The claimant is the Principal Employer in relation to the Arup UK Pension Scheme (the Scheme). In this Part 8 claim it seeks declarations as to the interpretation of the definition of "the Index" under the Rules of the Scheme, which is used for escalation of pensions in payment and various similar provisions, with the object of establishing that the defendants, who are the present Trustees of the Scheme, are obliged, or have the power, for the future either to change the relevant index from the Retail Prices Index (RPI) to the Consumer Prices Index (CPI) or Consumer Prices Index including Housing (CPIH), or, if they cannot change the Index itself, to make "adjustments" to calculations using RPI that would achieve the same effect.

The claimant's principal contention is that RPI has been "replaced" within the meaning of the Scheme's Rules by CPI (and subsequently by CPIH) notwithstanding that RPI continues to be published, because those indices, rather than RPI, became regarded as the main measure of consumer price inflation or of inflation for use by pension schemes. This is referred to by Mr Stallworthy on behalf of the claimant as the "functional" replacement of RPI.

The claim form seeks representation orders to the effect that the claimant is appointed to represent all those participating employers and members who may be interested in arguing for such declarations, while the Trustees represent all those interested in arguing the contrary. Such orders are now routinely made, and there is no reason not to do so in this case.

The Scheme was established in 1959. It was closed to future accrual in 2010, but still has a substantial number of members. The evidence at the hearing was that as at the last actuarial valuation date in 2016 the Scheme was in deficit on the "technical provisions" basis by some £232m, but that if the power contended for by the claimant exists and is exercised in respect of future increases in pension (it is not suggested it would be used to revisit increases awarded in the past) that deficit might be reduced by between £75-85m. This is because of the well known likelihood that annual inflation as measured by the CPI (or CPIH) will be of the order of 1% lower than as measured by RPI. The principal employer is of course expected to fund the deficit and so has an obvious interest in such a change being made. The corollary would be that pensioner members would be likely to receive lower increases in future than they would otherwise have done.

The provisions in issue before me were first introduced by a new Definitive Deed and Rules adopted in 1992. This was replaced, with few changes relevant for present purposes, by a revised Deed and Rules adopted in 1998. That in turn was replaced by the present Definitive Deed and Rules adopted in 2013 (the 2013 Rules). I refer to the relevant provisions by their numbering in the 2013 Rules.

Rule G3 provides for increases of pensions in payment as follows:

The Trustees must increase each pension in the course of payment annually in accordance with this Rule G3.01.

The increase must be made on the first day of the Scheme Year [i.e. 1 April] which next follows the date when the pension starts to be paid and on the first day of each subsequent Scheme Year.

Each pension as previously increased, must be increased by the lower of:

to the extent to which the pension is attributable to Pensionable Service completed before 1^st October 2006, five per cent per annum and, to the extent to which the pension is attributable to Pensionable Service completed on and after 1^st October 2006, 2.5 per cent per annum; and

the increase in the Index which has occurred over the period of 12 months ending at the end of the month which is three months before the start of the month in which the first day of that Scheme Year falls.

Where a pension comes into payment after the start of a Scheme Year, the increase to be made to that pension under Rule G3.01 on the first day of the following Scheme Year shall be adjusted to the extent that the Trustees so decide.

Where a person’s guaranteed minimum pension under the Scheme has come into payment at State Retirement Age, an increase shall be made under Rule G3.01 only on such part of that pension as exceeds his guaranteed minimum pension.”

"The Index" referred to is defined in Part V of the 2013 Rules as follows:

It is Rule H1.03 that the court is asked to construe. It provides:

The only material change in this wording since its introduction has been that the 1992 equivalent of R H1.03 provided that in the circumstances specified the Trustees

Counsel were agreed that the present law in relation to the principles of interpretation as applied to pension scheme documents is as set out by Lord Hodge in Barnardo's v Buckinghamshire [2018] UKSC 55:

A pension scheme, such as the one in issue on this appeal, has several distinctive characteristics which are relevant to the court’s selection of the appropriate interpretative tools. First, it is a formal legal document which has been prepared by skilled and specialist legal draftsmen. Secondly, unlike many commercial contracts, it is not the product of commercial negotiation between parties who may have conflicting interests and who may conclude their agreement under considerable pressure of time, leaving loose ends to be sorted out in future. Thirdly, it is an instrument which is designed to operate in the long term, defining people’s rights long after the economic and other circumstances, which existed at the time when it was signed, may have ceased to exist. Fourthly, the scheme confers important rights on parties, the members of the pension scheme, who were not parties to the instrument and who may have joined the scheme many years after it was initiated. Fifthly, members of a pension scheme may not have easy access to expert legal advice or be able readily to ascertain the circumstances which existed when the scheme was established.

Judges have recognised that these characteristics make it appropriate for the court to give weight to textual analysis, by concentrating on the words which the draftsman has chosen to use and by attaching less weight to the background factual matrix than might be appropriate in certain commercial contracts. In [Safeway Ltd v Newton [2018] Pens LR 2] , Lord Briggs stated (para 22):

The emphasis on textual analysis as an interpretative tool does not derogate from the need both to avoid undue technicality and to have regard to the practical consequences of any construction. Such an analysis does not involve literalism but includes a purposive construction when that is appropriate.

Counsel are further agreed that as the relevant provisions of the Rules have remained essentially unchanged since 1992 the starting point for any consideration of how surrounding circumstances would have affected the objective observer's understanding of the words used is the context in which they were employed in 1992. The court must then consider how that may have been affected by any material change in circumstances when the Rules were successively re- adopted, but in the present case neither counsel argued that there had been any subsequent change of circumstances that would affect the meaning of the relevant provisions.

