Atos IT Services UK Ltd v Atos Pension Schemes Ltd

I have before me a Part 8 claim which raises questions of construction in relation to a pension scheme known as the Atos UK 2011 pension scheme (“the Scheme”). The Claimant, Atos IT Services UK Ltd, is the Principal Employer for the Scheme and appeared by Mr Andrew Spink

This claim is another example of what is by now a fair number of cases raising the question which index should be used to increase pensions in payment (and sometimes for other purposes as well), and in particular, whether that should be RPI or some other index, typically CPI, although in this case the options canvassed have included not only CPI but a variant of RPI called RPI-J and an index derived from CPI called CPIH. As Mr Spink said, all these cases turn on the construction of the particular terms used in the scheme in question and, save insofar as they lay down general principles, no direct assistance can be obtained from them. Indeed, there is very longstanding authority that on questions of construction reference to other decisions on different words in other instruments is a practice to be deplored: see the classic statement of Sir George Jessel MR in Apsden v Seddon (1874-75) LR 10 Ch App 394 at 397:

“No Judge objects more than I do to referring to authorities merely for the purpose of ascertaining the construction of a document. That is to say, I think it is the duty of a Judge to ascertain the construction of the instrument before him and not to refer to the construction put by another Judge upon an instrument perhaps similar but not the same.”

Mr Spink’s written submissions included a selection of indexation provisions found in the reported cases and in other Atos schemes. Such provisions often have a family resemblance consisting of a reference to the index to be used followed by circumstances in which some other index can be substituted (what can be called a trigger provision), but the detailed drafting shows a wide variety, and I entirely accept that this case has to be determined on the wording of the provision in question and not by comparing other cases in which other words have been construed. I was therefore, quite rightly, not taken to those other cases, although some of them are very familiar.

I will proceed at once to set out the provision to be construed (“the Definition”). The Scheme is still governed by an Interim Deed dated 30 June 2011 (“the Interim Deed”). This contained, in the usual way, an undertaking by the parties to execute a definitive deed within 24 months, but, in the usual way, the parties have not complied with their undertaking – I do not know of any other species of undertaking that is so routinely broken. As explained in more detail below, the Definition is not technically part of the Interim Deed but is contained in a document attached to it and it is common ground that it applies for the purpose of calculating annual pension increases for some of the benefits payable under the rules. The Definition reads as follows:

“Retail Prices Index means the general index of retail prices (all items) published by the Office for National Statistics, or, where that index is not published, any substituted index published by that Office (or its successor) as the Principal Employer and the Trustees may agree. Where the retail prices index ceases to exist, the Principal Employer and the Trustees may agree any substituted index published by that Office (or its successor).” It is one of a number of definitions found in that document.

This provision, unlike some of the others that have come before the courts, does not to my mind pose any real difficulties of construction at all. There are three questions that arise on the wording, but none of them is difficult to answer. They are as follows:

s. 21(1) SRSA 2007 reads as follows (where the reference to “the Board” is now to be read as a reference to the UKSA):

“21 Retail prices index

(1)

The Board must under section 20-

(a)

compile and maintain the retail prices index, and

(b)

publish it every month.”

(The UKSA in fact operates through the Office of National Statistics (“ONS”) which is an executive agency of the UKSA and it is not generally necessary to distinguish between them; I will therefore in what follows refer to the UKSA as including the ONS where appropriate.) RPI is in fact the only index which the UKSA is under a statutory obligation to publish. Unsurprisingly, they comply with this obligation, and in practice they put the RPI figure on their website (or ONS’s website) every month. The evidence in this case makes it perfectly clear that the UKSA would prefer not to have to publish RPI because they regard it as flawed and last year (2019) they wrote to two successive Chancellors of the Exchequer, Mr Philip Hammond and Mr Sajid Javid, inviting them to consider repealing s. 21. Mr Hammond did not in the event have the opportunity to reply before he was replaced by Mr Javid. Mr Javid replied that he had no current intention of doing so. So I accept that the UKSA does not want to have to compile, maintain and publish RPI, but as at today, and for the foreseeable future, that is something they are obliged by primary legislation to do, and hence something they do, and will continue to do. In those circumstances the suggestion that the trigger condition in the Definition, namely “where that index is not published”, has been met is to my mind a wholly impossible view to sustain.

Those were the views that I reached on first reading the Definition last Wednesday. I am conscious, however, that first impressions can be misleading and all judges have the experience of having their initial views changed by the course of argument. I therefore read with care Mr Spink’s very detailed written submissions, and listened with interest and attention to his oral submissions which extended from 12 noon on Thursday to 4.30 on Friday. But I mean no disrespect to Mr Spink, who advanced his arguments with his usual clarity, forcefulness, skill and charm, that I remained wholly unpersuaded that the definition posed any difficulty of construction at all, or that there was in truth any real room for doubt.

Further confirmation for that view came from a point that emerged from Mr Stallworthy’s written submissions. This is that the wording of the Definition bears a striking resemblance to statutory definitions in the tax legislation. s. 833(2) of the Income and Corporation Taxes Act 1988 (“ICTA 1988”) provided as follows:

“References in the Income Tax Acts to the retail prices index are references to the general index of retail prices (for all items) published by the Office for National Statistics; and if that index is not published for a month which is relevant for the purposes of any provision of those Acts that provision shall be construed as referring to any substituted index or index figures published by that Office.”

Comparison of the wording of the statutory definitions and the Definition shows that although not verbatim they are remarkably close. In each case the index is defined as:

“the general index of retail prices (for all items) [or in the case of the Definition “(all items)”]

published by the Office for National Statistics”.

In each case the trigger provision is:

“if [or in the case of the Definition “where”] that index is not published”

(although the tax definitions carry on by referring to publication for a relevant month). In each case there is then a reference to:

“any substituted index [and in the case of the tax definitions “or index figures”] published by that Office.”

