IBM United Kingdom Holdings Ltd & Anor v Dalgleish & Ors

In 2009 IBM’s UK businesses had pension schemes which provided either final salary (defined benefit – DB) or defined contribution (DC) benefits. The DB parts of the schemes had been closed to new members for years but they still had about 4000 employee members. Under pressure from the IBM group’s commitment to investors to increase its earnings per share, and from the effect on asset values of the financial crisis of 2008, IBM decided to close the DB parts of the schemes altogether, except for some employees who were contractually entitled to such benefits, and to do so quickly. The exercise which it undertook was known as Project Waltz. It involved three essential elements: (a) excluding employees from membership of the DB parts of the schemes as regards future service from April 2011, (b) bringing to an end a long-standing policy of allowing early retirement, from age 50, on beneficial terms for employees, as from April 2010, and (c) ensuring that future increases in salary would not count towards the employee’s final pensionable salary for the purposes of DB rights referable to past service. When this was announced it caused great anger and resentment among employees, particularly because there had been two recent previous changes to IBM’s pension schemes, in 2005 (Project Ocean) and 2006 (Project Soto). The trustee of the schemes was in serious doubt as to the legal validity of what was proposed, and the employees resisted the proposals strongly. In proceedings brought to test the lawfulness of the proposals, Warren J held that they were not lawful, by reason of breaches of relevant obligations owed by each of two separate IBM companies. These appeals are brought by the IBM companies against the judge’s orders consequent on his decision. This is the judgment of the court on those appeals.

This case is the first in which a duty which bears upon the exercise of an apparently unfettered non-fiduciary discretion under an occupational pension scheme, which is often referred to as the Imperial duty, has had to be considered as a matter of decision at an appellate level. It also requires an examination of the duty of trust and confidence, implied into every employment contract, as it bears upon the provision and alteration of pension benefits. Because of these features, the case has a degree of general relevance which adds to the significance which it has in any event for IBM and for those employees of IBM who are (or would be) affected by the pension benefit changes in issue.

We will have to go into a lot of detail in order to consider and examine the circumstances of the case, the judge’s several judgments, and the various arguments deployed before us, and to explain our own view of the relevant law. For the reasons which we set out below, our conclusion is that IBM’s appeals should be allowed and the cross-appeal by the beneficiaries should be dismissed. On a separate point, which the judge left to be argued and decided on appeal, we will not order IBM to carry out a further consultation exercise before it can implement Project Waltz.

The litigation concerns two occupational pension schemes established in the UK for IBM’s employees, referred to as the Main Plan and the I Plan. Under each scheme the First Claimant and Appellant (Holdings) is the Principal Employer, as defined, but it does not employ any relevant individuals. The Second Claimant and Appellant (UKL) is the employer of the relevant members of the schemes. The First and Second Defendants and Respondents (together, the RBs) are representative beneficiaries under the two schemes. The Third Defendant and Respondent (the Trustee) is the trustee of both schemes. The Main Plan comprises both Defined Benefit (DB) and Defined Contribution (DC) sections; the former includes four different structures. In general it is not necessary to distinguish for the purposes of the appeal between the Main Plan and the I Plan or between the different DB parts of the Main Plan. We will therefore refer only to the Main Plan, and to its distinct DB and DC sections, except where a point requires specific reference. If it is not necessary to identify the particular corporate entity involved we will refer just to IBM or, as regards the UK companies, to IBM UK. Besides the two UK companies which are the Claimants, we will refer as necessary to IBM UKI, IBM’s abbreviation for its business operations in the UK and the Republic of Ireland, and to IBM Corporation, the ultimate parent, and we will use the same label as the judge did, CHQ, to refer to the group headquarters in the USA.

The proceedings were brought in order to obtain a declaration from the court that certain actions which Holdings and UKL sought to take in relation to the pension schemes were valid. The Trustee had declined to put the proposals into effect without such a declaration being obtained. The changes have been effected but only on a provisional basis pending the court’s decision; that remains the case pending these appeals.

The proceedings came to trial before Warren J in February 2013, the hearing lasting 30 days. He decided some points in favour of the Claimants, but on the most contentious issues he held that the Claimants’ actions were in breach of relevant duties. His judgment following that trial (the Breach Judgment) was delivered on 4 April 2014: [2014] EWHC 980 (Ch). Then the consequences of the breaches which he had found were debated in a further trial over 9 days in July 2014, upon which he gave judgment on 20 February 2015: [2015] EWHC 389 (Ch), the Remedies Judgment. (Where we use the formula of either B or R followed by a number, as in B1526, this is a reference to the paragraph so numbered in the Breach Judgment or the Remedies Judgment as the case may be.) He dealt with some supplementary points after a further hearing in April 2015, in judgments delivered on 18 and 19 May 2015: [2015] EWHC 1385 (Ch) and 1439 (Ch). In order to give effect to his conclusions he made two orders relevant to the appeals, dated 8 June 2015 (the Breaches Order) and 30 June 2015 (the Remedies Order). He himself granted permission to appeal to the Claimants and to cross-appeal to the RBs.

As can be seen from the length of the hearings, the issues in the proceedings were highly contentious as between the Claimants and the RBs. Although the case had been commenced by a Part 8 Claim Form, the parties’ cases were set out in pleadings, IBM having first given extensive disclosure of documents. It was common ground that the burden of proof lay on the RBs on the disputed issues of breach of duty, and they therefore took the position of Claimants in the conduct of the trial.

One of the points dealt with in the First Supplemental Judgment is the error in the Breach Judgment where (except in B3) the judge proceeds on the basis that Holdings was not only the Principal Employer under the pension schemes but also the employer of the relevant employees, and therefore that Holdings was subject not only to the Imperial duty but also to the contractual duty of trust and confidence. In fact Holdings had no relevant employees; UKL was the employer and therefore subject to the contractual duty. When reading the Breach Judgment it is necessary to bear this confusion in mind. The judge discussed the consequences of this error in paragraphs 2 to 18 of the First Supplemental Judgment. His conclusion may be summarised adequately for present purposes as being that the Breach Judgment and the Remedies Judgment should be corrected to record that the breaches of contractual duty attributed in the Breach Judgment to Holdings should instead be attributed to UKL, and correspondingly in the Remedies Judgment the breaches of contractual duty and the consequent remedies should be attributed to UKL only, not to Holdings.

The court had to consider a series of steps in relation to the pension schemes which together were labelled Project Waltz. This was formulated and announced in 2009, to take effect, as regards its separate elements, between late 2009 and April 2011. Different aspects of it involved action by Holdings, as Principal Employer under the schemes, and by UKL, as employer of the relevant employees. Among the issues which the judge had to decide was whether the actions by Holdings were within the ambit of one of the powers under the schemes which was to be relied on, and whether they were a proper exercise of that power. On those points the judge decided in favour of Holdings, but the RBs challenge by their cross-appeal his rejection of their contention that the power was used for an improper purpose.

Otherwise the issues turned primarily on two related duties incumbent on Holdings and UKL respectively: the duty of trust and confidence implied by law into every employment contract, and the similar duty (sometimes referred to as a duty of good faith) imposed on a party such as Holdings which has a non-fiduciary discretionary power under an occupational pension scheme. This is known as the Imperial duty, after the decision in Imperial Group Pension Trust Ltd v Imperial Tobacco Ltd[1991] 1 WLR 589, at 596-9 in which the duty was first formulated. The judge held that each of Holdings and UKL was in breach of its respective duty in implementing Project Waltz. The Claimants appeal against that decision and against some of the judge’s conclusions as to the consequences of those breaches.

The consequences of the Project Waltz changes in pension benefits varied as between different employees, but they could be substantial. We were shown three illustrative examples. Details of one of them are set out in Appendix Two to this judgment. This concerns a hypothetical male employee who had joined the C Plan (the main section of the DB Part of the Main Plan) at the age of 25, was paid £40,000 p.a. on 6 April 2010 and would be 55 years of age on 5 April 2011. Expressing the figures in capital terms rather than as pension per year, if he were to retire at 55, he would have taken a pension worth

We have been greatly assisted by the array of advocacy talent deployed before us on behalf of all parties, and by the careful and thorough preparation of the case by Counsel and solicitors on all sides. The appeal hearing stretched over 10 days, and was conducted according to a tight timetable agreed between the parties on our direction. The provision of daily transcripts enabled the advocates, in some cases, to proceed much more quickly than would otherwise have been possible, and the provision of virtually all of the documentation in the case (including authorities) in electronic form has been of considerable assistance to us in the preparation of this judgment.

The judge had a massive and unenviable task, both at the first trial and at the later hearing concerned with remedies, in coping with the huge amount of material placed before him, as regards both documentation and factual and expert evidence, and the complex and sometimes confusing interaction of many difficult and diverse issues. His extensive specialist experience of the law and practice in relation to trusts, and in particular to occupational pension schemes, was of great assistance to him in his command of the issues arising, as is apparent from the judgments. In our judgment, inevitably, we focus on aspects of his judgments with which one or another party sought to take exception. On a good many of these points we do not agree with the judge’s reasoning or conclusions as set out in his judgments, and we have sometimes found that the complex structure of, in particular, the Breach Judgment made his treatment of some of the issues somewhat difficult to grasp and therefore the subject of doubt or dispute as to their meaning. We should, however, point out that there is a great deal in those judgments which no party sought to challenge. For example, his consideration of the consultation process, in the course of which he made trenchant criticisms of aspects of IBM’s conduct, and of the issues arising as regards consultation, stand unchallenged. Moreover, his discussion of the true nature and scope of the Imperial duty, though not in issue before us, seems likely to provide a valuable contribution to the development of the law in this respect.

It is convenient to set out at this stage, first the provisions of the pension scheme documents on which issues turn, and secondly the main elements of Project Waltz and of the two other previous projects by which IBM altered the arrangements for pension benefits, Projects Ocean and Soto.

Both the Main Plan and the I Plan contained a power which has been referred to as the Exclusion Power. It is in the following terms, so far as material:

The Main Plan allowed active members to retire before Normal Retirement Date on advantageous terms but (except in the case of certain members who had attained the age of 60) this required the consent of the Principal Employer. In practice IBM had a long-standing policy of allowing active members to retire early, sometimes as early as at the age of 50, on generous terms which did not involve such a discount as to make the acceleration of payment of the pension cost-neutral for the scheme. (The Normal Retirement Date differed for different sections of the Plans, with dates between 60 and 65.)

Project Waltz had five main elements:

All but certain specified members of the DB Plans were to be excluded from membership of those plans with effect from 6 April 2011, by virtue of notices served by Holdings under the Exclusion Power on 28 May 2010, thereby bringing to an end their accrual of DB pension rights for future service.

A new early retirement policy was to be introduced as regards the DB section of the Main Plan with effect from 6 April 2010 under which consent would only be given on less advantageous terms, which involved an actuarial discount such as to make the process cost-neutral for the scheme.

The opportunity was offered to active members of the Main Plan DB section to retire early under the previous policy, if they registered their interest in doing so within a specified time; this is known as the Early Retirement Window.

Pay increases from 2009 onwards for DB members were to be nonpensionable, so that these increases would not count towards their final pensionable salary for the purposes of their accrued DB service up to 5 April 2011.

Active DB members of the Main Plan, though excluded under the Exclusion Notices, were offered the opportunity to join the DC section, and to do so on terms which gave their accrued DB benefits a new “hybrid deferred” status. They were also offered alternative opportunities including to join a new personal pension scheme set up by IBM and administered by Standard Life.

The fourth of these points was implemented by asking employees who were members of the DB sections to sign Non-Pensionability Agreements (NPAs), expressed to agree that future increases in their pay would not count towards their pensionable salary for the purposes of the Plans. If they signed such a document they would receive pay increases; if they did not they would not. Initially, when these were proposed to employees in November 2009, there was no limit in time to the period during which no salary increase was to be paid if the NPA was not signed. Soon afterwards this was corrected and the member was told that this would only affect salary for the years 2009 and 2010. Some members had signed such forms before the second communication. Some members signed under express protest, whether at the time of signature or afterwards, but before any pay rise was implemented. For these reasons, there are several categories of employee to which points concerning the NPAs are relevant. NPAs were sought, and some were signed, during 2010 and also during 2011, the latter, it seems, also on the basis of an indefinite period of deprivation from any pay rise for those who did not sign.

As already mentioned, IBM had made changes to the pension schemes on two recent occasions: Project Ocean in December 2004, put into effect as from 6 April 2005, and Project Soto in late 2005 and 2006, implemented as from 6 July 2006.

Project Ocean involved an increase in the employees’ contribution rate from 4% to 6% (or in the I Plan from 4% to 5%), a commitment by IBM to pay £200 million each year for 3 years towards the deficit in the schemes revealed by the latest actuarial valuation, and a guarantee by a US company (which was a parent of the UK companies) of the funding liabilities of Holdings as Principal Employer, to last until the first quarter of 2014.

Project Soto offered members of the DB schemes a choice between remaining in the DB Plan, but on the basis that only two-thirds of any future salary increases would be pensionable, or transferring to a newly created DC part of the Main Plan, accruing enhanced DC benefits for future service, while retaining final salary-linking for their accrued DB benefits. Under Project Soto, IBM paid additional funds into the schemes so as to extinguish the then funding deficit and to fund the 2006 service costs.

A central part of the judge’s basis for deciding the case in favour of the RBs was that IBM had created what he termed Reasonable Expectations for the active members of the schemes, which would be thwarted by the implementation of Project Waltz. We will examine these in some detail, but the genesis of most, if not all, of the Reasonable Expectations lay in communications made by IBM to members at the time of Project Ocean and Project Soto. Correspondingly, although many members objected strongly to the terms of Project Waltz in themselves, a major factor in the sense of anger and betrayal expressed by members in relation to the announcement of Project Waltz was that it followed so hard upon the previous changes, and appeared to be inconsistent with them.

Before Project Waltz could be implemented, a statutory duty of consultation had to be complied with. That duty did not require consultation in respect of all aspects of the proposals, but IBM carried out a consultation on the whole scheme. However, the judge held that the consultation process was seriously flawed, so much so as to constitute a breach of duty in itself. That conclusion is not challenged on appeal. The judge did not, however, decide what were

The parties agreed a list of issues arising on the appeals from the points taken in the Appellant’s Notices and the Respondent’s Notice. As presented to the court shortly before the opening of the appeal there were 40 issues, though in the end four of these were not pursued. The issues that remain live are set out in Appendix One to our judgment. Their subject-matter can be summarised as follows:

A threshold issue concerning the basis on which the judge decided that Holdings and UKL had breached their respective legal duties (Issue 1).

Reasonable Expectations as a legal concept limiting (1) a Principal Employer’s freedom to cease accrual in the DB scheme, (2) a Principal Employer’s freedom to change its policy for the exercise of its discretion to allow early retirement benefits under the scheme and (3) an employer’s ability to offer pay rises on a non-pensionable basis (Issues 2, 3, 3a, 5 and 7).

The Reasonable Expectations found in this case (Issues 8 to 13).

Additional Reasonable Expectations not found by the judge but contended for in the Respondent’s Notice (Issues 14 to 16).

Additional factors on which the RBs contend the Judge should have placed reliance (Issues 17, 20 and 21).

Justification in this case for disappointing Reasonable Expectations (Issues 22 to 28).

Validity of non-pensionability contractual terms as between employer and employee (Issues 29 to 33).

Issues on remedies: effect of invalidity of non-pensionability contractual terms (Issues 34 to 36).

Supplemental issue: changing early retirement policy (Issue 37).

Consequential issue on remedies: effect of deficient consultation (Issue 38).

Cross-appeal: whether the Exclusion Power was used for an improper purpose (Issue 39).

In the remainder of this judgment we will deal with the matters arising as follows.

A discussion of the Imperial duty and the contractual duty of trust and confidence: from paragraph [27].

The burden of proof: from paragraph [47]. iii)Reasonable Expectations: from paragraph [59].

The main facts: from paragraph [72].

The details of Project Waltz: from paragraph [142].

The validity of the Exclusion Notices: the cross-appeal (Issue 39): from paragraph [160].

The challenge to Project Waltz as a breach of duty, dealing first with the threshold issue (Issue 1) as to the basis on which the judge decided the case, and with Issues 2 and 5: from paragraph [176].

Issue 10: Reasonable Expectations and significant change in financial and economic circumstances: from paragraph [235].

Issues 11, 12, and 16: was the judge justified in finding the Reasonable Expectations as to DB accrual and early retirement policy proved? This starts at paragraph [248], deals with the Reasonable Expectation as to the early retirement policy from paragraph [249] and then deals with the Reasonable Expectation as to DB accrual from paragraph [261].

Reasonable Expectations generally: Issues 7, 8 and 9: from paragraph [268].

Issue 13: the Early Retirement Window: from paragraph [274].

Issue 37: notice of a change of the early retirement policy: from paragraph [278].

Respondent’s Notice points. Issues 14 and 15 as to nondiscrimination in relation to salary increases: from paragraph [283].

Issue 17: the aggressive investment policy adopted by the Trustee: from paragraph [290].

Issue 20: the rectification proceedings: from paragraph [293].

Justification for breach of duty, starting at paragraph [296]. This includes Issues 23 and 24, treating the IBM group as one, starting at paragraph [341], Issue 26, an alternative course of action, starting at paragraph [374], Issue 27, the sequence of events, starting at paragraph [391] and Issue 26, the judge’s finding that IBM was disingenuous, starting at paragraph [395]. xvii)NPAs: Issues 29 to 36: from paragraph [407].

The so-called Imperial duty was first formulated in the Imperial Group case already mentioned. It was framed by reference to a duty which had already been identified as an implied term of contracts of employment. That implied duty was then expressed as an obligation on the employer not, without reasonable and proper cause, to conduct itself in a manner calculated or likely to destroy or seriously damage the relationship of confidence and trust between employer and employee. Browne-Wilkinson VC gave this duty the label “the implied obligation of good faith” in his reasoning in the Imperial Group case.

He held that this obligation “applies as much to the exercise of his rights and powers under a pension scheme as they do to the other rights and powers of an employer”: [1991] 1 WLR 589, at 597H. He went on to say (at 598A):

Thus, the essence of the obligation was said to be the same whether as a duty on the employer under a contract of employment or as a constraint on the exercise of a power such as the Exclusion Power, vested in Holdings in the present case as Principal Employer under the pension schemes, or the power to consent to early retirement, both of which are non-fiduciary discretionary powers. Strictly speaking the phrase “the Imperial duty” is a misnomer or is at least capable of being misleading: it is a constraint or limitation on the use of an apparently unlimited discretionary power, as the judge recognised at B372. The judge did say, however, that it could properly be classified as a duty, though a negative duty rather than one which could require the person subject to it to act in a given way: see B471, his discussion between R364 and R379, again between R393 and R396, and his further observations at R431. He held that breach of this obligation could provide the basis for a claim for equitable compensation. These aspects of his judgment have not featured among the issues on the appeal, and we say no more about them than that, in a case in which such issues arise for decision on the facts, what he has said in those various passages is likely to be of value in carrying the debate forward. Without prejudice to the intricacies of the point, we will continue to use the label to denote the relevant legal principle, as the judge did, for ease of reference.

The contractual duty (which is owed each way under a contract of employment) has been affirmed at the highest level of authority: Malik v BCCI[1998] AC 20; Johnson v Unisys Ltd[2001] UKHL 13, [2003] 1 AC 518. In each of those cases passing reference was made to the Imperial Group case. The Imperial duty itself has been considered directly on a few occasions. It was followed and applied by Knox J in Hillsdown Holdings plc v Pensions Ombudsman[1997] 1 All ER 862. It was referred to with approval (but not as a matter of decision) in National Grid Co plc v Mayes in the House of Lords [2001] UKHL 20, [2001] 1 WLR 864. It was followed and applied by Newey J in Prudential Staff Pensions Ltd v The Prudential Assurance Co Ltd[2011] EWHC 960 (Ch), [2011] PLR 239, to which we will refer later. It was touched on in the Privy Council on appeal from Jamaica in Air Jamaica Ltd v Charlton [1999] 1 WLR 1399 and in UC Rusal Jamaica Ltd v Miller[2014] UKPC 39, [2015] PLR 15, but neither of these cases calls for further reference in relation to the present appeals.

We had submissions as to the incidents and features of the two duties, as the judge had. Browne-Wilkinson VC in the Imperial Group case made it clear that the test is not whether the company is acting reasonably; the company is perfectly entitled “to look after its own interests, financially and otherwise, in the future operations of the scheme”: [1991] 1 WLR 598H. He said that the obligation of good faith required that the company should exercise its rights “(a) with a view to the efficient running of the scheme established by the fund and (b) not for the collateral purpose of forcing the members to give up their accrued rights in the existing fund subject to this scheme”. In that case an existing scheme prohibited the payment of any surplus to the employer. Following a take-over the new owner wished to transfer members to a new scheme under which any surplus could be applied for the employer. That was held to be an improper purpose vitiating the exercise of powers in such a way as to coerce members into agreeing to a transfer to the new scheme.

Knox J’s decision in Hillsdown Holdings plc v Pensions Ombudsman[1997] 1 All ER 862 also concerned steps taken by a new owner to transfer its pension liabilities to a scheme under which any surplus could be paid to it. The judge held that it would be a breach of the Imperial duty for the employer to bring in a substantial number of new members to the scheme but to decline to make contributions in respect of such members, thereby resulting in the surplus being reduced for its own benefit: [1997] 1 All ER 890f-g.

Newey J reviewed and applied the relevant authorities in Prudential Staff Pensions Ltd v The Prudential Assurance Co Ltd[2011] EWHC 960 (Ch), [2011] PLR 239, which was about increases to pensions in payment. Under the rules these were at the discretion of the employer. It had maintained a policy over years of making increases so as, broadly and over time, to keep pace with inflation, but in 2005 it changed that policy. It was argued for the beneficiaries in that case that the company must have regard to members’ legitimate expectations and must deal fairly with the members. The judge rejected the notion of a test of fairness, but he said this at paragraph 146:

The judge held that there was a general understanding among members that, except when inflation had been particularly high in the 1970s, pension increases had fully compensated for inflation, and there was a widespread expectation that that position would continue (see paragraphs 52 and 184). He noted that the company had been alive to the existence of such expectations in its deliberations (paragraph 185). He then said this, at paragraph 186:

Accordingly the judge rejected the contention that, in making the decision not to pay increases matching inflation, Prudential had acted irrationally or perversely or otherwise in breach of the obligation of good faith.

The promulgation of the test as requiring irrationality or perversity was influenced by developments in employment law and elsewhere concerning the exercise of contractual discretions. Relevant employment cases included, in particular, some concerning the grant of discretionary bonuses, including

The latter case is particularly useful, even though the factual context was very different. The case concerned a death in service benefit under a contract of employment which was not payable if, in the opinion of the employer, the death resulted from suicide. The employer had concluded that it was a case of suicide. The issue for the court was the test to be applied in deciding whether the employer was entitled to come to that conclusion. The judgments in the Supreme Court differed as to the outcome but not significantly as to the correct legal test. Baroness Hale observed at paragraph 18 that where a contractual term gives one party to the contract the power to exercise a discretion, or to form an opinion as to relevant facts, the court cannot rewrite the bargain nor substitute itself as the decision-maker, but it will seek to ensure that such contractual powers are not abused. In the same paragraph she aligned a unilateral power to exercise a discretion with a contractual power to find facts, such as was directly at issue in that case. That applies all the more so, she said, “where there is a significant imbalance of power between the contracting parties as there often will be in an employment contract.” In order to seek to prevent abuse of such powers the courts have implied “a term as to the manner in which such powers may be exercised, a term which may vary according to the terms of the contract and the context in which the decisionmaking power is given.”

As to the correct approach, she drew an analogy with the court’s task in relation to a statutory or prerogative-based decision-making function. Having said at paragraph 28 that the contractual implied term “is drawing closer and closer to the principles applicable in judicial review”, she spoke at paragraph 29 of the duty to exclude extraneous considerations and to take into account those considerations that are obviously relevant to the decision in question. She went on to say this:

At paragraph 30 she said that, absent a context in which an objective standard of reasonableness can be implied, “the court will only imply a term that the decision-making process be lawful and rational in the public law sense, that the decision is made rationally (as well as in good faith) and consistently with its contractual purpose”. She also said that, in applying the test of rationality,

Lord Hodge expressed the same opinion at paragraph 53. Neither of them sought to set out a general rule as to the application of the Wednesbury approach to implied contractual terms, but both said that it was appropriate to apply this approach to employment contracts: paragraph 32 and paragraph 54. Lord Hodge also said at paragraph 55 that “the personal relationship which employment involves may justify a more intense scrutiny of the employer’s decision-making process than would be appropriate in some commercial contracts” and said in the following paragraph that the intensity of the scrutiny would depend on the nature of the decision to be made. A decision as to the cause of death is far removed in this context from, for example, decisions about contractual bonuses, where the employee is entitled to a bona fide and rational exercise of the employer’s discretion. Such an entitlement is capable of enforcement by the court, but does not allow much scope for an intensive scrutiny of the decision-making process (paragraph 57). Lord Kerr agreed with both Baroness Hale and Lord Hodge. Lord Neuberger expressed the same view at paragraph 103, although he (and Lord Wilson, agreeing with his judgment) came to the opposite conclusion when applying the test to the facts of the case.

The implied duty of good faith in an employment contract is relevant to many different types of situation, not only to those involving a decision by an employer under an express or implied discretionary power; Malik v BCCI is an extreme example of a different situation where the duty is relevant. In that case the employer’s conduct (running a fraudulent business) was not aimed at any particular employee although, once discovered, it affected many employees by way of a stigma of association. Other cases very different from those of contractual discretion include conduct which is aimed at a given employee or a group of employees, of the kind that can lead to a claim of constructive dismissal, such as harassment or other objectionable behaviour. For such cases the Wednesbury test is hardly likely to be directly relevant.

We were shown some cases concerning the denial of pay increases to

That case was not cited to us but it lies behind others that were cited, including in particular Transco v O’Brien[2002] ICR 721. In that case an employee was not given a pay increase that was given to all others in the same category, because the employer treated him (erroneously, as the Employment Tribunal found, though in good faith) as being in a different category, namely not being a “permanent employee”. The Employment Tribunal had spoken of employers being under a duty to treat employees in a fair and even-handed manner. In the Court of Appeal the employer challenged the use of that test (which the court agreed was incorrect), but also submitted that the duty of trust and confidence could not require the employer to offer any employee a variation of his contract. Pill LJ said this at paragraphs 16 and 17:

… There may in law be a breach of the implied term of trust and confidence in a decision to refuse to offer an employee a new contract in circumstances such as the present, just as in a decision to refuse to offer a variation. …

… To single out an employee on capricious grounds and refuse to offer him the same terms as were offered to the rest of the workforce is, in my judgment, a breach of the implied term of trust and confidence. There are few things which would be more likely to damage seriously (to put it no higher) the relationship of trust between an employer and employee than a capricious refusal, in present circumstances, to offer the same terms to a single employee.

Longmore LJ agreed and Sir Martin Nourse said much the same at paragraph 28. Thus the test of arbitrariness or capriciousness, as an aspect of the duty of trust and confidence, has been applied to that type of situation, so as to impose on an employer an obligation either to offer to the particular employee or group of employees the same or a similar pay increase as that offered to others who are comparable, or at least to justify not having done so on some basis which is not arbitrary or capricious.

