact on men. The Court of Appeal dismissed appeals by the applicants. On appeal by the applicants – Held, dismissing the appeals, (per Lord Scott of Foscote, Lord Rodger of Earlsferry and Baroness Hale of Richmond) that article 141 EC sought to ensure that men and women in the same position in the workforce, doing equal work, received equal pay, and the appropriate group for comparison, in relation to the statutory provisions denying the right to compensation for unfair dismissal and redundancy pay to those over 65, comprised all those still in the workforce at age 65; that, as those provisions applied to the same proportion of women in that group as men, there was no indirect sex discrimination; (per Lord Walker of Gestingthorpe) that the pool for the purposes of comparison comprised all those 24. 25. employed persons on whom rights were conferred by the 1996 Act, and, making a comparison between the proportions of men and women in that group advantaged by the cut-off age, which was the appropriate approach, there was no significant disparity; (per Lord Nicholls of Birkenhead) that in the make-up of the group comprising those not adversely affected by the cut-off age there was virtually nothing to choose between men and women, and, although a higher proportion of men than of women continued in employment after age 65, the ratio of about 1:1.4 in a group comprising only about 1.2% of the workforce was not sufficient to establish the degree of disparity necessary to establish that the legislation had an adverse impact on a substantially higher proportion of men than of women…” It is evident that, in the absence of any mechanism in article 141 for determining adverse impact, domestic courts applying that provision have to resort directly to an analysis of systemic adverse impact. Because there is no personal equation, the choice of pool is critical. But that is not the case in an Equal Pay Act claim. The Equal Pay Act sets out its own mechanism for determining whether the terms of a contract of employment breach the principle of equal pay for equal work – a mechanism which initially has nothing to do with systemic discrimination but depends entirely on a straightforward comparison between the woman’s contract and that of a man employed on like work: only two people are required. Where the comparison reveals a disparity, it falls to the employer to show, if he can, that there is a valid reason for it other than the sex of the particular employee, which ex hypothesi cannot be admitted as a valid reason. The harmonious construction of the legislation to which Lord Browne-Wilkinson’s speech in the Strathclyde case was directed means, as he indicated, that if the disparity turns out to be either directly or indirectly sexually discriminatory, the defence will fail unless the disparity can be independently justified. The reason for this lies in s.1(3) itself: a discriminatory contractual provision, absent independent justification, is due to the difference of sex. But is the converse true: does proof that there is no systemic discrimination in the contractual disparity afford a complete defence regardless of justification? If the question were free of authority one might well have difficulty in seeing why it should. Not only does nothing in s.1(3) seem to suggest it; it appears inconsistent with the statutory scheme, which is to afford a remedy for individual pay discrimination subject only to a defence of justification which, as Lord Browne-Wilkinson explains, will fail by reason of the proviso in s.1(3) (“not the difference of sex”) if the justification itself depends on discrimination. It might have seemed that, in giving its chosen statutory effect to the open-textured requirement of the Treaty, Parliament had left no room for any resort to an argument that there is no systemic indirect discrimination by way of defence to a claim by a woman who is being paid less than a man for doing the same work. It is only in non-contractual relationships governed by the Sex Discrimination Act that Parliament has required affirmative proof of either direct or systemic discrimination. But we are bound by the proposition in Lord Nicholls’ speech in Marshall, explained in the course of the speech, that “if the employer proves the absence of sex discrimination he is not obliged to justify the pay disparity.” It appears to follow that an employer who is paying a woman less than he is paying a man to do the same job will have a good defence if his workforce is large enough for her case to be statistically insignificant, while a small employer in the same situation will fail. 26. 27. I return therefore to the question presented by this appeal. I do so in the light of Lord Nicholls’ statement in Marshall that whether a pay disparity has a disparately adverse impact on women is primarily a question of fact. This is how the employment tribunal in the present case approached it. But, as Lord Nicholls’ use of the word “primarily” reminds us, like all questions of fact it is bounded by principles of logic, departure from which is an error of law. What are these principles? In Rutherford the pool was held to be the entire national over-65 workforce. No universal principle of law dictated this. It was the pool which made most sense - indeed the only pool which made sense at all - in the light of the issue being canvassed. The consequence was that the claims had to fail, since everyone in the pool regardless of gender was shut out by the material condition. Such a pool will frequently be available, and with the same consequence; but no principle of law