It is impossible to believe that this almost word for word reproduction of the statute is coincidental, and it seems an obvious inference that the draftsperson took the wording, directly or indirectly, from one or other of the tax definitions. (As appears below, this wording first appeared in a 2007 deed, not long after ITA 2007 was passed, and was repeated in 2011.) But in the tax definitions the definition clearly means RPI and nothing else, and there is no reason to think it could ever come to mean anything other than RPI, unless RPI ceased to be published, nor any reason to think that “published” should be given anything other than its ordinary meaning of “made public”. Now, of course, as a matter of legal theory, the words when transposed to a trust deed do not necessarily have the same meaning as they did in the statute, but one would expect them usually to do so unless there was good reason to conclude otherwise. There is in the present case, to my mind, no reason at all to think that they should be given a different meaning. Since it is common ground that in 2011 the draftsperson did intend to refer to RPI, there is nothing surprising in him or her reaching for a convenient statutory definition of RPI and adopting it wholesale. That does not begin to suggest that he or she intended a different meaning. I regard this point as strongly confirmatory of my initial impression of what the Definition meant, and indeed as putting the matter beyond any real doubt.

In those circumstances I made it clear to Mr Stallworthy at the conclusion of argument on Friday afternoon that I did not require him to advance any oral submissions on Question 1 on the Claim Form. That question reads as follows:

“A declaration as to whether on the true construction of the 2011 TDR, and in the events that have happened, in the definition of Retail Prices Index in Schedule 1 to the draft Siemens Rules:

a.

the expression “the general index of retail prices (all items) published by the Office for National Statistics”:

(i)

means RPI; or

(ii)

means, or meant until its discontinuance in March 2017 RPIJ; or

(iii)

means CPI; or

(iv)

means CPIH.”

My answer is (i): it means RPI.

As to Question 2 on the Claim Form this reads as follows:

“If 6.1(a) [ie Question 1] is answered in the sense “(i)” - ie RPI - the expression “or where that index is not published” means:

(i)

“where that index (although it has not ceased to exist) is not published in any given month or at any time required for its use under the 2011 TDR”; or

(ii)

“where that index is not published for any purpose”; or

(iii)

“where that index is not published for the purpose it was at the time the Scheme was established”; or

(iv)

“where that index is not published for the purpose it was at the time the index was first published”; or

(v)

“where that index is not published as an Official Statistic by the ONS”; or

(vi)

“where that index is not published with National Statistics status”; or

(vii)

“where that index is not published as an appropriate or alternatively as a preferred or alternatively as the lead measure of consumer price inflation by the ONS”; or

(viii)

“where that index is not published as an appropriate or alternatively as a preferred or alternatively as the lead measure of consumer price inflation for pension indexation purposes by the ONS”; or

(ix)

“where that index is not published by the ONS as an appropriate or alternatively as a preferred or alternatively as the lead measure of inflation used for uprating compensation and benefits”;”

Mr Stallworthy’s written submissions argued for sense (i) as his preferred answer, with sense (ii) as “the least implausible” of the alternatives. Mr Spink, who espoused some of the other answers, advanced a number of reasons why answer (i) was wrong, notably that it would lead to a conflict with other provisions providing that in the case of a temporary unavailability of RPI the Trustees should have a power, with the agreement of the Principal Employer, to substitute “such percentage as they consider to be a reasonably likely figure on the basis of information available to them”. That gives all the appearance of being intended to be a temporary fix for a temporary problem; whereas in the Definition the trigger condition permits, by contrast, the Trustees and Employer to adopt a substitute index. That looks more like a permanent fix to a permanent problem. But whether that is so or not, something I need not decide, it is clearly different from adopting “a reasonably likely figure”.

For that reason I indicated to Mr Stallworthy at the end of Mr Spink’s submissions that I was not inclined to think that the answer to Question 2 was in sense (i), but that I myself preferred, for reasons I have already explained, sense (ii), and that I would hear from him on Monday if he still wished in the circumstances to argue for sense (i), since sense (ii) would in practical terms mean that the trigger condition had not been met. Over the weekend he kindly indicated that he would not push for answer (i), although he reserved the right to resurrect it in case the case went any further. On that basis it was unnecessary for him to address me on Question 2 either. And I therefore answer Question 2 in sense (ii), namely “where that index is not published for any purpose”, and in those circumstances since the index is still published by the UKSA, as I have already explained, the trigger condition has not been met.

That is sufficient to dispose of the questions raised in the Claim Form as the following questions, Questions 3 and 4, do not in those circumstances arise. I could, I think, stop this judgment here as I believe that what I have already said is adequate to explain why I have reached the views I have, and hence to fulfil the obligation to give reasons as explained by the Court of Appeal in English v Emery Reimbold & Strick Ltd [2002] EWCA Civ 605 at [17]-[19], which makes it clear that there is no obligation on a judge to deal with every argument advanced; all he or she need do is explain to the parties, and if necessary to the Court of Appeal, the basis on which he or she has acted. But out of deference to Mr Spink’s careful argument I will say a little more. As I said on Friday afternoon, if I had had the leisure to write a lengthy judgment I would, I think, have enjoyed doing so, as the account that Mr Spink gave of what he called “the demise of RPI” is a fascinating one, both in terms of the underlying mathematics and in terms of the statistical, economic and political consequences. If I had nothing better to do I would have set it out in all its captivating detail, but it would have been mere indulgence to do so. The pressure of business in the Division is such, and the numbers of other litigants waiting to have their cases heard or their judgments finalised are such, that I thought it better to resist the temptation to turn Mr Spink’s account into a lengthy and polished reserved judgment that in truth would not have added anything of significance to the legal analysis I have already set out. Hence I have decided to give this unreserved, comparatively brief, and unpolished judgment instead, and will confine myself as much as possible to the essential points in what follows.

I will say a bit more about the Scheme. There was no dispute as to the facts and I heard no oral evidence. The Scheme was established by the Interim Deed (dated as I have said 30 June 3011) with effect from 1 July 2011. It was set up as a result of the acquisition by the Atos group of a company then called Siemens IT Solutions and Services Ltd (“Siemens IT”) from the Siemens group. The Interim Deed was executed by Siemens IT as Principal Employer, and by the Trustee, then called Atos Origin Pension Schemes Ltd. The Claimant was later substituted as Principal Employer by a deed dated 22 December 2011.