In cases of that kind, concerning pay rises, the employer is not exercising an express or implicit discretion under the employment contract, but they show that the court’s approach to the employer’s freedom of operation in relation to the contract may be constrained by the implied duty so as to require the employer to act in a manner which is not arbitrary or capricious. Such cases may not be susceptible to the full application of the Wednesbury test. However, in cases which do involve the exercise of an employer’s discretionary powers, whether express (as in many of the bonus cases, and in Braganza) or implied, then, in our judgment, the effect of the recent case law is that, in order to decide whether the employer’s act is or is not in breach of the implied duty, a rationality approach equivalent to the Wednesbury test (including both its limbs) should be adopted, taking into account the employment context of the given case. Such an approach is required because the court does not and must not substitute its own decision for that of the decision-maker, in these cases the employer.

Correspondingly, such a rationality approach should be applied in deciding whether a person who has a non-fiduciary discretionary power under an occupational pension scheme, such as those vested in Holdings in the present case, has respected the constraints imposed by the Imperial duty in relation to the exercise of the power, in addition to asking the question whether the power has been exercised for a proper purpose.

Having held that the Imperial duty required that the relevant power (a) should be exercised for its proper purpose and (b) should not be exercised perversely or irrationally, and that the test for the contractual duty of trust and confidence in the circumstances of the present case was also one of perversity or irrationality, Warren J also proceeded on the basis that the burden of proof lay on the RBs on all these issues (see, for example, B425 and B467). So much was common ground before him and on the appeal, but there was an issue before the judge as to whether there were two distinct elements to the test, or only one. The RBs argued for what was referred to as a “two-pronged” approach, involving two distinct issues: first, whether Holdings or UKL had conducted itself in a proscribed manner and, if so, secondly, whether it had done so without reasonable and probable cause. The judge rejected this approach in favour of a unitary approach: B385. It is accepted that, if the employee makes out a case of improper conduct, an evidential burden shifts to the employer to show what its reasons were, and in some of the employment cases, particularly those concerned with discretionary bonuses, the employer has not attempted to show that. However, leaving aside that proposition, the judge held that the test was unitary and that the legal burden of proof remains on the employee throughout.

For the RBs Mr Cavanagh Q.C., who dealt with this aspect of the argument on the appeal, submitted that this was not correct. He relied on what was said in three cases: Hills v Niksun[2016] EWCA Civ 115, [2016] IRLR 715, at paragraph 23, Attrill v. Dresdner Kleinwort[2013] EWCA Civ 394, [2013] 3 All ER 607, at paragraph 130, and Braganza at paragraph 37. In response on this point Mr Simmonds Q.C. also showed us Keen v. Commerzbank [2006] EWCA Civ 1536, [2006] CLC 844. We will start with that case, as the earliest at appellate level. Mr Keen claimed to be entitled to a larger bonus payment than had been made to him. He asserted and relied on an implied term of the employment contract that the bank would not exercise any discretion it had as regards his bonus irrationally or perversely; that duty was admitted. The bank sought summary judgment on the basis that he had no reasonable prospect of showing a breach of the term, and appealed (successfully) against the judge’s dismissal of that application. Mummery LJ recorded the submission for the bank that Mr Keen must show that no rational employer bank would have awarded bonuses less in amount than those recommended by his line manager, which were significantly higher than those in fact awarded, “in other words that the bank had exercised its discretion to award bonuses irrationally, arbitrarily or perversely” (paragraph 52). He accepted that submission and observed at paragraph 59 that:

Moses LJ addressed the question of burden of proof more directly at paragraph 110. He accepted that the implied duty of trust and confidence between employer and employee does generally require an employer to give his reasons for the exercise of a discretion to pay or withhold a bonus. But he said that failure to give reasons will not necessarily establish irrationality.

Jacob LJ agreed with both judgments. Because of Moses LJ’s proposition that, even if the employer remains silent, it is still for the employee to show irrationality, that decision is not consistent with Mr Cavanagh’s submission.

Attrill v Dresdner Kleinwort, the next relevant case in point of time, was also about bonuses. The claimants had been employed by a company in the Dresdner Bank group. In August 2008, in order to motivate staff to stay, a senior manager had announced that a large guaranteed minimum bonus pool had been declared for 2008, to be distributed “come what may” on a discretionary basis as between the relevant employees. However, Dresdner was then taken over by Commerzbank, which took a different view as to the appropriateness of paying bonuses on this basis. It introduced a “material adverse change” clause into the arrangements for the bonus, and then it reduced the bonus pool by 90%. The Court of Appeal affirmed the judge’s decision on the first point argued, namely that the guaranteed bonus pool was a matter of contract as between employer and employee, from which the bank was not entitled to resile. The court went on to consider the second question, though on this basis it did not strictly arise, namely whether the introduction of the clause was in breach of the duty of trust and confidence. Elias LJ (with whom Maurice Kay and Beatson LJJ agreed) said at paragraph 102 that this involved two questions: first, what was the employer’s reason for so acting, and second, did its reason constitute reasonable and proper cause? No witness had given evidence as to the employer’s reasons for introducing the clause; some documents were in evidence, but none from Dresdner’s original owners (Allianz) or from Commerzbank. The judge inferred that the reason for the change was pressure from Commerzbank. Elias LJ said at paragraph 130:

Those observations assist Mr Cavanagh’s case, but they are not altogether compatible with the approach in Keen. On the facts it is not difficult to see why Elias LJ put it as he did in Attrill. Keen was cited in that case, but the forensic position was very different. Moreover, it does not appear that any argument was addressed to whether the two-stage approach was correct or not.

Hills v Niksun was also a bonus case, where the relevant contract said “you may from time to time be eligible to participate in such commission plans as Niksun may in its absolute discretion determine”. There was a relevant plan, whose terms included this: “Niksun reserves the right to determine what level of incentive compensation, if any, is fair and reasonable under the circumstances and is in the best interests of Niksun”. Mr Hills sued on the basis that he had been awarded too small a bonus, relying on a conversation on the subject of bonus, in relation to a particular transaction, between his line manager and that manager’s immediate boss. Niksun called no evidence from which the court could ascertain why it had come to the challenged decision. Mr Hills submitted that the burden of proof lay on the employer as to whether the decision was reasonable. He relied on a statement by Baroness Hale in Braganza at paragraph 37:

Vos LJ said of the submission made, at paragraph 23:

Applying that approach, at paragraph 26 he said that, in the absence of evidence from the decision-maker, the judge would have been justified in

Mr Simmonds Q.C. for IBM showed us that the judge had adopted the same approach in the present case, recognising that the legal burden of proof lay with the RBs throughout: see his discussion at B379 to B385, B425 (third sentence) and B467. He also pointed out that the judge’s approach on this is not among the points challenged by way of the Respondent’s Notice. Furthermore, he reminded us that the RBs were treated as being in the position of claimants, so that they opened the case at trial and they had the last word in reply. That was the consequence of their being subject to the burden of proof. He submitted that, in addition to all the other points made, it would be wrong (and now impossible) to go back on that basis on which the trial had proceeded, by treating any aspect of the burden of proof as lying on Holdings and UKL.

In our judgment Mr Simmonds is right on this issue, and so was the judge. In order to decide whether an employer’s decision in a given case satisfies the rationality test we have described (in paragraph [45] above) the court may need to know what the employer’s reasons were and may also need to know more about the decision-making process, so as to assess whether all relevant matters, and no irrelevant matters, were taken into account. The legal burden of proof lies with the claimants throughout. If, however, the claimants show a prima facie case that the decision is at least questionable, then an evidential burden may shift to the employer to show what its reasons were. In such a case, if no such evidence is placed before the court, the inference might be drawn that the decision lacked rationality. However, in all cases the legal burden of proof rests on the claimant.

In the present case, so unlike Attrill, for example, or Hills v Niksun, IBM put in a great deal of evidence as to the reasons lying behind the decision. They satisfied the evidential burden, and the legal burden remained with the RBs to show that the decision taken was in breach of duty on the rationality test, whether as regards Holdings in respect of the Imperial duty or as regards UKL under the contractual duty.

A passage of central importance in the Breach Judgment, for the judge and for the appeal, is his discussion of Reasonable Expectations from B450 to B478. This topic had appeared in his discussion of the content of the duties at B386(iv), recording and commenting on a submission of Mr Tennet Q.C. for the RBs. Mr Tennet argued that reasonable expectations of the members of a scheme were relevant, whereas “mere expectations” would not be. He distinguished these two categories in this way, as the judge recorded at B386(iv):

A “mere expectation” is one which an employee may have in fact as to the future, in the sense that they anticipate, assume or expect that something (e.g. a discretionary increase) will happen in the ordinary course of events if things “carry on as they are”. Employees may have a mere expectation independently of any encouragement by the employer.

A “reasonable expectation” means an expectation as to what will happen in the future engendered by the employer’s own actions (and in relation to matters over which the employer has some control), which gives employees a positive reason to believe that things will take a certain course.

The judge went on in that paragraph to use Reasonable Expectations as a defined term (“a term of art”) with the meaning set out above, so as to distinguish it from any other kind of expectation.

Because of the importance that Reasonable Expectations had for the judge’s judgment, and in the course of the appeal, we need to review the judge’s analysis on this aspect of the case in some detail. It is clear from his judgment that, but for his findings about Reasonable Expectations, both as regards the facts and their legal relevance, he would not have decided the case in favour of the RBs as regards the principal attack on Project Waltz, nor indeed was it argued for the RBs that he could do so: see for example B1519.

Mr Simmonds submitted to Warren J that a Reasonable Expectation could not be relevant if it was only engendered by a representation as to intention (leaving aside the possibility of a dishonest statement of current intention), as distinct from one which amounted to a promise, commitment or guarantee as to future conduct or events. The judge concluded, first, that a misrepresentation relied on by employees could be relevant even if it was not deliberate or reckless but merely negligent or even wholly innocent (B465), and secondly that conduct could not be ruled out as irrelevant even if it did not amount to a promise, commitment or guarantee: B466. He said:

The significance of that observation can be seen later in the Breach Judgment, after the judge had said something about the change in the corporate culture of IBM in the early 1990s, and posed the question “to what extent the new culture can properly be invoked to override any such Reasonable Expectations as can be established”: B530. He agreed with Mr Simmonds that the assessment of commercial matters and the making of business decisions is a matter for the company, on which the court cannot second-guess the business judgment of management. The judge did not accept Mr Simmonds’ next submission, that “where there is a coherent, rational (in terms of making business sense in the context of IBM’s business) and a bonafide case for the pension changes, the Court should not seek to go behind that case”: B532. The judge accepted that the existence of such a business case would be highly relevant and “other things being equal” the court should not go behind it, “not least because it is ill-equipped to do so”. But he postulated a case in which the company had bound itself by contract with employees, for example, to keep the DB Plans open until 2014. He said at B533 that the company could not override that contractual obligation, and that, by analogy, a proposed course of conduct might, as a matter of fact, override and disappoint Reasonable Expectations and therefore be in breach of duty. In the following paragraph he said that, however rational and strong might be the business case for the changes on the part of an employer “untrammelled by commitments to its workforce”, the test of irrationality and perversity under the Imperial duty “is focused on a course of conduct which no reasonable employer would pursue. Irrationality, in that sense, may have nothing or little to do with the underlying financial and economic business case.” He said that he would revert to IBM’s business justification for Project Waltz later in the judgment (B535).

The case for the RBs was that Reasonable Expectations had been engendered by IBM, by its conduct in relation to Projects Ocean and Soto, and in particular by communications made at the time of each of those exercises. We have summarised those projects at paragraphs [21] and [22] above. The judge held that there were relevant Reasonable Expectations as a result of what was said to employees at the time of these two projects. He set them out at B1052 to B1056 but it is convenient to take them from his recapitulation later in the judgment at B1510, as follows:

In relation to future service, an expectation that benefit accrual (in accordance with the Soto changes) would continue into the future; but such an expectation would be subject to an appreciation that a significant change in financial and economic circumstances (including trading and competitiveness) might cause Holdings to make further changes to the Plans and that to do so might be a decision which Holdings could reasonably take. Subject to that qualification, the expectation would last until at least the operative date, 6 April 2011, of the cessation of benefit accrual as a result of the service of the notices under the Exclusion Power.

In relation to past service up to the time of the implementation of the Ocean proposals, an expectation that a member would be able to take advantage of Holdings’ then current early retirement policy until 2014 unless there was a relevant justification for a change in policy.

The judge clarified the first of these in the Remedies Judgment, at R6, by saying that, because members had a Reasonable Expectation that DB accrual would continue at least to 6 April 2011, they also had a Reasonable Expectation, not that salary increases would be awarded, but that, if they were awarded, they would be pensionable as to two-thirds. He described this as a necessary part of the Reasonable Expectation as to continued DB accrual. He also held that this Reasonable Expectation lasted until 31 March 2014: R426.

He went on in the Breach Judgment to say that the Project Waltz changes clearly conflicted with those Reasonable Expectations, and that those changes were on any view very significant and capable of giving rise to a breach of the Imperial duty and of the contractual duty (B1512). At B1514 he said:

Correspondingly at B1515, of the NPAs, he said:

We will return to the matter of the global and local strands, but in context they are a reference to IBM’s business reasons for making the changes, referring to the view taken of the concerns of IBM as a whole, and in turn to the view taken of the interests of the UK companies on their own.

At B1524 he posed what was for him the critical question, as follows:

In the next paragraph he answered that question:

He then went on in B1526 to set out in 13 sub-paragraphs what he said were his principal reasons for reaching that conclusion. Having done so and referred to a couple of other points, he said this at B1528:

The judge had to consider a vast mass of documentary and oral evidence about the circumstances which led up to Project Waltz, including the previous projects, Ocean and Soto, and the background to those projects. Farther back in time, to set the scene, he considered the history of IBM’s business and its corporate culture, as well as having to examine and explain a number of technical matters such as the treatment of occupational pension schemes for the purposes of corporate accounting in the USA. Having dealt first with the issues as to the validity of the Exclusion Power in the Main Plan, and whether it was exercised for a proper purpose, he then set out at B295 the headings under which he would deal with the remaining issues under the Imperial duty and the contractual duty, which included passages to which we have already referred about the nature of the duties.

The appeals do not involve any challenge to the judge’s findings of fact, except on two discrete points where IBM contends that the judge made findings that were not open to him because they were not in issue on the statements of case. The judge’s task was massive and his reasoning and conclusions are set out at substantial length in the judgments, particularly the Breach Judgment. We can be much shorter, given that the facts are not in issue and by no means so many are relevant to the issues before us as were to those which the judge had to decide. Nevertheless, the complexity of some of the points and of the development of the history is such that it cannot be told as simply and succinctly as we would wish. What we say about the facts in this judgment does, inevitably, involve over-simplification as regards much of the detail, where it is not directly germane to the issues on the appeal.

IBM is a worldwide group with its ultimate holding company based in the United States. It has subsidiaries in many parts of the world, including those in the UK and Ireland. Being based in the US, the group has to comply with US standards in terms of financial accounting. The group had a system called “powers reserved” under which, for internal decision-making purposes, certain decisions which formally were to be taken at local levels had to be approved at headquarters level, by CHQ. This meant that the process of discussion and formulation of the several pension changes at issue in this case were conducted both at CHQ level and in the UK management, and individuals from both levels were involved in the process, in different ways. The judge heard evidence from several IBM witnesses. Six were called by IBM:

Mr William Chrystie, Chief Financial Officer of IBM UKI from Spring 2009 onwards;

Mr Jonathan Ferrar, from 2007 to 2010 Director of Human Resources for IBM UKI, and part of the executive leadership team of IBM UKI, and since 2010 Vice President, Human Resources, Workforce Analytics, IBM Worldwide;

Mr David Heath, Human Resources Director for IBM UK from September 2003 to January 2007 when he left the company;

Mr Lawrence Koppl, who was at the times material to the proceedings Director of Pensions Analytics within the Finance function at CHQ;

Mr James Randall MacDonald, Senior Vice-President, Human

Mr Brendon Riley, who was General Manager of IBM UKI from April 2008 to January 2010.

The RBs called Mr Stephen Wilson who left IBM in 2009 having been Vice President and Chief Financial Officer of IBM UKI and, before 2008, CFO for IBM North Region (that is to say UK, Ireland, Netherlands and South Africa); he had also been a director of both Holdings and UKL. In addition the RBs submitted a witness statement from Mr Larry Hirst who was unable to attend the trial for health reasons. He had been a director of Holdings and of UKL. From 2001 to 2008 he was General Manager of IBM North Region and then of UKISA (UK Ireland and South Africa). The RBs also called member witnesses, but we do not need to deal with them or their evidence for present purposes.

Another important witness, called by the Trustee, was Mr James Lamb, who had been employed by IBM UK for many years, and had been a director of the Trustee since 1994. He retired from IBM’s employment in 2002 and from directorship of all companies other than the Trustee at the same time, when he became chairman of the Trustee.

Because of the impact of US standards of financial accounting on the IBM group, the judge had to consider and explain concepts which, as he said, are complex and counter-intuitive even for someone versed in the approach to pension costs in the UK. He covered this at B494 to B517.

Companies have to recognise in their accounts each year the accounting cost of their pension schemes. This is designed to accrue the cost of employee pensions payable over the approximate service life of employees (so as to match the cost to the related service), based on an actuarial estimate of the present value of future pension payments offset by the expected investment returns from pension plan assets. In relation to DB schemes in particular, where the employer has to bear the burden of any shortfall in the funding of the liabilities, this may be a positive or a negative figure depending on the level of funding of the scheme and on investment returns; the “cost” may therefore be a profit figure.

A crucial element in the calculations required for US accounting purposes is the Net Periodic Pension Cost (NPPC). This is the annual accounting expense or income that a company must recognise in its Income and Expenditure Account. There are several elements to the NPPC. One of importance for present purposes is the Expected Return on Assets (EROA): the return that the company would expect on its pension plan’s assets over the accounting period, calculated by reference to the Market-Related Value of Assets and the expected long-term rate of return on assets.

Another topic which the judge identified as important is that of volatility in relation to pension schemes. As he explained at B518, the word is straightforward, referring to a likelihood of sudden, unexpected or unpredictable changes, but what matters is to what it is applied. An obvious application is to the value of scheme assets, as values change on relevant markets. It can also apply to the amount of scheme liabilities, though these will not normally change so suddenly as asset values can. Volatility can also be spoken of as regards the gap between the value of scheme assets and the amount of scheme liabilities, and it can also be applied to NPPC. The judge observed that this last aspect of volatility was of greatest concern to CHQ because of its effect on the group’s accounts and on the level of Earnings Per Share (EPS), which was perceived as being of importance for the group in its relations with actual and potential shareholders.

In a section of the Breach Judgment headed Culture Clash, B521 to B535, the judge described something of IBM’s corporate history. He referred to IBM’s pre-eminence in technology in the 1980s, which generated significant profits and allowed for generous provision for employees, but also to its failure to respond to competition by the early 1990s such that by 1993 it was on the verge of insolvency. A new CEO, Mr Gerstner, began to change the corporate culture, limiting employment costs and focussing on increased shareholder value as a key indicator of success. The judge recorded that Mr MacDonald gave evidence that many senior IBM executives who survived the “near-death experience” of the early 1990s remained determined to retain tough controls on employee costs, as part of the bottom line (B529). He also noted that this change of attitudes led to resentment and opposition on the part of other longstanding employees of IBM, with which the resistance to the pension changes which we have to consider is entirely consistent. Having quoted an email sent

The particular background to Project Ocean was the crash in world financial markets in 2000 to 2002, which caused substantial falls to the value of the assets of the Plans and a significant shortfall on EROA. Over the three years 2000, 2001 and 2002 the actual Return on Assets (ROA) fell short of EROA by more than £2 billion. Recovery started in 2003, but the figures for previous losses were spread over following years, by way of smoothing the effect of such major changes. A triennial actuarial valuation of the Plans was due as at 31 December 2003. It became clear in 2004 that there would be a substantial funding deficit; in the event it turned out to be £900 million for the Main Plan and £19 million for the I Plan. Moreover, the IBM UKI businesses were seen as underperforming by comparison with other parts of the worldwide business, so that the UK management team started to prepare a business recovery plan. A review of pension provision formed part of this. This led to the formulation of Project Ocean.

Under the “powers reserved” arrangement within IBM, changes resulting from this review would have required approval by CHQ in any event. More specific involvement from above arose from the fact that a guarantee by IBM World Trading Corporation was to be given of Holdings’ funding obligations in respect of ongoing accrual, deficit repair obligations and the statutory employer debt (under section 75 of the Pensions Act 1995) up to March 2014. Speaking of Project Ocean, the judge said at B542 that while CHQ retained ultimate control:

The judge referred to the involvement of individuals from CHQ in the process and then said this at B544:

The significance of the guarantee was that it strengthened the employer covenant under the Plans, and thereby enabled the Trustee to take a less cautious approach than it might otherwise have done both as regards the assumptions relevant to actuarial valuation and generally, including on the ever-important issue of asset allocation in the investment of the Plan assets.

In 2001 the allocation of assets for the Main Plan was in accordance with a policy of 80% equities and other assets (mainly property) and 20% bonds. This was regarded as appropriate by the Trustee at the time though it was recognised that, as the Plan matured, the allocation would need to change in favour of bonds, which are regarded as less volatile in terms of value and more closely matched to the nature of the Plan liabilities. A modest change, to 79/21, was made in terms of policy and reality in 2002. Further change was deferred until after the 2003 actuarial valuation. The asset allocation policy had an immediate impact on NPPC since the EROA on equities and other property would be higher than on bonds.

By the middle of 2004, as a result of declines in share values, the actual allocation was 75/25. Forecasts of the likely outcome of the actuarial valuation at that time showed a deficit figure that was alarming for IBM, which would bring into focus the question of the strength of the employer covenant on the part of Holdings, and the amount and timing of the contributions that would be required of Holdings towards making good the deficit.

Project Ocean was devised to cope with this problem. In order to reduce the cost to IBM of future service benefits, employees were to be asked to pay higher contributions. To tackle the funding deficit for past service, the parent company guarantee was offered, together with payments of substantial sums (£200 million) towards the deficit each year for three years. The Trustee agreed, in the light of this, that the valuation could be undertaken on a less cautious basis (the “best estimate basis” rather than the “central basis”). This would also allow the Trustee to agree to a longer period for the change in asset allocation in favour of bonds.

The Project Ocean proposals were put by IBM to the Trustee formally in

The Webcast featured largely in the trial, as the principal communication to employees about Project Ocean. The judge was not able to listen to the Webcast in the way that employees heard it, but he did have a transcript (which may have been the script from which it was read) which he appended as Annex D to the Breach Judgment, in a form adapted for ease of reference. In the Webcast Mr Heath outlined the proposed changes, which he described as action taken “with the intention of securing the sustainability of our defined benefit pension schemes”. He dealt first with the increase in employee contributions in respect of future service benefits, stressing that this had nothing to do with the past service deficit to which employees were not being asked to contribute, and making the point that much of the increased cost of future benefits arose from greater longevity. He then turned to the past service deficit and the figures from the actuarial valuation. He explained the additional cash contributions of £200 million each year for three years and the guarantee by the parent company, which he described as “excellent news for members” which “enhances the security of members’ benefits and demonstrates IBM’s continuing commitment to the Plans and their members”. At the end of the Webcast he referred to there having been consideration of other proposals including changes to benefit design, but said that the particular changes proposed were thought to be right in terms of a balance between the need to ask members to contribute more and greater security. He said that these changes convinced him that IBM’s DB pension schemes would in this way be put “on a firm footing for the future”, and said that he believed the proposal “demonstrates IBM’s commitment to underpin the sustainability” of the UK DB schemes.

The judge examined and discussed the Webcast, the related evidence and the submissions made about it at B614 to B697. The RBs’ case was that the Webcast engendered a Reasonable Expectation that IBM would not be making further changes to, or closing, the DB Plans for the long term. IBM contends that nothing in the Webcast amounted to a guarantee or commitment that DB accrual would remain open or that other changes to the Plans’ benefit structure would not be made, for any particular period. The judge found that Mr Heath and Mr Hirst honestly believed at the time that there were no further plans to cease DB accrual or to make other changes such as were later proposed as part of Project Waltz (B625). But the judge did not regard that as sufficient to

After a further close examination of the rival submissions, the judge rejected Mr Tennet’s contention that the reasonable member could understand and expect, from the Webcast, that benefits would remain unchanged even if markets moved adversely (B666 and B667) but did accept that in the short term the members could reasonably expect that no further changes would be made in the absence of a significant change in financial and economic circumstances (B668), and in particular that the DB Plans would remain open for at least 3 years (B669). He declined in the Breach Judgment to say how long a period that expectation would cover, except to say (at B696) that it would not extend to future accrual until 2014.

The judge then digressed in order to examine the position as regards IBM’s favourable early retirement policy (B680 to B687). In the course of that he held that members had a Reasonable Expectation that they would be entitled to retire early on favourable terms. This was a widespread perception from IBM’s long-standing practice, engendered by IBM and of which IBM was aware: see paragraph B687, quoted at paragraph [250] below. The practice could be changed for the future, but since it affected the amount of pension earned by past service, it might be open to question whether and in what circumstances it could be changed so as to affect the pension attributable to past service. When the judge returned to his discussion of the Webcast he held (at B689) that its contents confirmed and reinforced the Reasonable Expectation as to the early retirement policy, but that a significant change in financial and economic circumstances might be such as to justify a departure from the Reasonable Expectation without the Imperial duty being engaged or, if engaged, being breached (B690).

In itself Project Ocean does not seem to have been controversial. Its importance for present purposes lies in its being the first stage in the sequence of changes, and in the content and effect of the communications made in relation to it to the employees. Project Ocean having taken effect as of 6 April 2005, Project Soto followed later that year, coming into effect eventually on 6 July 2006. This had a greater impact on DB members and their pension rights, as we have already described, at paragraph [22] above.

By some time into 2005, NPPC in respect of the UK Plans was forecast to increase substantially into 2006, at least partly because of the working out of the effect of the stock market crash earlier in the decade. This was of concern to CHQ in terms of the contribution of the UK Plans to IBM’s overall NPPC, both for 2006 in particular and on a longer-term basis. CHQ also wished to reduce volatility in terms of the gap between the amount of the pension liabilities and the value of the pension scheme assets. The judge said that the Project Soto changes were initiated by CHQ, not in the UK. CHQ had identified similar problems of an acute nature in several countries: Canada, Germany, Japan, Netherlands, UK, US and Switzerland. Solutions were to be sought in each country. CHQ introduced the subject to UK management in September 2005, with cessation of DB accrual as a possible solution to be considered. UK management resisted this, and in the end did not propose it. The UK proposals put to CHQ by way of response would have made substantial reductions in 2006 NPPC, but were not accepted as adequate by CHQ. The proposals were therefore developed further before being approved by CHQ and then put to the Trustee in November. Mr Lamb reacted strongly and adversely to the proposals, coming so soon after Project Ocean.

Early in January 2006 revised proposals were put to the Trustee and discussed at a meeting between Mr MacDonald from CHQ and Mr Lamb and Mr Newman for the Trustee, attended also by Mr Hirst and Mr Wilson from the UK management team. By the time of that meeting the proposals on offer to the Trustee were those eventually implemented as Project Soto, including a substantial payment (£500 million) towards the past service deficit. Mr Newman made a note of the meeting in which he said this:

The judge discussed in some detail what could and should have been drawn from what Mr MacDonald said in the circumstances. He noted that Mr MacDonald’s statements to Mr Lamb could not have contributed to any Reasonable Expectation on the part of members, since they did not know what had been said. But he said that such an assurance was a factor to be brought into account in deciding whether the Project Waltz changes gave rise to a breach of the Imperial duty because of its influence on the Trustee in agreeing to the Project Soto changes.

