United Kingdom House of Lords Decisions

Barry v. Midland Bank Plc [1999] UKHL 38; [1999] 1 WLR 1465; [1999] 3 All ER 974 (22nd July, 1999)

LORD SLYNN OF HADLEY

My Lords,

    The respondent Bank and the Banking, Insurance and Finance Union entered into a Security of Employment Agreement, the version of which is relevant for present purposes being dated August 1991. It provided in clause 20(b) that where an employee was accepted for voluntary early retirement "The terms for Voluntary Early Retirement, whether in circumstances of Redundancy or otherwise, shall be those under Appendix I of the Agreement with cash payments in [Part IV of Appendix I]" By that Appendix cash payments were to be based on a number of weeks' pay depending on the number of years' service and "Service is actual continuous service or continuous pensionable service, whichever is the greater, calculated as at the date of termination". Moreover "The calculation of a week's pay includes normal contractual remuneration" including certain specified payments. It is common ground that the calculation of such a payment is based solely on years of service and final pay. In such a calculation no account is to be taken of fluctuations in pay or hours of work.

    Mrs. Barry, in whose contract of employment these provisions were incorporated, worked for the Bank from 2 July 1979. She worked full-time (35 hours a week) until she took maternity leave; on return from such leave she worked from 8 October 1990 for 35 hours on alternate weeks so that she was treated as working the equivalent of 17.5 hours a week and she was paid half of the full-time salary. On 15 December 1992 her application to be made voluntarily redundant was accepted and her employment was terminated with effect from 30 April 1993. In the result she had worked 11 years full-time and 2½ years part-time. The total continuous service (which draws no distinction between full-time and part-time service) was thus 13½ years, and that produced a cash payment of 42 weeks' pay based on her final part-time salary giving a total of £5,806.08.

    She contended that this was discriminatory and led to her being treated less favourably than she should be contrary to the Sex Discrimination Act 1975, section 1 of the Equal Pay Act 1970 and to Article 119 of the EC Treaty. The Industrial Tribunal, the Employment Appeal Tribunal and the Court of Appeal for different reasons rejected her claim: she now appeals but no longer relies on the Sex Discrimination Act 1975.

    At first sight it may seem surprising and even unfair that just because she changed to part-time work her full-time salary should be ignored in calculating this payment. If the full-time salary for her final post had been taken into account, but treating the 2½ years part-time as 1¼ years full-time, she would have received £8,080.80, a substantial increase.

    But the fairness of a general scheme may have to take into account the staff as a whole, and alternate schemes fair to Mrs. Barry might have produced unfairness to other employees. In any event the question is not simply one of fairness but whether she has established a breach of section 1 of the Equal Pay Act 1970 or of Article 119 of the EC Treaty with which section 1 must be read.

    By the former where, as here, it is agreed that a woman "is employed on work rated as equivalent with that of a man in the same employment" then any term of her contract which "is . . . less favourable to [her] 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". By the latter a member state is required to ensure that "men and women should receive equal pay for equal work".

    There seems no doubt that as far as the Act is concerned there is no term in her contract which is less favourable than a term in the contract of a man doing the same work or work of equal value. All relevant employees were covered by the same scheme in the same terms whether they were men or women, fulltime or part time. Nor for the same reason is there any direct discrimination for the purposes of Article 119.

    The question is thus whether Mrs. Barry can establish indirect discrimination by showing (a) that she belongs to a group of employees which is differently and less well-treated than others, and (b) whether that difference affects considerably more women than men and, if she can, (c) whether the Bank can show that the difference in treatment is objectively justified. (Stadt Lengerich v. Helmig [1994] ECR I-5727, Kowalska v. Freie and Hansestadt Hamburg [1990] ECR I-2591 and Bilka-Kaufhaus GmbH v. Weber von Hartz [1986] ECR 1607).

    All employees receive a payment based on their final salary and the number of years' service. Prima facie there is no discrimination but Mrs. Barry says that her group consists of employees who have changed from full-time to part-time treatment and they are less well-treated than those who have changed from part-time to full-time or those who have always been full-time.

    There is no doubt on the basis of a long line of authority that this severance payment is pay within the meaning of Article 119 of the Treaty and that different treatment of part-time workers may constitute indirect discrimination where the great majority of part-time workers are women.

    The first question which arises is whether there is a difference in treatment at all between full-time and part-time workers for the purposes of the Act and the Treaty. In that regard it is not sufficient merely to ask whether one gets more or less money than the other. It is necessary to consider whether, taking account of the purpose of the payment, there is a difference in treatment. The purpose of the payment here is to provide support for lost income during the period immediately following redundancy. As the Industrial Tribunal put it, it is "to cushion employees against unemployment and job loss" (decision 15 June 1995). It is not to remunerate for past service (when it would be necessary to have regard to actual service at different periods) even if the payment takes into account years' of service to reflect loyalty to the employer. See Kowalska v. Freie and Hansestadt Hamburg case [1990] E.C.R.-1 2591 and Barber v. Guardian Royal Exchange Assurance Group [1990] ICR 616. In the latter case at page 668 the European Court of Justice said that a redundancy payment "makes it possible to facilitate his adjustment to the new circumstances resulting from the loss of his employment and which provides him with a source of income during the period in which he is seeking new employment."