The acquisition of Siemens IT of course meant that the employees of Siemens IT, although not changing their employer, would now be employed by a company in the Atos group rather than a company in the Siemens group, and as I understand it (although I was not taken to the arrangements in any detail) the Scheme was established both to provide for the continued accrual of their pension rights and as a vehicle into which their past service rights could be, and in due course were, transferred. They had formerly been members of a Siemens scheme called the Siemens Benefit Scheme (“SBS”), which in the usual way contained a number of different benefit structures, some on a defined benefit (or “DB”) basis, some on a defined contribution (or “DC”) basis, and the Scheme, as set out in the Interim Deed, was intended to provide mirror benefits.

Clause 4.3 of the Interim Deed provided, so far as relevant, as follows:

“Until the execution of the Definitive Deed the Trustees must administer the scheme in accordance with ...

4.3.2

the provisions of this deed and the Appendices and any subsequent amendments made thereto.”

Appendix 1 to the Interim Deed, headed “Contribution and Benefit Provisions” provided, inter alia, for what were called “Former DB members”, and paragraph 3.2 of that Appendix provided as follows:

“Subject to the conditions and modifications set out in paragraphs 3.3 to 3.11 below, in relation to each Member, his contributions payable to the Scheme, benefits payable to or in respect of him from the Scheme and the terms and conditions applicable to his membership of the Scheme shall (with the necessary alterations to points of detail) be the same as those applicable under the SBS Terms immediately prior to the Commencement Date (referred to in this paragraph 3 as the “Relevant Terms”).”

“SBS Terms” is a defined term, being defined in paragraph 1 of the Appendix as follows:

“ “SBS Terms” means, in respect of a Former DB Member his contributions payable to the SBS, benefits payable to or in respect of him from the SBS and the terms and conditions applicable to his membership of the SBS immediately prior to the Commencement Date as set out in the Saver Plan Plus Section or the Tower Plan Section, as applicable, and shall, where applicable include:

[and then various things are set out]

and for the purpose of interpretation of the SBS Terms the applicable definitions contained in schedule 1 of the SBS Rules shall apply.”

“SBS Rules” was itself defined in the same paragraph as follows:

“ “SBS Rules” means the definitive trust deed and rules governing the SBS dated 18 August 2008 (the “2008 SBS Deed”) as amended by deeds of amendment dated 21 December 2009 and 29 June 2010 and also including further changes contained within a draft definitive trust deed and rules (the “Draft 2011 SBS Deed”), a copy of which is attached to the Interim Deed but does not itself form part of the Interim Deed, and which is intended to replace the 2008 SBS Deed in order to consolidate the amendments made by the aforementioned deeds of amendment and to make certain other amendments to the 2008 Deed (and, for the avoidance of doubt, the SBS Rules shall not include any other amendments made to either the 2008 Deed or to the draft 2011 SBS Deed on or after the Commencement Date.”

As that indicates, the draft deed and rules for the SBS was annexed to the Interim Deed. I will call it “the Draft 2011 SBS Deed” as the definition of the SBS Rules does. Schedule 1 of the SBS Rules as amended by the Draft 2011 SBS Deed included definitions, among which is the Definition (of Retail Prices Index) with which I am concerned. I was told that the Draft 2011

I will give an illustrative example. The SBS contained a number of different benefit structures called Plans. In the case of a member with DB rights under a Plan called the Booster, Saver and Foundation Plan the benefit structure was set out in schedule 3 of the Draft 2011 SBS Deed. That provided for pension increases at paragraph 9 as follows:

“such part of the pension which relates to Pensionable Service prior to 6 April 2006 shall be increased by the percentage increase in the Retail Prices Index during the previous 12 months or, if less, by 5%; and ...”

and then there are some more provisions, but paragraph 9.1.3 states:

“The increases payable under this Rule shall take effect on 1 April each year and shall be calculated as at each 1 April by reference to the number of completed months since the date when the pension became payable, or, if later, from the date of the last calculation. For the purposes of this Rule, the Trustees shall use the Retail Prices Index published for the 12 month period ending on the preceding 31 December (normally published in the preceding January). If the Retail Prices Index is not published in respect of the relevant period (or is published too late to be used for the purposes of this Rule the Trustees may substitute such percentage as they consider to be a reasonably likely figure on the basis of information available to them, such figure to be agreed with the Principal Employer.”

That is an example of what I referred to earlier as what appears to be a temporary fix for a temporary problem. Very similar provisions are applicable under some of the other Plans set out in other schedules.

Since the Definition is contained in a draft deed amending the SBS, I was referred to some of the history of the SBS which sheds some light on the evolution of the wording now found in the Definition. The SBS was established by a definitive deed in 1990. That contained a provision for pension increases as follows:

“11.1.

Increases in current pensions

Any pension ... shall be increased by:-

(a)

the proportion by which the Government’s Index of retail prices figure (published in the preceding January) has been increased during the previous 12 months, or if it is less,

(b)

5% a year compound with yearly rests on the anniversary of the Member’s retirement or death (as appropriate).”

There was no definition of Retail Prices Index as such but there is no doubt that RPI was what was meant. CPI was not introduced until 1997, and no one could have thought that the reference to “the Government’s Index of retail prices figure” in 1990 referred to anything except RPI, which had been in existence in one form or another since 1947, and established as the Retail Prices Index in 1956. It was by 1990 undoubtedly the primary measure of price inflation in the UK. There may have been sub-indices or derivative indices derived from it by 1990, but even if there were (something the evidence in this case leaves rather unclear), only RPI could properly be described as “the Government’s Index of retail prices figure”. The contrary was not suggested.

By Deed dated 14 December 2007 the SBS Deed and Rules were re-stated. This contained for the first time a definition of Retail Prices Index as follows:

“Retail Prices Index means the general index of retail prices (all items) published by the Office for National Statistics, or, where that index is not published, any substituted index published by that Office (or its successor) as the Principal Employer and the Trustees may agree.”

By Deed dated 18 August 2008 the SBS Deed and Rules were again re-stated. They contained a definition of Retail Prices Index in identical terms to that in the Deed of 14 December 2007.

By Deed of Amendment dated 9 June 2011 the definition of Retail Prices Index was amended.