The proposals were put to a Trustee board meeting on 19 January 2006 and were approved at that meeting. A consultation process was undertaken between January and April 2006, and eventually the proposals were implemented, with effect from 6 July 2006.

Having described the evolution of the Project Soto proposals between B737 and B834, the judge went on to examine the submissions made on each side about the statements made, expressly and by implication, and their effect on members’ expectations, between B835 and B1061. In doing so he also dealt with the process of consultation, appropriately since that was the context for most if not all of the relevant communications. The extent of that discussion

Another element is that it was at the stage of Project Soto in the history that a divergence of approach, motivation and knowledge as between CHQ and the UK management became not only more marked but also, in the judge’s view, more significant for the case as a whole. Thus at B835 the judge recorded the RBs’ case as being that the Soto changes were approved by the Trustee on the basis that the Plans would have a sustainable long-term future, and that IBM contended that this was the expectation of UK management at the time. The judge accepted that this was indeed the expectation of UK management, but said that it was difficult to believe that Mr Koppl or Ms Salinaro at CHQ, for instance, thought that the DB plans had a long-term future.

Communications about the proposals to managers made reference to volatility as being one of the issues sought to be addressed by the proposals. The judge held that what was said at these meetings (which took place before communication to members generally) made no difference in terms of expectation to the understanding that could reasonably have been derived from what was said in relation to Project Ocean (B863 and B865).

The proposals were announced to members in January 2006. Mr Hirst sent an email setting the proposals out in which he introduced them by referring to Project Ocean as having been designed “to share the increased cost of longevity in the future service costs” of the DB Plans, and described the new proposals as “a balanced package of proposals which address broader cost and volatility issues”: see B866. He said that there was no proposal to close any of the DB Plans, and that the proposals looked to the future, “ensuring that the pension you have already saved is protected” and offering a choice for the future. He described the proposals in outline, but also said that further material would be made available (as it was, including the presentation that had already been made to managers) from which the proposals could be ascertained and assessed in detail.

Mr Lamb for the Trustee also wrote to members saying that the Trustee had agreed in principle to the changes: see B875-881. He said that the Trustee was disappointed that IBM had had to make these proposals only a year after the last changes, but that it had been explained that:

Having referred to the principal features of the changes and to it being the best possible deal for members in the circumstances, he concluded as follows:

The judge said that the view so stated was Holdings’ view, but also that some of those at CHQ knew that this view was being expressed: B884. He went on to say that “CHQ must be taken as approving of, and adopting, that view. Or if that is not so, it may nevertheless be highly relevant when considering a breach of its Imperial duties by Holdings.”

Mr Lamb’s statement was no doubt based in part on his meeting with Mr MacDonald referred to at paragraph [96] above. The judge said that drafts of his letter had been the subject of comment from and discussion with CHQ as well as with other members of the Trustee board: B888. He recorded that Mr Heath made a suggestion that would have led to the words “long term” as a description of the changes being omitted, but that Mr Lamb had rejected this firmly, referring to Mr MacDonald’s statement of the position, and making it clear that if the changes were not long term then this needed to be confirmed and the matter would then be reopened within the Trustee board. His challenge was not taken up, and the statement remained in his letter.

The member consultation process extended over almost 6 months. For employees who were members of one of the DB Plans the critical feature of Project Soto was the choice they were offered, between staying in the DB Plan, but on the basis that only two-thirds of future pay increases would be pensionable, on the one hand, and on the other hand joining the DC Plan on advantageous terms (as regards the employer contribution) and with the whole of their future pay increases counting towards their ultimate pension. It is easy to see that the balance of advantage would be affected by how long the DB Plan would remain open. It had recently been announced that the US DB Plan was to be closed and there was no doubt speculation that the UK DB Plans might be at risk of closure: B868.

The judge referred to some of the content of Q&As supplied to members. Prompted no doubt by reaction to the fact that this was IBM’s second set of proposals within a short time, there were questions about the prospect of a further set of changes, for example: “How do we know IBM will not have to do this again one year later?”. Mr Simmonds relied before the judge particularly on one exchange:

Another Q&A addressed in terms the issue of closure: “How do we know IBM will not close the DB plans in the future?” The answer was almost exactly the same as the answer quoted in the last paragraph.

The judge had said, at B886, that a reasonable member reading the words “long term” and “no plans for future change”, in Mr Lamb’s letter, in isolation would “be entitled to form a Reasonable Expectation that, in the long term (whatever that may mean) there would be no changes absent a significant change in economic and financial circumstances.” Having reviewed the communications with members extensively, he found little more which he regarded as the basis for any Reasonable Expectation. As regards future benefit accrual he found that the Reasonable Expectation which he regarded as arising from Project Ocean was modified by Project Soto, because not all of future pay increases would count towards a DB pension, but subject to that he made findings as to the existence of Reasonable Expectations (at B1045, B1052, B1053 and B1054) which we have already quoted from his later summary in B1510: see paragraph [64] above. We have also referred to his clarification of those findings at paragraph [65] above. He also observed that Project Waltz would rightly be seen as conduct contrary to the Reasonable Expectation as to benefit accrual, absent a change of circumstances.

Before setting out those conclusions as to Reasonable Expectations, the judge had spent some time examining the contention on behalf of the RBs that the Project Soto proposals were presented to employees and members on a misleading basis. There were several aspects to this case on behalf of the RBs. One, which is important in the light of what the judge said about it, is that Project Soto would have had little effect on volatility, and that Mr Koppl and other executives at CHQ knew it would not solve the problem of volatility, but that a contrary message was given to UK management who in turn conveyed an inaccurate message to members. This in itself was said to be a breach of the Imperial duty. The judge said at B967 that it was clear that Project Soto did not solve the volatility problems facing the DB Plans, but he was satisfied that “IBM could reasonably have thought that there could be a positive impact on both breadth of variance volatility [that is, volatility as regards the gap between the value of the scheme assets and the amount of the pension liabilities] and on downside risk [that is, the worst possible projected outcome in a given year]”.

CHQ sought the advice of Towers Perrin on an ongoing basis as regards the IBM pension problems and what might be done about them. In September 2005 (see B970) an email from Towers Perrin to Mr Koppl and Ms Salinaro at CHQ about the Soto options referred to the current efforts as involving a lot of effort and noise and “not really solving our future volatility in each case”, and suggested that if the changes did not resolve the problem of the 2006 profit

Having addressed what was said in the oral evidence on this aspect of the case, the judge concluded that there was a basis for the view that Project Soto would have a modest positive impact on volatility, but said that this did not justify “expansive statements” made to members about what Project Soto would achieve: B984. He referred to the secrecy maintained by CHQ, as referred to in B971 (quoted above), and then went on to consider what was known by whom in IBM’s management. Until then he had been dealing with Mr Koppl and Mr MacDonald, both of CHQ, of whom both had material contact with the UK management and the latter also with the Trustee. He said that Mr Greene (IBM Corporation’s Treasurer and Chief Financial Risk Officer, and a member of the Trustee board and Investment Committee from 2002-2011), Mr Loughridge (IBM’s global CFO), Mr Castellanos (Vice President Human Resources at CHQ) and Ms Salinaro (also of CHQ) were fully aware of Mr Koppl’s projections and Towers Perrin’s advice: B985. He then said it was unclear what other senior executives, whether in CHQ or in the UK, actually knew, but that they relied on Mr Koppl and others at CHQ to inform them of the position, and in particular that UK management depended on CHQ for an understanding both of the effect of Project Soto on volatility and of what level of volatility was acceptable to CHQ.

Communications between CHQ and the UK showed that the UK management did understand that reducing volatility was one of the aims of the project, and that they put forward proposals intended to achieve that objective: B987(iii), (iv) and (v). Mr Tennet for the RBs submitted to the judge that Holdings had no basis on which to say what it did to members about sustainability and commitment, that those at CHQ who knew the position could not properly have said those things and that Holdings was not entitled to shelter behind the

The judge then digressed from his review of the justification, or otherwise, for statements made to members in relation to Project Soto, in a passage headed “Whose duty?” from B999 to B1021, concerned with the respective knowledge of UK management and of CHQ, and of the relevance of disparities of such knowledge on relevant points. This is an important part of his judgment to which we will return.

He said that CHQ was the driver for the pension changes as regards each of Project Ocean and Project Soto, as it was later for Project Waltz. Not all the relevant people at CHQ had the same knowledge: Mr Koppl was not aware of some of the things said by CHQ to UK management or by UK management to UK employees. Nor did UK management always know of information that Mr Koppl passed to his colleagues at CHQ. The judge said this at B1000:

In the course of his review of this subject the judge expressed this conclusion at B1007, which is significant not only because of the issue as to treating the IBM Group as one entity, which we discuss later, but also because of his formulation of the question to be posed:

Reverting then to the case based on negligent misrepresentation about the prospective effect and therefore the justification for Project Soto (at B1022), the judge stated his view that it was relevant to consider whether statements made in relation to Project Soto (and Project Ocean) were inaccurate, whether as matters of fact or of implied representation about the beliefs of Holdings and CHQ as to expense and volatility, because such statements, if inaccurate, could be relevant to a determination of whether the disappointment of a relevant Reasonable Expectation gives rise to a breach of the Imperial duty. In the light of his consideration of the question of corporate hierarchy, he said that it was necessary to ask not only whether Holdings had reasonable grounds for making the statements it made, but also whether CHQ had reasonable grounds for allowing the statements to be made or, in other words, whether CHQ could properly have made the statements itself (B1033). The judge held that no criticism could properly be made of Holdings and the UK management for having made any relevant statements. However, he noted that Mr Simmonds had accepted that Mr Koppl would have agreed that he would have had no reasonable grounds for saying that the DB Plans were sustainable or that volatility had been addressed in a significant way (B1036). Mr Koppl did not keep his knowledge of the position to himself, of course, but shared it with other senior staff at CHQ. At B1020 the judge recorded that Ms Salinaro, Mr Loughridge and Mr Greene knew that Project Soto was not solving the volatility problem and that, on that footing, CHQ knew that to be the position. It would therefore not be correct to describe the DB Plans as sustainable in the absence of a solution to the volatility problem. A partial solution might be sufficient but only if it made an actual contribution to reducing volatility.

The judge went on to say that Mr Koppl’s knowledge of the position was important not in itself but because it had been shared with others at CHQ, including at least Ms Salinaro, Mr Loughridge and Mr Greene. He said (B1037):

On that basis he said that CHQ’s knowledge was to be taken into account in assessing whether Holdings acted in breach of its Imperial duty not on the basis of attribution of a shareholder’s knowledge to the company but “because of the factors which I have identified in my discussion of the issue”, which we take to mean, principally, the fact that CHQ was the driver for the project and that it knew of and approved the statements made by Holdings to the UK employees and members.

Having identified the Reasonable Expectations which he considered had arisen from the communications with employees and members in relation to Project Ocean and Project Soto (quoted from his later summary at paragraph [64] above) he came back to the knowledge of CHQ in an important paragraph, B1060, as follows:

IBM challenge that finding as to the state of mind of CHQ on the basis that it was no part of the RBs’ pleaded case that anyone at CHQ (or indeed anyone else on the part of IBM) acted in bad faith in any respect, and that therefore the issue had never been addressed in evidence and the judge could not properly make such a finding.

Thus, Project Soto was put in place and members made their choice as to whether to remain in the DB Plan but with only two-thirds of future pay increases being pensionable, or to switch to the DC Plan on the advantageous terms on offer.

The next important aspect of the history is what is called the 2010 EPS Roadmap. This was developed by Mr Loughridge and presented to investors in May 2007: B1075. It set out a target of achieving double digit earnings per share growth over the long term; illustrations given to investors related to the period to 2010. EPS had been $6.06 in 2006; the target was $11.01 in 2010. There were several elements to the plan, but a reduction of pension costs was to be a significant factor, estimated to generate $0.90 per share. In B1076 the judge showed how this was to be seen in context. Prospectively it required a reduction of worldwide retirement related costs, which in 2006 had been $2,428 million, to $707 million, which would require a profit figure of $428 million for NPPC.

The judge observed at B1096 that keeping to the 2010 EPS Roadmap became a corporate imperative. A fairly strong message had been given to the market and although it could not be regarded as a promise, a failure to deliver what the market had been led to expect might have had a serious impact on IBM’s share price. It left IBM with little room for manoeuvre if market performance were to fall short of expectations: B1097. So IBM was faced with a major problem when the global financial crisis occurred in 2008, leading to a collapse in asset values. All areas of the IBM business had to take steps to achieve the level of savings set out in the 2010 EPS Roadmap.

In this situation, attention was devoted at CHQ to retirement related costs (and, no doubt, also to other aspects of costs). At a meeting late in July the forecast contribution from pensions was down from $0.90 to $0.70 and a month later it was $0.66. That was at the time of a meeting between Mr Loughridge, Mr MacDonald and Mr Koppl, who was given the task of considering the various options. They had discussed reducing DC benefits and medical plan benefits, but this was regarded as inconsistent with IBM’s general policy of encouraging DC membership, and Mr MacDonald also had in mind the recent closure of the US DB plans. Mr Koppl was therefore left with changes to DB plans worldwide as being the only other available course. By 10 October 2008, after particularly severe falls in the market (which may have been the lowest point in the market), the projected contribution from pensions was no more than $0.21. It was at this time that consideration of the project was embarked on which, for countries outside the US, became Project Waltz. Three countries were identified as giving rise to particular problems because of their high service cost: Germany, Japan and the UK. By the end of October examination and development of the project was well under way in relation to these three and three other countries, with the benefit of preliminary work by Towers Perrin but on the basis that local teams and local UK consultants would take over the development of the proposals for the UK.

Mr MacDonald came to the UK to meet Mr Riley (general manager of IBM UKI since April 2008) and Mr Stephen Wilson (IBM UKI CFO). They discussed the closure of the DB Plans. Mr Wilson said that this would present difficulties as regards relations with the Trustee and with employees, particularly in the light of the two recent sets of changes. Mr MacDonald understood what he said as even stronger, namely that closure would not be possible. He reacted by telling Mr Loughridge that Mr Wilson ought to be moved on from his role as CFO. The judge interpreted this as confirmation that CHQ had by then decided that DB accrual in the UK Plans should be terminated. Mr Wilson was eventually removed from discussions about changes to the UK Plans in late January 2009, and left IBM in April.

CHQ regarded the achievement of the project as a very high priority and adopted a hard position in relation to local management teams. Mr MacDonald and Mr Koppl came to the UK for further meetings in February 2009. Three options were under consideration: a mandatory move to DC accrual; freezing of pensionable salary and a restriction on early retirement; and a total compensation approach to determine employee pay increases, plus a restriction on early retirement; the latter, however, was least favoured by CHQ. As a result of the meeting these were all to be worked on for a further meeting in early March.

At that meeting the UK management presented proposals of which the principal features were these (see B1190):

Close the DB Plans to future accrual as at 5 April 2011 or 2012; ii)No further pensionable salary increases up to closure; iii)Restrict approval of early retirement requests from 5 April 2010; iv) Offer an early retirement opportunity for 600 employees on the

The various points were discussed and agreement was reached subject to more detailed formulation. The resulting proposals were formally approved by CHQ on 23 March 2009, in substance by Mr MacDonald. Closure of the DB Plans was to take effect as at 5 April 2011.

Before proceeding further with this part of the story we must mention a parallel process, following a review conducted by Mr Riley on his appointment as general manager for UKI in April 2008. He identified ways in which UKI needed to change in order to meet the current economic circumstances. The resulting project, planned in the second half of 2008, was launched in January 2009 and was called UKI Transformation. It aimed to improve the underlying competitiveness of the UK business and its profit contribution, to ensure that the reward structure was fair as between different groups of employees, and to reduce the volatility of and risk attached to pension liabilities and (in order to assist competitiveness and profitability) to reduce future pension costs: see B1140. The initiative was driven by the need for UKI to compete with other parts of the IBM empire for investment, in relation to which, before Mr Riley arrived, the UK operation was not seen as especially attractive.

Those involved in formulating Project Waltz in the UK foresaw that there would be resistance on the part of the Trustee to the proposals, and that one reaction might be to change the asset allocation of the investments in the Plans, which would itself affect the NPPC through the EROA. Attention was therefore given to how the proposals would be presented to the Trustee, and in due course to the employees with whom consultation would be carried out in accordance with IBM’s normal practice, quite apart from the fact that it would be required by statute in this case because of the nature of the proposals.

A final version of the Project Waltz proposals for the UK was sent by Mr Ferrar to CHQ (Mr Castellanos) on 22 April for approval, which was given that day. It was on the agenda for a board meeting of Holdings on 29 April, at which it was approved. The judge examined the evidence about this meeting in detail, between B1212 and B1230. Mr Chrystie, the new CFO for IBM UKI, played a significant part. He was aware of two motivations for the changes, one local to the UK and one global. (These are what were referred to as the local and the global strands.) His focus was more on the local position. The judge accepted his evidence as to his state of mind: B1222. The judge also accepted (at B1226) that the board was presented with a business case for the changes, both in terms of the UK business and also of CHQ’s involvement and the need to make savings so as to assist in meeting the 2010 EPS Roadmap target. While the judge accepted that CHQ “was in practical terms calling the shots”, he said (at B1230) that he was satisfied that the board of Holdings:

Mr Riley met Mr Lamb on 1 May to give him advance notice of IBM’s financial position and the Project Waltz proposals. Mr Lamb’s reaction, unsurprisingly, was to oppose the idea of further changes. The Project Waltz proposals were presented formally to the Trustee board on 7 May 2009. One feature was that they were presented on the basis that the DB Plans would close on 5 April 2010, whereas Holdings had already decided that the relevant date would be 2011. Discussions with the Trustee were not easy. Holdings took the view that it could implement the proposals without Trustee consent, but clearly wished to obtain that consent on the basis that it would make life easier in a number of respects. The Trustee did not agree to the proposals, taking the view that, in the circumstances, they were legally questionable, and that they could not properly be implemented without the court ruling on their validity.

The proposals were then announced to the membership in July 2009; there was a strong adverse reaction. A consultation process followed which is the subject of distinct issues in the proceedings. That consultation was carried out without distinguishing between the roles of Holdings and of UKL, and in the judge’s supplemental judgment about the roles of those two companies, he said that it should be taken to have been carried out by both of them.

A Pensions Consultation Committee (PCC) was set up for the purpose, with member-elected representatives as well as IBM representatives. The consultation process was to run from 5 August to 5 October 2009, though in the end it started on 10 August and was supposed to end on 20 October. Employees were told that no final decisions would be made until the end of the consultation period. IBM posted information for the consultation on an internal group web page. From the outset it stated that the changes on which it was consulting were intended to take effect on 6 April 2010, although it was clear from the evidence before the judge that the only real choice under consideration by IBM was between start dates in 2011 and 2012. The member representatives and the Trustee both sought information from IBM as to the details of the proposals and as to the rationale for them, without much success. The judge described the points made at some of the PCC meetings at B1273 to B1286. At B1276 he said that additional suggestions had been put forward at the sixth session on 9 September and observed that, given the very serious effect that the proposals would have on some members, “one might reasonably expect that IBM would give genuine consideration to any sensible proposal”, and that the time had not arrived at which it would be too late for new suggestions. So far from that, the judge described what happened next as follows:

The seventh session was mainly concerned with suggestions on the part of a particular class of members, the Enhanced M Plan members (a category which emerged from Project Soto). At the eighth meeting, on 23 September, IBM said that the employee suggestions were not accepted, and set out its proposals (including an effective date of April 2011). These proposals were not well received, and employee representatives continued to press for information including as to the business rationale for Project Waltz, which up to that time had been given in only such general terms as not to provide a worthwhile basis for comment and real consultation. Despite this request, of which the judge said it was an eminently reasonable stance for the employee representatives to take, IBM declined to offer any more than had already been said. That position was made clear at the ninth meeting on 30 September 2009. The judge observed at B1288 that there was no hint in any communication with the Trustee or the employee representatives that a significant factor (to say the least) for IBM was the 2010 EPS Roadmap.

Before the tenth and last session of the PCC, IBM and the PCC were made aware of the possibility that the Main Plan Trust Deed, executed in 1983, had been drawn up in error because it appeared members of the then new C Plan (the largest group of DB members, and, with particularly long service) of a right, which had been communicated to them by announcement on introduction of the C Plan, to retire at 60 without any discount and without the need for Holdings' consent. IBM’s initial position about this suggestion was that it was not justified, and although it made certain concessions later as regards members who might have had such a right, it did not modify its position in the Project Waltz consultation, and pressed ahead with the proposals without regard to this matter, despite (for example) a letter from the Trustee dated 14 October expressing concern about this issue.

Eventually proceedings were brought against Holdings and UKL by the Trustee, and they resulted in the judge holding that the 1983 Deed and Rules had been incorrectly drawn in this respect and that they ought to be rectified accordingly: IBM United Kingdom Pensions Trust Ltd v IBM United Kingdom Holdings Ltd[2012] EWHC 2766 (Ch), [2012] PLR 469 (the rectification proceedings).

At the tenth and last session of the PCC IBM provided (for the first time) a draft of the proposed new early retirement policy, together with the proposal of an early retirement window.

The judge held that IBM’s decision to proceed with Project Waltz was taken on 13 October (B1302). It was reflected (though not in specific terms) in IBM Corporation’s announcement of its third quarter results on 15 October 2009 (B1304). The consultation period finished on 20 October and on that day Mr Riley announced the decision to proceed to all managers, and later that day to all members of the Plans. On 23 October the Trustee responded stating its position that it would require a ruling from the court as to whether the Project Waltz proposals were or were not in breach of Holdings’ Imperial duty.

We will now describe Project Waltz in more detail, before coming to the judge’s findings about it.

The first element was to exclude active DB members from membership of the Main Plan and (with the specific exception of certain members who had a contractual entitlement to continued DB accrual) also from the I Plan with effect from 6 April 2011, by the use of the Exclusion Power (quoted at paragraph [16] above). The Notice said:

In the case of the Main Plan it went on to make this statement:

…

This does not constitute a direction that the persons so ceasing shall not be eligible to join the M Plan pursuant to paragraph 1 of Schedule B to the Money Purchase Rules.

The juxtaposition of those statements gives rise to one of the issues on the cross-appeal.

The second element of Project Waltz, as listed by the judge, was the NonPensionability Agreements (NPAs). The point of these, for IBM, was to ensure that future increases in salary did not increase the cost of the DB benefits. The termination of DB accrual would mean that future service would not count for the calculation of a member’s DB pension, but, without more, the pension attributable to past service would be calculated by reference to the member’s final pensionable salary, and would therefore be affected by pay rises, whether attributable to promotion or otherwise. This is what is known as the final salary link. IBM wished to eliminate this element of the cost of service during membership of the DB Plans.

Mr Riley’s email announcement on 7 July 2009 gave the first notice of IBM’s intention that future salary increases would be non-pensionable in relation to DB benefits. On 22 October 2009 Mr Riley emailed employees to say that salary increases would be offered in 2009 and 2010. The increases to be offered were known as an “Employee Salary Programme” or “ESP”. However, only 5 days later, on 27 October 2009, Mr Ferrar emailed those employees to be affected by Project Waltz in the following terms:

am writing to you to explain that any salary increases offered as part of this and any future ESP will not be pensionable as long as you remain a member of a Defined Benefit pension plan, even if such salary increases are backdated. …

On 9 November 2009, Mr Ferrar wrote again extending the deadline for decision by employees to 16 November and saying that employees who did not agree to the terms, or who failed to respond, would retain the option to change their minds after the deadline. Mr Ferrar also told members that the ineligibility for salary increases, in the absence of NPAs, would only apply to the period 2009 to 2011 after which the position would be reviewed: B1326 and B1535(ii).

According to the judge (B26(iv)) 3798 active DB members were asked to agree to the NPAs in 2009, of whom 3066 accepted, 27 refused and 705 did not respond. It is not clear, we think, from the materials opened to us, which acceptances were made before the extension of the deadline and the “clarification” that eligibility for salary increases would be reviewed in 2011.

In 2011 (by email dated 3 March) IBM invited employees who were members of the DB Plans to enter into NPAs, so as to be eligible to receive salary increases. They could reject non-pensionability but would then sign a document in the following terms:

do not agree to this ... proposal and recognise that I shall not be eligible for any future increase to remuneration that would otherwise increase my DB Pensionable Pay.

Although we do not know the numbers, it is clear that some employees received salary increases having agreed to an NPA and others did so having accepted under protest.

The judge did not address all of the issues arising from the 2009 NPAs in the Breach Judgment. He went into them further in the Remedies Judgment. In that he also considered NPAs entered into in 2011.

The third element of Project Waltz was the Early Retirement Window. As we have mentioned, early retirement required the consent of the Principal Employer, but IBM’s long-established practice had been to allow early retirement on advantageous terms. There was a discount for early payment, but it was noticeably lower than the rate which would have made early payment of the pension cost-neutral for the scheme. It therefore added to the cost which had to be met by the employer’s balance of cost obligation. The details varied from one section of the DB Plan to another, and they do not matter for our purposes.

Project Waltz would involve a change to this generous policy, to take effect from 6 April 2010. With this in prospect, IBM offered an opportunity to active DB members who would be 50 or older by 5 April 2010 to take advantage of the old policy: B30 to B32. The first mention to employees came in a Factsheet published on 25 September 2009, which said there would be an early retirement window in the fourth quarter of 2009 but gave no further detail. It was announced in detail by Mr Ferrar on 21 October by email to all relevant employees inviting them to express interest in taking early retirement before the change in the policy. They were given until 16 November to express interest in early retirement, with support by way of seminars, a modeller and one to one consultation in the meantime. Those expressing interest by then would be given IBM’s decision by 4 December and would have until 11 December to decide whether or not to proceed with early retirement on the basis of the terms offered. The judge recorded at B31 (iii) that of 2324 employees eligible to apply, 1162 did apply and 861 of those took early retirement.

The fourth element was the new early retirement policy. The judge set out the relevant part of the new policy at B33. It recorded that, where IBM’s consent was needed to early retirement, and where early retirement would be on terms more generous than cost-neutral, then IBM would normally only give that consent in exceptional circumstances, giving compassionate or medical grounds as possible examples, as well as possible situations to do with business restructuring or divestment.

It is to this change of policy that the rectification issue, described above at paragraph [138] above, was relevant. As already mentioned, eventually the judge heard the trial of rectification proceedings to determine whether the C Plan Trust Deed and Rules should be rectified to accord with this understanding, and he held that they should be so rectified: see paragraph [139] above. Thus, the new policy would not affect active members of the C Plan who wished to retire at or after the age of 60, even though the NRD for that Plan was 63. As also mentioned above, IBM’s response to the rectification issue being raised in the course of the consultation was an issue before the judge (it is Issue 20 on the appeal). He discussed this at B1290 to B1303.

The fifth element of Project Waltz, not in itself controversial separately from the issues about the project as a whole, was that DB members of the Plans who were to be excluded from membership were at the same time offered the option of joining the DC section of the Main Plan (the M Plan) once their DB membership came to an end. Members of the DB sections of the Main Plan would therefore be excluded as members by the exercise of the Exclusion Power, but allowed to rejoin as members of the M Plan. Members of the I Plan would be excluded from that membership under the relevant Exclusion Notice, and offered membership of the M Plan of the Main Plan instead.

Members who accepted this offer or invitation would have a special status in the M Plan, referred to as “hybrid deferred”, as regards their past service DB benefits. The judge described the benefits of this status at B41 and B42. In practice they were of limited advantage, except for the promise of enhanced employer contributions to the DC schemes in respect of them for 2011-2 and 2012-3. Other features which might have been benefits would be negated or substantially reduced by the effect of the NPAs and the new early retirement policy, if those were valid.