    The weekly amount lost during the redundancy period is thus the amount of salary being paid at the end of the employment; it is not therefore a relevant difference in treatment to base all employees' severance payments on their final salary. In principle the position is the same here as under the statutory scheme for redundancy pay and the statutory scheme for payment in lieu of notice when payment is related to years of service but is based on final salary. The payment reflects the actual salary the employee would have received during the notice period and not some notional amount calculated on types and hours of service over the whole period of employment. In the present case there is no relevant difference in treatment because all employees, men and women, full-time and part-time, of all ages, receive a payment based on final salary. To relate severance pay to the number of hours proportionately worked at redundancy is, as I see it, consistent with the judgment of the European Court in the Kowalska case. (supra)

    This approach seems to me also to be consistent with the approach of the European Court in Helmig (supra) at paragraphs 21--25 where it was held that there was no difference in treatment where part-time workers received the standard rate for basic hours of work but an extra rate was paid for hours above 38 per week. The full-time workers got the increase over the normal contractual hours whereas the part-time workers did not get the increase until they did more than 38 hours a week and so did not get it when they had completed their normal contractual hours. The crucial factor, however, was that they got the same rate for the first 38 hours.

    This approach is not inconsistent with the decision of the Court of Justice in Kuratorium fur Dialyse und Nierentransplantation e. V. v. Lewark [1996] E.C.R. I-1243 where the sole question was whether it was discriminatory for full-time workers to be paid but for part-time workers not to be paid for hours spent at training courses which they were required to attend.

    This conclusion is the same as that of the Employment Appeal Tribunal [1997] I.C.R. 192 at page 207-208 with which in substance I agree. I would accordingly decide this appeal on that basis alone.

    If I had come to the opposite conclusion I would have accepted that the two groups for comparison are not simply (a) those full-time against those part-time workers, or (b) those full-time men against full-time women, or (c) part-time men against part-time women. The disadvantaged group is, as the Court of Appeal held, those part-time workers whose hours of work at termination were less than the average of their hours of work throughout their service. On that approach the Court of Appeal was entitled to say:

 "It seems to us logical to look at all to whom the relevant provisions of the Security of Employment Agreement applied at the relevant time (viz. the termination of Mrs. Barry's employment), there being no distinction in the scheme between employees in the various grades. . . . In logic Mr. Elias [counsel for the bank] must be right to say that the disadvantaged group is those part-time workers whose hours of work at termination were less than the average of their hours of work throughout that service.

 "It is therefore necessary to ask what proportion of women fell into that category as compared with the proportion of men that fell into that category. . . . In our opinion it would be necessary to look at all part- time workers at the time of Mrs. Barry's termination of employment and the average of their hours of work throughout their service and to compare the men and the women in the advantaged and disadvantaged groups. But there are no such statistics available, as we understand the position."

    The Employment Appeal Tribunal and the Court of Appeal found that if there was prima facie indirect discrimination it was objectively justified. In view of the conclusion to which I have come on the first point it is not necessary to consider this matter or the respondent's contention that the only remedy available to the appellant could be a claim for compensation under the European Court's judgment in Francovich v. Italian Republic [1991] ECR I-5357.

    I would dismiss the appeal.

LORD NICHOLLS OF BIRKENHEAD

My Lords,

    Mrs. Jacqueline Barry worked for Midland Bank, ultimately as a research and adjustment tracer clerk. When she left in April 1993 she received a severance payment of £5,806 under the bank's security of employment agreement. The amount payable under this scheme principally comprises a multiple of final salary, the multiple depending on years of continuous service. Mrs. Barry claims that this manner of calculation of her severance payment constituted unlawful sex discrimination. So far, although for differing reasons, the claim has failed at every level: an industrial tribunal, the Employment Appeal Tribunal and the Court of Appeal. The course of the somewhat complicated proceedings before these tribunals and in the Court of Appeal appears in the judgments of the Employment Appeal Tribunal, reported at [1997] I.C.R. 192, and of the Court of Appeal, reported at [1999] ICR 319.

    The facts can be stated shortly. Of Midland Bank's total workforce (as at December 1992) of 45,420 employees, about two-thirds were women. The bank seeks to retain its female employees after they have had children. It takes steps to facilitate their return to full-time work, but it assists them to work for a reduced number of hours as 'key-time' workers if that is their wish. About one-sixth of the bank's staff are key-time workers, and these are very largely (96 per cent) women.

    Mrs. Barry worked for 35 hours per week as a full-time employee for eleven years from July 1979. After October 1990 she worked key-time. She worked full-time alternate weeks, which is the equivalent of working 17.5 hours per week. This was following maternity leave for the birth of her first child. In 1992 her division moved to a new address, and she accepted an invitation from the bank to apply for voluntary severance. Her employment ceased at the end of April 1993, and she became entitled to a severance payment.