I bear in mind also the often repeated judicial statements that in matters of interpretation, each document is to be interpreted in its own context, and decisions as to the meaning of words in one document are not of direct assistance in construing similar words in another document, unless they lay down general principles. This did not prevent both counsel from referring me to instances in which, in the numerous cases that have considered whether a similar change to CPI could be undertaken by

Rule H1.03 operates in two sets of circumstances, namely :

"If the composition of the Index changes…" or ii) [If] "the Index is replaced by another similar index…" which were referred to as "trigger events".

The amended Claim Form sets out 9 questions, with alternative answers, on which the claimant seeks a decision. They are divided into two groups, the first five focussing on whether RPI has been replaced and if so what are the powers or duties of the Trustees, and the remaining four on the question whether there has been a change in the "composition" of the RPI by virtue of various events (and it is agreed that there has been at least one such change of composition) and if so the nature and extent of the adjustments the Trustees may make to calculations as a consequence.

The issues raised by the 9 questions are interlinked, and I agree with Mr Simmonds that they boil down to two main issues, which I frame in slightly different terms, ie

Has either of the two trigger events occurred?

If so does the power to make adjustments authorise, or require, the Trustees either to base its calculations on CPI (or CPIH) instead of RPI, or to achieve the same effect by "adjusting" the change in RPI to correspond with the change in CPI (or CPIH)?

The Government began publishing a Cost of Living Index in 1914, intended as a measure of price changes in the main items purchased by working class households.

Following a review after the second world war, that index was discontinued and replaced in 1947 by the Interim Index of Retail Prices.

In 1956 that index in turn was discontinued and replaced by the Index of Retail Prices. That index was in due course renamed as the Retail Price Index or RPI.

All of these indices, and indeed all other similar indices, are subject to regular review and evolution of their methodology, in particular of the range of goods and services measured (the so-called "basket") and the weightings given to each price measured, but also in relation to their methodology, for instance in relation to the range of types of households surveyed in order to establish what prices to monitor and the weightings given. In the 1960s, variants of the RPI were produced for different user groups, including state pensioner indices introduced in 1969.

In 1979 a Tax and Prices Index was introduced, measuring changes in after tax income required to purchase the notional basket of goods, and in the 1980s a Rossi index began to be published, which excluded most housing costs and was used to uprate state benefits.

In 1992 the Government set its first explicit inflation target, which was based on RPIX, a variant of RPI that excluded mortgage interest payments.

By the time of adoption of the 1992 Rules there were therefore in addition to RPI four indices of prices that might have been adopted by the Scheme, ie

The state pensioner indices

RPIX

TPI

The Rossi index.

The Scheme was not bound by legislation or requirements of Inland Revenue approval to adopt RPI specifically as its reference index. Prior to April 2006 a scheme required to be approved by the Inland Revenue, which provided guidance on criteria for such approval (known as IR12) including limits on benefits and escalation of benefits after retirement. The latter provided that benefits should not be increased by more than the increase in RPI "or any other index agreed for use by a particular scheme by the Pension Schemes Office".

The CPI began to be published in 1997. It was then called HICP (Harmonised Index of Consumer Prices) and intended to meet obligations on member states of the EU under the Maastricht Treaty to produce harmonised statistics. It did not include owner-occupiers' housing costs.

In 2003 the Government's inflation target was amended so as to be based on CPI rather than RPI. It continued to use RPI as the basis for uprating pensions and indexation of state benefits, and for issues of index linked gilt stocks.

Following the Statistics and Registration Service Act 2007 responsibility for publishing indices such as RPI and CPI was transferred to what is now known as the UK Statistics Authority (UKSA). They are now administered by the Office for National Statistics (ONS) which is an agency of the UKSA.

That Act required the UKSA to continue to publish RPI and not make certain material changes to it save with the consent of the Chancellor of the Exchequer. This is explained by the fact that by that time the UK had issued some hundreds of billions of pounds of index linked bonds based on RPI, with dates extending far into the future, the terms of which might entitle the bond holders to immediate repayment and compensation if RPI ceased to be published or was changed in a way that materially adversely affected the interests of bondholders.

In June 2010 the government announced that it "will adopt the CPI for the indexation of benefits, tax credits and public service pensions from April 2011". From that date, for instance, CPI was used to determine the rate of annual escalation of public service pensions in payment specified in Pension Increase (Review) Orders.

In March 2013 a variant of RPI known as RPIJ was introduced alongside RPI. The "J" stands for Jevons, and represents the introduction of the Jevons formula for measuring changes in a basket of prices (as used in CPI) rather than the Carli method used by RPI. Differences between these methods were known to augment inflation as measured by RPI by around 0.5 percentage points per annum (referred to as the "formula effect").

Because RPI continued to use the Carli formula, which was no longer considered compliant with international standards, it was by announcement made on 14 March 2013 removed from designation as a "national statistic", notwithstanding its continued use for many government and private purposes.

At the same time the UKSA announced that it would effectively "freeze" further development of RPI's basic formulation and confine future changes to what were described as routine changes to baskets and weights and improvements to data collection and validation.

The claimant submits that for the purposes of the 2013 Rules RPI was replaced by CPI on that date, 14 March 2013, because from then on CPI was regarded by the ONS as "the main measure of inflation for the UK". Alternatively it had been so replaced by the government with effect from April 2011 because of its decision from that date to use CPI as its main measure for indexing state pensions and other benefits (witness statement of Mr Reece, para 37.1).

CPIH was also introduced in March 2013, as a variant of CPI but including a measure of owner-occupiers' housing costs. It was designated a "national statistic" in November 2013, but because of disagreement as to how housing costs should be taken in to account, that status was suspended in August 2014 and not reinstated until 31 July 2017.