In the Draft 2011 SBS Deed the definition of RPI was amended again to read as follows:

“Retail Prices Index means the general index of retail prices (all items) published by the Office for National Statistics or, where that index is not published, any substituted index published by that Office (or its successor) as the Principal Employer and the Trustees may agree. Where the Retail Prices Index ceases to exist, the Principal Employer and the Trustees may agree any substituted index published by that Office (or its successor).”

That is, as already explained, the Definition that I have to construe. At the same time the schedules were amended to deal with the temporary non-availability of the index in the form of which I have read an example in paragraph 9.1.3 of schedule 3.

Finally, as already referred to, the Draft 2011 SBS Deed was in fact executed on 30 June 2011 in identical terms to the draft attached to the Interim Deed. I am told that the Definition remains part of the SBS rules to this day, but I am of course not concerned with any question of construction of the current SBS rules, or the subsequent history of the SBS provisions, which cannot affect the meaning of the Definition which I am construing.

The above history can be summarised as follows. The first sentence of the Definition (“Retail Prices Index means the general index of retail prices (all items) published by the Office for National Statistics or, where that index is not published any substituted index published by that Office (or its successor) as the Principal Employer and the Trustees may agree.”) can be traced back to the Deed of 14 December 2007 and is identical to the definition as it then stood. The second sentence of the Definition (“Where the Retail Prices Index ceases to exist, the Principal Employer and the Trustees may agree any substituted index published by that Office (or its successor).”) can be traced to the Deed of Amendment dated 9 June 2011, and again is in identical terms to the last sentence of the definition as there amended. Other parts of the definition as amended by the Deed of 9 June 2011 have not survived into the 30 June 2011 Definition. Part of it, as I have explained, found its way, in effect, into the pension increase provisions in the schedules, and part of it was dropped altogether.

I do not think I need to say anything more about the history of either the Scheme or the SBS. I was told that the Scheme remains open to further accrual on both a DB (that is, final salary and CARE (career average revalued earnings)) basis, and a DC (or money purchase) basis. At the latest full actuarial valuation as at 31 March 2018, the Scheme had assets valued on a Technical Provisions basis of £585m, and liabilities valued on the same basis of £528m, so the Scheme was in surplus on a Technical Provisions basis. It is not, however, in surplus on a discontinuance basis, something which I do not find surprising as employers do not in general fund for discontinuance. The scale of the impact of the present question can be gauged from the fact that the Scheme Actuary estimates that if the Scheme had switched from RPI to CPI as at 31 December 2018, there would have been a reduction in the Scheme’s liabilities valued on a Technical Provisions basis of some £65m. That figure has since reduced but this is sufficient to indicate the scale of the impact and the precise figures are unimportant.

I was provided with expert reports from experts instructed by the parties. Counsel were agreed that it was not necessary to cross-examine them, and I agree. In those circumstances I heard no oral evidence. The respective experts were as follows: the Claimant’s expert was Dr Bill Robinson, formerly the chief economist at KPMG and now a senior adviser to the firm; the Defendant’s expert was Ms Jill Leyland who has a long career as an economist and statistician. Both have suitably impressive CVs which it is not necessary to detail, and their reports were of high quality. I read all their reports (other than their first reports which were superseded by their second reports). Between them they explained the history of indices in the UK for measuring consumer price inflation, principally, but not wholly, RPI and CPI; the differences between them both in composition and, more significantly, in the way in which they calculate what are called the elementary aggregates; the practical effect of this difference in calculation techniques, known as the formula effect; the impact of a particular group of changes in 2010 in the collection of clothing prices; and the increasing loss of confidence in RPI as a good measure of inflation.

This was immensely helpful by way of background and as an introduction to the recent decline in RPI’s fortunes, which Mr Spink took me through in considerable detail, and I read the reports with both interest and gratitude. I am particularly grateful for the clear explanation between the three approaches used for calculation of elementary aggregates from the dataset of individual prices, namely what are called the Dutot (ratio of arithmetic means), Carli (arithmetic mean of ratios) and Jevons (geometric mean of ratios) methods, and why it is that the Carli approach used for many items by RPI will always produce a higher figure than the Jevons approach used by CPI: see the Appendix. This, the formula effect, is the main but not the only reason for the well known phenomenon that RPI is typically higher than CPI by between 0.5% and 1% pa, which is a significant difference in periods of sustained low inflation.

If it were necessary to do so to resolve any of the issues in this case I would of course set out the effect of this evidence in great detail but it is not necessary to do so and I will confine myself to recording the matters that Mr Spink referred to as the eight most significant points to be drawn from the expert evidence. They were as follows:

As I have said, I found the expert reports both enormously helpful by way of background and immensely interesting, but they do not to my mind in the end affect any of the questions I need to decide, and I need not resolve any of the differences of opinion between them, which were in any event very slight. I should however just mention one particular point of disagreement. Dr Robinson in his second report referred to RPIJ as having in November 2013 become “the official version of” RPI. That was not language that could be found in any of the contemporaneous documents that I was shown, and Ms Leyland vigorously disagreed with it, saying:

“I strongly disagree that RPIJ became “the official version of the RPI” in 2013. RPIJ was always described as “an improved variant” of the RPI by the ONS. It was always clear from the January 2013 announcement following the 2012 consultation that RPI would be continued to serve users’ needs.”

Mr Spink explained that what Dr Robinson meant was just shorthand for the sequence of events in 2013, namely that in March 2013 RPI lost its status as a National Statistic, and then in November 2013 RPIJ obtained National Statistics status. I will therefore simply say that while I accept entirely the underlying facts I do not accept Dr Robinson’s characterisation of RPIJ as

“the official version” of RPI. The language used by the UKSA to refer to RPIJ was, as the

experts agree, as “an improved variant” of RPI. It does not to my mind matter but my reading of the history is that RPIJ, which uses Jevons instead of Carli in the construction of the elementary aggregates but is otherwise identical to RPI, was initially developed by the UKSA in the hope that it might in due course come to displace RPI, being based on the same underlying data but calculated in a less flawed way. But it did not in fact gain any widespread acceptance and the UKSA downgraded its ambitions for it to being an index that quantified the formula effect. In the event, a review by Mr Paul Johnson, the Director of the Institute for Fiscal Studies, (“the Johnson Review”), concluded that the use of Carli instead of Jevons was not the only flaw that RPI suffered from, and that RPIJ suffered from those other flaws as well, and in due course RPIJ was discontinued.