Holdings also offered the excluded DB members a different option, namely to join a new personal pension plan, the IBM UK Personal Pension Plan, a contract-based arrangement administered by Standard Life as a DC arrangement entirely outside the Plans, with effect from 6 April 2011. Of course a given employee could also choose not to have any IBM-related pension provision at all for the future.

At the trial the RBs mounted a sustained attack on the use of the Exclusion Notices on a number of grounds. The judge rejected all of these, and only one of them is pursued on appeal, by way of a cross-appeal by the RBs. Their contention is that the Exclusion Power in the Main Plan was used for a collateral and improper purpose, so that the Notices were invalid. This argument does not apply to the I Plan. We deal with this Issue first although it is last on the list of Issues, because, if the RBs’ contention were correct, the Exclusion Notices would have been completely ineffective in relation to the Main Plan. Correspondingly, this was one of the issues with which the judge dealt first in the Breach Judgment, before coming to the issues of breach of duty.

We have quoted the relevant provision of the Trust Deed (which is in clause 4 of Part V) at paragraph [16] above. It can be exercised in relation to a person or a class of persons, and is to be exercised by notice in writing. The notice may direct that the person or class “shall not be eligible for membership, or shall cease to be a Member or Members”. Membership means membership of the scheme as a whole. Such a notice will terminate the membership of a relevant person who is currently a Member, will preclude someone who is not a member from becoming one, and, so the judge held (at B270), will also preclude someone who ceases to be a Member by the effect of the notice from being eligible for membership. Thus the excluded Member cannot just re-join any part of the scheme which is open to new members (as the DC parts are, but the DB parts are not).

However, the Rules of each part of the Plan do contain provisions under which someone who is not, but has been, a Member can be re-admitted, with consent. The Rules relating to the DC part of the scheme provide for eligibility for membership in Schedule B rule 1. A person satisfying conditions set out in rule 1(1) becomes a Member unless rule 1(2) applies, which covers cases of the person opting out of membership, but also (among other cases) the person being precluded from membership by a notice served under Clause 4 of Part V of the Trust Deed. Rule 2, correspondingly, says that a Member shall cease to be a Member, among other cases, if he is precluded from continued membership of the Plan by the Employer under a notice under clause 4. However, rule 3 provides that a person who has ceased to be a Member by reason of any of the events mentioned in rule 2 (and who otherwise still satisfies the conditions of eligibility) “may resume his membership of the Plan”, but this is at the sole discretion of the Trustee. There is also rule 1(5) under which the Trustee, at the request of the Principal Employer, may waive any or all of the eligibility and entrance conditions set out in the rule. Probably this provision relates, in practice at any rate, only to persons who have not yet become Members, and not to someone who is a former Member whose membership came to an end by way of a notice under clause 4.

Similar provisions are set out in the rules for the DB parts, but since these have been closed to new Members for years, they are both more complicated and less relevant than those of the DC part rules.

On the basis of these provisions the judge recognised, at B271, that a Member who is excluded from the DB parts by a relevant notice could join the DC part by virtue of rules 1(5) and 3, so long as the Trustee gives consent.

The notice served by Holdings was expressed to direct that all Members currently accruing benefits under the DB rules “shall cease to be Members of the Plan with effect on and from 6 April 2011” (with immaterial exceptions) but it went on to say that “this does not constitute a direction that the persons so ceasing shall not be eligible to join” the DC part. For the RBs Mr Stallworthy Q.C. submitted that this discloses an inherent flaw in the exercise.

Furthermore, he contends that Holdings intended by the Notices to break the final salary link (as described at paragraph [146] above), but the judge has held that the exercise of the power was implicitly subject to a proviso such that it could not do so: see B289(iii). Therefore, he argues, Holdings sought to use the power for a purpose beyond its ambit, which is also an improper purpose.

The judge considered the arguments on this aspect of the case at B281 to B285. He rejected the submissions for the RBs at B285. In our judgment he was right to do so. He pointed out that Holdings could have offered the DB Members the option of joining a new and free-standing money purchase scheme with the same benefits and other provisions as the DC part of the Main Plan, and could have used the Exclusion Power to terminate their Membership of the Main Plan in that case without any problem. If so, he could not see why it would be improper to achieve the same result by providing benefits through the DC part of the Main Plan.

It is to be noted that the points are not run in relation to the I Plan because, first, the Members of that Plan were indeed to be excluded from that Plan, which did not have a separate DC part. Instead, those persons were offered the same choices as former DB members of the main plan: to join the DC part of the Main Plan, or to join a new and distinct personal pension plan, a DC arrangement outside the Plans: see B39 and B40. Thus, what was on offer to the I Plan members was very similar to the judge’s hypothetical case in B285. Moreover, the Exclusion Power in the I Plan was not subject to the implied limitation as regards the final salary link, because it was not introduced into that scheme by amendment under a power which itself was limited. It was in the documentation for that scheme from its outset.

Mr Stallworthy challenges the judge’s conclusion at B285, on the basis that it by no means follows, if you can achieve a given result in two steps, that you can elide those distinct steps and do the same in a single exercise. In support of this, in written submissions, observations were relied on of Millett J in Re Courage Group[1987] 1 WLR 495 at 511D. That concerned a very different situation. The judge said this:

We accept the proposition that, because a given result can be achieved if you go about it in one way, it does not necessarily follow that you can also achieve it if you proceed in a different way. The judge referred to this proposition earlier in his judgment at B206-7. We also accept the submission of IBM that pension scheme documents “should be construed so as to give a reasonable and practical effect to the scheme” and that “technicality is to be avoided”: see Arden LJ in Stevens v Bell, Re British Airways Pension Scheme:[2002] EWCA Civ 672, [2002] PLR 247 at paragraph 28.

For IBM Mr Simmonds’ response to Mr Stallworthy’s submission as to the alleged improper purpose of excluding members but at the same time inviting them back in is that there was nothing improper in what it set out to do by

We agree. In our judgment the judge was right to reject the contention that the service of the Exclusion Notices was vitiated on the basis that the power was used for an improper purpose. The statement as to eligibility to join the DC part was collateral to the function of the notices, which was to exclude members from membership of the Plan. The Notices were therefore neither self-contradictory nor internally inconsistent.

As regards the other part of Mr Stallworthy’s argument, relying on IBM’s intention, which it could not fulfil, to break the final salary link, this is a different kind of situation. It is not really a case of improper purpose at all but, at most, of what is sometimes called excessive execution, that is to say a purported exercise of a power which, for some reason, cannot take effect in full. That is quite unlike the classic cases of improper purpose where the defect lies not in the terms of the execution of the power but in the motive lying behind it. The judge considered this very line of authority earlier in his judgment when addressing the first issue, whether the exclusion power had been validly introduced into the Main Scheme trust deed at all. He referred at B199 and following to several authorities, including Bestrustees v Stuart: [2001] EWHC 649 (Ch), [2001] PLR 283, which Mr Simmonds also showed to us. The judge held that the Exclusion Power was validly introduced, but was subject to an implied limitation such that it could not be used to break the final salary link: B289(i) and (iii).

By the same reasoning, it seems to us that there is no reason why the exercise of the Exclusion Power by the notices actually given in this case should not be held valid to the extent permitted by the implicit limitation on the power. If one were to ask whether Holdings would have given the same Notices if it had been aware that it would not be able thereby to break the final salary link, the answer would have to be that it would. The object of terminating DB accrual was the principal reason for using the power. That it could not break the final salary link would perhaps have been seen as a disadvantage, but not at all as a reason for not exercising the power to the full extent available, not least because that feature was also to be dealt with by the NPAs as a separate element of Project Waltz.

In reply Mr Stallworthy showed us passages from Thomas on Powers, 2^nd edition (2012), Chapter 8, where the author suggests that it is sometimes difficult to distinguish between cases of excessive execution and other defects including fraud on a power. That may be so in some cases, but in the present case it seems to us that Holdings’ intention was to use the power to exclude DB Members from their membership of the scheme, which is within the scope of the power. The fact that IBM also hoped, but failed, thereby to break the final salary link was incidental. We therefore dismiss the cross-appeal.

In the list of issues agreed between the parties, the first is described as a threshold issue, namely on what basis did the judge decide that Holdings breached its Imperial duty and UKL breached its contractual duty? IBM argued that, although the judge identified, at least in principle, the correct legal test on which to decide each of these matters, the reasoning he expressed when deciding whether the two companies were in breach shows that he did not apply the correct test. The RBs contend that it is absurd to suggest that, having directed himself correctly, at length and repeatedly as to the legal test, he failed to apply it when he came to apply the law to the facts. We must therefore analyse in some detail what it was that he said when enunciating the test and when applying it.

The separate roles of Holdings, as Principal Employer under the Plans, and of IBM UK as employer of the relevant individuals who were members of the Plans, make it clear that different duties owed by the different entities are relevant to separate aspects of Project Waltz. Holdings was subject to the Imperial duty in relation to the exercise of the Exclusion Power, and to any other relevant exercise of a non-fiduciary discretionary power under the Plans, in particular its position as regards early retirement. IBM UK owed the contractual duty of trust and confidence to its employees in respect of its part in the proposals, above all the NPAs.

Although Project Waltz has to be seen as a composite whole, as it was presented by IBM to employees, we will focus first on the position of Holdings and therefore on the elements of terminating future DB accrual and changing the policy as regards early retirement.

Warren J reviewed the cases and the submissions made to him about the nature and content of the Imperial duty at some length, between B353 and B446. He accepted that the correct test was one of irrationality or perversity.

He had to consider what matters are relevant in determining whether the employer’s conduct is such as to destroy or substantially damage the trust and confidence between itself and its employees. The RBs submitted that there was no limit to the type of conduct which is capable of being destructive of the relationship of trust and confidence between employer and employee, and the judge accepted that proposition, so long as the conduct is such as to impinge on that relationship, and its impact is to be decided objectively: B420. All such facts and matters should go into the “melting pot”. IBM does not dispute that proposition as such.

In particular the RBs relied on expectations which they contended had been raised by IBM’s conduct with which Project Waltz as implemented would be incompatible. The judge accepted that what he called Reasonable Expectations could be relied on in this context, having referred to paragraph 146 of Newey J’s judgment in Prudential, cited at paragraph [33] above: see our discussion of Reasonable Expectations starting at paragraph [59] above.

In applying the Wednesbury test, the judge did not examine whether Holdings had considered all relevant matters and had excluded from its consideration all irrelevant matters. That is because the RBs did not advance a case on the basis of an error of that kind: see B1531. He therefore had to consider only whether the decision was one which no reasonable decision-maker could have reached.

Quite early in his consideration of the correct test, he set out five points which he regarded as clear (B359):

First, the exercise of a discretion, such as the exercise of the Exclusion Powers, requires “a genuine and rational” as opposed to an “empty or irrational” exercise of the discretion.

Secondly, the correct test is not one of fairness.

Thirdly, whatever the test is, it is a “severe” one.

Fourthly, the test, whatever it is, is objective.

Fifthly, the employer’s financial and other interests are relevant.

He did not put these forward as being in any respect exhaustive, but he did treat them as elements in the identification of the correct test. He then reviewed relevant cases that had been cited to him, and the submissions made on the basis of them, and came to express his conclusions at B441 to B446. At B441 he said this:

For IBM Miss Rose Q.C. submitted that, although expectations on the part of employees can be relevant to be taken into account by a decision-maker such as Holdings, what the judge said at B441 is an early indication of what came to be his error in deciding the case, namely that he accorded to the Reasonable Expectations a special status which obliged, or at any rate might oblige, the decision-maker to give effect to those Reasonable Expectations unless there had been some special change in circumstances.

After referring to the judgment of Rix LJ in Socimer Bank Ltd v Standard Bank Ltd, the judge said this at B443:

He then expressed his conclusion as follows, at B444:

However, he added (in B445) that Reasonable Expectations must not only be brought into the balance, but they may, in a given case, be critical in the sense that conduct which disappoints such Reasonable Expectations will amount to a breach of the Imperial duty.

Moreover, in B446 he observed, of the proposition that an employer is able to take account of his own financial interests in deciding how to exercise the powers vested in him without thereby putting himself in breach of his Imperial duty, that

He came back to the subject in the passages to which we have referred at paragraph [63] above.

For IBM Miss Rose submitted that although the judge’s self-direction at B444 may appear to be correct on its face, the words “in the sense that I have described it”, understood as a reference back to B441, show that the test so promulgated is not a true statement of a rationality test equivalent to that in Wednesbury, which we have held to be the correct test in a case of this kind (see paragraph [45] above), so that already at this stage the judge had begun to misdirect himself in law.

Then the judge turned to his consideration of the facts. In the course of that he came to the question of the relevance of the position of CHQ. That is an aspect with which we will deal later, but in the context of considering the judge’s self-direction, it is relevant to notice what he said at B1007, which we have quoted at paragraph [117] above, with his use of the concept of whether Project Waltz was “an appropriate response to the problems facing IBM Corporation, taking into account” any Reasonable Expectations..

That was said in the context of IBM’s argument that the proposals were justified and rational having regard to the so-called “global strand”, that is to say the requirement by CHQ of substantial and immediate cost savings in the light of the 2008 financial crisis: B1330(i). The judge then turned to the alternative justification by reference to the “local strand”, namely Holdings’ need to address its current and future lack of competitiveness: B1330(ii). In that context, he set out the same approach at B1011:

We note that, in the immediately following paragraph, the judge referred again to the question as being “whether Holdings would be acting perversely and irrationally (in the sense which I have described)”, so it does not seem that he intended, by using the word “appropriate” in B1007 and B1011, to abandon the “perverse and irrational” test.

Later in the judgment, towards the end of a long passage in which he examined IBM’s arguments as to the business justification for Project Waltz, the judge came to consider the relevance of IBM’s commitment to investors. In an important passage at B1391 he said this:

Such commitments [i.e. IBM’s commitments to investors] may need, however, to compete with other commitments. Suppose, for example, that Holdings had, with CHQ’s knowledge and approval, made the clearest possible commitment (not amounting to a contractual promise or giving rise to an estoppel) giving rise to Reasonable Expectations that it would not, during the currency of the Funding Agreement, close the Plans to future accrual or alter the early retirement policy. Adopting the test in relation to the Imperial duty which I have explained, the question would be whether making changes to the terms of pension provision would be within the range of decisions which a reasonable employer could take. In this example, if IBM sought to close the Plans or change the early retirement policy on the basis that it needed to do so in order to achieve savings required to meet its commitment to investors, a close examination would be required of precisely what it had committed to investors and whether, in the light of its Imperial duty, it could properly have committed to what it did commit. Even if that commitment had been properly given, a close examination would also be required to see whether IBM could meet that commitment whilst at the same time acting in compliance with the Reasonable Expectations which it had engendered. Only if it could not do so, it seems to me, would IBM have an argument for saying that the commitment to investors trumped its obligations to members.

For IBM it is argued that this passage shows the judge setting out a test under which, in given circumstances, the Principal Employer could only escape a finding of breach of the Imperial duty if, first, the commitment to investors had been properly given, having regard to the Reasonable Expectations of employees, and, if so, secondly, if there was no other course open to IBM by which it could meet its commitment to investors. That approach, it is said, is inherently quite different from the correct rationalitytest, and shows the judge failing to apply the correct test in his reasoning.

The judge went on at B1392 and B1393 to say this:

This is not to say that a significant change in financial conditions would not be capable of justifying changes to the Plans: but that is a different question. The point is that a deterioration in financial conditions which produces problems for IBM in meeting its commitment to investors would not necessarily be a sufficient justification for changing the terms of pension provision. It will all depend on the precise circumstances.

On the facts, the 2010 EPS Roadmap was first presented to investors in 2007. This was after the date of the events said to give rise to the Reasonable Expectations relied on by the RBs. Since certain Reasonable Expectations are, as explained, established, it follows, in my view, that IBM must do more than demonstrate that meeting the 2010 EPS Roadmap was a rational, commercially sensible, course but must show why the confounding of those Reasonable Expectations was within the range of decisions which a reasonable employer would take.

At B1394 he contrasted this with a situation in which the RBs had not established any Reasonable Expectations. In that case there would be no competition between meeting the 2010 EPS Roadmap and giving effect to Reasonable Expectations, and the RBs would need to show that the Project Waltz changes standing alone were not such as a reasonable employer would have adopted, or that the ways in which they were consulted on or implemented were ways in which no reasonable employer in IBM’s position would have adopted. No such case was advanced for the RBs.

Before coming to the passage in which the judge set out his conclusions, we must go back to examine what he said about the distinct contractual duty. His analysis of the position as regards that duty, owed by UKL to its employees, started (at B357) with the formulation of the duty set out first by BrowneWilkinson J, in the EAT, (as we have stated it at paragraph [27] above) and approved by Lord Steyn in Malik v BCCI. He then moved on to address the Imperial duty, pointing out that, although the Imperial duty was first enunciated by reference to the contractual duty, the content of the principle in one case is not identical to its content in the other case, and “the two situations must not be elided” (B358); he reiterated this point at B421, pointing out that there are several different types of case under employment law, in relation to which the test has been differently formulated, for example as regards contractual bonuses, on the one hand, as distinct from the BCCI case, by way of example, on the other. At B366, he accepted a proposition for the RBs that “at the end of the day the only relevant test is whether the impact of conduct complained of has, objectively, seriously damaged the relationship of trust and confidence”.

The judge’s examination of the debate about whether the test was “twopronged” (see paragraph [47] above) applied both to the Imperial duty and to

Much of his discussion of the features of the Imperial duty also applied to the contractual duty. He accepted in relation to both the relevance of any conduct which objectively can impinge on the relationship of trust and confidence (see paragraph [180] above), and that the employer’s conduct may not be otherwise actionable. Moreover, although his propositions are stated principally in relation to the Imperial duty, it seems to us that what he said at B441 to B444 (quoted at paragraphs [184] and following above) applied both to the Imperial duty and to the contractual duty; this follows, in particular, both from his reliance on contractual cases including Clark v Nomura, and from his use of the phrase (in B443) “in the context of employer/employee (and I would say employer/member) relationships”. This is consistent with his repeated references to Reasonable Expectations as being relevant in the context of the contractual duty as well as that of the Imperial duty.

After examining the rival contentions as to the business case for Project Waltz in itself, the judge reached the point at which he started to bring the threads together and to set out his conclusions, beginning at B1506. He restated the test for the Imperial duty, in general terms, as being “one of irrationality and perversity in the sense that no reasonable employer could act in the way that Holdings has acted in the present case” (B1507). As regards the contractual duty he put it differently, but observing that the different tests may not lead to different results:

His reference to B407 is to a quotation from Lord Steyn in Malik, speaking of a power to suspend an employee.

At B1508 he identified as the subject of the contractual duty, first the alleged failure to consult properly and secondly the issues about the NPAs. He dealt with the issues about consultation separately, but because the NPAs were an integral part of Project Waltz he dealt with them alongside his treatment of the Imperial duty which was relevant in relation to the termination of DB accrual and to the early retirement issues.

At B1509 he reiterated what he had said at B1011 (quoted at paragraph [193] above) namely that the question to ask was “whether Project Waltz was an appropriate response to the combination of problems facing both IBM Corporation (including in particular the requirement to meet the 2010 EPS Roadmap) and Holdings (including the need to improve competitiveness) taking into account the Reasonable Expectations of members.”

He then set out the Reasonable Expectations that he had found, which we have quoted at paragraph [64] above, and said at B1511 that there could be no doubt that the Project Waltz changes conflicted with those Reasonable Expectations. (That conclusion is challenged by IBM, on separate grounds to which we will return.) He followed up his identification of the Reasonable Expectations by saying, at B1513, that the communications to members in relation to Project Ocean and Project Soto “were not simply statements and communications; they were intended to be, and were, acted upon by the members in making the choices which they did”. As we have mentioned at paragraph [65] above, he did later clarify the Reasonable Expectation as to DB accrual by saying that it included an expectation that, if there were any later salary increases, they would be pensionable as to two-thirds.

Next he looked at the Project Waltz changes separately. He said that, if it had not been for the business case relied on by IBM, both the global and the local strands, the cessation of DB accrual and the change to the early retirement policy would have given rise to a breach of Holdings’ Imperial duty (B1514, quoted at paragraph [66] above). Correspondingly, he held that, absent a business case, UKL’s conduct in relation to the 2009 NPAs would have given rise to a breach of the contractual duty of trust and confidence. He did, however, observe that this conclusion only followed from treating the NPAs as a part of Project Waltz. If they had stood on their own they would not have been in breach of any duty (subject to one separate point) because there was no Reasonable Expectation about salary increases (B1516). In particular, he rejected an argument that, albeit that there was no Reasonable Expectation about any particular salary increase, there was at least a Reasonable Expectation that there would be no discrimination as regards salary increases by reference to the choice made at the stage of Project Soto as between staying with the DB Plan and moving to the DC Plan: B1517-8. This point is renewed on appeal by the RBs as Issues 14 and 15.

The separate point about the 2009 NPAs is that, when originally promulgated, the basis was that for those who did not sign there would be no salary increase at all at any time in the future. This was later changed so as to apply only to 2010 and 2011. The passage also needs to be read in the light of R18 to R22. The judge did not make a finding as to whether the later proposition involved a breach of the contractual duty, but he did hold that the earlier proposition was in itself a breach of that duty: see B1535(ii).

He also considered another hypothetical situation, one which he had identified in B1012, namely that CHQ’s decisions and conduct in relation to the Project Waltz changes were not constrained by any Reasonable Expectations, so that the only question would be whether Holdings was acting perversely or irrationally by agreeing to changes required by CHQ, even though they would disappoint Reasonable Expectations. At B1523 the judge said that, if that were the correct analysis, it could not be said that Holdings would have been

Having looked at the case from two hypothetical positions, he then moved on to his conclusion on what he saw as the correct basis, on the footing that IBM did rely on the global and local strands for its business case justification, but that Holdings could not just shelter behind CHQ’s demands “but must instead address the validity of the business case”. We have already quoted B1524 and B1525, at paragraphs [69] and [70] above, in which he stated his conclusion that Project Waltz involved breaches of the Imperial duty, as well as of the contractual duty.

What he said were his principal reasons are set out in thirteen sub-paragraphs of B1526. Having held Reasonable Expectations to be established, and that Project Waltz was clearly inconsistent with them, he said that the disappointment of the Reasonable Expectations was “a very serious matter going to the heart of the relationship” of employer and employee, and he repeated what he had said at B1513 about the status of the communications (see paragraph [206] above) and added that members had made decisions on the basis of those communications.

Then he came to consider the justification relied on by IBM as regards the local strand. He said this at sub-paragraph (iv):

He then turned to consider the justification advanced on the basis of the global strand, as to which he said that the target set by CHQ for the UK DB Plans was the principal driver for the Project Waltz changes in the UK. We set out sub-paragraphs (vi) and (vii) later, at paragraph [318] below. At subparagraphs (viii) and (ix) he went on to say this:

Had CHQ acknowledged that the members had the Reasonable Expectations which I have held they were entitled to hold and had CHQ recognised the need for Holdings to act, so far as possible, consistently with those Reasonable Expectations (and its own need, therefore, to enable Holdings to do so), CHQ should have given consideration to developing proposals which would meet the twin objectives of (i) meeting the EPS target in the 2010 EPS Roadmap and (ii) allowing effect to be given to the members’ Reasonable Expectations.

The evidence demonstrates that this was not done. Instead, CHQ determined that delivery of NPPC savings and increases to PTI were to be effected in certain ways. For the UK, that included pension savings of such a scale and having such attributes that the target set for the UK would be met. I am wholly unconvinced that proposals could not have been developed which would have met those twin objectives; indeed, I am satisfied on a balance of probabilities that they could have been. In the first place, the necessary savings to meet the EPS target (in contrast with the savings in NPPC cost) did not need to come from DB Pension plans, let alone from the UK DB Plans. If pension cost was an essential element, DC Plans could have been targeted: as the evidence (particularly from Mr Koppl) shows, that was an option and would have provided an easier solution. Another option might have been to impose heavier burdens on other DB Plans in countries where members had no similar Reasonable Expectations or no remedies arising out of them.

As regards the early retirement changes, he said this at sub-paragraph (xii):

We have already quoted at paragraph [71] B1528 in which he held, following on from these thirteen sub-paragraphs, that no reasonable employer in the position of Holdings [or UKL] would have adopted the Project Waltz proposals as they did, and accordingly that there were breaches of both the relevant duties. He reinforced this passage in R15 where he said that the 2009 NPAs, viewed as a separate element of Project Waltz but in the context of Project Waltz as a whole, gave rise to a breach of the contractual duty.

The judge’s reasoning at B1526(viii) and (ix) is challenged by IBM on several grounds. One is that it was not open to him to find, as he did in that passage, that there was a way of achieving the objectives of meeting the EPS 2010 Roadmap which was consistent with satisfying the Reasonable Expectations. The RBs had never alleged that there was such an alternative course of action and IBM had therefore never dealt with such a case in its evidence. We will address that point under Issue 26 at paragraph [374] below.

Leaving that aside, the judge’s reasoning in that passage, and likewise in subparagraph (iv), does show him approaching the issue of justification as regards a business case on the basis that, in order to establish such a justification, IBM had to prove that there was no other way of achieving its business objectives than by making the Project Waltz changes. That is clear from his use of the phrases “so far as reasonably possible” in sub-paragraph (iv) and “so far as possible” in sub-paragraph (viii). It is also shown by what he said in subparagraphs (x) and (xi). In the former he spoke of IBM not having a free hand to act as they chose. Because of the Reasonable Expectations, he said, they did not have a free hand, so that if they went beyond such freedom as they did have, the court could hold that there was a breach of duty on the part of Holdings. In turn at sub-paragraph (xi) he said that he was “not satisfied that CHQ was unable to develop proposals consistent with (i) meeting its concerns about headwinds and (ii) acting consistently with the members’ Reasonable Expectations”.

For IBM Miss Rose submitted that these several passages show that the judge’s approach to the final question was to start from the proposition that the Project Waltz changes were inconsistent with the Reasonable Expectations that he had held existed, and then to examine IBM’s case for a business justification on the basis that, unless the Project Waltz changes were the only possible way of resolving the problem that IBM sought to resolve, or maybe the only reasonably possible way of doing so, then the business case could not be an adequate justification, and the decision to disappoint the Reasonable Expectations was a breach of the Imperial duty and of the contractual duty. She further submitted that such an approach would have been erroneous in law, because it would not have been a true application of the Wednesbury test. That test, properly understood, requires no more than that the discretion will be rationally exercised having regard to all the circumstances relevant at the time of the decision.

For his part Mr Tennet accepted that, if the judge had done what IBM contended that he had done in this respect, he would have erred in law. However, he contended that this was not how the judge had proceeded. He pointed to the judge’s many references to irrationality and perversity, and to the concept of a decision that no reasonable employer could reach: to take only two examples see B471 and B1507. He asked, rhetorically, how it could be supposed that, having directed himself correctly and repeatedly as to the right test in law, the judge could fairly be read as having departed from this test at the moment of his actual decision. He also pointed to the fact that the sub-paragraphs in B1526 are said to be the judge’s principal, not his only, reasons, and that he referred to the “no reasonable employer” test in subparagraph (xiii) (quoted at paragraph [396] below) and again in B1528. As for the “so far as possible” phrases in sub-paragraphs (iv) and (viii) he submitted that all the judge was doing was seeking to refer to ways in which IBM could have proceeded without disappointing the Reasonable Expectations, and he was not saying that it followed that Holdings or UKL was in breach of duty as a result.