compels tribunals of fact to select it. Indeed, if there were such a principle, it is hard to see how indirect discrimination claims could ever succeed. The correct principle, in my judgment, is that the pool must be one which suitably tests the particular discrimination complained of: but this is not the same thing as the proposition that there is a single suitable pool for every case. In fact, one of the striking things about both the race and sex discrimination legislation is that, contrary to early expectations, three decades of litigation have failed to produce any universal formula for locating the correct pool, driving tribunals and courts alike to the conclusion that there is none. One need go no further than the detailed examination of the statistical issues in the speech of Lord Walker in Rutherford (No2) to see why. 28. Possibly the most helpful exegesis we have been shown is that of Cox J in the EAT decision in Armstrong v Ministry of Defence [2004] IRLR 672. Having explained why, in the EAT’s judgment, “the concept of indirect discrimination under the Equal Pay Act, when read together with European equal pay legislation and case law, is broader than that which applies under the Sex Discrimination Act 1975”, Cox J held (§46): “The fundamental question for the tribunal is whether there is a causative link between the applicant’s sex and the fact that she is paid less than the true value of her job as reflected in the pay of her named comparator. The link may be established in a variety of different ways, depending on the facts of the case. It may arise, for example, as a result of job segregation or from pay structures or pay practices which disadvantage women because they are likely to have shorter service or to work less hours than men, due to historical discrimination or disadvantage, or because of the traditional social role of women and their family responsibilities.” Carrying this broad methodology into the assessment of adverse impact, the tribunal will be concerned to make a comparison which illuminates such of these questions as seem to them potentially critical (here, for instance, the need for female cabin crew with childcare responsibilities to have shorter and more flexible working hours), and to find a pool which best helps them to do this. A pool so narrow that no comparison can be made at all is unlikely to serve this end; nor a pool so large that the comparison is no longer of like with like; though in rare cases the nature of the claim may drive the tribunal to one of these extremes and make the claim self-defeating. 29. 30. 31. 32. In Chaudhary this court, albeit obiter, has gone to the opposite extreme to Rutherford: it has held that the relevant pool for testing a bar imposed by the BMA on financially supporting race discrimination claims was not (as had been argued) the membership as a whole but only those members wanting to bring race discrimination claims. Since this excluded all members regardless of race, there could be no adverse racial impact. Here, too, the chosen pool – this time, however, a very small one – was such that the claim had to fail. The dilemma for fact-finding tribunals is that they can neither select a pool to give a desired result, nor be bound always to take the widest or narrowest available pool, yet have no principle which tells them what is a legally correct or defensible pool. The EAT’s brave attempt in the present case to distil a principle simply enhances the uncertainty. But it is arguable that the single proper pool in Chaudhary was neither of the pools which were canvassed: it was members seeking support for legal claims. If within that pool there was a blanket refusal to support race discrimination claims, it might well have had a substantial adverse impact on ethnic minority members; but it would have been open to the BMA to justify this on the ground – for example - that experience showed the ratio of cost to success in race discrimination cases to be so poor that funds could legitimately be withheld from these cases (apart possibly from exceptionally strong ones) as a class. Rutherford (No 2) seems to me to be a striking illustration of Lord Nicholls’ proposition that the assessment of disparate impact is a question of fact, limited like all questions of fact by the dictates of logic. In discrimination claims the key determinant of both elements is the issue which the claimant has elected to pose and which the tribunal is therefore required to evaluate by finding a pool in which the specificity of the allegation can be realistically tested. Provided it tests the allegation in a suitable pool, the tribunal cannot be said to have erred in law even if a different pool, with a different outcome, could equally legitimately have been chosen. We do not accept that Rutherford is authority for the routine selection of the widest possible pool; nor therefore that any question arises of “looking at” a smaller pool for some unspecified purpose short of determining the case. A practical example is Price v Civil Service Commission [1978] ICR 27. The claimant, who had married young and brought up children before seeking employment, found herself excluded from the Civil Service by a rule that applicants had to be under 28 years old. The EAT rejected the Civil Service Commission’s argument that the pool for determining whether the rule discriminated indirectly against women was the entire national workforce (in which the proportion of women to men was about equal). It considered that there was much to be said for the view that the pool was those men and women who were qualified for civil service employment. That was a pool which made sense of the issue: was a significantly smaller proportion of women than of men able to comply with the requirement to be under 28 in order to be a candidate for a civil service job? To have taken the national workforce would have been to empty the issue of reality; to have taken only those who were 28 or over would have assumed the legitimacy of the very rule that was in issue. There was therefore a logical reason to take neither, and a sound reason to take a pool – potential applicants both over and under 28 - which enabled like to be compared with like, but excluding the disputed criterion. 