In those circumstances, while I accept that RPIJ was introduced as an improved variant of RPI it never did replace RPI for any purpose, and in particular it was not the RPI published by the UKSA in fulfilment of its statutory obligation under s. 21 SRSA 2007, nor, as I explain later, could it have been, and I do not think that in those circumstances it is accurate to describe it as the official version of RPI.

I should refer to what Mr Spink called “the demise of RPI”, already foreshadowed in some of the propositions he drew from the expert reports. I was tempted to refer to this section as “The strange death of RPI”, but that would be inaccurate because RPI, as I have already explained, is very much alive despite the increasingly strong criticisms levelled at it. Much of the story has been told before, by Mr Spink as it happens, in British Telecoms Plc v BT Pension Scheme Trustees Ltd, and can be found in the judgment of Zacaroli J at [2018] EWHC 69 (Ch). For that reason, and because it does not to my mind affect the answers to any of the questions I was asked, I do not intend to set it all out in the same detail as that in which Mr Spink so carefully laid it before me. But I should give some account of the main points.

As already referred to, RPI’s origins can be traced back to 1947 when an index called the

In 2007 the SRSA 2007 was passed. It created a new body, initially called the Statistics Board, but shortly after re-named the UKSA, with statutory responsibilities (s. 1), as well as the statutory posts of National Statistician and Head of Assessment (s. 5). It introduced a formal status of National Statistic, and a formal process of assessment as to whether official statistics (broadly statistics produced by Government and other official bodies (s. 6)), met the necessary standards to qualify as National Statistics (s. 12). It also of course imposed a special statutory obligation on UKSA to compile and maintain the RPI and publish it every month, as already referred to. RPI was the only statistic treated in this way. s. 21(2) and (3) provide as follows:

“(2)

Before making any change to the coverage or the basic calculation of the retail prices index the Board must consult the Bank of England as to the whether the change constitutes a fundamental change in the index which would be materially detrimental to the interests of the holders of relevant index-linked gilt-edged securities.

(3)

If the Bank of England considers that the change constitutes a fundamental change which would be materially detrimental to the interests of the holders of relevant index-linked giltedged securities, the Board may not make the change without the consent of the Chancellor of the Exchequer.”

As those definitions indicate, the explanation for this special treatment of RPI is that the Government had issued index-linked gilts by reference to RPI, and up until July 2002 the prospectuses of index-linked gilts contained an option for redemption conferred on the gilt holder as follows:

“If any change should be made to the coverage or the basic calculation of the [RPI] which, in the opinion of the Bank of England, constitutes a fundamental change in the Index which would be materially detrimental to the interests of the stock-holders Her Majesty’s Treasury will publish a notice in the London Gazette immediately following the announcement to the relevant Government Department of the change, informing stockholders and offering them the right to require Her Majesty’s Treasury to redeem their Stock in advance of the revised index becoming effective.”

As at today, there are three issues of index-linked gilts whose prospectuses contain that wording, one of them maturing later this year, the others in 2024 and 2030. The value of those outstanding issues is over £30 billion.

Until 2010 Government had used RPI for almost all purposes of up-rating, but in June 2010 the new Coalition Government announced in the budget that it would switch to using CPI for indexation of benefits and tax credits in April 2011. There were, no doubt, sound technical reasons for this but my recollection, for what it is worth, is that it did not escape attention that it had the convenient benefit, from the point of view of Government, that it would tend to lead to lower payments by Government, something that was of course a priority following the global financial crisis and its effect on the UK banking system and public finances. It did not, in my recollection, prompt an immediate move away from RPI for all purposes.

In 2010 the ONS had made changes to the way in which clothing prices were collected. These were intended to be methodological improvements but had a particular and marked consequence on the formula effect. Before the change, the formula effect had accounted for roughly 0.5% of the difference between the annual rates of RPI and CPI but the clothing changes added a further 0.3% to that. The technical reasons for this striking change include the fact that the Carli method is particularly poor at dealing with large fluctuations in prices, something that characterises clothing prices.

This prompted a number of investigations by the ONS and the UKSA. In 2012 a review by Erwin Diewert of the Department of Economics of the University of British Colombia, commissioned by the ONS, recommended that the ONS cease to use Carli because it was inferior compared to Jevons and led to upwards bias. A public consultation by the ONS, however, saw a large response in favour of keeping RPI unchanged, probably not because of any disagreement on technical grounds but because the respondents stood to lose out if RPI was reduced. The ONS therefore decided to keep RPI as it was, with the use of Carli, and with future changes:

“limited to issues such as the annual update of the basket and weights, improvements to data validation and quality assurance etc.”

This was put in place in January 2013 and subsequently referred to as the “freezing” of RPI. At the same time the ONS developed a new index, RPIJ, which was based on RPI but used Jevons instead of Carli.

In March 2013 the UKSA de-designated RPI as a National Statistic, on the basis of two key reasons, namely (i) the finding that the methods used to produce the RPI:

“are not consistent with internationally recognised best practices”

and (ii) the decision to freeze methods used to produce the RPI and only to contemplate routine changes. In November 2013 RPIJ was designated as a National Statistic.

As already referred to, a further review was commissioned by the UKSA from Paul Johnson; the Johnson Review was published in January 2015. It concluded that RPI was a statistically flawed measure of inflation and included statements such as:

“clear evidence that the current methodology is flawed… headline RPI is not a robust measure of inflation…

[RPI is] a statistic that is no longer fit for purpose …

[RPI is] known to be statistically flawed the use of Carli is statistically flawed and can result in an upwards bias in recorded inflation.” But it also identified further problems with RPI beyond the use of Carli, saying:

“it is not just the use of the Carli which is problematic…

there are further weaknesses in the RPI beyond those identified when its National Statistics status was revoked. For example [its treatment of] … insurance premiums and second-hand car sales … The RPI excludes certain households … The treatment of owner occupied housing costs in the RPI is also not the best available. Addressing any of these would arguably breach the National Statistician’s commitment to not change the RPI.”