We cannot accept Mr Tennet’s submissions on this point. We accept that the points made in B1526 were not the judge’s only reasons for coming to his decision, but they were the main reasons, and it is these reasons that we must examine to see how he came to his conclusion. The issue is central to the case and we cannot suppose that if the judge had other substantial reasons for coming to the conclusion that he did on this point, besides what appears to be the necessity test indicated in the passages to which we have referred, he would not have expressed them in B1526 as being among his principal reasons. We must proceed on the basis that all of his reasoning of any significance on this point is set out in that paragraph.

The references to “so far as possible” do not make sense in context if they are only referring to ways in which IBM might have proceeded without disappointing the Reasonable Expectations. This would not have been the appropriate context in which to enter into a consideration of that question. In our judgment, Miss Rose is right to contend that these passages show the judge proceeding from his finding of Reasonable Expectations, as set out in B1510, and his finding that Project Waltz was inconsistent with those Reasonable Expectations, to a position in which the business case advanced by IBM as the justification for Project Waltz was held to be inadequate unless there had been no alternative course of action open to IBM. It is true that the judge does not articulate this in terms in sub-paragraphs (iv), (viii), (ix) or (xi), but he says that IBM should have sought alternative courses of action which would not conflict with the Reasonable Expectations, and that IBM could have found some such, and in sub-paragraph (x) he describes Holdings as not having a free hand, because of the Reasonable Expectations, and therefore as being constrained in its freedom of action.

This is also consistent with the indication given in B441 that expectations may have been engendered in such a way that to disappoint them would, in the absence of some special change in circumstances, involve the employer in

The judge also indicated this approach in B1391, quoted at paragraph [195] above, expressing the need to consider first whether a commitment had been properly given to investors and, even if it had, then to examine whether IBM could meet that commitment while at the same time acting in compliance with the Reasonable Expectations, and saying that only if it could not do so might IBM have an argument for saying that its obligations to members were overridden by its commitment to investors.

Thus, whereas the judge had, as it seems, accepted the proposition that all relevant factors should be taken into consideration, he did not approach the case on the basis that it was for the decision-maker to assess the weight of the various competing elements. Rather he concluded that, if Reasonable Expectations were established, then effect must be given to them unless there was no other possible, or reasonably possible, course open to IBM than to disappoint them. We venture another possible explanation at paragraph [326] and following below, but we prefer the analysis that we have set out in this paragraph.

This approach on his part was also foreshadowed in B533 where he takes two different situations, one where IBM was bound by contract to keep the Plans open to DB membership until 2014 and the other where IBM had engendered Reasonable Expectations to that effect. Pointing out that in the first case, IBM could not override the contractual obligation except by persuasion, he said that to disappoint the Reasonable Expectation in the other case might constitute a breach of the Imperial duty. To put the existence of Reasonable Expectations on the same footing as the existence of a separate binding legal obligation seems to us to be revealing of the judge’s attitude, which formed part of his eventual reasoning, that the existence of Reasonable Expectations has a special significance in relation to the decision-making process, so that it is not merely one among many relevant factors which should be considered and taken into account in the course of the decision-making process.

Accordingly, as we read the judge’s judgment, he failed to apply the Wednesbury test in relation either to Holdings as regards the Imperial duty or to UKL as regards the contractual duty. It seems to us that, in referring to the reasonable employer test, as he often did, he may have incurred the risk identified by Baroness Hale in Braganza at paragraph 29, quoted at paragraph [38] above, that “concentrating on the outcome runs the risk that the court will substitute its own decision for that of the primary decision-maker”. In particular, reference to the reasonable employer may lead to the application, even if unconsciously, of a test diluted and distorted from the true test of irrationality, as enunciated, for example, by Lord Diplock in Council of Civil Service Unions v Minister for the Civil Service [1985] AC 374, 410:

In what Baroness Hale (quoting both passages at paragraph 23 of Braganza) described as an obvious echo of this, Lord Sumption said this, in Hayes v Willoughby: [2013] UKSC 17, [2013] 1 WLR 935:

Thus, in order to apply the rationality test correctly, it might have been more helpful for the judge to refer to the rational employer, rather than the reasonable employer. In any event, his view as to whether any “reasonable employer” might have done what Holdings and UKL did needed to be tested by reference to one or both of these formulations: was the decision by Holdings or by UKL so outrageous in its defiance of logic or accepted moral standards that no sensible person who had applied his mind to the question to be decided could have arrived at it, or is it completely lacking in any logical connection between the relevant circumstances and the ostensible reasons for the decision?

In our judgment the judge did approach the issue on an incorrect basis, by according to the Reasonable Expectations which he had found to be proved a paramount significance, such that Holdings and UKL could only be found to have acted in accordance with their relevant duties either by giving effect to the Reasonable Expectations or by showing that there was no other course open to it in the given circumstances. The existence of the Reasonable Expectations, or at any rate the history of the communications to employees in the course of Project Ocean and Project Soto from which the Reasonable Expectations were said to arise, were relevant factors to be taken into account by the decision-maker. But to elevate them to a status in which they had overriding significance over and above other relevant factors was erroneous in law, and therefore vitiates the judge’s decision that Holdings was in breach of the Imperial duty and UKL of the contractual duty.

The judge held in terms (although referring incorrectly to Holdings) that UKL was in breach of its contractual duty of trust and confidence: B1525 and B1534. That breach of duty consisted of its part in Project Waltz, namely procuring the 2009 NPAs. He did not differentiate as regards this holding between the earlier and the later 2009 NPAs. In the Breach Judgment his holding in this respect proceeded from the fact that Project Waltz was presented to employees, and was to be taken, as a composite whole. This aspect of the changes could not stand on its own. Since the other essential features of Project Waltz, as the judge held, involved breaches of the Imperial duty by Holdings, so, in his view, this element must be regarded as involving a breach of duty by the entity relevant in this instance, namely UKL. This conclusion is reinforced by his clarification, in the Remedies Judgment R6, that the Reasonable Expectation as to DB accrual carried with it an expectation that any salary increases that were awarded would be pensionable as to two-thirds (see paragraph [65] above).

For the reasons set out above, we answer Issues 1, 2, 3, 3(a) and 5 together as follows. The basis on which the judge decided that Holdings was in breach of the Imperial duty, and UKL in breach of the contractual duty of trust and confidence, was that relevant employees had Reasonable Expectations, as to DB accrual and as to the early retirement policy, which would be disappointed by the Project Waltz reforms, and that, although to adopt the reforms was otherwise a rational commercial decision, they could not be justified, given the existence of the Reasonable Expectations, unless there was no way of achieving the company’s legitimate business aims which was compatible with giving effect to the Reasonable Expectations.

In so deciding, the judge erred in law. The correct approach is to apply a rationality test equivalent to that in Wednesbury (see paragraphs [45] and [46] above) in order to decide whether a decision by a decision-maker such as Holdings, as Principal Employer under a pension scheme, or UKL as employer, is valid and lawful having regard to the Imperial duty and the contractual duty of trust and confidence. Both limbs of the test can apply, but it was not argued in the present case that any irrelevant matter had been taken into account, or any relevant matter left out of account. Therefore the question was whether the decision taken was one which no rational decision-maker could have reached. Although the judge directed himself that the test to be applied was one of capriciousness, perversity or arbitrariness, which is close to the rationality test, he accorded an overriding substantive significance to the Reasonable Expectations such that they could only lawfully be disappointed in a case of necessity, which is not compatible with the correct approach. Members’ expectations, even if they satisfy the judge’s criteria for a Reasonable Expectation, do not constitute more than a relevant factor which the decision-maker can, and where appropriate should, take into account in the course of its decision-making process.

It follows that the judge was wrong to hold that Holdings was in breach of the Imperial duty in implementing Project Waltz by failing to give effect to the Reasonable Expectations as to DB accrual and as to the early retirement policy, and that UKL was in breach of the contractual duty of trust and confidence by procuring the NPAs as part of Project Waltz and thereby failing to give effect to the aspect of the Reasonable Expectation as to DB accrual that two thirds of any future salary increase would be pensionable.

On this basis, Issues 7 to 12 do not arise for decision. However, the significance of Reasonable Expectations was debated extensively before us and, given that this is the first case at appellate level in which the Imperial

When formulating the Reasonable Expectations that the judge held to have been proved, he expressed them to be subject to the effect of a significant change in financial and economic circumstances. At B1510 (quoted at paragraph [64] above) the Reasonable Expectation about continued DB accrual is formulated as subject to that qualification, and the Reasonable Expectation about the early retirement policy was qualified by the words “unless there was a justification for a change in policy”. Already in B666 the judge had said:

As regards the early retirement Reasonable Expectation, he had said at B690 that a significant change in financial and economic circumstances might be such as to justify departure from the Reasonable Expectation. At B1045 he had referred to the position after Project Soto and referred to the Reasonable Expectation about DB accrual as being affected by Project Soto, but that, as so altered, it “would not be further changed for some period of time in the absence of a significant change in financial and economic circumstances”. In turn, at B1392, having put the case about the effect of the Reasonable Expectation in terms which foreshadowed his reasoning that they should prevail unless there was no other alternative course (in B1391), he said that a significant change in financial conditions might be capable of justifying changes to the Plans, but that that was a different question: see paragraph [197] above. This had been foreshadowed by what he had said at B697 and B1052. In particular, at B697, dealing with expectations about continued DB accrual, he said this:

For IBM Mr Simmonds argued, first, that the indication in the final sentence of that paragraph that the two different approaches identified should not lead to different outcomes was belied by the judge’s conclusion, namely that if the Reasonable Expectation was in fact to be disappointed by IBM’s actions, then this could only be justified (so as to have no consequence) if there was no other course of action open to IBM. We agree that, of the ways in which the judge puts it at B697, the more logical formulation is that significant changes in financial and economic circumstances can override the expectation, so that no separate justification is needed. The member is not led to expect that (in this case) DB accrual will continue for the period which the judge identified as material (eventually, until 2014) if in the meantime there is a significant change in financial and economic circumstances. That fits with the judge’s formulation of that Reasonable Expectation at B1052 and B1510.

Mr Simmonds also submitted first that, consistently with the essence of the rationality test, whether there had been a significant change in financial and economic circumstances was a matter for the decision-maker to consider, not an objective question for the court to decide, and secondly that the judge never dealt with that question, on whatever basis, as a matter of decision.

The judge did refer to some of the relevant factual material, of course, while examining the business case advanced by IBM as justification for Project Waltz, at B1329 and following. At B1100 the judge had noted that the collapse in asset values which was part of the financial crash in 2008 was projected to have a dramatic effect on pension costs (as measured by NPPC) and this had an adverse effect on the projection for the EPS in 2010. By early October 2008 the contribution of NPPC towards the improvement of EPS figures under the 2010 EPS Roadmap, which had been put at $0.90 in 2007, had fallen to $0.21. A month later it had recovered a little, to $0.27. At B1111, having referred to the severity of the market falls by October 2008, the judge said that those falls did play a causative role in the decision to proceed with Project Waltz and that IBM was entitled to rely on those falls in the market as justifying its actions, though he left open for separate consideration whether, even with that reliance, IBM’s “actions can be justified”.

The global strand of IBM’s justification was the need for cost savings in all areas of IBM’s activities, worldwide, in order to meet its commitment to investors under the 2010 EPS Roadmap in the light of the global financial crisis. From the position, which was not contested, that the assumptions as regards NPPC which underlay the 2010 EPS Roadmap as formulated in 2007 were reasonable best estimate assumptions (see B1340(i)), and the further position that IBM regarded it as vital to adhere and live up to that commitment despite the financial collapse in 2008, the parties entered into a debate as to whether, as IBM said, the crisis in 2008 was of almost unprecedented and unpredictable magnitude or, as the RBs argued, the effects of the 2008 market falls were neither unprecedented nor unpredictable: B1343-4. The judge rejected the proposition that the 2008 crash was foreseeable (B1348), and he accepted that in October 2008 it was widely thought that the crisis would deepen still further (B1353). He said that, if it had not been for the Reasonable Expectations, the response of IBM and its group operating companies to the 2008 crash “made perfectly good commercial sense”, and that “the commitment to the 2010 EPS Roadmap was an entirely rational reason for the Project Waltz changes”: B1354. However, he drew a distinction between the position as CHQ saw it and the position of Holdings (and UKL) given the Reasonable Expectations which he had held to be established. He said that separate consideration was needed, having regard to the position of Holdings and UKL, as to whether the Project Waltz changes could properly be implemented.

The judge then turned to address the local strand, which was concerned primarily with improving the efficiency, productivity and economic performance of the UK business. That passage in the judgment does not take the matter forward as regards the assessment of the significance of the financial collapse in 2008.

He then dealt with the several heads of criticism advanced by the RBs in relation to IBM’s business case for the Project Waltz changes. This includes B1391 to B1393 which we have quoted at paragraphs [195] and [197] above. In B1392, as we have noted already, the judge identified a significant change in financial conditions as a separate issue, implicitly for separate treatment. However, he did not return to address that point as such later in his judgment.

For the RBs, Mr Tennet accepted that the judge never dealt in terms with the question which he had identified in B1392 by saying either that there had been a significant change in financial and economic circumstances, or that there had not been such a change, or indeed either that IBM was entitled to take the view that there had been such a change or that it was not so entitled. He argued that the judge had examined the business justification contended for by IBM, which depended heavily on the effect of the 2008 crash, and had considered the crash as part of that process. He submitted that the relevant issue was not the significance of the crash on a global basis but its impact on IBM, that is to say whether, as a result of it, IBM had undergone a significant change in circumstances. He drew attention to the fact that the crash had little effect on IBM’s operational performance, as distinct from its effect on the pension funds. He also relied on the graph at B938 which shows that, although the Return on Assets in 2008 was significantly below the Expected Return on Assets, it had been almost as much below EROA in the wake of the dotcom crash in 2002. From that he submitted that the judge was entitled to take the view that there was no substantial difference between the two cases and therefore the events of 2008 did not represent a significant change in relevant circumstances. However he accepted that the judge did not say anything to that effect in his judgment, and his ultimate submission on this issue was that

He did submit that IBM had encouraged the Trustee to adopt a high-risk investment policy in the hope of significant sums being earned on the invested funds which would exceed the annual outgoings and would therefore make a positive contribution to NPPC and thereby to the annual accounts as regards profit or loss, that this policy, in its nature, carried higher risks, and that it was not right for IBM to take the profit when things went well but to expect members to bear the loss when they did not. This was referred to as the “Heads I win, tails you lose” argument. As Mr Simmonds pointed out, this was an argument that the judge rejected, and which is revived by way of the Respondent’s Notice (Issue 17). It therefore does not answer the question whether, and if so how, the judge dealt with the distinct question of significant change.

We accept that IBM’s business justification for the changes does overlap with the question of significant financial and economic change, but it seems to us that, given the way the judge formulated the Reasonable Expectations that he found to exist, the two questions are necessarily distinct, as the judge recognised at B1392, following on from what he had said at B1391. If there had been a significant change in financial and economic circumstances, then the expectations that he had held to exist would give way to the effect of the change, so as to be overridden. Accordingly, for the judge to hold that the changes made by Project Waltz did disappoint the Reasonable Expectations, it was necessary for him (on his own approach) to come to a decision that there had not been a relevant change in financial and economic circumstances. On what we see as the correct approach, as submitted to us for IBM, he should have considered and decided whether or not IBM could rationally have taken the view that there had been such a change. Certainly the judge did not consider that question. We do not accept that the judge can be regarded as having decided the point, on either basis, without having said so expressly. On this point, as on the other ingredients of their case, the burden of proof lay on the RBs. It follows that, in the absence of a decision on the point, they failed to make out an essential ingredient of their case.

On Issue 10, therefore, we conclude that the judge did find that the Reasonable Expectations (both on DB accrual and on early retirement) were qualified so as to give way in the event of significant financial and economic change, and that he failed to address the question whether the Reasonable Expectations had been, in effect, overtaken by significant change at the time of the financial crash in the autumn of 2008. In particular, his statement in B1526(i), that Project Waltz was inconsistent with the Reasonable Expectations, was not justified in the absence of a finding on the question of significant change. He was right to identify it as a distinct question at B1392, but he did not come back to that question and answer it later in his judgment.

There is a good deal of force in Mr Simmonds’ submission that, if he had asked himself the question, he must have found that there had been a relevant significant change, or, properly, that IBM could rationally have concluded that there had been such a change. Whether or not that is so, it seems to us that the

At B696 of the Breach Judgment the judge explained his assessment of the message conveyed by the Webcast in November 2004. There had been no mention in the Webcast of early retirement, but there had been talk about past service. IBM was to take responsibility for the past service deficit, but members were being asked to make a larger contribution toward the cost of future service benefits. The judge said that the central message of the Webcast as regards past service was that “nothing is changing or will change without justification at least prior to 2014”. The basis for the reference to 2014 was that this was the end of the period covered by the guarantee by the American parent company of Holdings’ liability as regards the deficit.

At B687, in the section of his judgment headed “Automatic entitlement to an early retirement pension”, the judge said this:

Thus, in this passage the judge held that there was already a Reasonable Expectation as to early retirement in 2004, at the time of Project Ocean, based on the application for many years of a consistent policy by Holdings that those who wanted to retire early, at any age from 50 upwards, could do so on favourable terms as to the discount for early payment, because although it was a matter for the discretion of Holdings, that company had engendered the perception that consent would not ordinarily be refused, and it was aware of that perception on the part of scheme members. He contemplated that the policy might be changed, without breach of the Imperial duty, as regards a relatively young employee with only a few years’ service, but not necessarily so if the change is made with little warning in relation to a member aged 58 with 20 years’ service.

Then at B689 the judge held that a member who had that Reasonable Expectation as to early retirement would find it reinforced and confirmed by the Webcast, and that such a member would as a result have a Reasonable Expectation that the early retirement policy and practice would remain in place until the first quarter of 2014.

The only basis for identifying the date of April 2014 was that it was the end of the period for which the funding guarantee was to remain in force. However, the judge had already said at B646 that, although references in the Webcast to commitment and sustainability could be reinforced by the existence of the guarantee, its presence was not of itself enough to give rise to any Reasonable Expectation as to future accrual or continuation of current pension practice.

The judge’s finding as to a Reasonable Expectation as to early retirement existing in 2004 is striking. At B386 he explained how a Reasonable Expectation differed from a “mere expectation” (see paragraph [59] above). In essence the difference is this. A mere expectation is an assumption that something (for example a discretionary increase) will happen in the ordinary course of events, if things carry on as they are. A Reasonable Expectation differs from this in that it is an assumption that things will take a certain course, this assumption having been engendered by the employer, in relation

Mr Simmonds submitted that it is difficult to see why the undoubtedly longstanding and known policy and practice of Holdings, normally, consenting to early retirement on favourable terms should be found to be more than a mere expectation, and, further, that if the application of this policy and practice did give rise to a Reasonable Expectation such that, at any rate in some cases, it could not be changed (absent special circumstances) without giving rise to a possible breach of the Imperial duty, this would be worrying for employers and would discourage them from applying a generous policy as a matter of discretion, lest it be found to be legally binding in some way for the future and not open to being changed.

Mr Tennet submitted that the judge’s finding was based on his review of the evidence as a whole, and he pointed to statements made within IBM’s UK management which showed an awareness that members did regard early retirement as an entitlement. He also relied on the fact that employees’ benefit statements, issued each year, showed figures for retirement at different ages which, as regards dates before NRD, applied the current policy for early retirement discount factors. From 1997, we were told, these statements did include a statement somewhere in the text that early retirement was at the discretion of Holdings. However, he submitted that although it was, of course, a policy, that is not how either members or at least some in management regarded it.

Although Mr Tennet submitted that we should not assume that everything the judge relied on for his conclusion that members had a Reasonable Expectation as to the application of the early retirement policy until 2014 was identified in the relevant passage in his judgment (B680 to B687), it seems to us that what the judge did refer to in that passage justifies his statement (which we think was not really in dispute) that members had a widely-held perception that they would be able to retire early on favourable terms, because consent would not, in practice, be refused, and that Holdings was well aware of that perception. The basis for saying that this perception was “engendered” by Holdings was the fact that this was, and had for a long time been, its policy, about which it was quite open and which it did apply openly and consistently.

The difficulty arising from the judge’s own distinction set out in B386(iv) is to see, on the one hand, how that situation goes beyond what he described as a mere expectation and, on the other, if it does, where the essence of the distinction lies. As it seems to us, the situation described by the judge in B687 can fairly be categorised as one in which members do expect that in future they will be able to retire early on favourable terms, through the application of the company’s policy, because they expect that in the ordinary course of events the policy will continue to be operated and the favourable discount factors will continue to be made available. All that Holdings had done to engender this expectation was to adopt the policy, to continue to apply it, and to do so openly. That is, of course, something within its control but so is a practice such as that to which the judge refers by way of example in the context of a mere expectation, namely discretionary increases.

Nor can we see how the application of a policy as to a matter within the discretion of Holdings, however consistently and openly and for however long a period, can be said to provide a positive reason that the policy will continue to be applied, still less to be applied for any given period into the future, so that it could not be changed in respect of past service. Of course, a policy as to the discount factor to be applied on early retirement inherently affects benefits earned by past service, but it does so by allowing the member more by way of pension for that past service than he or she is strictly entitled to, by allowing the pension to come into payment early, but without discounting the amount of pension to the full extent appropriate to allow for the acceleration of payment. A rule which allowed for that would be one thing (and certain C Plan members were entitled to the benefit of such a rule from age 60 by virtue of the rectification of the Trust Deed and Rules) but the case we are considering is one which depends on the exercise of an unfettered discretion by Holdings. In our judgment what the judge said at B687 as to a Reasonable Expectation about the early retirement policy existing already in 2004 is not justified by his own definition of Reasonable Expectation at B386(iv).

In B687 the judge was considering the position in 2004, when Project Ocean was launched. Project Ocean had no effect on early retirement as such. However, as we have said, the judge did hold that a Reasonable Expectation as to early retirement policy arose from the Webcast and from Project Ocean on the basis that because the deficit attributable to all past service benefits provided by the DB Plans (including early retirement policy and practice) was to be guaranteed up to early 2014, therefore the reasonable member would be entitled to expect that those benefits would continue to be provided. Nothing was said in the context of Project Soto that had any bearing on early retirement at all, so the position as regards expectation, reasonable or not, in respect of early retirement cannot have changed at that stage. Accordingly, it seems to us that the judge’s finding as to a Reasonable Expectation relating to the early retirement policy is almost entirely based on what he said in paragraph 687 about there being a Reasonable Expectation already in 2004. That, however, is not justified even on the judge’s own definition of Reasonable Expectation. If there was no more than a mere expectation in 2004 on the continuance of the early retirement policy, it does not seem to us that anything said in relation to Project Ocean, the changes under which did not affect or relate to early retirement policy in any way, can have converted the mere expectation into a Reasonable Expectation as defined by the judge. We therefore conclude that the Reasonable Expectation as to the early retirement policy was not made out on the judge’s own test.

As for the Reasonable Expectation as to continued DB accrual, one of the striking features is the judge’s process of reasoning as to its duration. At B652 he had recorded a submission by Mr Tennet that members had a Reasonable Expectation that benefit accrual would continue for the foreseeable future, but said at B655 that he had no idea what that period was. At B657 he said that the guarantee provided no indication of the period of the foreseeable future.

In R23 he clarified and reinforced this by saying that it was implicit in his reasoning elsewhere in the Breach Judgment that the Reasonable Expectation endured for a period after 6 April 2011 sufficient for the Project Waltz change to have been incompatible with the Reasonable Expectation and therefore a breach of duty. At R426 he went further and decided that the Reasonable Expectation as to future accrual covered the period up to 31 March 2014. His reasoning in that paragraph explicitly ties the date to the duration of the other Reasonable Expectation as to the continuance of the old early retirement policy. If, however, there was no proper basis for his finding of a Reasonable Expectation as to the early retirement policy, for the reasons we have given above, then this reasoning in turn is not soundly based. Moreover, the date of 2014, which is derived solely from the Webcast, is the date when the funding guarantee comes to an end, but the judge had said that the guarantee was not a proper basis for any Reasonable Expectation. He said so, for example, in B628, B646, B648-9, B710 and B878.

In the light of those statements by the judge it seems to us that he could not derive any support from the existence or terms of the guarantee for his conclusion that the Reasonable Expectation as to accrual would last until 2014.

That would not matter, other things being equal, if his prior conclusion, that the Reasonable Expectation would last at least until some time beyond 6 April 2011 were justified. Going back to that, there is no reasoning as the basis for that conclusion in the course of the Breach Judgment, nor of the Remedies Judgment. He proceeds from the point where he says that the foreseeable future is a wholly uncertain period, which may not be more than three years, to the position of saying, in effect, that it is rather more than five years, and that it starts from 2006, the time of Project Soto, so as to run as far as, and some way beyond, 6 April 2011. With respect to the judge, it seems to us that this is an unreasoned conclusion, the basis of which it is impossible to discern.

In their Respondent’s Notice, the RBs contended that the judge ought to have relied on additional matters in support of his conclusion as to the end date (referred to as the Longstop Date) of the Reasonable Expectation as to DB accrual. This is Issue 16 in the agreed list of issues. They argued that “long term” is a phrase that has a meaning, and that it must be clear that, looking forward from 2006, the date of Project Soto, to stop further accrual by way of salary increases only three years later (by way of the NPAs in 2009) and stopping all DB accrual five years later by the use of the Exclusion Power in 2011 must on any fair basis be regarded as well within the period denoted by the phrase “long term”.

The judge did refer to Holdings’ promise, under Project Ocean, to contribute £200 million each year for three years in order to reduce the deficit, as an indicator of the duration of Holdings’ commitment to keep the DB Plans open: see B669. However the judge said that this would only justify an expectation of continued DB accrual for “at least” three years. The RBs argued that that reference by the judge gave insufficient weight to the guarantee and to other statements about continuing support made in relation to Project Ocean. Since Holdings was in any event liable to make up the deficit, which the latest actuarial valuation showed to be £900 million, it is not clear how the programme of deficit reduction could be seen as the basis for anything at all. The RBs also relied on the fact that the deficit as so calculated presupposed the continuance of DB accrual and of the current early retirement policy, and built on that proposition the contention that members could therefore reasonably expect that the policy would indeed continue for the future. The flaw in that argument is that an actuarial valuation is, of course and necessarily, based on assumptions as to what will be the case in a number of respects, some of which are within the employer’s control, such as pay policy and early retirement policy. However, these are no more than assumptions, which will be reviewed every three years as successive actuarial valuations fall to be made. They cannot properly be seen as a commitment by the employer not to change the basis of the assumptions, either within the three years for which, in effect, they apply or at any stage in the future. They fall into the judge’s category of “mere” expectations, as explained at B386(iv), since they involve assumptions that things will carry on as they are in the ordinary course of events.

For these reasons, we would hold that the Reasonable Expectation as to early retirement policy as found by the judge was not justified on his own formulation of the distinction between a mere expectation and a Reasonable Expectation (described at paragraph [59] above), and that, even if he was justified in finding a Reasonable Expectation as to future accrual, his conclusion that it would still be subsisting a significant period beyond 6 April 2011 is not explained by his reasoning and is not otherwise justified. Nor can his conclusion be justified by resort to the arguments relied on by the RBs under Issue 16.

Issue 7 raises the question whether, to be legally relevant as the judge held, Reasonable Expectations must meet minimum standards of clarity, certainty and assumption of responsibility, and if so what those criteria should be. Issues 8 and 9 depend on the answer to Issue 7. The issue does not arise on the basis of our decision that the judge was wrong to accord to Reasonable Expectations a substantive legal effect so as to require special justification, whether or not that justification must be that no other course was reasonably available. As we have said, expectations held by employees and other members of a pension scheme may be legitimate matters to be taken into account by a decision-maker, such as someone in the position of Holdings, when considering the exercise of a non-fiduciary discretionary power under a pension scheme, or in that of UKL acting as employer with a general discretionary power (express or implicit) under the contract of employment. We agree with what Newey J said about this in paragraph 146 of Prudential (see paragraph [33] above).