In 2013 the ONS published a document entitled "Consumer Price Indices- a brief guide". This was a non- technical document designed for lay readers to "[describe] the main consumer price indices in the UK and [explain] how they are put together". In a brief section of explanation of what was meant by consumer price inflation it said "In the UK, the main measure of inflation is

The Johnson report recommended that the ONS should move towards making CPIH its main measure of inflation, but in the meantime continue to use CPI as its main measure. It should make clear that RPI was no longer regarded as a credible measure of inflation and government should work towards ending use of RPI wherever practicable.

In the period since the Johnson report government has continued to expand the number of purposes for which it uses CPI or CPIH as a measure in place of RPI. The claimant refers to a number of instances in which RPI is described as a "legacy" measure or used for legacy purposes.

The claimant submits that if CPI replaced RPI in March 2013 by virtue of the actions of the ONS (as distinct from government generally) CPI was by the same means itself replaced by CPIH with effect from 21 March 2017. That date is based on an ONS Information Paper entitled "Quality and Methodology Information" in which CPIH was referred to as its "preferred" or "lead" measure of inflation.

Because the questions posed in the claim form are interlinked, I do not propose to set them out and answer them entirely sequentially. But the first asks "By whom is replacement effected" for the purpose of the 2013 Rules and whether for instance (implying there may be yet other possibilities) it is by the Trustees, the Principal Employer, the authority responsible for publishing the Index (ie presently the ONS) or by the Government more generally.

The parties are agreed that R H1.03 does not contemplate "replacement" by either the Trustees or Principal Employer acting separately, or by both acting together. It is in a section of the Rules allowing the Trustees to make limited changes to the calculation of benefits without the consent of the Principal Employer, but it would be very unlikely that the parties to the creation of the Scheme contemplated a unilateral decision by either Trustees or Employer simply to adopt a different index in place of RPI. Such a change could have significant implications for the cost of funding the Scheme and the value of the benefits it provides, which neither party would be willing to have subject to the unilateral decision of the other. There are other detailed provisions for alteration of the Rules in general, so there would be no reason for this provision to introduce, by implication only, a separate mechanism for joint determination adopting a different index.

By contrast, the fact that the "change in composition" limb of the Rule plainly contemplates something that can only be done by the body responsible for producing the Index clearly suggests that the "replacement" limb refers also to action by that same body, ie (presently) the ONS. Thus, both parties agree that replacement must be by action of the ONS. I agree. I have not therefore considered further the submissions related to replacement occurring by government more generally.

The second question asks whether "replacement" requires that RPI ceases to be published, or whether it may be replaced "functionally" while still continuing to be published.

In support of his argument, Mr Stallworthy submits that dictionary definitions of "replace" or "replacement" do not require that the thing replaced ceases to exist; thus the Concise Oxford Dictionary definition of "replace" in its 1990 edition included "1. Put back in place. 2. Take the place of, succeed, be substituted for…". A new thing might be put in the place of an old thing without the old thing having to cease to exist; he gave the example of replacing a bulb in a desk lamp, the old bulb does not thereby cease to exist although its place in the lamp has been taken.

Mr Stallworthy refers also to instances in which various persons or bodies have referred to CPI as replacing RPI despite the latter remaining in existence, such as a Financial Reporting Council document issued in December 2010 entitled "Accounting Implications of the replacement of the RPI with the CPI for Retirement Benefits". He points to instances in judgments in the many cases already brought on similar questions in which he submits judges have referred to, for instance "discontinuance or replacement" of RPI, implying that they did not consider replacement inevitably to involve discontinuance.

Acknowledging that there were two instances prior to 1992 in which an inflation index had been discontinued and replaced, he submitted that this did not compel the conclusion that discontinuance was always necessary. All of the references to the Index in the 2013 Rules were capable of a construction including circumstances in which the previous Index continued to exist. Even if an index was discontinued, some consideration of functionality would be necessary to identify which of the potentially multiple indices then introduced or remaining was to be regarded as a "similar" index "replacing" the old one. That consideration, he said, could only be done by the publishing authority. If for instance RPI had simply been discontinued in 1992 there were four other potential candidates that might be regarded as similar and therefore as potential replacements.

What is important, Mr Stallworthy submits, is to identify the "place" occupied by the old index, from which it can then be identified which of any potential candidate indices has been put in to that place, so "replacing" the old index. It is obvious, he submits, that RPI was selected when the first Rules were drafted in 1992 because it was at that date the UK's main measure of inflation. Its "place" therefore was as the index so regarded, and when in 2013 the ONS announced that in future CPI would be regarded as that main measure, CPI thereby took that place and so "replaced" RPI.

I would agree that, considered as a matter of language, the concept of "replacement" of one thing by another requires consideration of the "place" that is occupied. But the flaw in Mr Stallworthy's argument, as it seems to me, is that it fails to follow the logic of the fact that, as he concedes, "replacement" of the Index is an act to be done by the producer of the index, and not by a person who uses it for a particular purpose.

Consider for example the example of a model of car. One may say that a particular model (say the Ford Escort) is "replaced" by the manufacturer with another (the Ford Focus). That usage implies that production of the Escort is discontinued, and instead the manufacturer produces the Focus, which it regards as targeted at the same market

There could also be "replacement" of Ford Escorts by a user. Suppose for instance a fleet manager had decided that from a point in time he would purchase Vauxhall Astras instead of Ford Escorts. It could be said that he had "replaced" the Escort with the Astra as his fleet car, even if the Escort as a model continued to be manufactured (and although, presumably, the particular cars the user had previously purchased continued to exist).