    The bank's security of employment agreement is a collective agreement negotiated between the bank and the Banking, Insurance and Finance Union. It applies to virtually all employees of the bank. It was incorporated into the terms and conditions of Mrs. Barry's employment. The scheme draws no distinction between men and women or between full-time and key-time workers. After one year's service employees qualify for a severance payment of twelve weeks' pay. Each year of continuous further service entitles the employee to a higher payment, based on a further two or three weeks' pay. The ceiling is 90 weeks' pay, for 35 or more years' service. The relevant pay is the employee's weekly pay at the date of termination. The scheme makes adjustments for employees over 40 years of age, those close to retirement age and those with children in full-time education.

    Having worked for the bank continuously for over thirteen years, Mrs. Barry was entitled to a payment equal to 42 weeks' pay. Her complaint is that this method of computing her severance payment failed to recognise that for eleven years she had worked full-time. She was, wrongly, treated on an equal footing with an employee who had worked key-time throughout.

    The bank's scheme is similar to many redundancy schemes entered into by employers, including some other banks. The key features of this scheme are also similar to those of the statutory redundancy scheme, now contained in section 162 of the Employment Rights Act 1996, but the amounts payable under the bank's scheme are significantly more generous. The amount of £5,806 paid to Mrs. Barry included £1,520 payable under the statutory scheme. Payments under the statutory scheme are a factor of age, length of service and weekly pay at the date of termination. In Regina v. Secretary of State for Employment Ex parte Equal Oportunities Commission [1993] I.C.R. 251 the commission, in its second application, challenged the lawfulness of this mode of calculating statutory redundancy pay. The point at issue was the same as in the present case. The challenge failed on a variety of grounds, but both the Divisional Court and, in the Court of Appeal, Hirst L.J. held that the Secretary of State had made good his defence of objective justification.

    Mrs. Barry's claim is not one of direct discrimination. Nor is it a case where comparison with a male comparator would assist Mrs. Barry. The severance pay of a male employee in her position would be calculated in the same way as Mrs. Barry's payment. Rather, the claim is based on article 119 (now article 141) of the E.C. Treaty as interpreted in several well-known decisions of the European Court of Justice. Severance pay is a form of deferred pay, within the meaning of article 119, to which an employee becomes entitled in respect of his employment: see Barber v. Guardian Royal Exchange Assurance Group [1990] ECR I-1889, 1949 and Kowalska v. Freie und Hansestadt Hamburg [1990] ECR I-2591, 2611.

    A preliminary point is to be noted. Although severance pay is a form of pay, not all forms of pay have the same purpose. The purpose of a payment in lieu of notice is to compensate for loss of the salary an employee would have received during the notice period. The previous employment history is irrelevant when calculating such a payment. An employer who pays a flat-rate bonus to all his employees at the end of the year seeks thereby to recognise and reward the equal commitment of each employee. That is the purpose of a payment in that form. The senior employees, predominantly male, could hardly advance a claim that a flat-rate bonus payment was unlawful and that the bonus must be proportionate to salary. Bonuses can be, and often are, calculated proportionately to salary, but a proportionate bonus has a different purpose: recognising and rewarding the value of each employee's contribution. A flat-rate bonus and a proportionate bonus share the same broad purpose, of rewarding employees for their work during the year, but they seek to reward different aspects of the employees' work. Thus, and this is a further point to be noted at the outset, in some instances a change in the method of calculating a payment may be achievable only by changing the purpose of the payment. The method of calculation and the purpose sought to be achieved may go hand in hand. When this is so, a crucial issue is likely to be whether the object sought to be achieved was objectively justified in all the circumstances.

    The established approach to cases of indirect discrimination involves a three-stage review: (1) whether there was a difference in treatment between two groups of employees; (2) whether this difference had a disparately adverse impact on women; and (3) whether there were objective factors, unrelated to discrimination on grounds of sex, justifying the difference in treatment: see, for instance, Stadt Lengerich v. Helmig [1994] ECR I-5727, 5753-5754, at paragraphs 22-25.

Disparate treatment

    The complaint calls for careful examination. The complaint is that an employee whose hours of work remained constant throughout receives due recognition for her past record by a severance payment calculated by reference to a multiple of final salary and years of service. This is so with an employee who worked full-time throughout. It is equally so for a key-time worker whose hours of work remained unchanged throughout her employment. But, and this is the heart of the complaint, the scheme makes no allowance for employees whose hours of work fluctuated. If an employee's average hours of work per week while employed exceeded her hours of work when her employment ended, she receives no credit for the excess. She is treated the same is if she had throughout worked the lesser number of hours she was working at termination. Conversely, it should be added, an employee's severance pay is not reduced when her average hours while employed were less than her hours of work at the end of her employment.

    Although not so expressed in argument, the claimant's case is an attack on both of the principal factors used in the bank's scheme when calculating severance pay. Her complaint is that when the average hours worked exceed the employee's final hours of work final salary should not be used in the calculation. Instead, key-time work should be converted into its full-time equivalent, and this should be applied to the full-time equivalent of the key-time worker's final salary. Thus, and this is the way she formulated her financial claim, the amount payable to Mrs. Barry would be arrived at as follows. Eleven years full-time work and two years key-time work at 17.5 hours per week are equivalent to twelve years full-time work. She should have been paid an amount calculated by reference to twelve years work and, not her actual final salary, but the salary she would have been paid had she been working full-time when her employment ceased. This calculation yields a figure of £8,080, a difference of £2,274.