Consequently, the Johnson Review recommended that the ONU should make it clear to the public that the RPI was not fit for purpose, saying:

“ONS and the [UKSA] should re-state its position that the RPI is a flawed statistical measure of inflation which should not be used for new purposes and whose use should be discontinued for all purposes unless there are contractual commitments at stake…

The [UKSA] and ONS should make it clear to users that the RPI is not a credible measure of consumer price change…

The [UKSA] and ONS should also be very clear in explaining that the RPI is not a credible measure of consumer price change. They should make it clear that the RPI is not fit for purpose and should not be used except where existing legal contracts, for example index-linked gilts, demand it.”

He also recommended that RPIJ should be discontinued, as already referred to, because it suffered from the same flaws, other than the use of Carli, as the RPI itself.

The ONS was already describing the CPI as “the main measure of inflation” in 2013. From 2015 the ONS and the UKSA regularly referred to RPI in unflattering terms. Thus, the ONS in March 2015 described RPI as “found not to meet the required standards for designation as a National Statistic.” In 2015 the UKSA said “there are basic statistical flaws in the construction of the RPI.” In March 2016 Mr John Pullinger, then the National Statistician, wrote to Sir Andrew Dilnot, then the chair of the UKSA. Among other things he said:

“CPIH should become the ONS preferred measure of consumer inflation and the focal point of ONS commentary in due course…

Put simply, I believe that the RPI is not a good measure of inflation and does not realistically have the potential to become one...

I strongly discourage the use of the RPI as a measure of inflation as there are far superior alternatives.”

He announced that RPIJ would no longer be published from March 2017 and that the reason it was not possible to stop producing and promulgating RPI entirely, despite its significant flaws, was that:

“RPI is still used for a number of legacy purposes and its production is mandated by legislation”.

In November 2016 the ONS published a statement from Mr Pullinger as the National Statistician which again said that the ONS intended to make CPIH the preferred measure from March 2017 and described CPIH as “the most comprehensive available measure”. It also repeated that the ONS would cease publication of RPIJ from March 2017, as although it addressed the Carli problem it shared RPI’s other shortcomings and was not a good economic measure of consumer inflation. By March 2017 the ONS was describing CPIH (that is CPI with the addition of housing costs, which were excluded from CPI) as “our most comprehensive measure”, and by 2017 as “our lead measure”. Then in March 2018 the ONS published a further report, this time by Mr Pullinger himself, titled “The Shortcomings of the RPI as a Measure of Inflation”, in which he repeated his views, saying, among other things:

“Overall the RPI is a very poor measure of general inflation at times greatly overestimating and at other times underestimating changes in prices…

Our position on the RPI is clear: we do not think it is a good measure of inflation and

discourage its use…

RPI has a number of significant weaknesses. While the formula issue has been at the centre of the debate ... there are a number of other problems. Housing costs are produced using a poor proxy for depreciation … Overall we do not view the RPI as a good measure of inflation…

For the reasons set out in this document, RPI does not have the potential to become a good measure of inflation.”

The House of Lords Economic Affairs Committee (on which a number of impressive individuals sat) heard evidence during 2018 and produced a report in January 2019 called “Measuring Inflation”. It was overtly critical, both of RPI and of the continued use by Government of RPI for selected purposes, something that it described as “index shopping”. It described RPI as flawed and recommended that the UKSA should try and fix the issue with clothing prices. It recognised that s. 21 SRSA 2007 might require the UKSA to consult the Bank of England and obtain the consent of the Chancellor of the Exchequer, but recommended that the UKSA attempt to fix the issue with clothing prices and that if it requested the change, the Chancellor of the Exchequer should consent to it, saying:

“It is untenable for an official statistic, that is used widely, to continue to be published with flaws that are admitted openly.”

It further recommended that the UKSA and the Government should agree on a single measure of inflation for official use within the next five years and that:

“to prevent index shopping, in the interim, the Government should switch to CPI from RPI in all areas of present use that are not governed by private contracts.”

The chair of the UKSA, Sir David Norgrove, therefore wrote to the Bank of England on 26 February 2019 with a proposal to adopt the CPIH methodology for RPI. The Bank of England replied on 4 March 2019 that the proposal would constitute a fundamental detrimental change for the holders of index-linked gilts. That could scarcely have come as a surprise, and on the same day Sir David wrote to the Chancellor, Mr Philip Hammond. He said in that letter:

“We are in a position where we and the ONS, who are the producers of the UK’s consumer price statistics, are clear that the RPI is not a good measure of inflation…

[RPI’s] production and the legislation around it sits uncomfortably with our legal obligation to promote and safeguard of the quality of official statistics…

…publication of the RPI should cease. Better alternatives exist.”

And he suggested both replacing RPI in its entirety, which would mean primary legislation to repeal s. 21 SRSA 2007, and, in the alternative, changes to the methodology of RPI so as effectively to turn it into CPIH.

As already referred to, Mr Hammond did not in the event have an opportunity to reply before he left office and was replaced by Mr Javid. On 30 July 2019 Sir David wrote to Mr Javid reiterating his requests. On 4 September 2019 Mr Javid replied. He said:

“I recognise that there are flaws in RPI and that maintaining public trust in official statistics is important, but RPI is used widely across the economy. UKSA’s first proposal to end the publication of RPI will potentially be highly disruptive for the wide range of users of RPI. In turn this could be damaging to the economy and the public finances. As we have discussed, the next few months are a critical period for the UK as we ready ourselves to leave the EU on 31 October. Given the potential disruption for users of a change to RPI and the Government’s focus on Brexit, I am not minded to promote legislation that would remove the requirement for UKSA to produce and publish RPI.”