The point is that where replacement is considered from the perspective of a user of a thing, the relevant "place" is the use he makes of that thing, and he may put a new thing in that place without necessarily destroying the old. But when considered from the perspective of the producer of the thing, the "place" is the producer's place in his range of goods or functions, and replacement tends to imply discontinuance of production of the old thing.

Mr Stallworthy's other examples, it seems to me, are all examples taken from the perspective of a user, of replacement for a particular use or purpose of that user- thus the FRC document he referred to addresses replacement of RPI "for retirement benefits", ie benefits provided by the state, not for all purposes, and the instance he refers to of the 1992 Rules "replacing" the then existing rules of the scheme as regards accrual after April 1988 but preserving them in respect of earlier service merely shows that the term is used in the context of that identified purpose.

The hypothetical desk-lamp owner might well say that he had "replaced" his incandescent bulb with an LED bulb. The old bulb, and incandescent bulbs generally, could still be available for use. He would be speaking as a user of the bulb. But if he said that the manufacturer had replaced incandescent bulbs with LED bulbs, he would be taken as meaning that the manufacturer was no longer making or selling incandescent bulbs, but LED ones instead.

I do not go so far as to say that as a matter of language "replacement" by a producer could only ever happen in circumstances of discontinuance, but only that that is what is strongly indicated. One must look at the context (being the context in 1992) to determine whether the draftsman of these Rules is likely nevertheless to have intended another meaning.

Such considerations in my judgment also point against the claimant's contention. First, in 1992 it would have been known to the parties that it could not be assumed that the RPI would continue to be produced indefinitely, not least because there had been two previous occasions on which its predecessors had been discontinued and replaced. The most obvious contingency at which the "replacement" limb would be aimed was a further similar occasion.

As Mr Stallworthy himself says in his skeleton, "no doubt the parties would not have envisaged the present circumstances, in which RPI continues to be published despite

It is in my view a further indication against the notion of functional replacement of an index that continues to exist that there were other indices in existence in 1992 that could, in principle, have been chosen but were not. It indicates that as between those potentially available (five, on the claimant's evidence) the choice of RPI was made by the parties themselves, and it was not described in terms of any assessment of RPI's relative suitability or particular purpose by the body or bodies producing those indices. This does not suggest that they intended that if a sixth index began to be produced, and was described by the producer as better or more suitable in some respect than one or more of the previous five, that index would thereby automatically become the Index for purposes of the Scheme. Any differences between the new Index and RPI would inevitably have implications for the funding of the Scheme and members' benefits, and switching to it when RPI was still available would be much more likely to be regarded as a change to the Rules to be achieved by amendment in accordance with those Rules.

There are other reasons why the objective observer in 1992 would not have concluded that the parties intended a concept of functional replacement of RPI:

There is nothing in the 1992 Deed and Rules (or any of its successors) referring in any way to the function or purpose of RPI as compared with any other index actually or potentially available, and certainly no reference to it being the "main measure of inflation in the UK" which is the function now relied on.

I do not agree with Mr Stallworthy that it is "obvious" that RPI must have been selected as it was then the UK's main measure of inflation. There is no evidence either in the Deed and Rules or extrinsically of the reasons why RPI was chosen. One may speculate that it could have been because the parties themselves at the time considered it to be the UK's main measure (as distinct from the government or publishing body having identified it as such), or that they considered it to be the index that was most widely used in practice by private pension schemes (not the same thing) or that it would be the easiest to justify when seeking Revenue approval, or that irrespective of the views of government or other pension providers they considered it was the best available balance between the interests of members and the funding employer, or simply that it was as good as any other option but the one most likely to be available, in one form or another, for the long term over which the Scheme might be expected to continue. There could be many other considerations, no doubt. One cannot assume any of them to have been decisive in the absence of admissible evidence.

There is no evidence that in 1992 there was any concept, formal or even informal, of what was the main measure of inflation in the UK. I have not been referred to any contemporaneous statement that there was such a measure,

The concept of a main or preferred measure of consumer inflation seems to have emerged subsequently, in light of the dissatisfaction of professional statisticians with the technical basis of compilation of RPI and the development of international standards with which it did not comply. That concept has however never been given any formal status, and has always been a qualitative assessment by the ONS as to which measure it regarded as more satisfactory for statistical purposes, and/or which measure was now most widely used by government as a user of statistics for its own purposes.

This lack of formal status is made clear by the disparate range of sources relied on by the claimant as showing that CPI (or CPIH) has achieved "main measure" quality. For instance it selects 14 March 2013 as the suggested replacement date because on that day RPI was removed from designation as a national statistic, but the concept of national statistics does not appear to have existed, even informally, until 2000 at the earliest (witness statement of Mr Reece para 25) and had no official status until the 2007 Act came into force. The principal document relied on as announcing that the ONS regarded CPI as its main measure is not any formal statement by it but an informal document produced in March 2013 aimed at explanation of price statistics to lay persons. Thereafter there are various documents explaining why the ONS regarded CPI as technically superior, such as a letter from the National Statistician in 2016 stating that he does not regard RPI as a good measure of inflation, but none of these, it seems to me, creates any formal position of a "main measure", still less indicates that such a position existed or would have been likely to be in contemplation in 1992. The date selected for the putative replacement of CPI by CPIH is based on similarly tangential references by the ONS to CPIH as its "preferred" or "lead" measure (which in any event is a change in terminology) in two documents published in March 2017, "Guide to changes in consumer price statistics" and "Quality and Methodology information", neither of which could reasonably be taken as an official announcement of a particular status.