    A scheme structured in this way would be a significantly different scheme from the present scheme. The present scheme is based on years of service and actual final salary. A scheme structured as claimed would, rationally, have to make corresponding downward adjustment in the calculation of the payments to employees whose average hours of work were less than those being worked when employment ended. Thus the re-structured scheme would be based neither on years of service nor on actual final salary. It would be based on hours of service and rate of final salary.

    For better or for worse, a scheme structured in this way would depart from the aims of the present scheme. The broad purpose would be the same, but the method chosen as the most appropriate means to achieve the purpose would be different. This is not in itself an answer to the claim, but it is a step towards identifying the core issue. The overall purpose of the bank's scheme is to deal with reorganisations and redundancies as fairly as possible and to cushion employees against unemployment and job loss. This would equally be true of the restructured scheme. But there is more than one way of calculating the amount of the financial cushion. Factors which may be taken into account include hours of service, years of service, rates of pay throughout service, actual pay throughout service, final rate of pay, actual final salary, age and financial responsibilities. The selection of which factors shall be used in preference to others depends upon the objective sought to be achieved. The bank's preferred aim is to have regard to the financial loss suffered by the employee on severance, measured by actual final salary, and to length of service and loyalty, measured by years of service. Those are the objectives of the scheme. Past employment history is not otherwise a relevant factor. For all employees severance payments are made irrespective of salary, grade and place of work before termination.

    In contrast, the restructured scheme would disregard the amount of the salary actually lost on severance. The restructured scheme would reward service, but it would take no account of actual years of service. Instead, it would have regard to hours of service. An employee who had worked full-time for two years would be treated the same way as a key-time employee who had worked 10 hours a week for seven years. Thus, the restructured scheme correlates severance pay with work done in the past. Some employees might regard this as a fairer scheme, as it places more emphasis on rewarding past service. Other employees might regard it as less suited for the overall purpose of a severance payment, because it fails to take into account the actual financial loss suffered on severance. The question raised by Mrs. Barry's claim is whether it was unlawful for the bank to adopt a scheme with the objectives described, at any rate where this would have a disparately adverse impact on women when compared with a restructured scheme having a different objective.

    Miss Booth Q.C. contended that this question falls to be considered at stage (3) (the objective justification stage). The preferred approach of Mr. Elias Q.C. was that this issue should be addressed at stage (1) (the difference in treatment stage). He advanced a submission to the following effect. The first stage raises the question whether the principles of discrimination are applicable at all. Article 119 is concerned with discrimination: equal pay for men and women for equal work or work of equal value. Discrimination means treating like cases differently or, as is claimed in the present case, treating unlike cases the same: see Hill v. Revenue Commissioners [1999] ICR 48, 68, paragraph 22. Mrs. Barry worked full-time for eleven years and key-time for 17.5 hours per week for two years, but she is treated the same as a key-time worker who had worked for 17.5 hours per week for thirteen years. Cases are 'unlike', and ought to be treated differently, when the difference between them is material for the purpose in hand. The purpose in hand here is the calculation of severance payment. Whether the difference relied upon by Mrs. Barry as a material difference should be so regarded depends upon whether the bank could properly adopt a severance payment scheme with the objects of this scheme. If the bank could do so, the difference relied upon is immaterial. Her hours of work history is irrelevant. In that event, discrimination not having been established, Mrs. Barry's claim would fail at the first hurdle. The question whether the alleged discrimination had a disparate impact on women would not arise. Nor would the further question whether the alleged discrimination was objectively justifiable. The materiality issue can be addressed at the objective justification stage, but a more natural approach is to treat an immaterial difference as negativing the existence of discrimination rather than justifying its existence.

    The argument has some attraction but I am unable to accept it, for two practical reasons. First, an evaluation of the legitimacy of a severance pay scheme using the factors adopted by the bank is better made after the existence and extent of any disparately adverse impact on women have first been considered. A responsible employer takes into account such disparate impact, if there be any, when considering which scheme to adopt. The second reason is the important procedural consideration that at stage (3), unlike stage (1), the burden of proof rests upon the employer. When an employer asserts that an apparent difference between two cases is immaterial, the burden of proof properly rests upon him. Unless, therefore, the immateriality of the difference relied upon is plain and obvious, which is not so in the present case, this issue will usually be better considered at stage (3).