If, contrary to our view, Reasonable Expectations were to have a special legal status in relation to the making of such a decision, we agree with IBM that it would be necessary for them to meet standards of clarity and certainty, so that it should be clear enough to all parties concerned (including, in the case of a pension fund, the trustee) whether the particular legal rights and responsibilities that could arise have or have not in fact arisen. We think there is much to be said for IBM’s proposition that the standards ought to be equivalent to what is required for a contract or an estoppel. We note the judge’s comparison at B533 between a case where there is a binding contract to keep a scheme open for DB accrual until a given date and a case where a Reasonable Expectation to the same effect has arisen from actions of the employer or Principal Employer. There is a clear legal difference between those cases which has nothing to do with the Imperial duty, since in the first case to act inconsistently would be a straightforward breach of contract and in the second it would not. But it seems to us that the basis of this comparison, if it is of any legal relevance, should be that the conduct giving rise to the Reasonable Expectation must meet similar tests to that which would give rise to a contractual obligation, and the content of the Reasonable Expectation must also be comparable to that which would be required for a valid contract.

We also note that, in B1526(iii), the judge’s language is redolent of estoppel. Communications giving rise to the Reasonable Expectations, he says, were not merely expressions of intent, but they were to be the basis for members to take decisions as part of the implementation of Project Ocean and Project Soto of

Later, at B1391 the judge postulated “the clearest possible commitment (not amounting to a contractual promise or giving rise to an estoppel)” giving rise to a Reasonable Expectation that it would keep the plans open for DB accrual and not change the early retirement policy for a given period. That is stated as a hypothetical case since not even the RBs contended that what had been said and done amounted to the clearest possible commitment to the given effect. If Holdings had given a commitment which met the standard of clarity and certainty of content which the judge envisaged there, then either it would satisfy the tests for contract or estoppel (if acted on) or it would at least be a highly material factor for the decision-maker to take into account. It might be such that a decision not to fulfil the commitment would not satisfy the applicable rationality test (see paragraph [46] above), in which case there might be a breach of the Imperial duty. That is not this case, and it is interesting to note that the judge recognised that IBM was careful to make it clear that it was not giving a commitment to, or guarantee of, any particular course of action. For example, the word “commitment” was used in the Webcast, for example, but only in two rather general passages: Holdings’ “continuing commitment to the Plans and their members” and its “commitment to underpin the sustainability of” the DB Plans in the UK. We have already quoted the last paragraph of Mr Lamb’s letter to members in January 2006 in which he said in terms that IBM was not giving a commitment that there would be no further changes to pension benefits: paragraph [104] above.

At B1047 the judge referred to statements which showed that Holdings was not undertaking any commitment about the future, and he made the same point at B1049 and B1050. However, he contrasted these with the proposition that “this was not the message which IBM wanted members to receive”: B1048. This, and his review of the communications generally, led him to make the finding at B1052 that there was a Reasonable Expectation as to future benefit accrual, subject to changes in financial and economic conditions, which he reiterated at B1510(i), quoted at paragraph [64] above, and clarified in the Remedies Judgment as explained at paragraph [65] above. It seems to us that, quite apart from the issue of duration which we have already discussed, this is

Since the question is hypothetical on our view of the correct legal approach we say no more about this issue than this. If a different legal approach were to be thought appropriate and applicable, so that something amounting to a Reasonable Expectation (whether along the lines defined by the judge or not) would have substantive legal effect on a decision-maker’s reasoning process, rather than being a relevant factor proper to be taken into account along with all others, there would need to be minimum standards of clarity and certainty, as regards both the nature of the representation, commitment or assurance and its content.

If the judge was right to hold that there was a Reasonable Expectation as to the continuance of the old early retirement policy until 2014, then it necessarily follows that Holdings should not have given members a limited time in the autumn of 2009 to take advantage of the old policy before it was changed the next year. However, he also considered the Early Retirement Window in the alternative context of there being no relevant Reasonable Expectation, at B1526(xii) after the first sentence: see paragraph [214] above. In this passage the judge seems to hold that to change the early retirement policy, by withdrawing the previous practice of allowing a favourable discount on early retirement, and to do so with effect before 2014, was “not consonant with” the Imperial duty, and that the Early Retirement Window allowed was unreasonably short, and would have been a breach of duty even if there had been no relevant Reasonable Expectations.

However, at B1535(v) he said that, viewing the Early Retirement Window in isolation, that is to say absent any Reasonable Expectation and absent any suggestion in the communications that the early retirement policy might not change [the judge omits that last negative but we think it is required by the sense], then the members’ complaint could only be that they were not given enough time to consider their position. He held that, in that case, though the time was short and might be considered unfairly short, it would not satisfy the perversity or irrationality test. Thus, in that case he would have held that the Early Retirement Window did not involve any breach of duty, despite its brevity.

The judge came back to this subject in his second supplemental judgment at paragraphs 20 and 21, considering whether notice was required of the change of policy. If no notice was required, then it could not be necessary to give members the opportunity to take advantage of the old policy before any change. At paragraph 21 he said, first, that there was no implied restriction on the exercise of the discretion to determine an early retirement policy to the effect that it can only be exercised upon giving some reasonable period of notice. Accordingly, a change without notice could only be challenged as a breach of the Imperial duty. As to that he said that “the Imperial duty does

We consider that his later observations on this point were correct, and his earlier statement that the shortness of the Early Retirement Window constituted a separate breach of the Imperial duty was not correct. Mr Tennet sought to support that statement on the basis that it was based on communications in connection with Project Ocean and Project Soto even if they fell short of giving rise to a Reasonable Expectation. We do not see that this can be correct, given the terms in which the judge dealt with the point in B1535. Moreover, Mr Tennet was not able to point to any communication relating to Project Ocean or Project Soto which could have been relevant for this more limited purpose, nor did the judge identify any. Therefore, the answer to Issue 13 is that the judge was wrong to hold at B1526(xii) that there was a separate breach of the Imperial duty as regards the duration of the Early Retirement Window.

It is convenient to move now to this relatively minor issue, which arises from paragraph 22 of the judge’s second supplemental judgment. He said this:

We were told that this point had not been argued before him at the hearing which led to this judgment. On the judge’s findings, Holdings was able to adopt a new early retirement policy from 1 April 2014, but not earlier. This supplemental judgment having been handed down on 19 May 2015, Holdings did give a general notice of its new early retirement policy on 29 May 2015. Therefore, while the issue depends on the judge’s conclusion as to the Reasonable Expectation standing (since otherwise the notice it gave in 2009 of the change would have been effective), it only affects anyone who took early retirement between 1 April 2014 and 29 May 2015. It is not in dispute that there will have been some such people.

The judge did not base his proposition on anything in the Trust Deed or Rules of the Plan. The basis for his conclusion seems to be the statement in the third sentence of paragraph 22, namely that someone might have taken early retirement without realising that the new policy was to be applied. Mr Simmonds submitted that there was no evidential basis for this statement. No party had advanced the contention and none had sought to adduce evidence on the point. Mr Stallworthy for the RBs defended the judge’s conclusion by saying that a member who knew of the policy (that is, the pre-Project Waltz policy) and had not been told that it had been changed, was entitled to proceed on the basis that it had not. He accepted that a member who was told, before he or she opted for early retirement, that the new policy would be applied after 31 March 2014 would not be entitled to complain of the absence of any more general notice.

Asked whether it was likely that an employee would take the steps necessary for early retirement irrevocably without contacting the company first to find out what his or her pension entitlement would be, he contended that there might be such a person, who would take the decision on the basis of the annual benefit statements sent out which, unusually, gave figures for retirement at different ages up to NRD. However, in reply Mr Simmonds showed us a benefit statement dated in 2011, which had been before the judge. This does not give figures for retirement at ages other than on NRD, and it says in terms that a new early retirement policy has been adopted, and sets out the discount rates that will be applied under the new policy.

Accordingly, it seems to us that there was no evidential basis for the third sentence in the judge’s paragraph 22 quoted above, and therefore no factual basis on which the judge could conclude that Holdings could only properly (that is to say, consistently with its Imperial duty) adopt its new early retirement policy by giving notice to members generally that it was doing so. We therefore find for Holdings on Issue 37. The judge was wrong to decide that an announcement to members after 31 March 2014 was necessary for the change in the early retirement policy to come into effect.

The first point in the Respondent’s Notice (apart from the cross-appeal) seeks to support the judge’s decision on the additional ground, which he rejected, that DB members who chose to remain in the DB Plan at the stage of Project Soto, so as to continue DB accrual, had a Reasonable Expectation that UKL would not discriminate against them as regards salary increases, by comparison with employees who were members of a DC part of the Plan, and that UKL was in breach of its contractual duty in acting contrary to this Reasonable Expectation.

At the stage of Project Soto a member of the DB Plan had a choice, as we have described, between continuing in the DB Plan, with ongoing DB accrual for

That might be all very well, so the RBs contend, but if the employee’s ability to obtain the salary increase depends on agreeing that it should not be pensionable, by way of an NPA, if he or she is in a DB Plan, but not for a DC Plan member, that subverts the reassurance given as to the manager’s decision. The RBs accept that decisions by managers were indeed made in ignorance of the particular employee’s status as regards pension benefits, but that is not of any real value to the employee, it is said, if the actual award of the increase depends, in the case of a DB member, on agreeing to an NPA.

The judge considered and rejected this argument at B1517 and 1518. He recorded that nothing was said in the communications about Project Soto to the effect that members might be treated differently in respect of salary increases according to the choice made as between DB and DC benefits. Thus, the Reasonable Expectation which the RBs contend for in this respect has to be based on the absence of any warning that things might change in different ways for employees in different categories, in other words on silence rather than on any positive representation.

For IBM, however, Miss Rose pointed out that the concept of equality of treatment as between employees with different kinds of pension provision is inherently problematic, because of the intrinsic differences in the pension benefits. For a DC member, salary is not pensionable, so it is difficult to see how any meaningful comparison can be made between the treatment of a DB member’s pay as pensionable or not, or partly pensionable, and that of a DC member, whose pay is not pensionable at all.

This issue does not arise for decision, given our conclusion that the judge’s approach as regards Reasonable Expectations was erroneous. On that basis, the Reasonable Expectation contended for in this respect (which is relied on only in respect of the NPAs and therefore only against UKL) would fail for the same reason. Even apart from that, by itself it could not justify a finding that UKL acted, in this respect, in a perverse and irrational way, so as to be in breach of the contractual duty.

For these reasons we reject this aspect of the Respondent’s Notice.

The RBs contend that Holdings and UKL were in breach of the relevant duties in one respect which the judge did not accept, namely by reason of the investment policy adopted by the Trustee for the investment of scheme assets. This is, or includes, what became known as the “Heads I win tails you lose” point, by which the RBs argued that Holdings got the Trustee to adopt an aggressively mismatched investment policy, in the hope of positive returns, but as soon as the policy failed to generate such returns Holdings turned round and imposed the consequences on the members by way of the Project Waltz benefit changes. So IBM took such benefit as there was from the policy but the members had to bear the consequences when, as was an inherently predictable risk, its downside materialised in a significant way.

The judge held that the investment strategy was not unusual and was itself legitimate in the circumstances. The RBs do not challenge that, nor, of course, do they question the proposition that, as regards DB service benefits, Holdings had to bear whatever shortfall there might be at the end of the day and in the meantime as shown by the regular actuarial valuations of the Plans.

The judge also referred to this issue at B1527, where he said that, because the investment strategy was “within the range of a permissible and proper spread of investments” it was “very difficult to suggest … that Holdings was in breach of duty in implementing Project Waltz simply because that investment policy had not produced the expected returns”. We agree. Moreover, we do not see how this aspect of Holdings’ conduct as regards Project Waltz could properly be regarded as irrational or perverse, so as to fail the rationality test.

At paragraphs [138] and [139] above we referred to the discovery, during the consultation process, of the possibility that an error had been made in the preparation of the Main Plan Trust Deed and Rules as a result of which certain employees who were already then members of the Main Plan were entitled as of right to retire early from the age of 60 on favourable terms, and to the rectification proceedings in which it was established that such an error had occurred and ought to be put right. We also referred to IBM’s initial refusal to accept this, and to its pursuing the consultation process for Project Waltz according to the original timetable notwithstanding this discovery.

Issue 20 arises from paragraph 2(c) of the Respondent’s Notice. The RBs argue that when he was deciding whether Holdings had acted in breach of the Imperial duty, the judge should have taken this conduct of Holdings into account.

At paragraph 1530 the judge said, of this point, “I do not need to rely on that point as supporting the conclusion that there has been a breach of duty and, in the light of IBM’s case on that, I do not do so”. Mr Tennet criticised this sentence for its reference to IBM’s case. It is not altogether clear to what the judge was referring by that phrase. However, if the judge were right on his principal conclusions about the effect of the Reasonable Expectations, it would be difficult to find serious fault with his declining to decide this point as an additional factor in the RBs’ favour. Conversely, if he was wrong on the main issues, as we have held, then it is impossible to see that this point on its own could justify a conclusion that Holdings was acting irrationally or perversely in these respects so as to give rise to a finding of breach of duty. Accordingly, we reject the RBs’ contention on this issue.

Issues 22 to 28 are about justification. It seems to us that they were drafted to reflect IBM’s submission as to the broad architecture of the judge’s reasoning on justification. IBM’s submission is that the judge’s reasoning went as follows:

Stage 1: he pierced the corporate veil between Holdings and CHQ;

Stage 2: he found that there was an alternative route to achieve relevant objectives even though they never formed part of the RBs’ case;

Stage 3: he relied on the fact that Reasonable Expectations were engendered before the business needs to depart from them arose;

Stage 4: he relied on the disingenuous nature of certain communications by CHQ.

We prefer to call Stage 1 “Treating the IBM group as one”.

These four stages are reflected in the issues under this head as follows:

We will deal with those issues below. However, we have first to address issues 22 and 25, which are on this analysis preliminary issues.

We would observe that, as will now be clear, the list of issues takes the question of justification separately from that of breach of duty. No doubt this is convenient for the purposes of analysis but, as we have said at paragraph

There is no doubt that IBM contended at trial that its actions were justified by its business case for Project Waltz. IBM’s case was that there was a business case for the following reasons. Project Waltz was driven by CHQ’s need, in the light of the 2008 crisis, to make NPPC savings in order to meet its commitment to investors to improve earnings per share from $6.06 in 2006 to $10-11 in 2010 as set out in the 2010 EPS Roadmap (B1330(i)). Further, CHQ’s need to make NPPC savings resulted in the imposition of PTI (profit before tax and interest) and NPPC targets at the local level including in the UK. Project Waltz was Holdings’ response to, among other things, the targets imposed by CHQ. Meeting the targets set by CHQ assisted Holdings to obtain internal investment by CHQ in the various IBM business units in the UK with a result that larger amounts were available for employee reward programmes. Holdings’ decision to implement Project Waltz in those circumstances was a rational decision that could not constitute a breach of the Imperial duty.

Moreover, the judge accepted that he should not second-guess IBM’s commercial decisions:

So, as the judge saw it, the critical question whether to adhere to the 2010 EPS Roadmap lay with CHQ. He added a qualification as follows:

IBM’s case on appeal is that the judge accepted the factual basis for IBM’s submission. In this regard, IBM relies on B448, where the judge held that the Imperial duty was owed by Holdings alone, B1142, B1355 and B1367-8 where the judge held that Holdings’ need to comply with targets set by CHQ to secure inward investment was a commercial reality and a matter of “central significance”. IBM also relies on Holdings’ management being acutely aware

The judge summarised IBM’s case as follows:

The global strand of IBM Corporation requiring costs savings in all areas (including pensions). This was required in order to meet its commitment to investors to improve EPS from $6.06 in 2006 to between $10 and $11 in 2010, contained in the 2010 EPS Roadmap and in the light of the global financial crisis of 2008 which was projected to have a dramatic effect on IBM’s ability to meet that commitment.

The local strand of Holdings’ need to address its current and future lack of competitiveness, demonstrated by its failure to meet its PTI targets and its relatively lowly position compared to other IMTs. Holdings’ pensions costs were increasing by reason of the fall in asset values following the 2008 financial crisis. The answer to these problems was UKI Transformation under which it was sought to transform the UK business to increase its profit margin (and, hence, its contribution to the earnings of the global business) from 23% to 34% by the end of 2012. The Project Waltz changes served to assist in meeting the objectives of this project. Therefore, although the global and local strands of the business reasons for Project Waltz were separate they were nevertheless complementary. (B1329-30)

The judge was clearly very uneasy about the global strand. For example, he observed:

The RBs raised numerous reasons for rejecting IBM’s business case, which it is sufficient to list as the judge did:

IBM’s desire to meet EPS targets is not a reasonable justification for breach.

IBM UK’s performance targets and PTI were no justification to implement Project Waltz.

Headwinds and IBM’s concerns as to the future: Project Waltz was premature.

The Proposal to make I Plan changes was even less justifiable.

The other reasons IBM have cited are also not reasonable justifications for Project Waltz.

Parity.

Preference for DC as a matter of principle/policy. (B1388)

The judge made lengthy findings on each of those issues. No party has challenged any of those intermediate findings. For instance, he found that everyone knew that IBM, which included Holdings, was not meeting its performance targets set by CHQ (B1475). His ultimate finding was that, but for the question of the Reasonable Expectation, there were good commercial reasons for proceeding as IBM did. Thus, he held:

The judge repeated this point in his conclusions, but we do not need to repeat the point (B1523).

The Reasonable Expectation, however, made all the difference. The judge continued:

Again, the judge held that the commitments to investors would make it necessary to investigate whether IBM could at the same time meet the Reasonable Expectations: See B1390-1392 quoted at paragraphs [195] and [197] above.

In this context, the RBs made a submission which is potentially relevant at Stage 2. The RBs submitted to the judge that it would have been possible for IBM to have excluded pension costs (other than current Service Cost) from the 2010 EPS Roadmap (B1395). Indeed, this is what IBM did from 2011. The judge added:

Mr Koppl did indeed say this in evidence recorded in B1497 to 1498, but he graphically made it clear that that was with the benefit of hindsight:

The judge pointed out that it was debatable whether Mr Koppl should have referred to Joseph Swan rather than Thomas Edison. He then went on to record Mr Koppl’s evidence that leaving out pension costs might have led to a loss of investor goodwill. Overall the judge concluded that this particular route was not relevant to his inquiry, a point he reiterated in his conclusions (B1526(vii)(b) set out below at paragraph [318]).

Even so, in his conclusions, the judge went on to hold that he considered that an alternative route could have been found: see B1526 (iv), set out below at paragraph [318].

The judge set out his conclusions in B1506 onwards. As already explained at paragraph [202] above, he formulated the duty as follows:

He did not place any reliance on any failure to take members’ interests into account and indeed that had not been argued: B1531.

The judge found the breaches of duty at B1525, and set out his principal reasons in the individual subparagraphs of B1526, of which the following relate to justification (the italics are our own):

As to the global strand, the most important element of that strand was the requirement for operating units around the world to deliver NPPC improvements to meet the pensions-related requirements of the 2010 EPS Roadmap in relation to which it will be remembered that the evidence is that it was a requirement of CHQ that each element of the Roadmap had to be met. The target set for the UK DB Plans was, in my judgement, the principal driver for the Project Waltz changes in the UK.

The 2010 EPS Roadmap was itself a reflection of

The first reminder is that the commitment to investors was given after the relevant communications to members had been made and the Reasonable Expectations engendered. To prefer investors is not, of course, of itself a breach of duty. That said, it does seem to me, however, that this is a factor which can be taken account of in deciding whether Holdings’ actions were such that no reasonable employer could take [relevant to Issue 27].

The second reminder is that since 2011, retirement-related costs no longer feature as a performance indicator. Again, as Mr

Had CHQ acknowledged that the members had the Reasonable Expectations which I have held they were entitled to hold and had CHQ recognised the need for Holdings to act, so far as possible, consistently with those Reasonable Expectations (and its own need, therefore, to enable Holdings to do so), CHQ should have given consideration to developing proposals which would meet the twin objectives of (i) meeting the EPS target in the 2010 EPS Roadmap and (ii) allowing effect to be given to the members’ Reasonable Expectations [relevant to Issues 22 and 25].

The evidence demonstrates that this was not done. Instead, CHQ determined that delivery of NPPC savings and increases to PTI were to be effected in certain ways. For the UK, that included pension savings of such a scale and having such attributes that the target set for the UK would be met. I am wholly unconvinced that proposals could not have been developed which would have met those twin objectives; indeed, I am satisfied on a balance of probabilities that they could have been. In the first place, the necessary savings to meet the EPS target (in contrast with the savings in NPPC cost) did not need to come from DB Pension plans, let alone from the UK DB Plans. If pension cost was an essential element, DC Plans could have been targeted: as the evidence (particularly from Mr Koppl) shows, that was an option and would have provided an easier solution. Another option might have been to impose heavier burdens on other DB Plans in countries where members had no similar Reasonable Expectations or no remedies arising out of them [relevant to Issues 22 and 25].

Mr Simmonds would say that this sort of decision was one for management to make and that the court should not second-guess management. But that would be to miss the point. I am certainly not second-guessing management. If they had a free hand, they may well have been entitled to act as they did; even Mr Tennet accepts that, absent Reasonable Expectations, the Project Waltz changes would not have given rise to a breach of duty even though many people would regard them as unfair. The point is that they did not have a free hand; so that if they went beyond such freedom as they did have, the court can say that there has been a breach of duty. I am guilty of some slightly truncated language there: IBM Corporation and CHQ did, of course, have a free hand in the sense that this court cannot control their actions; but they did not have a free hand in the sense (see paragraphs 1007ffabove) that their actions and decisions can have no impact on Holdings’ own position and whether it has acted in breach of its duties [relevant to Issues 22 and 25].

Issues 22 and 25 are awkwardly drafted. They place at the forefront of any discussion on justification an underlying and fundamental disagreement between the parties about what the judge actually decided or was entitled to conclude on this point. By their terms, they challenge the correctness of the judge’s rejection of IBM’s case on justification (a ground of IBM’s appeal), and, if that challenge succeeds, the question (raised by the RBs) whether the grounds for justification advanced by IBM at trial were in any event capable of being sufficient.

Justification arises because, as we have already held (paragraph [46] above), the test for breach of the Imperial duty in this case is whether IBM UK acted irrationally in adopting Project Waltz. The parties have to some extent used the opportunity afforded by these issues to repeat their earlier submissions on burden of proof. On that, we have also already held in paragraph [57] above that IBM had to discharge the onus of showing what its reasons were, which it did, but that the burden of showing that it did not act rationally remained on the RBs. We do not repeat those points here. Submissions

IBM contends that it was common ground that IBM’s business case for Project Waltz was to achieve a reduction in NPPC and to meet the earnings per share targets. This would ensure that IBM continued to invest in its UK business. This was broadly accepted by Mr Wilson, Chief Financial Officer of UKL in the early stages of Project Waltz: see B1070 to B1072.

The RBs’ response in part on justification is that the judge adopted a “melting pot” approach and issues about business case were within that pot and so a challenge to his findings is an impermissible attack on the judge’s findings of fact. We have dealt with the melting pot point in paragraph [180] above.

The RBs’ primary case is that the judge did not need to deal specifically with IBM UK’s business case as the lack of justification was just a melting pot factor. The judge in effect was satisfied that, even if it was a reasonable and proper course of action from IBM’s point of view, it was destructive of the trust and confidence of UK employees, particularly in the context of all the other factors that went into the ‘melting pot’.

The RBs go on to argue, if justification was relevant, that the objective which IBM was trying to achieve by Project Waltz was that of meeting profit targets, and that IBM’s attempts to meet them were part and parcel of the attempt to hit IBM global EPS targets (see, for example, B1384-6). The RBs further contend that IBM had failed to show that a desire to meet increased profit targets and avoid disinvestment was in fact the motivation of those whom it alleges took the decision on behalf of UKL to implement Project Waltz. Thus, the judge held that any alleged disinvestment risk “does not detract… from the fact that the main driver for change was the need to reduce NPPC” (B1369).

It seems to us that the judge’s reasoning in deciding the question whether an employer had rationally departed from the Reasonable Expectations may have been that this was to be determined simply by reference to the employer as an employer, disregarding factors relevant to other aspects of the employer’s undertaking. For brevity, we will call this an ‘as employer’ approach. On this basis, considerations arising from adverse trading conditions for (say) the employer’s business as a manufacturer of widgets are to be left out of account when determining rationality under the ‘as employer’ approach.

This could make sense of the judge’s reasoning. Thus, for example, at B1354, the judge held that, leaving the Reasonable Expectations to one side, the response of the IBM group operating companies around the world to the global financial crisis in

A few further examples from the judge’s judgment will help to elucidate the judge’s approach. At B1100, the judge held that although the 2008 crisis was projected to have a dramatic effect on pension costs and thus on IBM’s ability to meet its EPS targets, and that IBM had to take steps as a result, that did not mean that those steps were justified.

Moreover, at B1391, the judge considered that the Imperial duty would have been satisfied if Holdings had departed from the Reasonable Expectation only so far as it was necessary for it to do so. The ‘as employer’ approach explains why, on the positive side, the judge felt he had to consider whether there was any alternative course Holdings could have taken – for otherwise the ‘as employer’ approach would have been elevated into an absolute obligation - and, on the negative side, why the judge did not consider whether, in response to the financial conditions in which the group found itself, IBM could not rationally seek to effect Project Waltz, even though (at B1111) he took the view that the adverse effects of the global financial crisis were among the causes of Project Waltz. Indeed, IBM relied upon the unpredictable and unprecedented characteristics of the crisis as one of the causes (B1339, B1340). The judge accepted that IBM could rely on the crisis for this purpose (B1348) Nonetheless, the question, as the judge saw it, was:

It is true that, in referring to the possibility that the Reasonable Expectations would not subsist in the event of significant financial and economic change, the judge did contemplate the possibility that other factors affecting the employer’s business could be relevant. In particular, at B1510 (i) he referred to significant changes in circumstances including trading and competitiveness. That suggests that the judge was not necessarily limiting his focus to the position of the company (here UKL) as employer. Moreover, in relation to the Reasonable Expectation of which he there spoke, the relevant entity was Holdings, not UKL. He dealt with Holdings and the Imperial duty in the same terms as for breach of the contractual duty, although Holdings was not an employer at all, so it would make no sense to limit the consideration to its position ‘as employer’. It is, however, notable that, as regards UKL which was the employer, the judge (as we have already discussed) did not address substantively the issue of whether a relevant significant change had occurred.

The RBs’ case on these issues is that, if justification is relevant, the evidence such as it was shows that UKL acted so as to promote the interest of CHQ, not those of itself. As the judge held, it was the 2010 EPS Roadmap which was the principal driver for Project Waltz. There is no real evidence that there was any risk of disinvestment. IBM did not give evidence that it had seriously considered the possibility. Any decision by UKL to approve Project Waltz lacked any reasonable cause. That is why Mr Wilson had challenged Project Waltz as not justified by any change in UKL’s business. The economic pressure placed on UKL by CHQ would not restore the breach of trust in confidence.

Although the RBs make several objections to UKL’s action to approve Project Waltz, approve it, it did. Moreover, the judge held that it was “not a rubber stamp” (B1230). He held:

The judge can only have meant that it was an independent decision of its board. Accordingly, the RBs’ objections to UKL’s business case on that ground must fail.