As to the second proposal to fix RPI by aligning its methodology with CPIH, Mr Javid said that he could see the statistical arguments for the proposal. He, however, recognised that his consent as Chancellor was required to change the calculation of RPI as proposed, because of its use as the reference index for some government debt. He continued:

“The requirement to seek my consent to certain changes to RPI expires in 2030. You have signalled that, while you cannot commit your successors, it is unlikely that the UKSA in 2030 will take a different view from your proposal to align RPI with CPIH. In coming to a decision I consider the integrity of the statistical system, the effect on the public finances and on the holders of specific index-linked gilts.”

After indicating some of the considerations he had had in mind, he said he was:

“... unable to consent to the introduction of the change you have proposed any earlier than February 2025, based on the information I have available.”

He then proposed a public consultation on whether the change should be made other than at 2030, and if so when between 2025 and 2030. That consultation was initially due to take place in the early part of this year but, I was told, has been put off, so that it is now due to take place in March or April of this year.

The current position can be seen from the statement on RPI issued by Sir David on behalf of the UKSA on 4 September 2019 which, having explained that the UKSA had written to the Chancellor of the Exchequer recommending that publication of the RPI be stopped at a point in the future, said that the UKSA:

“... have consistently urged all – in Government and the private sector – to stop using it. However, the RPI is unique as we need consent from the Chancellor to make certain changes such as the one we have proposed”

The statement explained that the Chancellor had not agreed to stop RPI, and Sir David stated:

“We regret that no change will occur before 2025.”

In those circumstances I fully accept that the reason, and the only reason, that the UKSA continues to compile, maintain and publish RPI is because of its statutory obligation to do so, and that were it to have a free hand it would cease its publication. Nevertheless, as I have already referred to, it is under that obligation; it does comply with it; RPI is still published; and it appears that it will still be published until at least 2025 and possibly until 2030.

The principles applicable to the construction of written instruments in general, and pension schemes in particular, have been expounded in numerous cases. I see no need to repeat them as they are very familiar. Mr Spink referred me in particular to two of the Supreme Court decisions, namely Wood v Capita Insurance Services Ltd [2017] UKSC 24, and specifically to [13] in the judgment of Lord Hodge JSC, and Barnardo’s v Buckinghamshire [2018] UKSC 55 at [13] to [15], [23] and [26], again in the judgment of Lord Hodge JSC, the latter identifying certain characteristics of pension schemes that bore upon their construction.

To that can be added the well known observation of Millett J, as he then was, in Re Courage Group’s Pension Schemes [1987] 1 WLR 495 at 505F, that the provisions of a pension scheme should be construed so as to give “reasonable and practical effect” to the scheme (and a similar statement by Warner J in Mettoy Pension Trustees Ltd v Evans [1990] 1 WLR 1587 at 1610DE), and the comments of Arden LJ, as she then was, in Stena Line Ltd v Merchant Navy Ratings Pension Fund Trustees Ltd [2011] EWCA Civ 543 at [29] to [35]. What I take from the latter is although what I am construing are the words as they appear in the Draft 2011 SBS Deed (that is a document of 30 June 2011), I can have regard to the fact that those words were first introduced into the SBS rules in 2007, and although they might bear a different meaning that is only likely to be the case if the context is materially different. Mr Spink confirmed that he was not here suggesting that they did bear a different meaning in 2011 from what they had borne in 2007.

I should however deal with one case on which Mr Spink placed considerable weight, Lloyds TSB Foundation for Scotland v Lloyds Banking Group Plc [2013] UKSC 3. Here the defender, Lloyds Bank, had covenanted in 1997 to pay a certain percentage of its pre-tax profits for each accounting reference period to the pursuer, a Scottish charity. “Pre-tax profit” was defined as the “group profit before taxation … shown in the audited accounts” subject to certain adjustments. A change in accounting practice instituted at EU level required that when Lloyds Bank acquired HBOS in 2009 it was obliged to book in the group consolidated income statement a figure for gain on acquisition of over £11 billion (thereby turning a loss of over £10 billion into a profit of over £1 billion) even though this was not a real or realised gain. (Footnote: ¹)

Lord Mance JSC held that the £11 billion unrealised gain was not part of Lloyds Bank’s pre-tax profits. It is not possible adequately to summarise his closely reasoned judgment, but the essence of it is that such a “gain”, which would never have been treated as profits under the accounting standards applicable in 1997 when the covenant was given, was not part of the group’s “profits” for the purpose of the covenant even though it appeared, as it now had to, in the 2009 accounts with that label. If I can put that in my own words, the decision is that what “profits” meant in the covenant was to be determined by what the parties would have understood profits to be in 1997, and did not include something that would never then have been regarded as a profit simply because a change in accounting practice required it to be included and labelled profits in 2009.

Mr Spink relied in particular on the following passages:

On the basis of those passages Mr Spink submitted that although RPI was properly described as the general index of retail prices in 2007 or 2011, it could no longer be so described today, and indeed had ceased to be capable of being so described in 2013, being replaced either by RPIJ or (his preferred answer) first CPI and subsequently CPIH. Picking up the language used by Lord Mance at [23] that the question is how the language of the covenant best operates “in the fundamentally changed and entirely unforeseen circumstances and in the light of the parties’ original intentions and purposes”, and similarly the language of Lord Clarke at [50] to the effect that the changes had been “unthinkable” and that the problem is how to construe the contract “in the context of changed circumstances which were unforeseeable when the contract was entered into”, he suggested that the changes in the status of RPI were so profound and extraordinary, and so unthinkable and unforeseeable, that the problem was how to construe the Definition in these changed circumstances. On that basis he suggested that because the draftsman can be assumed to have initially selected RPI as an accurate measure of price inflation, once it had become apparent that it was no longer regarded as such, the language of the Definition must be regarded as referring to some other index.

With all respect to Mr Spink this argument seems to me as unsound as it is ingenious. It is to my mind a good illustration of the danger of taking the words of a judgment out of context and trying to apply them to a quite different situation. In the Lloyds case the definition of pre-tax profits needed to be applied to each accounting year to identify the relevant figure for that year. The question was what the parties meant by “profits” in 1997. Did they mean whatever was shown as profits in the accounts for 2009, or did they mean profits as understood in 1997? The answer was the latter. That is not, as Mr Spink would have it, really a case of a definition that changes between 1997 and 2009. Properly analysed it is a case where the Supreme Court were reinforcing the proposition that what “profits” in the covenant meant in 2009 was what it had meant in 1997, and therefore did not include an unrealised gain that could never have been regarded as a profit in 1997.