In 1992 then, the objective observer contemplating the possibility of another index being created alongside RPI but regarded as "the main measure" in its place, would have concluded:

There was no precedent for production of two indices of general price inflation operating in parallel and covering the same or similar ground and no reason to think it would be likely to happen in future (it has not been suggested that the reasonable observer would have been aware of the future requirement to produce harmonised statistics and that it would be impossible in practice to discontinue RPI),

If two such indices were produced, it would be impossible to foresee how they would be differentiated, whether one would come to be regarded as the "main measure" of inflation, by whom or for what purposes, and

It could not be anticipated how any such "main measure" status would be determined, with the degree of certainty required for operation of the Scheme.

If such a concept were considered at all, and given these uncertainties, it is likely it would have been dealt with by detailed provisions and definitions in order to achieve precision, as other contingencies were in this Scheme's documentation and that of other professionally drafted schemes. It would I consider be most unlikely that professional draftsmen would have intended to provide for such an eventuality only by the chain of inference from what is omitted that the claimant is now compelled to rely on.

For these reasons, then, I conclude that on the true construction of R H1.03 RPI is "replaced" only if it is discontinued and another similar index is introduced or declared by the responsible body to be in its place, and that the Rule does not contemplate any form of "functional" replacement.

The third question asks how "replacement" is to be recognised. If as I have held replacement occurs only if RPI is discontinued, this is unlikely in practice to present a difficulty. It is not likely that the UK would abandon entirely production of an index of consumer price inflation, and the parties could reasonably have expected that if RPI were discontinued, the responsible body or bodies would identify at the time which index was regarded as its replacement. In the unlikely event that they did not, there would have to be an application to the court or an amendment to the Rules.

The fact that the question has to be asked in the postulated circumstance that RPI might be replaced without being discontinued, and that for the reasons given above it could not be foreseen in 1992 how it would be answered, reinforces the conclusion that the parties did not intend to provide for that eventuality.

The fourth question asks whether, if RPI could be functionally replaced in the manner contended for, on the facts it has been so replaced. Given the conclusions I have reached, that does not arise.

The fifth question is also framed in terms conditional on my having found that RPI can be, and has been, functionally replaced. It asks whether in such circumstances the Trustees are obliged, or merely permitted, to use the replacement index as the Index, and whether any adjustments may be made to calculations starting from either such index.

In the circumstances this question too does not arise. I would only observe that I agree with both counsel that however the Index is replaced, R H1.03 necessarily implies that the Trustees must thereafter base their calculations on the replacement index, subject to the power to make adjustments.

The remaining four questions address what amounts to a change in the composition of the Index, and what the powers of the Trustees are in relation to making adjustments in consequence of any such change. For convenience I repeat the relevant wording from R H1.03:

Question 6 asks, in effect, how far back in time the Trustees may look to find a change in composition on the basis of which they may act, and poses possible answers ranging from the date of inception of the Scheme in 1959 to the date on which the Index was last used to escalate (or limit) benefits. Question 7 asks whether the "composition" of the Index means only the notional basket of goods used to compile it, or includes any other aspects of its methodology.

The claimant's preferred solution on the timing issue is that the Trustees may look back to the last time they undertook an informed deliberation and reached a conclusion about whether the composition had been changed. If they determine it to be appropriate they may then make adjustments to calculations at any time thereafter, so not being limited to adjustments for one year but potentially indefinitely into the future. That preference is I think prompted by the fact that the Deed and Rules do not require the Trustees to undertake any regular review of whether changes have been made to the Index or if so whether they should make any adjustment in consequence, and the Trustees have not as a matter of practice actually done so. If so, the suggested test would not in the circumstances of this case impose any limit on how far back the Trustees could now look.

The defendants' contention is that the power is only exercisable at any pension increase date by reference to changes made in the 12 month period over which the indexation increase is to be calculated- ie, in broad terms the period from 15 months to 3 months before the relevant pensions increase date. The provision is thus for a once off adjustment to counteract, as Mr Simmonds put it, any spike or trough attributable to the Index switching from one basis to another during that period, and thereafter calculations must follow the index as amended.

No time limit is expressly imposed by the wording of the Rule, so the question is whether one can be inferred from the context. In this respect, the relevant context includes of course not only the other provisions of the Rules, but also considerations of what is likely to have been intended as necessary or appropriate for the proper administration of the Scheme, ie in the words of Millett J a construction "to give reasonable and practical effect to the scheme".

As I have said above, it is clear that RPI and all other similar indices are not static constructs. They evolve over time, not only in relation to the content of the "basket of goods" (which in fact includes prices of services and other items of expenditure) but also as to the weighting given to different prices, the methods by which prices are collected and the statistical methods used to measure changes in prices and compute

In 2010 the ONS made changes to the way it collected data relating to changes in prices of clothing and shoes (see the Johnson report, para 3.8 and Annexe B.3). These apparently related to the way prices were compared of goods which were "on sale" at one date but not at another. At the time these changes were considered to be routine methodological improvements. It emerged however that they had the unintended consequence of magnifying the "formula effect", ie the additional apparent rise in prices caused by use of the Carli method (in the RPI) as compared with the Jevons method (in CPI). In consequence of these changes the formula effect was estimated to have increased from around 0.5 percentage points to around 0.9 percentage points pa. This led to an urgent investigation of the suitability of the Carli method, resulting in the ONS concluding that so long as RPI continued to use that method it could not be considered to comply with current international standards. Proposals to amend it to adopt the Jevons method foundered on the obstacle that this would arguably trigger repayment of the government's RPI linked bonds, and led to the somewhat unsatisfactory position that the separate "RPIJ" index was produced.

The decision in 2013 to "freeze" technical development of RPI and thereafter make only routine changes to matters such as basket contents, weighting and data validation.