    This course is consistent with the approach adopted by the European Court in cases such as Bilka [1986] ECR 1607, 1627, and Nimz v. Freie und Hansestadt Hamburg [1991] ECR I-297, 319, paragraph 12. In Kowalska [1990] ECR I-2591, 2611-2612, paragraphs 13 and 14, part-time workers were excluded from severance grants. The defendant contended that part-time workers did not need this temporary assistance because they did not provide for the needs of themselves and their families solely from their employment income. The court considered this contention as an argument going to objective justification. By way of contrast, the Helmig case, mentioned above, is an example of a plain case. Part-time workers complained that pay at overtime rates was only available once an employee had worked for longer than the ordinary working week of a full-time worker. They contended they were treated differently from full-time workers because, unlike full-time staff, they were not paid overtime for each hour in excess of their contracted hours. The court held there was no difference in treatment because part-time and full-time workers received the same amount of money for the same number of hours worked. In other words, the purpose of pay in the form of salary was to remunerate for hours actually worked. The court decided that, in the light of this conclusion at stage (1), it was not necessary to proceed to consider objective justification.

Disparately adverse impact on women

    In order to decide whether the bank's scheme has a disparately adverse impact on women, a comparison must be made between, on the one hand, the respective proportions of men in the bank who are not disadvantaged by the difference in treatment of which complaint is made and those who are disadvantaged and, on the other hand, the like proportions regarding women in the workforce: see Regina v. Secretary of State for Employment Ex parte Seymour-Smith [1999] IRLR 253, paragraph 59. These proportions by themselves can be misleading, because they are affected by the comparative sizes of the non-disadvantaged group and the disadvantaged group. The smaller the disadvantaged group in proportionate terms, the narrower will be the differential. Take an employer whose workforce of 1,000 comprises an equal number of men and women. Ten per cent. of the staff (100 employees) work part-time, and of these 90 per cent. are women. A scheme which disadvantages part-timers will disadvantage 10 men (2 per cent. of the male employees) and 90 women (18 per cent. of female employees). If the figures were the same save that the total workforce was 10,000 employees, the disadvantaged part-timers would comprise 10 men (0.2 per cent. of male employees) and 90 women (1.8 per cent. of female employees). A better guide will often be found in expressing the proportions in the disadvantaged group as a ratio of each other. In both my examples the ratio is 9:1. For every man adversely affected there are 9 women. Absolute size, in terms of numbers, remains relevant, since a low ratio may be of little significance in a small company but of considerable significance in a large company.

    In the present case the figures needed to make these comparisons are not available. The allegedly disadvantaged group comprises all employees at the date of Mrs. Barry's termination whose average hours of work exceeded their current hours of work. There are no statistics on the size or composition of this group. The Court of Appeal regarded this as fatal to the claim, although the court was reluctant to decide the appeal on this ground. I consider that, although unsatisfactory, the figures available are sufficient to enable an inference to be drawn regarding the composition, but not the size, of the disadvantaged group. The figures show that, in the Midland Bank as elsewhere, a disproportionately high percentage of key-time employees are women. By disproportionately high I mean that the proportion of female key-time workers is considerably higher than the proportion of women in the entire workforce. The figures also show that in the two years (1992 and 1993) for which statistics are available, the employees whose hours of work fluctuated during the year were overwhelmingly women. It is a reasonable inference from these two facts that the disadvantaged group is composed very largely of women. A pre-requisite for entry to the disadvantaged group is that the employee's hours of work have fluctuated. This characteristic appears to be confined principally to women.

    I should add that it was on this point that the claim failed before the industrial tribunal. The tribunal took, as the groups for comparison, the full-time and the key-time employees who were of the same grade and in the same department as Mrs. Barry at the time of her redundancy. The proportion of women to men in both of these groups was much the same. Both parties now accept that these are not the relevant groups.

Objective justification

    I turn to the question of objective justification. In Bilka-Kaufhaus GmbH v. Weber von Hartz [1986] ECR 1607, a department store pursued a policy of excluding part-time workers, mostly women, from an occupational pension scheme. The employer's case was that the exclusion of part-time workers was intended to discourage part-time work, since part-time workers generally refused to work in the late afternoon or on Saturday. The European Court held, at 1628, it was for the national court to determine whether and to what extent the ground put forward by the employer might be regarded as an objectively justified economic ground. The court added:

 'If the national court finds that the measures chosen by Bilka correspond to a real need on the part of the undertaking, are appropriate with a view to achieving the objectives pursued and are necessary to that end, the fact that the measures affect a far greater number of women than men is not sufficient to show that they constitute an infringement of article 119.'

More recently, in Enderby v. Frenchay Health Authority [1993] ECR I-5535, 5575, the European Court drew attention to the need for national courts to apply the principle of proportionality when they have to apply Community law. In other words, the ground relied upon as justification must be of sufficient importance for the national court to regard this as overriding the disparate impact of the difference in treatment, either in whole or in part. The more serious the disparate impact on women or men as the case may be, the more cogent must be the objective justification. There seem to be no particular criteria to which the national court should have regard when assessing the weight of the justification relied upon.

    The industrial tribunal found that the objectives of Midland Bank's scheme were to compensate for loss of job and for loyalty to the bank. Length of service and loyalty to the bank which that denotes is an element of the scheme and this increases in importance as length of service increases. This is an unexceptionable finding. So far there is no difficulty. The tribunal also found that 'a scheme which fails to take into account, to a significant extent, the full service of an employee does not meet the objectives of the scheme'. On this I share the difficulty of the Court of Appeal in understanding what is the evidential or other basis on which the tribunal concluded, in effect, that because it did not take into account hours of service as distinct from years of service, the scheme failed to meet its objectives.