33. I mention the comparison of like with like because it seems to me a relevant guide. Section 5(3) of the Sex discrimination Act 1975 requires comparisons made for, inter alia, indirect discrimination purposes to “be such that the relevant circumstances in the one case are the same, or not materially different, in the other”. This seems to me a useful indicator of how the pool should be chosen. It needs to include, but not be limited to, those affected by the term of which complaint is made, which can be expected to include both people who can and people who cannot comply with it. In the present case both of the contending pools have this character, but in critically different proportions. 34. Mr Jeans has drawn our attention to the reasoning of Wall J, as he then was, in Rutherford (No 2) in the EAT. Wall J pointed out that the disadvantaged group, as it was presented to the tribunal, was so small as to make any reliable conclusion on disparate impact impossible, while the wider pool (the advantaged group) clearly showed none and was therefore conclusive. Such an approach is not dictated by any legal principle: it is a practical way – sometimes the only practical way - of looking at the available evidence. 35. In my judgment the pool taken by the employment tribunal in the present case passes this test. It has to be kept in mind, as I said earlier in this judgment, that the test is being applied in a case like the present one not in order to see if a complaint of indirect discrimination is made out but in order to see whether a disparity in pay is explained by a material and non-discriminatory factor. The tribunal’s task, subject to what the exercise will logically bear, is to identify a cohort within which the defence can be objectively tested. The Watford tribunal did this, looking at both the advantaged and the disadvantaged groups but focusing on the latter, and found systemic discrimination with the result that the defence failed. Mr Jeans’ case is that the reverse process would – as Ms McNeill accepts – show no systemic discrimination; but unless he can demonstrate an error of law in the tribunal’s choice, he faces a finding of fact which, like many findings of fact, may be debatable but is legally impregnable. In my judgment no such error can be shown. Contrary to Mr Jeans’ unqualified submission, there is no principle of law which requires the fact- finding tribunal always to base its test of disparity on the advantaged cohort. British Airways must live with the finding of disparate impact. 36. This leaves the issue of justification, on which the EAT indicated that it would have upheld the employment tribunal, and on which Mr Jeans therefore cross-appeals contingently on Ms McNeill’s appeal on disparate impact succeeding. Argument was reserved on this issue, and must now be heard. Lord Justice Carnwath: 37. I agree that the appellants should succeed on this part of the case for the reasons given by Sedley LJ. In deference to the impressive submissions we have heard (set out in 150 paragraphs of closely-argued “skeleton” arguments, and supported by 35 European and domestic cases in the joint bundle), I feel I should add a few words to explain why in the end the issues raised seem to me relatively straightforward. 38. On the facts of this case, and cutting through the legal complications, there seem to me to be in summary three questions: i) ii) Was there a difference of pay between Mrs Grundy and Mr Wynne? Was the difference substantial or considerable? iii) Was it justifiable? At this stage we are not concerned with (iii). 39. 40. 41. 42. The answer to the first question is easy. It comes from section 2(a)(i) of the Equal Pay Act 1970, which applies where a woman is employed on like work with a man, and the woman’s contract is less favourable. There is no dispute that in this case it is. She is paid £13,582 for the same job for which Mr Wynne is paid £17,499. The second question is a little more difficult. As Sedley LJ has explained, its statutory derivation is not straightforward. Section 1(3) of the 1970 Act provides a defence where the employer proves that the variation is due to a material factor which is “not the difference of sex”. Those words, it has been held, have to be read so as to give effect to the principle of equal pay under article 141 of the EC Treaty. That in turn leads to Article 2(2) of the Burden of Proof Directive 97/80/EC, under which (subject to justification): “… indirect discrimination shall exist where an apparently neutral... practice disadvantages a substantially higher proportion of members of one sex…” The words “substantially” and “considerably” appear to be used without distinction in the European and domestic cases. Whether the test is satisfied is a question of fact for the Employment Tribunal. In the relevant year 2002 (when the SCC contracts were brought to an end, and Mrs Grundy became a part-time cabin crew