In 2017 the ONS implemented a change to the way housing prices were included in the RPI, by switching from using data produced by a separate Housing Prices Index (HPI) that had been maintained specifically for the purposes of RPI and adopting instead the data from a separate index called UK HPI, which it regarded as technically more satisfactory (see witness statement of Mr Reece at para 39ff). The Bank of England assessed this as amounting to a "fundamental" change to the RPI, by reason of differences in coverage between HPI and UK HPI, including the fact that, unlike HPI, UK HPI did not include data from Northern Ireland but did include data from households in the top 4% by income and those of pensioners dependant on state benefits, which were deemed to be outside RPI's "reference population".

The last of these is accepted by the Trustees to amount to a change of composition. Mr Simmonds submits that it is not necessary to determine whether the others do or do not, because if he is right on the timing issue they are both out of scope. If he is wrong on the timing issue, the power to adjust exists by virtue of the changes in housing content in 2017 and it does not matter whether it would also arise by virtue of the other two contended changes (or any others). As will be seen, I would not accept that submission.

In approaching these questions, in my view one must step back and consider what purpose can be found or inferred for the existence of the provision to adjust index calculations. In 1992 (and indeed at all stages before and after) it seems to me that the parties must be taken to have known that the index they were choosing was not static, and that myriad changes were likely to be made to it continuously as the publishing

Further, the parties would have known that RPI was not, and probably no index could ever be, regarded as a perfect measure of inflation (itself a concept capable of multiple definitions). RPI was said to have been selected when there were four other potential options available, but it is of course well known that the UK authorities produce far more statistical indices than that. Some could no doubt be immediately seen to be unrelated to considerations of, in general terms, maintaining the value of pensions in payment, such as indices of wholesale prices or particular categories of goods. But others could well be argued to be potentially relevant, such as indices of average earnings. Such an index, for instance, forms part of the "triple lock" arrangement by which the government reviews state pension benefits.

In choosing RPI (or any other index) then, it seems to me the parties must be taken to have accepted these imperfections and the propensity of the chosen index to change. It could not realistically have been the intention to make adjustments to reverse every change that might be made to RPI in future; to do so would amount to the Trustees having in effect to maintain their own version of RPI exactly as it existed in (say) 1992. That would be impossible on many levels; at the most basic because it would involve the Trustees having to collect their own price data, following the 1992 RPI methodology, for items that were no longer included in the current RPI basket.

It is I think made clear from the fact that the power is to make adjustments that are considered fair and reasonable that the provision is aimed at circumstances in which the Trustees consider that, for one reason or another relating to a "change in composition" it would not be fair or reasonable simply to use the unadjusted movement in the index over the reference period. Use of an index is a balance between the interests of members, for whom the real value of a fixed amount of pension would decline over time, and those of the employer that is expected to fund the cost of benefits. Any change in the way the index was compiled that would have an unusual and significant effect on the end result might in principle have this result. A change to the index that, for instance, artificially increased the apparent rate of inflation in one year might have the effect of locking in the amount of that increase for all future years when in reality it was only a statistical effect, and amount to a windfall to members at a substantial and continuing cost to the employer, and vice versa.

The most obvious example of how this might arise is no doubt a one-off impact of a change made, resulting in a substantial and abnormal peak or trough in a particular period that would make it unsatisfactory to compare the index values at the beginning and end of that period. Not every change would be likely to justify an adjustment of course- it is in the nature of an index that is subject to amendment that there will frequently be cases in which the change between two dates involves, to some extent, not comparing like with like. But there may well be some changes of such a magnitude that the Trustees would consider that it would be fair and reasonable to make an adjustment.

But I would not limit the scope of the power to changes of that nature, or necessarily to making only once-off adjustments. In principle, it seems to me, there might be such a significant change in the way the index was compiled that the Trustees might consider that it no longer properly reflected what was required for scheme purposes, such that adjustments over a longer term would be fair and reasonable. Suppose, hypothetically, that the ONS decided that RPI should no longer include housing costs at all. The Trustees might (it would be a matter for them) conclude that this change had such an effect that it would be fair and reasonable to counteract it. They might in principle depending on the circumstances do so by deciding to make a once-off adjustment for the period in which the change occurred and thereafter following the annual change in the index in its new form. Alternatively they might take the view that excluding housing costs meant the index was likely to increase in future at a faster rate (or a slower rate) in future and, if considered to be practicable, seek to make adjustments year on year to reflect more closely what the movement in the index would have been but for the change made.

Making adjustments other than on a once-off basis would of course be fraught with practical difficulties. An adjustment to reintroduce housing costs and reflect future changes in such costs, for instance, would require an identifiable source of data on which to act. But I do not consider that would rule out such adjustments as a matter of principle- the Trustees might for instance be able to identify another index measuring changes in housing costs and, with the advice of the Actuary, an appropriate method of combining that with RPI to produce an adjusted index figure.

If the provision is, as I would hold, intended to serve this function, it sheds light on what is meant by "composition" and a change of composition. Mr Stallworthy refers me again to the Concise Oxford English Dictionary definition from the 1990 edition, including "the act of putting together, formation or construction… the constitution of … a mixture; the nature of its ingredients…". This he submits is in wide terms and it is unlikely that the draftsman would have intended to limit the term to the makeup of the basket of goods.

I agree; to achieve the purpose I have found the important point is that a substantial change must be produced in the end result, ie in the index figure calculated or the way in which that index figure is likely to change over time. Such an effect may come not just from changing the content of the basket of prices measured (which is not limited to prices of goods in any event) but by any other change in the methodology used, including changes in weighting, methods of data collection (including the geographical or demographic scope of data collected) or statistical techniques used to compile the data into the index figure.