The RBs submit that the judge considered IBM’s business case in sub-paragraphs (viii) and (ix) of B1526 (set out in paragraph [318] above). The judge there held that the evidence showed that CHQ had not considered the formulation of proposals which both would meet the 2010 EPS Roadmap and allow effect to be given to the Reasonable Expectations, although in the judgment of the judge, such proposals could have been developed.

The RBs submit that these conclusions were open to the judge. In our judgment, it is clear that the judge asked the wrong question and applied the wrong test.

If the judge is to be taken as having decided the case by looking at UKL merely as employer, then this would have been erroneous in law. It is clear from Imperial Group Pension Trust Ltd v Imperial Tobacco Ltd: [1991] 1 WLR 589 at 599B, that the employer must take into account all the circumstances existing at the date of its decision. It follows that all the circumstances of the case must be taken into account in determining rationality. So, to revert to the example we have already given, it would be relevant to take into account adverse trading conditions to which the employer was subject.

If the judge did not consider the position of UKL merely as employer, but did allow for the relevance of, for example, adverse trading conditions, or other factors impinging on the company’s economic position, nevertheless it seems to us clear that his conclusion on rationality cannot stand. As we have said already, although he held that all relevant factors should be taken into account (his “melting pot” approach – see paragraph [180] above), he did not treat all such factors equally, but held that Reasonable Expectations must prevail unless there was no other way for the company to achieve its business aims. That is also a legally incorrect approach.

We must next consider whether this Court can make a finding applying what it holds is the appropriate test. We do not consider that we could do so and reach a conclusion favourable to the RBs. This is because the judge has already decided that but for the Reasonable Expectation the decision to promote Project Waltz would have been a justifiable decision.

Moreover, it is not open to the Court to retake a commercial decision previously adopted by a commercial entity. Nor should the Court assess the legitimacy of IBM UK’s actions with the wisdom of hindsight. It is therefore not relevant that, as it turned out, a smaller contribution was needed from pension plans than Project Waltz envisaged and that the contribution actually made by the DB pension plan was only about 60% of what was planned (see B1408).

Accordingly, IBM’s appeal on justification for this reason succeeds.

These issues are connected and we therefore take them together. The essential complaint is that the judge treated CHQ and Holdings as one, which meant that IBM UK was affected by the motivations and actions of CHQ even though they were not those of IBM UK taken alone.

The judge held that it was not possible for Holdings to

As that citation shows, the judge’s reasoning turns on the fact that the targets had been imposed by CHQ. The judge went on effectively to treat CHQ as subject to the Imperial duty imposed on Holdings. Thus, as explained at paragraph [205] above, as the judge saw it, the appropriate question in those circumstances was:

The judge found that CHQ was the “driver” for Project Ocean and Project Soto and for Project Waltz. Holdings was part of IBM’s corporate structure, with IBM Corporation (and within it CHQ) at the top, whose governance resulted in the team at CHQ having significant control over certain aspects of Holdings’ activities including control of the direction of the pensions projects (B1000).

The judge held that CHQ had to take account of the Reasonable Expectations just as Holdings was obliged to do so (B1011). He further held that CHQ did not stand in relation to Holdings in the same position as an ordinary shareholder and that employees regarded IBM as a “single enterprise” (B1014).

The judge also considered whether the knowledge of Mr Koppl could be attributed to CHQ. Mr Koppl, IBM’s director of pensions analytics, considered that the DB plan was “doomed” and that the restructurings involved in Project Ocean and Project Soto would not produce a permanent solution. The judge found that Mr Koppl not aware of everything being said by UK management to members of its DB plans, but that CHQ as a whole knew what was happening in the UK in relation to pensions. The judge, having considered Meridian Global Funds Management Asia Ltd v Securities Commission [1995] 2 AC 500, held that, since Mr Koppl had passed the material information on to senior management, CHQ could be taken to know this information.

Holdings was not always aware of the information which Mr Koppl was giving to his colleagues at CHQ. Indeed, CHQ had decided that certain information should not to be given to Holdings (B971).

IBM contends that the judge effectively treated CHQ as if it, rather than Holdings, owed the Imperial duty to the members. He therefore purported to “pierce the corporate veil” in circumstances not permitted by law. The separate legal personalities of the UK companies must be respected.

The judge found in previous proceedings involving the C Plan (described in paragraphs [138] to [139] above), that CHQ operated a “powers reserved” policy (see IBM United Kingdom Pensions Trust Ltd v IBM United Kingdom Holdings Ltd[2012] PLR 469 and paragraph [74] above). This meant that IBM UK had to seek CHQ’s approval to any changes in the pension arrangements for UK employees. But the judge had also found in that earlier judgment, at paragraph 100, that this policy

Moreover, as IBM emphasises and as we have explained at paragraphs [133] and [332] above, the judge found that Holdings’ decision to implement Project Waltz was not a “rubber stamp.” (B1230, 1523).

Nonetheless, as a commercial matter, Holdings did not have complete independence (B1355). So, in the judgment of the judge:

IBM contend that the judge avoided the conclusion that Holdings was not in breach of the Imperial duty by holding that the question whether Holdings was in breach of its Imperial duty should be assessed by asking whether Project Waltz was an appropriate response to the problems facing CHQ, taking into account the Reasonable Expectations of members. Moreover, as we have already explained, the judge held that Holdings could not “shelter” behind a business case based on the need to meet targets imposed by CHQ unless, in turn, a business case could be demonstrated justifying the imposition of those targets (B448, B1509 and B1524).

On IBM’s submission, the judge justified his legal analysis by reference to CHQ’s influence and control over Holdings’ actions and in particular its responsibility for the message given to the members in the course of projects Ocean and Soto which resulted in their Reasonable Expectations. Thus, at B1005, the judge held:

the influence and control which CHQ had over the direction of pensions in the UK;

the manner in which that influence and control was in fact exercised; and

(although this may be part of ii)) the responsibility which CHQ must, in my view, take for the message which was given to members in both the Ocean and Soto communications. Just as IBM worldwide is a global business with its local arms answerable to Armonk, so IBM Corporation/CHQ must take responsibility for the representations made to members with its knowledge and approval, especially bearing in mind that those representations were made with a view to obtaining the assent of members to proposals, in each case, driven by CHQ’s own agenda.

The judge’s view was that IBM’s position was unattractive (B1013). His view was that CHQ was not simply an ultimate parent of Holdings because of the influence and control which it exercised (B1014), and that the IBM culture was that IBM worldwide was a “single enterprise” to which all employees owed their loyalty (B1014). He held:

IBM submits that the judge was wrong in his treatment of the CHQ/Holdings issue. The judge cited no authority for concluding that CHQ had to take responsibility in these circumstances. It amounted in effect to piercing the corporate veil and treating CHQ as if it too owed the Imperial duty to the members. IBM contends that it was not open to the judge to pierce the corporate veil in this case. Piercing the corporate veil is only justified in very limited exceptional circumstances, not present here. The judge did not refer to any of the authorities, in particular the decision of the Supreme Court in Prest v Petrodel Resources Ltd[2013] 2 AC 415, still less explain how they supported the judge’s conclusion. The correct principle is that set out by Lord Sumption:

Moreover, submit IBM, CHQ’s influence and control over Holdings did not justify piercing the corporate veil. There is no finding that the information given to members was given jointly or by Holdings as agent for CHQ. CHQ was not a party to the proceedings. IBM contends that IBM’s “culture” could likewise not justify ignoring legal distinctions. There is no evidential basis for the judge’s statement at B1014, quoted in paragraph [345] above.

UKL had its own corporate reason for implementing Project Waltz (to avoid the risk of lack of investment by CHQ) and it was irrelevant that members viewed IBM as a single corporate group. IBM points out that it was common ground that the objective rationale for Project Waltz was the protection of investment in the UK business.

The RBs disagree with IBM’s criticisms of the judge’s approach. They contend that the judge did not treat CHQ as if it, rather than Holdings or UKL, owed an Imperial duty or the contractual duty to members and that the judge did not pierce the corporate veil. The judge merely addressed the question whether Project Waltz was an appropriate response to the problems which faced both Holdings and IBM Corporation as a whole: see B1000 and B1009. Moreover, the RBs contend that Holdings adopted CHQ’s motivations and thinking and did not make its own independent decision to proceed.

As to whether the judge was entitled to disregard the separate legal personality of Holdings, the RBs contend that the judge was entitled:

to treat the knowledge of persons at CHQ as relevant to whether Holdings was in breach, given CHQ’s central role in Holdings’ decision to implement Project Waltz; and

to have regard to CHQ’s motivations because the members of the DB plans were encouraged by IBM to view, and did view, IBM as a single corporate group, and the question of whether there had been a breach of duty should be considered objectively with their point of view in mind.

The RBs contend that the judge’s approach was consistent with existing case law. In Attrill (discussed in paragraph [51] above), the employees brought claims against their employer for denying them discretionary bonuses. It was the parent company which had announced a minimum bonus pool and had later sought to introduce a term that in the event of material adverse change the bonuses might be reduced. It did so because of pressure from yet another company. That trust and confidence was broken by the actions of the parent company.

The RBs contend that CHQ’s responsibility for Project Waltz went far beyond that of a shareholder in setting performance targets for IBM UK. The judge found as a fact that the sole impetus for Project Soto came from CHQ and the performance of the UK business was not a driver for the changes. Certain individuals at CHQ (Messrs Koppl, MacDonald and Michaud) took the lead in developing the form of the Project Waltz proposals and then put pressure on IBM UK to adopt them: see B1138 and B1179. In the end, CHQ caused Mr Wilson to be removed from the Project Waltz implementation team when he refused to entertain it. The judge held that CHQ required (not requested) pension changes by IBM UK to meet its financial demands (see B1374). The judge’s finding that Holdings’ board did not act as a “rubber stamp” must be read in this context.

In the circumstances, the RBs contend that the knowledge of CHQ should still be attributed to Holdings (cf Hawksford Trustees Jersey Limited v Stella Global UK Limited[2012] 2 All ER 748).

We consider that the judge treated CHQ and Holdings as one when it came to finding a breach of the Imperial duty: see the passage cited from B448 set out in paragraph [342] above (the judge made the same point in B1509) and see B1000 to B1014. We are not persuaded that IBM UK simply adopted the motivations and thinking of CHQ in the light of the judge’s finding that Holdings was not a “rubber stamp” (see above, paragraphs [133] and [332]). There is no appeal against these important findings.

It is a fundamental principle of English law that a company is a separate legal entity, separate from its shareholders, and that its shareholders cannot be made liable for the company’s debts: see Salomon v Salomon [1897] AC 22. There are some exceptions to this principle, when, to use the familiar metaphor, the court may “pierce the corporate veil”, which the Supreme Court reconfigured in Prest. The Supreme Court regarded the old language of “façade", "sham" and "relevant impropriety" as unhelpful. It emphasised that the exceptions to Salomon v Salomon were limited and do not apply when some other remedy, for example, agency exists.

Lord Sumption enunciated a “concealment” principle in paragraph 29 of his judgment, where the corporate entity is used as a cloak for other persons, which does not involve piercing the corporate veil, and an evasion principle, which is set out in paragraph 35 of his judgment (paragraph [355] above). On this approach, the court may pierce the corporate veil. It applies when a person is under an existing legal obligation or liability or subject to an existing legal restriction which he deliberately evades or whose enforcement he deliberately frustrates by interposing a company under his control.The court may then pierce the corporate veil solely for the purpose

But it was not the unanimous view of the Supreme Court that the evasion principle reflected the only circumstances in which the court could pierce the corporate veil. Lord Neuberger endorsed the evasion principle but thought that the underlying policy ("fraud unravels everything") had a wider operation and that it could apply to a person who transfers assets to a spouse or civil partner, as opposed to a company (judgment, paragraphs 81 to 83). Baroness Hale thought that the cases might be examples of the principle that people who operate limited companies should not be able to take unconscionable advantage of the people with whom they do business (judgment, paragraph 92). Lord Mance and Lord Clarke endorsed the evasion principle but declined to limit the categories in which the principle might operate (judgments, paragraphs 100 and 103). Lord Walker thought that "piercing the corporate veil" was nothing more than a label, and doubted whether it operated independently of other remedies.

The common theme in Prest is that the court does not pierce the corporate veil unless there has been abuse of the separate personality of the company in question. What is also clear by implication is that there was no support for an “enterprise” theory, under which liability is imposed on the whole group of companies of which a company is a member. The judge placed emphasis on the culture at IBM of a “single enterprise” (B1014). But it follows from Prest that this could not strengthen the case for lifting the corporate veil.

The circumstances of the present case would not therefore, on any view of the English law doctrine of piercing the corporate veil, justify imposing the Imperial duty or the contractual duty on CHQ. CHQ never owed any Imperial duty to the members of the DB plans and was therefore not evading any legal obligation. Moreover, Holdings was not interposed between CHQ and the members. The members were at all times employed by UKL, not CHQ.

In addition, the unattractiveness or otherwise of CHQ’s position is irrelevant: separate legal personality cannot be ignored merely because the court may consider that justice so requires: see the observations of this Court in Adams v Cape Industries PLC [1990] Ch 433 at 536. This approach can be contrasted with that adopted by other jurisdictions with companies legislation similar to our own, which sometimes adopt more flexible approaches: see, for example, Arden, Corporate veil – old metaphor, modern practice? [2017] JCCL&P, pub pending).

The judge also attributed to IBM UK the motivations of CHQ. The fact that the knowledge of Mr Koppl was attributable to the senior management of CHQ, to whom he reported and conveyed his views, does not mean that that knowledge is to be attributed to Holdings, a separate legal entity, simply because it was a member of the IBM group whose parent exercised considerable control and direction over its subsidiary. In the light of the judge’s clear finding that Holdings was not a “rubber stamp” and thus, necessarily, took an independent decision to adopt Project Waltz, this approach cannot stand.

The RBs rely on Attrill, but that case is distinguishable because the employer acted under pressure from the parent and did not seek to justify its actions. Nor was any point taken in that case about distinct corporate entities being involved.

IBM UK did not confer any authority on CHQ to negotiate changes to its pension plans or to make decisions for it. In those circumstances, no principled basis has been suggested by the RBs for attributing the knowledge of CHQ to IBM UK.

Accordingly, we conclude that IBM is correct on this issue.

This issue raises a challenge to the judge’s holding that he was “wholly unconvinced” that proposals could not have been developed which would have met the EPS target in the 2010 EPS Roadmap and allowed effect to be given to the Reasonable

We have already explained that in our judgment the judge may have been led by an erroneous ‘as employer’ approach to consider whether there were alternative ways of delivering the aims of Project Waltz without disappointing the members’ Reasonable Expectations. While strictly therefore this Issue does not now arise, we will set out the principal submissions and our response to this Issue.

IBM argues that the judge was wrong to hold that IBM had breached the Imperial duty on a case that was never articulated by the parties bearing the onus of proof and which, accordingly, IBM did not address in evidence or in submissions. Although the RBs bore the burden of proof in respect of breach of the Imperial duty, they did not allege in their pleadings that there was any alternative route of achieving the aims of Project Waltz. IBM submits that the judge’s conclusion was not properly based on appropriate findings of fact. Therefore, the judge’s conclusion was speculation. IBM submits that the judge’s approach was surprising as he had previously directed himself that he should not speculate as to what might have happened if Mr MacDonald had resisted pressure from CHQ (B1156).

The judge postulated that there could be savings in non-pension areas. In fact, this was considered (B1110). However, it was not put to IBM’s witnesses that nonpension areas should have been subjected to even greater savings.

Moreover, submits IBM, the judge’s finding that an alternative route existed was inconsistent with his parenthetical observation at B1487 that:

The judge also postulated there could be savings from non-UK DB plans. In fact, savings were made in other five other countries. This case, again, was not put to IBM’s witnesses. There was no evidence about legal constraints operating in those countries.

The judge also considered that savings could be obtained from DC plans (in practice the US DC plans). In fact IBM did consider this but it was rejected on human resource grounds. It was not suggested to IBM’s witnesses that this was without foundation.

The RBs contend that the finding that there was an alternative route was open to the judge. It would be absurd if IBM could claim that it could make pension cuts whenever investment returns fell. That would render the plans vulnerable over the longer term and would be inconsistent with the Reasonable Expectations which the members were led to hold. The judge’s finding of an alternative course was in reality his acceptance of the RBs’ case on this point.

The RBs submit that in any event the judge’s comments on any ‘alternative route’ were a secondary (and not a necessary) part of his final conclusion. The alternative route point was simply a factor which went into the melting pot when deciding whether IBM had behaved in a way in which no reasonable employer would act.

The RBs reject the suggestion that at trial they accepted that DB pension fund cuts were inevitable.

In any event, submit the RBs, there was plenty of evidence that IBM UK could have achieved the aims of the 2010 EPS Roadmap without cutting pensions. For example, under Project Waltz, IBM in fact achieved a much greater saving in its pension cost than it needed to do.

Moreover, savings could have been made from DB schemes in other countries. Cuts in US DC pensions had been actively considered at the planning stage for Project Waltz. The judge was obviously unconvinced by the reasons for rejecting cuts in the US DC pensions.

There was some debate before us about a document which had showed pension costs in other countries, but which had been redacted to obliterate the information about countries other than the UK. Mr Simmonds relied on this to show that no-one regarded pension costs in other countries as relevant at the trial. Mr Tennet’s riposte was to show that there had been an order for the production of un-redacted copies of documents including this one, which showed, he submitted, that the information about pension costs in other countries was regarded as relevant. To that, in reply Mr Simmonds accepted that such an order had been made, but pointed out that, despite the order, it was the redacted document that had been included in the agreed trial bundle, which showed what the parties regarded as relevant at the time of preparation for trial. So, he argued, as matters proceeded before the judge there was no investigation of, and no intention to investigate, pension costs in other countries.

Further, IBM’s own pleaded case expressly recognised that “Had the 2009 Project Waltz Proposals not been implemented, IBM would have had to look elsewhere in the UK business for savings, potentially with a greater impact on employees” (ReAmended Response to AVPs, paragraph 5.1).

It was inherently improbable that, in an organisation the size of IBM, there was no possible alternative way of achieving the cost savings/earnings increase achieved by Project Waltz.

The finding that an alternative route for delivering the aims of Project Waltz could have been found was an integral part of the judge’s conclusions on breach. We do not consider that it was simply an additional “melting pot” factor.

We are satisfied that there was no pleaded case before the judge that Project Waltz could have been achieved by some route which was less onerous so far as the members were concerned. Certainly, there were suggestions recorded in the judgment to that effect and, as one would expect, steps were taken to investigate other possibilities. However, there was no pleaded case, nor was it put to IBM’s witnesses that this could have been done. In those circumstances, it was not open to the judge to make a finding that there was an alternative and less onerous means of achieving Project Waltz. The requirement that a material allegation be pleaded is not simply a technicality: it is a procedural guarantee that is there to ensure that the trial is fair. Accordingly, the judge ought not to have said what he did on this subject at paragraph 1526(viii) and (ix). IBM succeeds on this Issue.

At B1526 (vii)(a), the judge held:

IBM submits that the judge erred because the order in which an employer enters two or more commitments is irrelevant when applying a test of irrationality or perversity. Had the Reasonable Expectations been generated with knowledge of conflicting assurances to investors, it would no doubt have been said that they should still have been observed. The order in which the commitments were given was therefore irrelevant.

The RBs respond that the judge placed no special reliance on the order of events. He simply evaluated the unreasonableness of IBM’s conduct based on the actual facts before him, namely that the voluntary adoption of the EPS Targets in 2007 (and the commitments which IBM thereby regarded itself as having made) post-dated and were treated as taking precedence over the prior representations it had already made at the time of Project Soto and Project Ocean. IBM could also have been criticised if (counter-factually) it had given members assurances as to the future of the plans despite having already given potentially conflicting EPS ‘commitments’ to investors. That does not mean that the judge was wrong to criticise IBM on the basis that he did. The fact was that IBM had increased the volatility of the pension funds by committing to the 2010 EPS Roadmap.

In his conclusions at B 1526(xiii), the judge held:

This paragraph refers back to B1060, which reads:

In the final sentence of B1526(xiii) the words “it made” must mean “Holdings made”, but it did so on the basis of information provided by CHQ and without the knowledge that could have made the statements disingenuous on the part of Holdings. This conclusion cannot, therefore, stand in the light of our answer to Issues 23 and 24 on separate corporate personality, and in answering Issue 28 we disregard the erroneous attribution to Holdings of CHQ’s knowledge about the future prospects of the DB plans.

The judge’s conclusion at B1526(xiii) that the Reasonable Expectation had been engendered by disingenuous and misleading statements made by CHQ entails a severe criticism of IBM’s conduct at the time of Project Soto. It means that IBM provided information knowing it to be unfair or incomplete, even if it did not intend deliberately to mislead the plan members. In fact, the judge draws a distinction between disingenuous statements and deliberately misleading statements, which is hard to follow. Even, however, if there was a valid distinction to be drawn here, the criticism is still a grave one because to be disingenuous is to act in bad faith. Submissions

It is common ground that the RBs did not include in their statement of case any allegation that CHQ acted in bad faith. The type of bad faith necessary to constitute conduct disingenuous involves fraud, since it implies a deceptive practice for the purpose of gain. “Disingenuous” is not a word that is used lightly. It is well known that any allegation of fraud must be specifically pleaded. That requirement is now found in paragraph 8.2 of the Practice Direction supplementing Civil Procedure Rule 16. The furthest the pleading goes is to allege that, as a result of Project Ocean and Project Soto the DB members were led to believe a number of matters to the knowledge of CHQ (para 179 of the Re-Amended Particulars of the “good faith claim”, that is, of breach of the Imperial duty). This was insufficient to comply with the Civil Procedure Rules if an allegation of bad faith was being made in the context of engendering the Reasonable Expectations. It is also no answer that IBM knew that the RBs alleged that some IBM officers knew about certain matters not disclosed to IBM UK, for example, Mr Koppl’s opinion as to the sustainability of the proposals for restructuring the plans.

Consistently with the RBs’ pleaded case, IBM observed in its closing submissions at

The absence of an allegation of bad faith is also consistent with the RBs’ case at trial. They contended that IBM was “schizophrenic” in its attitude to Project Waltz. What

At the stage when IBM’s witnesses were giving evidence, IBM’s representatives could only ensure that their witnesses covered such matters as were alleged against them. In the absence of a pleaded allegation of bad faith, IBM had no reason to investigate whether other witnesses might need to give evidence. This is all the more important since such an allegation would have to include particulars of who it was in IBM that had the necessary knowledge such as to constitute a dishonest mind. That person or those persons would, if available, be likely to be crucial witnesses. The making of the finding without an appropriate pleading was, therefore, a serious procedural irregularity.

In the result the judge was wrong to hold that CHQ’s conduct had been disingenuous in relation to the Project Soto communications. It follows that CHQ’s conduct so described could not (even if it would otherwise do so) count against Holdings for the purpose of determining whether IBM UK had breached its Imperial duty.

Issues 29 to 33 inclusive concern points arising from the judge’s findings of breaches of contractual duty by UKL in procuring NPAs. Most of them proceed on the assumption that the judge had been right to hold as he did that the Reasonable Expectations were established and were disappointed by, among other things, the NPAs. It was accepted by the RBs that, had it not been for the Reasonable Expectations, Project Waltz would not have given rise to any breach of duty on the part of Holdings or UKL. On that concession by the RBs none of the 2009 NPAs (and we interpolate none of the subsequent NPAs) would have given rise to any breach of the contractual duty, whether viewed as part of the overall changes or viewed in isolation: see B1519.

However, Mr Stallworthy identified Issues 32 and 33 as being allegations of breach of the contractual duty independently of the existence of the Reasonable Expectations found by the judge. We will therefore address these now, starting with Issue 33.

Mr Stallworthy submitted that IBM’s case on this aspect of their appeal had to be that it could not be a breach of the contractual duty to withhold a pay rise to which an employee is not entitled under the contract. That must be the proposition, he said, because if it were not a clear rule of law, then it must be a question of the evaluation of the facts, and that is a matter for the judge, not to be interfered with, he submitted, unless it is plainly wrong. We will come back later to the question whether “plainly wrong” is the right test in such a situation. However, he went on to note that such an absolute rule of law could not stand with cases such as Transco (see paragraph [43] above). If the answer depends on the factual context, then IBM could only succeed if there were no arguable basis, on the facts found, for concluding that there could have been a breach of the duty.

Mr Stallworthy identified five elements in the judge’s reasoning as regards the NPAs having been procured in breach of the contractual duty of UKL: (i) the NPAs were part and parcel of the third cumulative set of restrictive changes to pension benefits made within four and a half years; (ii) they (or at least some of them) imposed total non-pensionability for future salary increases rather than partial, as under Project Soto; (iii) they denied DB members any salary increase at all if they did not submit to

Of these, the last is covered by Issues 29 to 31 and falls away on our conclusion that there was no such Reasonable Expectation. The fourth, supported by the second and third in part, is relevant to Issue 32. The first, again with some support from the second and third, is the basis for Issue 33, regarding the NPAs as an integral part of Project Waltz – as indeed they were.

In presenting the appeal on Issue 33, IBM treated the matter as being at one with their overall case on Project Waltz as a whole. Miss Rose, who argued this part of the case for IBM, did not present separate argument when opening this issue. In reply, however, she disavowed any contention that not to grant a pay rise could never be a breach of contract, and accepted that Transco and similar cases established that it could be in some circumstances. However, she submitted that these cases show that the test for whether it is in breach of duty is that of arbitrariness or capriciousness. Transco and all the other such cases cited to us had applied such a test, and only if it was truly perverse, arbitrary or capricious not to award a pay rise (almost invariably because such a pay rise was being given to other comparable employees) would the failure or refusal to make the offer constitute a breach of the contractual duty.

The judge had not applied such a test in the present case, she said, and if he had, there was, according to her, no factual basis on which he could have come to the conclusion that the test was satisfied.

In our judgment, Miss Rose’s submissions on this are correct. Failure or refusal to offer a pay rise to which the employee is not contractually entitled may in some circumstances be a breach of the implied duty of trust and confidence, as in Transco, but the circumstances have to be extreme. The test to be applied is the rationality test which we have already identified (see paragraph [45] above). There is no sign that the judge applied that test in addressing this question, and we do not consider that, if he had done, he could have found sufficient factual material to justify holding that the decision to impose NPAs as a condition of the award of a pay rise to DB members was itself a breach of the implied contractual duty.

Accordingly, we hold, on Issue 33, that the judge was wrong to find that for UKL to have procured the NPAs gave rise to a breach of the contractual duty on the basis that they formed part of Project Waltz.

Although the point does not arise on that view, we wish to touch briefly on the question whether the test on appeal, in a case such as Mr Stallworthy argued that this was, is whether the judge is “plainly wrong” or, simply, “wrong”. This was a point that was argued also by Mr Cavanagh QC for the RBs in the wider context of the appeal. His submission was that in testing the judge’s findings of breaches of duty, we had to be satisfied that he had been “plainly wrong”, before we could properly reverse those findings. What we say in the next few paragraphs, in our judgment, properly states the appellate court’s function in a case of the present character, although for the present we are concentrating on the point in the context of issue 33.