    The tribunal also held that the scheme does not 'need to be in the form it is', and that the 'scheme as drafted is convenient, rather than necessary'. It is clear that, in reaching this conclusion, the tribunal was considering whether the bank's severance pay scheme could have used different factors, as suggested by the claimant. So understood, the tribunal's conclusion on necessity is obviously correct. Not all employers use the same factors as this scheme. Your Lordships were told that at least one retail bank takes into account hours of service. But that is not the relevant question in the present case. In the Bilka case, from which the necessity test is drawn, the ground set up as justification was the need for shop staff in the late afternoon and on Saturday. Self-evidently, that was a legitimate business purpose, unrelated to any discrimination on the ground of sex. The question in that case was whether fulfilment of that purpose necessitated the exclusion of part-time workers from the pension scheme. In the present case the equivalent question does not in practice arise, because the objects of the bank's scheme and its terms are inseparable. The restructured scheme is not an alternative way of achieving the same object as the bank's scheme. No alternative scheme has been suggested which would correlate hours of service as sought by Mrs. Barry without detracting from the primary objects of the bank's scheme. Thus, as already remarked, in the present case the crucial question is whether a scheme with the objects chosen by the bank is lawful, despite its discriminatory consequence when compared with another scheme which has a different object and, hence, uses different factors in calculating the amount payable.

    I turn to this crucial question. I consider that the factors used in the bank's scheme are inherently apt for calculating severance pay. Severance pay is, or may properly be treated as, compensation for loss of a job. Loss of a job entails loss of the actual salary then being paid. The longer an employee held the job, the greater the disruption. Older employees may be expected to have more difficulty in obtaining a new job, and immediate financial problems may be more acute for employees with children at school. The bank's scheme gives some recognition to these considerations. There could perhaps be a case where volatility in hours of work was so marked that an employee's salary at any one time was not a sound guide to his current earnings. In such a case, use of final salary as a factor might produce capricious results. That is not this case. Mrs. Barry's hours of work, for instance, changed only once in thirteen years, and by April 1993 she had been working for 17.5 hours per week for over two years. As to loyalty, the loyalty of an employee who works key-time for many years is equal to that of an employee who works full-time for the like number of years. Indeed, as pointed out by Peter Gibson L.J., to award full-time employees a higher amount for loyalty than key-time workers might itself be discriminatory against key-time staff as a class and, hence, against women.

    Nor is this a case where the discrimination against women, when compared with a restructured scheme, is altogether clear-cut. There is a lack of evidence on the extent to which the bank scheme has a discriminatory effect on women. Further, under a restructured scheme there would be losers as well as winners. Some employees, such as Mrs. Barry, would benefit, but others would be worse off. By the same token that it is reasonable to infer that the disadvantaged group is predominantly comprised of women, so it is reasonable to infer that the presently advantaged group (those whose average hours are less that their hours at termination) is mostly comprised of women. There is no evidence showing how these groups compare in size. On this the burden of proof lay on the bank. In 1992 the majority of changes in hours by female employees were increases, not decreases, but this is not a satisfactory basis for drawing any clear conclusions about the size or extent of the two groups. However, even in the absence of evidence it is reasonable to infer that under the restructured scheme there would be a significant number of losers, and that these would be mostly women. They would be mostly women because, as already noted, fluctuating hours of work is a characteristic principally confined to women.

    In these circumstances I agree with the Court of Appeal that the bank's scheme is lawful. Its objects are of sufficient importance to override the weight to be given to the fact that under a different scheme with a different object a group of employees, mostly women, would be better off. To decide otherwise would be to compel Midland Bank to abandon its scheme and substitute a scheme where severance pay is treated and calculated, not as compensation for loss of a job, but as additional pay for past work. That could not be right. I am reassured in my conclusion by noting that the same point was decided in the same way by the German federal labour court (the Bundesarbeitsgericht) in case No. 10 AZR 129/92, Entscheidungssammlung zum Arbeitsrecht 247 sub # 112 BetrVG.

    If, contrary to my view, the present scheme were unlawful, a further and difficult question would arise on the nature of the relief to be granted to Mrs. Barry. There is more than one way a restructured scheme can be drawn. In the Equal Opportunities Commission case concerning the statutory redundancy scheme Dillon L.J. observed, at page 270, that there is no valid reference point against which the court can calculate redundancy pay for an employee in a position comparable to Mrs. Barry in the present case. Since the point does not arise I will say no more about it.

    Neither party has suggested there is a point of law or principle on which your Lordships should seek guidance from the European Court. I would dismiss this appeal.

LORD STEYN

My Lords,

    This appeal concerns the method adopted by the bank for calculating redundancy payments. The feature of the contract of employment which is challenged is the calculation of redundancy payments by multiplying salary as at the date of employment by the number of years of active continuous service. A part-time worker, who was previously employed full-time, alleges that in effect the scheme offends against the principle of equal pay for equal work. I regard this as a difficult issue. At the end of the hearing I was inclined to the view that in its actual operation the scheme has a discriminatory effect. On further consideration I have been persuaded that properly analysed the appellant's claim is that she has been treated less favourably than she would have been treated if a different scheme, with different objectives, giving greater credit to a feature to her advantage, had been adopted. I would now on the special facts of this case accept the argument of Mr. Elias Q.C. that the rules of discrimination are not engaged. In my view therefore the scheme does not offend against the principle of equal pay for equal work and is therefore not unlawful.