The parties must therefore, in my judgment, be taken to have intended a wide meaning to be given to "composition" in order that the Trustees could use the power given for the purpose intended. The Trustees are required to exercise their judgment as to which if any changes justify the making of adjustments, and their discretion as to what adjustments it is fair and reasonable to make.

It is also apparent from the evidence and examples canvassed of candidate changes in composition that it is not always that case that when a change is made in the way the index is compiled (a) that the effects of the change on the index result become apparent immediately, or (b) that those effects are accurately anticipated by the ONS, or (c) that the fact that changes have been made is necessarily brought to the attention of persons such as the Trustees or (d) that as and when those effects emerge, they and their potential significance are immediately appreciated by persons such as the Trustees or even professional advisers such as the Scheme's Actuary. This is not surprising, given the complexites of compilation of an index such as RPI and of the economy it seeks to measure.

In giving a power of this nature to the Trustees therefore, it seems to me the parties must be taken to have been aware of these matters, and accordingly in the absence of specific provision it would be unlikely that they intended that there be any particular time horizon for the exercise of the power to adjust, after a change has been made. Accordingly, while I can see the attraction from the point of view of clarity and certainty of the one off opportunity to adjust argued for by Mr Simmonds, which might have led a draftsman to include an express provision to that effect, I do not consider that in the absence of such express provision it can be justifiably inferred from either the language of the Rule or its purpose.

The words "if the composition if the Index changes" imply of course a change happening at a time in the future, so the material question is, from which point does that future start?

It does not seem to me that this can be a date earlier than the adoption of the present Rules. No doubt those Rules must be interpreted so that the words have the same meaning as the corresponding words did in 1992, but that does not mean that when the 2013 Rules speak of matters in the future they include references to the past. That would require the Rule to be read as if it said "if the composition of the Index has changed since 1992" which would be unjustifiable alteration to its language.

In my judgment therefore each adoption of new Rules in effect resets the clock; the Index as defined is the RPI as it exists at the date of adoption of the new Rules (12 April 2013) and the question whether its composition changes is considered by reference to changes made after that date.

Nor do I consider that it would be right in principle for the clock to be effectively reset as Mr Stallworthy suggests if and whenever the Trustees reach what he refers to as a considered decision whether or not to make any adjustments. First, there is no foundation for that in the Rules. Second, I doubt if in practice it would achieve any real degree of certainty- it could operate only when the Trustees had considered a specific change or changes, and raises the question what would amount to a "considered" decision. Would they have to commission advice on the likely impact of each change? Presumably so. Third, if they took advice but it later emerged that the effect of a change was significantly greater than had been envisaged, they would have fettered their discretion to act in the future.

Mr Stallworthy objects that an interpretation that introduces a fixed starting date when the Rules are adopted (a) prevents any action being taken in relation to a change in the

It is true that if the administration of the scheme is seen as a continuum, resetting the clock in this way when new rules are adopted means that an opportunity may be lost to make adjustments in respect of (most likely) changes made since the last date the index was applied under the preceding Rules. But it seems to me preferable to infer that in adopting the new Rules the parties must have satisfied themselves, to the extent that they wished to, that the index chosen was satisfactory for their purposes as it then stood, and to have accepted any risk that it might have some recently introduced change that they were not aware of. That risk would be present even if Rules referring to RPI were being adopted for the first time. The only alternative potentially available from the language, it seems to me, is to construe the Rules as permitting reference back to 1992, which would exacerbate Mr Stallworthy's second objection.

As to that second point, I do not see it as in practice creating such potential problems that the objective observer would conclude the parties could not have intended this meaning. In practice, it would be expected that changes likely to produce a substantial effect that the Trustees might wish to take in to account would emerge and become known about relatively quickly. If the Trustees in fact consider a change and conclude that an adjustment in consequence is not justified, it would be relatively unlikely, though not impossible, that they would come to a different conclusion at a later date. In the relatively unlikely circumstances that the existence of a change, or the magnitude of its consequences, are not appreciated until a later date, it does not seem unreasonable that the Trustees should in principle retain power to deal with it.

It is however clear from the wording of the Rule that there must be a "change" in the composition of the index, and it follows that the decision in 2013 to "freeze" the development of RPI could never have amounted to such a change. It was no doubt a change in the approach the ONS would take to development of the index in the future, but it did not immediately change the way RPI was compiled and to the extent it meant that thereafter changes that might otherwise have been made were not, it had the result that changes in composition were not made, rather than that they were.

The conclusions I reach on questions 6 and 7 are therefore that

The Trustees may in principle act upon any change in composition of the Index occurring after 12 April 2013 when the present Deed and Rules were adopted,

A change in composition of RPI is not limited to a change in the items in the basket of prices but extends to any change in its data or methodology,

The change in 2010 in the way in which price data for clothing and footwear were collected and analysed was capable of being such a change in composition, but it is now too late for the Trustees to act on it because it predated the adoption of the present Rules,

The "freeze" in development of RPI announced in March 2013 could never have amounted to such a change, and in any event it would now be too late for the Trustees to act on it, and

The change in 2017 in the way housing cost data were incorporated is, as has been conceded, a change in composition giving the Trustees power in principle to make adjustments, and it is not too late, in principle, for it to do so.

Question 8 asks whether, in the circumstances as they now stand, there has been a change in composition such that the Trustees may now make adjustments. Since I was not asked to make a determination on any set of facts other than the three examples dealt with under question 7, it does not seem to me that question 8 adds anything.

The most substantive issue in this section of the questions arises under the last of them, question 9 which asks what is the nature of the adjustments the Trustees may make and in particular whether they can make adjustments so as to reflect the increase in an index other than RPI.