The submission for the RBs was that an appellate court should only interfere with a trial judge’s primary findings of fact, or with his conclusions based upon an evaluation of facts, if it is satisfied that the judge was plainly wrong, exceeding the generous ambit within which a reasonable disagreement over the evaluation of facts is possible. This was said to be justified by the well-known case of Assicurazioni Generali v Arab Insurance Group[2003] 1 WLR 577 at paragraphs 9-10, 12 and 1617. The RBs also relied upon the decision of the Supreme Court in Henderson v Foxworth[2014] 1 WLR 2600 in which it was held that, in the absence of some identifiable error, such as a material error of law or the making of a critical factual finding which had no basis in the evidence, an appellate court would not interfere with the factual findings of the trial judge unless it were satisfied that his decision was “plainly wrong” in the sense that it could not be reasonably explained or justified and so was one which no reasonable judge could have reached. We invited the parties’ attention, in addition to these cases, to Re B (A Child) (Care Proceedings: Threshold Criteria)[2013] 1 WLR 1911.

Reflecting upon the helpful submissions made to us on these authorities, this is not a case in which challenges are made to the judge’s findings of primary fact at all. It seems to us that, even if they were put at their highest from the point of view of the RBs, the relevant issues would be largely akin to the question considered by the Supreme Court in Re B as to whether the trial judge’s evaluation of whether the “threshold” requirements of the Children Act 1989 in that case had been met on the findings of primary fact made at the trial. Lord Wilson characterised the issue as an “evaluative” determination (at 44) and said,

In Re B, some members of the court said that it is not possible to lay down any single clear rule as to the proper approach to be taken by an appellate court where the appeal is against an evaluation: see e.g. per Lord Neuberger at 60 and Lord Kerr at 110. However, in a case such as the present, as it seems to us, a question such as whether the judge’s view that there had been breaches of the Imperial duty is to be upheld, should be determined by asking simply whether he was or was not “wrong” rather than whether he was or was not “plainly wrong”.

In our judgment, Henderson v Foxworth, much relied on for the RBs, is not inconsistent with that conclusion, because, when read with care, it can be seen that the appeal involved a challenge to the findings of primary fact made by Lord Glennie, the Lord Ordinary, as trial judge. The issue on appeal was indeed, therefore, whether his decision was “plainly wrong”. The significant issue in the case was what was the true consideration for the transaction that was being challenged as having been at an undervalue. That was not a matter of evaluation; it was a matter of finding the primary facts, to be decided on the evidence that the Lord Ordinary, as trial judge, had heard and which the Extra Division of the Court of Session, from whom the appeal was successfully brought, had not.

With regard to Issue 32, which simply adds the element of “threat” (or as the RBs have it “threats or coercive conduct”) in telling employees that there would be no salary increases except on NPA terms, this draws on observations by the judge at B1521 and B1535(ii), but we do not see that that makes any real difference. In our judgment, it was not a breach of the contractual duty for the employer to say that it did not intend to award pay increases in future except on a non-pensionable basis. Absent the wider breaches of duty alleged by the RBs, arising out of the Reasonable Expectations, we cannot see that the firm terms in which the pay increases were to be offered constituted any breach of the relevant duty.

In this respect it seems to us that the case is not substantially distinguishable from the second of Warren J’s judgments in Bradbury v BBC[2015] EWHC 1368 (Ch). In that case the judge rejected a contention that, when the BBC offered Mr Bradbury a pay increase on the basis that only as to 1% would it be pensionable, this was in breach of the contractual duty by reason that it involved applying improper coercion.

Looking at the issues about the NPAs overall, if Project Waltz is not in itself objectionable because there were no Reasonable Expectations of the kind found by the judge, quite apart from the judge having applied the wrong legal test, then it is difficult to see that the NPAs could have given rise to contractual breaches by UKL in implementing that part of the Project which fell to it. It seems to us in reality that the breaches for which the RBs contended depended upon the success, or otherwise, of the overall attack on the Project. Once that attack had failed, because the case on Reasonable Expectations failed, we do not consider that these separate attacks focussed on UKL’s contractual duty could succeed.

Issues 34 to 36 inclusive raise a number of points of contention between the parties as to the judge’s conclusions about the appropriate remedies for the breaches that he had found of the contractual duty by UKL in these respects. While we had originally envisaged addressing all the issues in the list summarised in Appendix One, including these, we do not consider now that it is either necessary or expedient to address all of them, having regard in particular to the necessary length of this judgment on the other issues and the resultant time and expense to which the parties will be put in considering what we have said on those other issues.

We have said all that we consider we need to say about the liability issues as regards NPAs. In the circumstances, therefore, we do not intend to say anything about the remedies issues, 34 to 36.

We have left these issues to last. Except to the extent that the RBs succeed on these issues, the result of these appeals is that the RBs have failed to establish liability on the part of IBM. Both issues are about the legal consequences of the process by which IBM UK consulted the members of the Plans in preparation for Project Waltz. Each Issue has a distinct but related aim for the RBs. By Issue 21 they seek to establish that the defects in the consultation, and IBM UK’s implementation of Project Waltz despite these defects, were breaches of both the Imperial duty and the contractual duty such as vitiate Project Waltz entirely and render it voidable for breach of duty. By Issue 38, as a fall back, they ask the court to prevent Project Waltz from being implemented until a further and proper consultation has taken place. Success on either would preclude IBM from relying on the Project Waltz changes, though in different ways.

The judge did not need to decide the first of these Issues, having held that Project Waltz was in breach of duty on other grounds. However, he did discuss it at B1538 onwards. The second of these Issues he left, in terms, to the expected appeal. The importance of both Issues arises because we disagree with the judge on his other findings of breach of duty.

The Occupational Pension Schemes (Consultation by Employers and Miscellaneous Amendments) Regulations 2006 (“the Consultation Regulations”) made pursuant to section 259 of the Pensions Act 2004 required UKL to consult the members of the DB plan before the termination of future DB accrual. Consultation on all matters, including those outside the Consultation Regulations, potentially engaged the employer’s contractual obligation. Whether, and if so to what extent, the Imperial duty is relevant to consultation is a less straightforward question, to which we will refer later.

IBM UK set up a pensions consultation committee (“PCC”) to consider the changes involved in Project Waltz. As we have mentioned at paragraph [135] and following above, its members included representatives of plan members and representatives of IBM UK. IBM UK was to consult the members through the PCC.Consultation with members ensued, as the judge described. Despite the consultation, Holdings and UKL decided to implement Project Waltz. As we have mentioned, the Trustee declined to proceed with its implementation without court proceedings to determine its validity, hence these proceedings. In the meantime Project Waltz has been implemented, but on a provisional, reversible basis.

The judge held that Holdings and UKL acted in breach of the Consultation Regulations and UKL also in breach of its contractual duty of trust and confidence in providing information about Project Waltz to DB plan members. IBM does not seek to challenge these findings on appeal In particular:

They had failed to make it clear to members that the driver behind Project Waltz was the 2010 EPS Roadmap (B1567).

They had failed to provide information requested by the PCC (B1572).

They had deliberately misstated the intention as to when it was intended to close the DB plan to future accrual. The correct date was April 2011 even though they told the Trustee that it was intended to close it in April 2010. IBM was “[playing] hardball” with the Trustee (B1552).

They did not consult with an open mind: B1563. On the contrary, their minds had effectively been made up before the process began.

The judge gave careful consideration to the nature of the duty to consult. Again, these matters have not been challenged on appeal and so we are not concerned with those important passages in the judge’s judgment.

IBM submitted to the judge that, even if it was in breach of the Consultation Regulations, the RBs’ only remedies were those set out in Regulation 18, which provides:

an improvement notice issued under section 13 of the Pensions Act 2004 (improvement notices), and

a penalty imposed under regulation 18A.

IBM also relied on section 259 of the Pensions Act 2004, the section under which the regulations were made. Section 259(3) is as follows:

The judge rejected this argument. He held that Regulation 18 did not preclude a claim based on breach of the contractual duty of good faith because the breach was not limited to a breach of the Regulations. It followed that the RBs were entitled to the ordinary contractual remedies for breach of contract; those remedies included an injunction (if appropriate) as well as damages (R688). The judge left the question whether an injunction should be granted to this Court (R699). The RBs now seek that injunction (Issue 38).

It will be noted that the judge found only that defective consultation involved a breach of the employer’s contractual duty. In paragraph 2(d) of their respondent’s notice, the

IBM submits that the RBs did not argue at trial that the defective consultation was a breach of the Imperial duty. It is too late for that matter to be raised now.

The RBs contend that an injunction ought to be granted becausethere is no appeal from the finding that IBM UK was in breach of duty in relation to the consultation process. For IBM UK to proceed to implement Project Waltz would itself be a breach by UKL and Holdings of their respective duties given the defects in consultation. Moreover, the adequacy of damages is not relevant to this issue, because IBM was under a negative obligation and, in any event, damages would not be an adequate remedy to members because it would be difficult for members to pursue proceedings against IBM. Project Waltz has only been implemented on a provisional basis and a fresh consultation is not impossible or impractical.

IBM submits that the decision to implement Project Waltz was a separate exercise from the prior process of consultation, and that the RBs did not contend at trial that implementation following consultation would be a breach of the Imperial duty. The RBs argue that this submission cannot be correct, unless IBM had no intention of listening to the views expressed during the consultation. Employers cannot be permitted to take such an approach to consultation. IBM has brought the present situation on itself. Damages would be a pointless remedy.

For the RBs to succeed on Issue 21 in showing that Project Waltz as a whole is vitiated by the defects in the consultation, they must establish that it was a breach of the Imperial duty for Holdings to implement the changes after such a deficient consultation. This was not argued before the judge, as he noted at B1394, B1508 and B1537. In his First Supplemental Judgment, discussing the consequences of his having referred only to Holdings in the Breach Judgment, he referred specifically to the consultation issue. At paragraph 17 he said this:

The first of those caveats is relevant for present purposes. At paragraph 17(i) he said:

The RBs take that as an invitation to argue that Holdings was in breach of the Imperial duty, but not only in respect of its own conduct in the consultation. Rather they focus on Holdings’ decision to proceed with Project Waltz after and despite the consultation, with all its defects. That position is logical and correct, because the Imperial duty affects and constrains the exercise of discretionary powers under the Plans, rather than affecting the conduct of Holdings in general.

IBM contended that, as indicated by the judge, the RBs would require permission to appeal in order to raise this point, which they have never sought or obtained. To the contrary, the RBs submitted that no such permission is necessary, because they do not in this respect need, or seek, to vary the judge’s order. All they need to do (as they say they have done) is to advance this argument as another reason for supporting the judge’s decision and his order.

To deal with this we need to refer to the judge’s Breach Order. In paragraphs 1 to 3 he dealt with the three arguments that had been advanced for saying that the Exclusion Notices were invalid. Then at paragraph 4 he dealt with the breach of duty issues. Paragraph 4 is divided into two, a general and a particular provision, the latter being stated to be “without prejudice to the generality of” the former. Paragraph 4.1 is as follows:

Paragraph 4.2 sets out four instances of breach of duty by either Holdings or UKL, and one instance which would not have been a breach of duty. They do not include anything about consultation. Instead, consultation is covered at paragraph 4B of the order, as follows:

Thus, although the Breach Order, rather conspicuously, does not find Holdings to be in breach of the Imperial duty in any respect to do with the consultation, but only of the statutory duty, and does not refer to consultation as an issue in paragraph 4.2, the RBs contend that if they were able to show that Holdings’ decision to implement

We have no doubt that, if it had been argued before the judge, and if he had held, that Holdings was in breach of the Imperial duty in some respect connected with the consultation process, this would have been mentioned in terms in paragraph 4.2, and paragraph 4B would have been in different terms. We therefore consider that the RBs’ contention that success on this point does not require any change to the judge’s order is not correct. If the argument were pursued and were upheld the Breach Order would undoubtedly have to be varied, certainly as regards paragraph 4B and also as regards paragraph 4.2 which would otherwise be seriously misleading,

That could be regarded as a somewhat technical point, which could be overcome by a late application for permission to appeal. However, it is connected with a point of more substantive significance, namely that the point was not argued before the judge and, so IBM submits, it ought not to be allowed to be raised on appeal.

For the RBs, Mr Stallworthy, arguing these issues, contended that in fact this had been part of their case all along and had indeed been pleaded. We will not take time in this long judgment to examine the pleadings, but we have to say that it is very far from clear to us that Mr Stallworthy is right on the question of the RBs’ pleaded case. Whatever the position may be in that respect, it is clear that the judge did not understand this case to have been advanced before him. We find it impossible to suppose that it had been argued before him that the consultation defects amounted to or gave rise to breaches of the Imperial duty as well as the contractual and statutory duties, given that the judge says the opposite in both the Breach Judgment (B1394, B1508 and B1537) and the Remedies Judgment at R662. If he was wrong about that, it was a misunderstanding which ought to have been pointed out to him at the stage of circulation, in advance of hand-down, of the draft judgments. We proceed on the basis that the point was not argued at trial and it is therefore a new point which is sought to be raised for the first time on appeal.

If it had been addressed as such on an application for permission to appeal (as we consider it should have been, for reasons already given) the question whether it should be allowed to be raised might have been argued at that point, but even if permission had been given, the point would be open on the hearing of the appeal. The same is true in the present situation where the point comes before the court under a Respondent’s Notice seeking to uphold the judge’s order for additional reasons which had not been argued before the judge.

It is a matter of discretion for the court whether to allow a party to raise on appeal a point not relied on below. If it is a pure point of law arising entirely on facts which were already before the court, then the court may allow it to be taken if no possible prejudice would be caused to the other party as regards the conduct of the trial, as in Pittalis v Grant [1989] QB 605. However, in a case in which, if the point had been taken below, the evidence adduced or the course of the trial might have been different in some material respect, then the new point will not be allowed to be raised. Cases to this effect under the CPR include Jones v MBNA International Bank[2000] EWCA Civ 514 (see Peter Gibson LJ at paragraph 38 and May LJ at paragraphs 51 and 52), Crane v Sky In-Home Ltd[2008] EWCA Civ 978 (see Arden LJ at paragraphs 18 to

In Crane, Arden LJ said, at paragraph 22:

In similar vein, in Re Southill Finance Lloyd LJ said at paragraph 49:

Of course, the circumstances of cases differ widely. It is evident that, in the present case, the formulation of the case for the RBs at trial had been prepared and addressed with elaborate care and preparation. Mr Stallworthy contends that, since the consultation was in issue as such in the context of the allegations of breach of duty against both Holdings and UKL, it is not clear in what way the evidence adduced before the court would have differed if it had been made clear that a breach of the Imperial duty was alleged against Holdings for pressing ahead with Project Waltz, as well as a breach of the contractual duty against UKL for the deficient consultation as such. However, the burden is on the RBs to satisfy us that the trial could not have been conducted differently. Moreover, the judge would have had to have addressed the point, and to have made findings about it, which he did not. In those circumstances, given the very full and thorough way in which the case had been prepared and put at trial for the RBs, it does not seem to us that it would be right to allow them, as an afterthought, to widen their case now, by expanding the asserted consequences of the deficient consultation process, and therefore relying on it as an additional reason for setting aside the steps taken as part of Project Waltz. We therefore find for IBM on Issue 21, both because permission to appeal was necessary in order to put it before the court, and because, even if such permission had been sought (which it was not) we would not have granted it because it involves raising a point which had not been relied on below and the circumstances are such that allowing that point to be taken would not, we consider, be a proper exercise of the court’s discretion.

Both section 259(3) and Regulation 18 of the Consultation Regulations make it clear that the delimitation of remedies made by those enactments is only for breach of the Consultation Regulations. We agree with the judge for all the reasons that he gives (see R680 to 688) that those enactments do not affect remedies for breach of contract.

As to the grant of an injunction to prevent IBM from implementing Project Waltz, the judge made no decision on this and left the matter for this Court (R699). It is right

Thus, what the RBs contend for under Issue 38 is that Project Waltz should not be allowed to be implemented, despite it not being in itself in breach of any relevant duty, until and unless a proper consultation process, compliant with the relevant obligations, has been undertaken. Necessarily, such a consultation would have to be about new proposals. It would make no sense to ask in 2017 (or later) for comments about proposals designed in 2009 to take effect between 2009 and 2011. IBM would have to develop a new set of proposals. IBM also points out that no attempt has been made to suggest in what respect, if any, Project Waltz as implemented would have been different if the consultation process had not been deficient in any of the respects identified by the judge. Thus, as a consequence of UKL having broken the contractual duty by the defects in the consultation process, and both Holdings and UKL having broken their statutory duties, Holdings and UKL would be required to unwind the implementation of Project Waltz, a scheme which was in itself a proper and lawful exercise of the relevant powers, and start again to address what changes should now be proposed by way of changes of the relevant provisions of the Plans. Members of the Plans would be put in a very different position from that in which they would have been either under Project Waltz or under other different proposals if they had been put forward in 2009. They would certainly not be put back into the position they ought to have been in after a proper consultation in 2009 on the Project Waltz proposals.

In our judgment it would not be appropriate to put IBM into that position in the given circumstances. An injunction is a discretionary remedy. There are classes of case where an injunction may be granted readily for breach of a negative obligation, and where the court may not find it necessary to consider whether damages would be an adequate remedy. We were shown the case, about as different from the present case as one could imagine, of Araci v Fallon[2011] EWCA Civ 668, in which Mr Fallon was restrained by injunction from riding a different horse in the Derby, having agreed, for value, to ride Mr Araci’s horse in the race. We do not doubt the principle but do not find that case helpful in the present circumstances, where the relevant contractual obligation is indeed negative in its content, but far less specific than that undertaken by Mr Fallon. Moreover the remedy sought would not restrain the breach of duty as such, which is in the past, namely the defective consultation process on the part of UKL.

Mr Stallworthy submitted that damages would not be an adequate remedy because they would be difficult to assess and because of inequality of arms between the RBs and IBM. As for difficulty of assessment, that is often a problem. Miss Rose showed us, as an example, Hagen v ICI Chemicals & Polymers Ltd[2001] EWHC 548 (QB),

As for inequality of arms, it is clear that the RBs have not hitherto suffered from that in the conduct of the litigation. We were told that the RBs’ litigation expenses have until now been met out of the funds of the Plans, and that their protection, as regards both costs incurred and any contingent liability for other parties’ costs, has been covered by protective costs orders. (In practice, because the Plans are all in deficit, IBM has borne the burden of the RBs’ costs, because their payment out of the funds increases the deficit which IBM is liable to make good.) We have not seen those orders, but we proceed on that basis, and on the basis that, to the extent that the litigation continues, for example as regards a damages claim for breach of the duty we are now discussing, the RBs will be able to seek and, subject to normal considerations as to merits and the like, might obtain continued protection by way of such an order. We should make it clear that we make no assumption as to whether they could get such an order, and we also note that the RBs might not be appropriate as representative claimants in a claim for damages. Whether or not such an order could be obtained, we do not consider that inequality of arms is a determinative point in favour of the RBs’ claim for injunctive relief.

The principal reason for rejecting the claim for an injunction as a matter of discretion is that it would change the position of IBM and of the members of the Plans far too radically, by requiring Project Waltz (which on this basis is not legally objectionable in itself) to be unravelled and cancelled, and by putting IBM in the position of having to consider and formulate what would have to be entirely new proposals for the reform of the benefit provisions of the Plans, a long time after the relevant events, namely the consultation process in 2009, and in what are likely to be profoundly altered circumstances. It would not be a case of consulting again, in a proper manner, on the original proposals. It would not result in restoring the beneficiaries to the position they would have been in if a proper consultation had been carried out in 2009.

We bear in mind the statutory provision as to the consequences of default as regards the statutory consultation obligations. It manifests a policy in favour of legal certainty, which is desirable for the sake of all concerned: employers, beneficiaries and trustees. That is not in any way conclusive, though it is relevant that the only breach established against Holdings is breach of the statutory duty. But it seems to us that it is consistent with the approach that, even if a serious breach of contractual duty

For those reasons, on Issue 38, we hold that it would be wrong to make an order requiring a new consultation process to be undertaken before Project Waltz can be implemented. The beneficiaries will be entitled to claim damages against UKL for breach of the contractual duty in the conduct of the consultation.

Issue 1, the threshold issue: The judge decided the case on the basis that the Reasonable Expectations which he held to have been generated must be satisfied unless there was no other possible course open to IBM than to disappoint them (paragraph [176] and following, especially from [217]).

Issue 2: The judge was wrong to decide that a principal employer’s nonfiduciary discretion must be justified as a necessary and appropriate response to the circumstances. Instead he should have applied a rationality test, equivalent to that in Wednesbury as set out in paragraph [46]. Issues 3 and 3(a) do not arise.

Issue 5: The judge was wrong to hold that an employer’s ability to offer discretionary salary increases on the basis that they would not be pensionable was restricted by the existence of Reasonable Expectations to the contrary. He should have applied a rationality test equivalent to that in Wednesbury, as set out at paragraph [45].

The burden of proof was not one of the identified issues, but it was relevant to several of those that were agreed. The judge was right on this, both to adopt a unitary approach to the question whether Holdings or UKL had acted in a manner incompatible with the relevant duty and whether its conduct was justified, and also to hold that the burden of proof lay on the RBs throughout, although an evidential burden could shift to the employer or principal employer in a case such as this to show what its reasons had been for acting in the relevant respects (paragraph [47] and following).

Issues 7, 8 and 9: These issues do not arise, but if Reasonable Expectations were to be legally relevant on the basis adopted by the judge, they would have to meet minimum criteria of clarity, certainty and assumption of responsibility, which should be akin to those required for contract or estoppel (paragraph [268] and following).

Issue 10: the judge found that the Reasonable Expectations he relied on were qualified so as not to subsist after a significant change in financial or economic circumstances. He ought to have considered and decided, but did not, whether Holdings or UKL could properly have concluded that such a significant change had occurred in 2008 (paragraph [235] and following).

Issue 11: The judge was wrong to hold that the Reasonable Expectations in respect of future DB accrual he identified were generated by the facts he found. There was no adequate basis of fact or reasoning for his decision that such Reasonable Expectations had survived until 2011, still less until 2014 (paragraph [261] and following).

Issue 12: The judge was wrong to hold that the Reasonable Expectations in respect of the early retirement policy that he found had been generated by the facts he found. There was no adequate basis for his finding that a Reasonable Expectation as to early retirement policy existed already in 2004, and no other proper basis for finding that such a Reasonable Expectation existed (paragraph [249] and following).

Issue 13: The judge was wrong to find that there was a separate breach of the Imperial duty in respect of the duration of the Early Retirement Window, even if he had not been wrong to find a Reasonable Expectation about the early retirement policy (paragraph [274] and following).

Issues 14 and 15: The judge was right to hold that there was no Reasonable Expectation as to non-discrimination as regards pay policy between members who had elected to remain as DB members as part of Project Soto and DC members (paragraph [283] and following).

Issue 16: Statements at the time of Projects Ocean and Soto regarding the effect of the changes on the stability and sustainability of the Plans over the long term did not by themselves give rise to a Reasonable Expectation that DB accrual would continue until at least 2014 (paragraphs [265], [266] and [xr])

Issue 17: The judge was justified in not finding that the investment policy adopted by the Trustee provided an additional reason for holding that Holdings and UKL acted in breach of their duties (paragraph [290] and following).

Issue 20: The judge was right not to take into account, in deciding whether Holdings and UKL acted in breach of their duties in implementing Project Waltz, the fact that attention had been drawn to the facts which later became the subject of the rectification proceedings (paragraph [293] and following).

Issue 21: The judge was right not to decide that Holdings’ conduct in implementing Project Waltz after a defective consultation process which did not comply with statutory requirements and involved a breach of its duties as employer by UKL, was a breach of the Imperial duty on the part of Holdings. Such a case was not put to the judge, and it could not properly be raised on appeal (paragraph [439] and following).

Issues 22 and 25: In our view the judge held that there would have been adequate business justification for Project Waltz but for the fact of the Reasonable Expectations, which he considered would be thwarted by the changes, and for his view that IBM had to show (and could not do so) that no other possible course of action was open to it (see in particular paragraph [308] as to justification apart from the Reasonable Expectations).

Issues 23 and 24: The judge erred by treating the IBM group, including CHQ as well as Holdings and UKL, as one entity. The judge should not have had regard to the motivations and knowledge of CHQ (or of individuals within CHQ) in considering the motivation and justification for the acts of IBM UK (paragraph [363] and following).

Issue 26: The judge was wrong to hold that IBM could have developed proposals to deliver its business objectives in other ways. No such case was advanced by the RBs, on whom the burden of proof lay (paragraph [389] and following).

Issue 27: The judge was entitled to take into account the fact that (on his findings) the Reasonable Expectations had been generated before the commitments to investors were made by way of the 2010 EPS Roadmap (paragraph [394]).

Issue 28: The judge was wrong to find that CHQ was disingenuous in relation to the Project Soto communications to members and that this counted against Holdings. No such case was advanced by the RBs (paragraph [402] and following).

Issues 29 to 31: As these depend on there being a Reasonable Expectation as found by the judge, on which we conclude that he was wrong, these issues do not arise (paragraph [407] and following).

Issue 32: Insofar as the judge decided that the Initial 2009 NPAs and the 2011 NPAs were procured in breach of UKL’s Contractual duty because of a “threat”, he was wrong (paragraph [421]).

Issue 33: The judge was wrong to hold that the NPAs were procured in breach of UKL’s Contractual duty because they formed part of Project Waltz (paragraph [415]).

Issues 34 to 36: These issues as to remedies do not arise, and we say nothing about them.

Issue 37: The judge was wrong to hold that before the early retirement policy was changed, Holdings had to announce that such a change would be made (paragraph [278] and following).

Issue 38: Despite the unappealed breach of the Contractual duty by UKL and of the statutory duty by both Holdings and UKL, it would be wrong to require those companies to undertake a new consultation process before implementing Project Waltz (paragraph [454] and following).

Issue 39: The exercise of the Main Plan Exclusion Power by Holdings did not involve an exercise of the Exclusion Power for an improper purpose (paragraph [160] and following).

This is one of the three hypothetical examples supplied to us so as to illustrate the effect of Project Waltz. The capital value of the pensions is given for ease of comparison; it uses the 1:20 factor recommended by the Government Actuary’s Department for use in calculating the value of a pension for Lifetime Allowance purposes. The pension figures are taken from projections made for the Trustee in July 2010 by Towers Watson, the Scheme Actuary. As noted under Case Two below, the post-Waltz projections in that case are too low.

The hypothetical case is of a male employee who had joined the C Plan (the main DB part of the plan) at age 25, who was paid £40,000 p.a. on 6 April 2010 and who would be 55 on 5 April 2011. He would have the following pension benefits depending on his choice under Project Soto (PS), as between staying in the C Plan with continuing DB benefits, and moving to the Enhanced M Plan with DC benefits for future service. In the first two cases he would (or might) be affected by the new early retirement (ER) policy.

Thus, if the employee chose to remain in the DB Plan under PS, the pension taken at age 55 was projected to be lower by 30% because of the effect of the new ER policy and of the NPAs.

If he switched to the DC Plan under PS, his pension at age 55 was projected to be 23% lower because of the new ER policy.

If the employee chose to remain in the DB Plan under PS, this shows the pension taken at age 60 was projected to be lower by 21% because of the effect of the New ER policy, of the NPAs, and the cessation of DB accrual.

If he switched to the DC Plan under PS, his pension at age 60 was projected to be 14% lower because of the new ER policy.

However, these projections give too low a figure for the post-Waltz cases, because this hypothetical employee would be entitled to retire at 60 with the benefit of the pre-Waltz ER policy. Thus the reduction would in fact be less than is shown in the figures above.

If the employee remained in the DB Plan under PS, and remained in employment until NRD, his pension was projected to be 5% higher because the new ER policy would not affect him (even if it otherwise would have done), and the projected value of the DC element from 5 April 2011 would result in the higher pension figure.