member) there were 13,127 cabin crew in total. There were 45 in the disadvantaged SCC class: 42 women and 3 men. Although the numbers of SCC were relatively small, they were not insignificant, and the disparity is striking. Furthermore, it reflects a consistent pattern over the period from 1994 to 2002. Thus, in 1994 there were 495 disadvantaged staff out of a total of 9,915: 471 women and 24 men. The reduction in the relative numbers of SCC is unsurprising, since recruitment to the SCC stopped in 1994. The Tribunal found that this difference was substantial. At first blush that appears to be a pure conclusion of fact, which they were entitled to reach. The EAT found an error of law. They thought that the Tribunal had been wrong to have “reversed the focus… from the advantaged to the disadvantaged group” (para 59). That approach was entirely understandable in the light of the then most recent guidance from the Court of Appeal in Rutherford (2). The EAT had earlier quoted the judgment of Mummery LJ in that case, where he had said: “The primary focus is on the proportions of men and women who can comply with the requirement of the disputed rule. Only if the statistical comparison establishes a considerable disparity of impact, must the court then consider whether the disparity is objectively justifiable.” (para 25) He had commented on the risk of “seriously misleading results” which might be caused by concentration on the disadvantaged group, taking the hypothetical example of a requirement with which 99.5% of men can comply and 99% of women can comply: “If the focus is then shifted to the proportions of men and women who cannot comply (ie 1% of women and 0.5% of men), the result would be that twice as many women as men cannot comply with the requirement. That would not be a sound or sensible basis for holding that the disputed requirement, with which the vast majority of both men and women can comply, had a disparate adverse impact on women.” (para 28) 43. 44. Since the House of Lords judgments in Rutherford (2) it is clear at least that the guidance of the Court of Appeal can no longer be taken as authoritative on this point. There is plenty of room for argument about the precise ratio in the House. However, a common theme of the majority judgments, as I read them, is to emphasise the fact- sensitive nature of decisions in this field, and to lean against the application of general rules. (I agree in particular with Sedley LJ’s comments on that case at paragraph 31 of his judgment.) Even the minority judgment of Lord Walker emphasised the limitations of hypothetical statistical examples, of the kind used by Mummery LJ (see para 67). If there is no rule such as that relied on by the EAT, the only ground for questioning the Tribunal’s judgment of fact falls away. Finally, I hope I will be forgiven for adding a comment about the form of the House of Lords judgments in Rutherford (2). This goes back to an issue I raised in Doherty v Birmingham City Council [2006] EWCA Civ 1739. In that case we were faced with the need to analyse six substantive House of Lords judgments in the leading case, four majority and two minority, to extract a test for local authority possession cases against gypsies. In giving the judgment of the court (with the agreement of Tuckey LJ and Neuberger LJ, as he then was), I observed that we would have been assisted by a single majority judgment, “or at least a single judgment on those issues where there appears to have been no material difference within the majority”; we expressed the hope that there might be a wider debate on this and other similar issues, in the process of transition to the new Supreme Court (para 65). 45. More recently, in OBG Ltd v Allan [2007] UKHL 21, Baroness Hale expressed a similar sentiment. Having referred to the majority’s agreement with Lord Hoffmann’s leading judgment in that case, she said: “The least said by the rest of us who take the same view, therefore, the better. There should be no doubt, and no room for argument, about what has been decided and why. Any perceived inconsistency between what I say and what he says is to be resolved in favour of the latter. Indeed, there would be much to be said for our adopting the practice of other supreme courts in having a single majority opinion to which all have contributed and all can subscribe without further qualification or explanation. There would be less grist to the advocates' and academics' mills, but future litigants might thank us for that.” (para 304) 46. The present case might, with respect, have benefited from that approach. A single majority judgment might have given clearer and more manageable guidance for tribunals working in this difficult area. I am happy, however, to concur in the conclusion reached by our colleagues in Chaudhary (and shared by Sedley LJ) that the ratio can for practical purposes be taken from the headnote summarising the reasoning of Baroness Hale. Lord Justice Waller: 47. I agree that the appeal should be allowed for the reasons given in both judgments, which reasons I should stress, I understand to be the same. I say that because I would add my support for the plea of Carnwath LJ in the last paragraphs of his judgment to the effect that where possible there should be a judgment or judgments at the appellate level which make clear precisely what an appeal has decided. That must be as true in the House of Lords as in the Court of Appeal. We should all be conscious that it would be better, and saving of considerable time and expense, not to leave it to others to puzzle out what we are deciding.