It is in relation to this question that Mr Stallworthy contends that the Trustees have power to make adjustments that would in effect replace the percentage change in RPI over a given period with the change in CPI (or CPIH) over the same period.

Although the Rule provides a power to "make such adjustments to any calculations using the Index… as [the Trustees] consider fair and reasonable" Mr Stallworthy submits that this merely identifies the calculations to be adjusted and does not constrain the adjustments that may be made. The power should be given the widest ambit, he submits. If a change to the composition of RPI "reveals or emphasises flaws" in RPI, a change in calculations "to reflect a more appropriate index" is consequential upon the change in composition and still amounts to an adjustment to calculations using RPI. The Trustees may consider such a change fair and reasonable "in the light of the changes in composition of RPI and what those changes reveal or emphasise about the flaws in RPI." A change to reflect movements in CPIH would be consistent with the future proposals of the ONS in relation to RPI itself, since the ONS has now announced an intention to align the methodology of RPI with that of CPIH, though this requires the consent of the Chancellor of the Exchequer (and on the evidence appears unlikely to happen before at least 2025).

The starting point in my judgment is that, wide as I consider the discretion of the Trustees to be, the adjustments it makes must (a) be adjustments to calculations made using the present Index and (b) be logically related to the effects of the changes in composition of that index that give rise to them and designed to counter, mitigate or allow for the effects of that change. Given that, as both parties agree, the evolving nature of such indices means that multiple changes will inevitably and frequently be made, most of them individually having only a minor effect on the result, it would not be a reasonable construction of such a provision that it provided merely a threshold condition that some change in composition had been made, however minor and however insignificant its effect on the results of computation of the Index, upon which the Trustees had an unrestrained power to make any changes they wished to the index calculations.

Nor would it be any more reasonable to suggest that any change of any nature gave rise to a power in effect to adopt a different index that the Trustees considered generally preferable. Both parties agreed that one reason why "replacement" of the existing Index must be something done by the publishing authority and not by the Trustees (or employer) acting unilaterally was that the choice of Index had significant effects for members and for the employer as funder, such that it would be most unlikely that they would have intended to give a unilateral power to either of them to make such a change without concurrence of the other. They would equally not have intended the same effect to be achieved under the guise of a power of "adjustment".

It is no answer to say that a change in composition may reveal or emphasise flaws in RPI. The Trustees have no power to switch away from RPI merely because they consider it to be flawed, so it is irrelevant how they come to be aware of flaws or have an emphasised appreciation of such flaws. A change to the composition of RPI that (ex hypothesi) does not correct any perceived flaws cannot reasonably be relied on to grant a power to switch away from RPI (or effectively do so) in order to escape the flaws it does not address.

I do not say that there are no conceivable circumstances in which the Trustees could base any adjustments on movement in another index. Suppose again the hypothetical example I gave above of a change to RPI that excluded housing costs altogether. In such a case, the ONS might perhaps create a second index (called, say, RPI+H) that continued the former methodology of RPI, in order that users could still have access to an index that was in effect the old RPI. In such a case, the Trustees might in principle decide:

That no adjustment was necessary, or

That it would be fair and reasonable to make a one off adjustment for the transitional effect of the change to prevent "locking in" of such an effect, but thereafter no adjustment was necessary, or

That the exclusion of housing costs in future created a risk of long term divergence of the modified RPI from what had been envisaged when it had been adopted, which it would be fair and reasonable to counteract by adjusting RPI calculations to re-incorporate those costs, and that the most practicable way of doing that was in effect to follow RPI+H.

If the last course were adopted, that in my view could be potentially justified as a measure related to the effects of the triggering change in composition.

But that is far removed from what the claimant seeks to achieve here. The one matter that is agreed to amount to a change in composition involves a change in the way housing cost data is compiled into RPI, but the adjustments the claimant wishes to propose have little or nothing to do with reversing mitigating or allowing for that change. There is no readily available alternative index or source of information that would enable the Trustees to do so, even if they considered it desirable. Rather, the claimant simply wishes to use that change as a peg on which to hang a decision to replicate movements in CPI or CPIH instead of RPI.

It is not suggested that those indices use housing cost data that would restore the way such data were included in RPI (I assume they do not, since the announced reason for the change to RPI was a wish to cease collecting housing cost data in the form previously used by RPI). Nor are CPI or CPIH put forward on the basis that they are preferable because of the way they treat housing costs, as compared to RPI as revised in that respect. Their advantage, from the claimant's perspective, is that mainly because of the formula effect , their rate of increase over the long term is considered likely to continue to be significantly lower than that of RPI, resulting in a lower funding requirement.

A change for that reason would not be for the purpose of dealing with the effects of the change in composition of RPI but with the perceived flaws in RPI that exist irrespective of that change. It would not therefore be a proper exercise of the Trustees' power of adjustment.

I answer question 9 therefore that the nature of the adjustments the Trustees may make consequential on a change in composition of the Index is such as the Trustees in their discretion consider to be fair and reasonable for the purpose of counteracting, mitigating or allowing for the effect of that change on calculations using the Index, and that in principle they may take account of short term or long term effects, and do so either by a once-off adjustment or on a recurring basis. However, the admitted change in composition in relation to housing cost data in 2017 could not be a proper justification for switching from RPI to another index. Nor, in my view could either of the other two matters that were canvassed as being changes have been, even if I had found they were in principle available as a basis for exercising the power to adjust.

My answers to the questions posed by the claim form are contained in the judgment above.

This judgment will be deemed handed down remotely by email to the parties and release to BAILII. I invite the parties to agree the order resulting. If there are any matters that cannot be agreed, the parties should file brief written submissions two days before the handing down date and I will if possible deal with them without a hearing.