    I would make clear that if I had not been so persuaded on the first issue, I would have had no difficulty in ruling that the bank should fail on objective justification. And I would not have been troubled by the task of devising appropriate relief.

    I would dismiss the appeal.

LORD HOFFMANN

My Lords,

    I have had the advantage of reading in draft the speech of my noble and learned friend Lord Nicholls of Birkenhead. I agree with him that the appeal should be dismissed. But I have arrived at this conclusion by a somewhat shorter route which he has felt unable to use. I must therefore explain my reasons. For this purpose, I gratefully adopt my noble and learned friend's statement of the facts and issues.

    The general principle stated in Article 119 is that "men and women should receive equal pay for equal work." "Pay" is defined in the widest possible terms as any consideration, in cash or kind, "which the worker receives, directly or indirectly, in respect of his employment from his employer." The Article then gives two examples of what the application of this principle involves. If workers are paid at piece rates, they must be calculated on the basis of the same unit of measurement for both men and women. If they are paid at time rates, the rates for the same job must be the same for men and women. The jurisprudence of the European Court has applied these principles to cases in which the employer's method of payment does not expressly discriminate on grounds of sex, but has in practice an adverse impact upon members of one sex. The most common example is the case in which the method of payment makes no express reference to men or women but is discriminatory against part-time workers. If most part-time workers are women, then prima facie that method discriminates on grounds of sex. So, for example, if part-time workers are paid less per hour than full-time workers, there is prima facie discrimination on grounds of sex. Article 119 is infringed unless the employer can produce an objective justification for paying them a lower hourly rate.

    What Article 119 does not do, however, is to mandate any particular method by which pay should be calculated. So, for example, if part-time workers and full-time workers are employed on the same hourly rates, the part-time workers cannot complain that they would have earned more in relation to the full-time workers if they had been paid at piece rates. Article 119 says only that whichever method is used, it should not discriminate between men and women.

    If one then considers the very wide definition of pay, it will be obvious that some elements in the consideration which the employee receives from his employer will be determined by methods which cannot be described as either time or piece rates. They may be related to age, need, place of work, past service or simply the fact of being an employee. Take, for example, membership of a health scheme. The scheme might provide the same benefits for all employers, skilled or unskilled, full-time or part time. Article 119 would constitute no objection to such a scheme. The high-paid or full-time employees, if mostly men, could not complain that their work is worth more than that of the lower-paid, part-time women and it is therefore unfairly discriminatory not to give them correspondingly better health care. As long as the method used by the scheme is not discriminatory, one sex cannot object that they would have done better as against the other if a different method had been employed. In the example I have given, the method is not discriminatory because there is only one criterion for obtaining the same benefits and that is being an employee. So there is no discrimination between one class of employees and another. Similarly, there would have been no discrimination if the benefits of the health scheme were graded according to the number of weekly hours worked by each employee. This may be an eccentric way of distributing health benefits but it would not be discriminatory between full-time and part-time workers. Their "pay" in the form of the health benefit would be equal for equal work. On the other hand, if the health scheme had been confined to full-time workers, there would be discrimination. The method for distributing the benefits would be by reference to the work done by the employee but not on the principle of equal pay for equal work. The employer would have to produce an objective justification for not giving the part-time workers benefits at least commensurate to their work.

    The approach of the European Court of Justice to cases under Article 119 has consistently been to ask, first, whether the benefit in question is "pay" within the meaning of the article and secondly, whether the method by which it is earned has an effect which offends against the principle of equal pay for equal work. If the answer to the second question is no, there is no prima facie infringement of Article 119. If yes, the question of whether the employer can prove an objective justification is (with or without guidance on the point) remitted to the national court. What the court has never done is to say that although the method by which it is earned does not operate in a discriminatory manner, Article 119 is infringed because a different method would have been more advantageous to one sex or the other. I give some illustrations from the court's jurisprudence.

    In Case 170/84 Bilka-Kaufhaus GmbH v. von Harz [1986] ECR 1607 the benefit in dispute was an employer-funded supplementary pension. This was held to be deferred consideration in respect of employment and therefore "pay" within the meaning of the article. The benefit was paid only to employees who had worked full time for at least 15 out of the previous 20 years. Thus the method was related to the past work of the employee but in a way which disregarded the work of those who had been part-time for more than 5 out of the previous 20 years. As this class was mainly composed of women, the method was held prima facie to infringe the principle of equal pay for equal work. The court gave guidance to the national court on what might amount to objective justification for the discrimination.