Reverting to the Definition in the present case, for the reasons I gave at the outset I have no doubt at all that what the Definition referred to in 2011 and 2007 was RPI; what it referred to in 2013 and 2017 was RPI; what it refers to today is RPI; and so long as RPI continues to be published (that is so long as s. 21 SRSA 2007 continues in force and the UKSA complies with its statutory obligation under that section), will continue to be RPI. That is not because RPI is still the UKSA’s or ONS’s best or lead or preferred or main or most comprehensive measure of inflation – the evidence leaves it entirely clear that it is not – but simply because the words in the Definition mean the same today as what they meant in 2011. As a general principle, in English law a written instrument has a single meaning and that meaning is fixed when the instrument is executed, as indeed Mr Spink accepted. The purpose of a definition is to identify what the draftsperson meant by the defined term, in this case “Retail Prices Index”. Having identified that, that is what it means.

Mr Spink showed me a passage from Sir Kim Lewison’s book on the Interpretation of Contracts (6^th edn) at §5.1.5, referring to the difference between what is called a static and a mobile interpretation of a contract. I need not go into the detail: I accept that in some contracts words can be given a dynamic, or mobile, interpretation. Thus an example given by Sir Kim Lewison

In those circumstances I do not strictly need to answer the question whether RPI is still appropriately described as “the general level of retail prices (all items) published by the Office for National Statistics”, but for completeness I should make it clear that in my view it is. Mr Spink sought to invest the word “general” with some particular meaning. In written submissions he said that Atos contended that “general” connoted two related concepts, namely: (i) the index has to be of general application as opposed to a specialised variant; and (ii) the index must be produced and promulgated by the ONS as being suitable for use as a general measure of inflation. In oral submissions he referred to “the general index” as connoting the main or primary measure of consumer price inflation.

The difficulty I have with that is twofold. First, for reasons already given the question simply does not arise; but second, and in any event, the experts were agreed what the word “general” was doing in the definition. It was to distinguish the main or primary index in each family. In the RPI family that means RPI, as opposed to RPIJ or RPIX or any of the other variants or derivatives or sub-indices of RPI. In the CPI family it means CPI rather than CPIH. Since that is what the word is doing, I see no reason to make it do any more work than that. So the only question that arises is whether the family referred to is the RPI family or the CPI family. On that there seems to me no contest. I accept that both RPI and CPI were intended to measure consumer price inflation. What distinguishes them, however, is that one is labelled “Retail Prices Index” and one “Consumer Prices Index”. Here the draftsperson has referred to the

“index of retail prices”. Moreover, he or she has actually used as the term to be defined “Retail Prices Index”. The ordinary meaning of the term defined can assist in the interpretation of a definition: see Chartbrook v Persimmon Homes Ltd [2009] UKHL 38 at [17] per Lord Hoffmann. So even if one did need to keep asking whether this phrase still refers to RPI I think it obvious that it does.

That I think is sufficient to deal with question 1.

I can deal with Question 2 very briefly. As I read from the Claim Form, a large number of variants were put forward, although in oral submissions Mr Spink discarded some of them. His preferred answer was (vi) (“where the index is not published with National Statistics status”), followed by (iii) (“where the index is not published for the purpose it was at the time the scheme was established”), which he described as really another way of saying (vii) (“where the index is not published as an appropriate … measure of consumer price inflation by the ONS”). Mr Spink accepted that the other variants of (vii), and (iv), (v), (viii) and (ix) can be ignored.

A great deal of thought has gone into this but I do not understand why “published” bears any of these meanings. None of these words appear in the definition. The very first principle of construction is that ordinary English words should prima facie be given their ordinary English meaning: see Lewison at §5.01 where this is described as “the Golden Rule”. This is particularly, as we have seen, the case in pension schemes, where loyalty to the text is the first principle. I see absolutely no reason why “published” in the Definition should not mean published in the ordinary sense of being made public. That is what it evidently means in s. 21 SRSA 2007. I think it is also obvious that that is what it meant in s. 833 ICTA 1988, and means in s. 989 ITA 2007. In my judgment no reason has been given for it meaning anything different here. Apart from anything else that is how it would, it seems to me, strike the ordinary reader. It also makes for administrative convenience, being a simple and easy test to apply. Those are not the only reasons for adopting this construction, but they are reasons for not adopting any alternative and more complicated construction.

As to what had been Mr Stallworthy’s preferred answer, namely answer (i) (“where that index, although it has not ceased to exist, is not published in any given month or at any time required for its use under 2011 TDR”), I see no good reason to adopt that either. To adopt that answer would bring it into conflict with the provision that I read earlier from paragraph 9.1.3, found in schedule 3 and some of the other schedules.

Both sides thought that the final sentence of the definition, which I will read again (“where the retail prices index ceases to exist the Principal Employer and the Trustees may agree any substitute index published by that Office (or its successor)”), meant that some special meaning had to be given to the trigger condition of “where that index is not published”. But there are two reasons to reject that. First, that reasoning does not apply to the first iteration of this language in 2007 or in 2008, but there is no reason to think that when the draftsperson included identical words in the Draft 2011 SBS Deed he or she intended to change the meaning.

The second reason is that the point is in any event, to my mind, simply a bad one. Whatever meaning you give to “published” in the first sentence the final sentence is otiose. This is because if RPI ceases to exist it will cease to be published, whether that means temporarily or as a National Statistic, or as an appropriate measure of inflation, or simply published. On any view, therefore, one doesn’t need the final sentence at all because as soon as RPI ceases to exist it will cease to be published, and the trigger condition will be met. There is, to my mind, no answer to this and it means that the presence of the final sentence sheds no light at all on the meaning of “published”, nor requires it to be given anything other than its ordinary meaning.

In those circumstances my answer to Question 2 is in sense (ii), namely, “where that index is not published for any purpose”. Questions 3 and 4, as I have said, do not arise.

That is sufficient to dispose of the issues that need to be resolved.