    Case C-33/89 Kowalska v. Freie und Hansestadt Hamburg [1990] E.C.R. 1-2591 was a similar case. In that case the benefit was a severance payment on retirement. The court held that it fell within the definition of pay. The method for determining entitlement was that a retiring full-time employee received a grant of up to four months final salary. Part-time workers did not qualify. The payment was related to earnings at retirement and past service, but part-time employees got nothing. Their earnings and past service were disregarded. As the latter class was mainly women, this method was held prima facie to infringe Article 119.

    On the other hand, in Case C-399/92 Stadt Lengerich v. Helmig [1994] ECR I-5727 the benefit in question was overtime pay. This was obviously pay within the meaning of Article 119. The method of payment was that anyone who worked more than the standard working week of 38.5 hours was paid an enhanced rate for the additional hours. The result was that a full-time worker who was contractually obliged to work 38.5 hours a week was paid the enhanced rate for any additional hours. But a part-time worker who was contractually obliged to work only (say) 20 hours a week was paid normal rates for additional hours until her hours for the week exceeded 38.5. The part-time workers (again, mainly women) claimed that the system of payment discriminated against them.

    The court held that the scheme was not discriminatory. Part-time workers received the same remuneration for hours worked as full-time workers. A part-time worker, contractually obliged to work eighteen hours, received (at the ordinary rate) the same pay for a nineteenth hour as a full-time worker received for his nineteenth hour. The question of objective justification did not arise. The applicants' complaint was not that the actual scheme was discriminatory but that a different scheme would have been more advantageous to part-time workers. As Advocate-General Darmon said (at p. I-5733, para. 29):

 "for the uniform criterion of the weekly working hours fixed by statute or by collective agreement [the applicants] seek to substitute a fluctuating one which would vary with the contractual hours and which, on the pretext of removing an assumed inequality of treatment, would in fact give rise to a real inequality because, for the same number of hours worked, some workers would be paid the supplement, and others not."

    If one applies these principles, the answer to the present case would appear to me to be plain. The redundancy payment is clearly pay within the meaning of the Article. The next question is therefore whether the method of payment has a discriminatory effect. It uses two criteria. One is final salary. The other is prior years of service. Neither has any effect which discriminates between full-time and part-time workers. The final salary of all workers is taken into account, whether they are working full-time or part-time. Of course the part-time workers are earning less, but this is because they are working less. There is no infringement of the principle of equal pay for equal work. Likewise, the past years of service of all workers are taken into account, whether they worked full-time or part-time. Again there is no infringement of the principle of equal pay for equal work.

    The applicant's case is that she would be better off if, instead of using a non-discriminatory criterion of past years of service, the scheme had used an equally non-discriminatory criterion of past total hours of service. So she might, although it is worth observing that employees who had recently taken full-time employment after a long period of part-time employment would be worse off. But the question, as I have suggested, is not whether the applicant or even the majority of part-time workers (and therefore the majority of women) would do better under a different scheme. The only question is whether the actual scheme offends against the principle of equal work for equal pay.

    I quite accept that in examining the method of payment adopted by the employer to see whether it has a discriminatory effect, the court is concerned with substance rather than form. The question is how the scheme actually operates. But there is nothing artificial about the criteria adopted in the present scheme. The emphasis is upon final salary because the main purpose of a redundancy payment is to cushion the employer against the hardship of a sudden cessation of his salary, whatever it may have been. On the other hand, the inclusion of past years of service in the calculation recognises that the employer may feel a greater duty to provide such a cushion to those who have worked for him for a longer time. Both criteria are perfectly rational choices for the employer to make. Provided that neither has a discriminatory effect, it does not matter that a different scheme would have been more advantageous to one sex rather than the other.

    In my opinion the scheme does not infringe the principle of equal pay for equal work and for that reason I would dismiss the appeal.

LORD CLYDE

My Lords,

    I agree that the appeal should be dismissed.

    The question is whether the existing scheme is open to challenge on the ground of discrimination. In its terms it makes no distinction in its application between women and men and plainly involves no direct discrimination. I have not been persuaded that even indirectly it discriminates against the appellant. It recognises a distinction between part-time and full-time employees, and on the assumption that the majority of those who finish their working career on a part-time basis will be women it might appear that by using a multiplier based on the number of years worked rather than the hours worked women might be worse off then men. But the primary objective of the scheme was to cushion employees against unemployment. In that context the calculation by reference to the level of final salary seems to me entirely appropriate and it is applied to men and women without distinction. If the multiplicand in the case of a woman is lower than it might generally be in the case of a man, that is due to the fact that the man will in general have been working full-time and not part-time during the final period of employment and will have been earning a larger weekly salary. The multiplicand is a direct reflection of the relative level of weekly salary which the employee will have been earning and that in turn reflects the respective extent of work which the employee has been performing. A further purpose was to give some recognition to past service. Again the decision to do that by reference to years of work regardless of the sex of the employed person does not seem to me to involve anything discriminatory. Different schemes will produce different results. But it is only with the existing scheme that we are concerned. That the use of a multiplier based on hours rather than years of work would produce different results, or even results more favourable to the appellant, does not mean that the actual scheme is discriminatory.

    Even if in some way the existing scheme was to be seen as discriminatory I agree with the view reached by my noble and learned friend Lord Nicholls of Birkenhead that it would be objectively justified.