England and Wales Court of Appeal (Civil Division) Decisions

IN THE COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM THE HIGH COURT OF JUSTICE, CHANCERY DIVISION
Mr Justice Warren
HC13C05161

		Royal Courts of Justice Strand, London, WC2A 2LL
		2 November 2016

Lord Justice Lewison:

1. Under the 1988 rules of the Barnardo's pension scheme pensioners are entitled to annual increases in their pensions. The annual increase is the lower of 5 per cent and increases in prices. Increases in prices have hitherto been measured by reference to the Retail Prices Index ("the RPI"). The main issue on this appeal is whether the trustees of the scheme have the power under the 1988 rules to substitute the Consumer Prices Index ("the CPI") or some other index for the RPI. Warren J held that in current circumstances they had no such power. With his permission Barnardo's, as sponsoring employer, appeals. The judge's very full judgment explains the factual background in detail. It is available at [2015] EWHC 2200 (Ch), [2015] Pens LR 501.
2. The answer that we give to the main issue will affect other versions of the rules applicable to the scheme, and also the method by which pensions not yet in payment are revalued. The question is an important one because if the trustees have that power and choose to exercise it the substitution of the CPI or some other index for the RPI will on the one hand significantly reduce the deficit in the pension fund; but, on the other hand, is likely significantly to reduce future increases in pensions payable to pensioners.
3. The answer to the question posed by the appeal depends on the interpretation of the rules. The two most immediately relevant rules are the definition of "the prescribed rate" in rule 30.1.3 and the definition of "Retail Prices Index" in rule 53. The first of these provides:

"... an increase at the rate of the lesser of:-

(a) 5%, and

the percentage rise in the Retail Prices Index (if any) over the year ending on the previous 31 December".

4. The second provides:

"Retail Prices Index means the General Index of Retail Prices published by the Department of Employment or any replacement adopted by the Trustees without prejudicing Approval. Where an amount is to be increased "in line with the Retail Prices Index" over a period, the increase as a percentage of the original amount will be equal to the percentage increase between the figures in the Retail Prices Index published immediately prior to dates when the period began and ended, with an appropriate restatement of the later figure if the Retail Prices Index has been replaced or re-based during the period."

5. "Approval" standing alone is not defined in the rules, although it is defined in the Appendix to the rules. The rules define "Tax Approval". Tax Approval is defined to mean approval as an exempt approved scheme under Chapter I of Part XIV of the Income and Corporation Taxes Act 1988. "Approval", standing alone, is similarly defined in the Appendix.
6. The critical words in the definition of the RPI are "or any replacement adopted by the Trustees without prejudicing Approval." Does the definition mean:

i) The RPI or any index that replaces the RPI and is adopted by the trustees; or

ii) The RPI or any index that is adopted by the trustees as a replacement for the RPI?

7. The first of these is a two stage process in which first, the RPI is replaced by something else which the trustees then adopt. The second is a single step by which the trustees simply choose another index, whether or not the RPI itself has been replaced. I should also record that it is no part of Barnardo's case, presented by Mr Rowley QC, that the RPI has in fact already been replaced independently of any exercise of discretion by the trustees. Any replacement will come about if and when the trustees exercise the discretion which he says they have to choose a replacement index.
8. There is no significant dispute about the applicable principles of interpretation. The rules of a pension scheme are, in principle, to be interpreted in the same way as any other written instrument. As the Supreme Court said in Arnold v Britton [2015] UKSC 36, [2015] AC 1619 at [15] the court must focus on the meaning of the relevant words in their documentary, factual and commercial context.

"That meaning has to be assessed in the light of (i) the natural and ordinary meaning of the clause, (ii) any other relevant provisions of the [instrument], (iii) the overall purpose of the clause and the [instrument], (iv) the facts and circumstances known or assumed by the parties at the time that the document was executed, and (v) commercial common sense, but (vi) disregarding subjective evidence of any party's intentions."

9. Reliance on background and commercial common sense must not be allowed to undervalue the importance of the words of the instrument. In addition commercial common sense cannot be invoked retrospectively.
10. There are, however, at least three points of special relevance to the interpretation of pension schemes. First, all or almost all pension schemes are intended to be tax efficient and to comply with Inland Revenue requirements. So Inland Revenue requirements are relevant to their interpretation. Second, pension schemes should be interpreted to have reasonable and practical effect. Third, since the rules of a pension scheme affect all those who join it (in some cases many years after its inception) other background facts have a very limited role to play.
11. One further point of general application is that in principle one would expect words and phrases to be used consistently in a carefully drafted instrument. Judicial statements to this effect are legion, but I take as representative the observations of Tomlinson LJ in Interactive Investor Trading Ltd v City Index Ltd [2011] EWCA Civ 837 at [29] that "it should ordinarily be presumed that language is used consistently within the four corners of an agreement."
12. The importance of Inland Revenue requirements is reflected in the drafting of the pension scheme itself. First, the rules contain many prohibitions on things that would prejudice Tax Approval: for example limits on additional contributions under rule 6.2; dates when pension may become payable under rule 12.2; limitation of benefits under rule 32, and winding up under rule 48, all in addition to the uprating of pensions under rule 30. Second, the rules contain an Appendix which sets out the Inland Revenue limits which were understood to apply at the date when the rules came into force. Those limits include, at paragraph 6 of the Appendix:

"The maximum pension … may be increased whilst in payment at 3% p.a. compound or (if greater) in line with RPI."

13. Although the rules themselves contain a definition of "in line with the Retail Prices Index" the Appendix contains its own definition in very similar terms:

"'in line with RPI' over a period means in proportion to increases between figures in the General Index of Retail Prices published by the Department of Employment (or a replacement of that Index not prejudicing Approval), immediately prior to the dates when the period began and ended with appropriate restatement of the later figure if the Index has been replaced or re-based during the period."

14. At the time when the rules came into force Inland Revenue requirements were contained in a statement of practice known as IR12. The particular point to emerge from IR12 was that it permitted pensions in payment to be increased and deferred pensions to be revalued in certain specified ways, including by reference to "increases in the cost of living". A footnote to that statement said:

"Increases in the cost of living may be measured by the index of retail prices published by the Department of Employment or by any other suitable index agreed for the particular scheme by the Superannuation Funds Office."

15. It is now necessary to say something about the indices that were in existence at the time when the 1988 rules came into force. The history of measuring changes in prices is contained in a report by Mr Paul Johnson, the Director of the Institute for Fiscal Studies, addressed to the UK Statistics Authority. Although the report itself long post-dates the rules, its historical parts are uncontroversial. The UK Government had been collecting data on prices since 1914. At that time the index was intended to measure the costs faced by working class households and was called the Cost of Living Index. A committee convened after the end of the Second World War was asked to consider the future of the Cost of Living Index. It recommended stopping the index in 1947, and replacing it with the Interim Index of Retail Prices. The recommendation was accepted and the Interim Index of Retail Prices replaced the Cost of Living Index. In 1956 the Interim Index was itself stopped and replaced by the Index of Retail Prices. It was a comparison of changes in prices for a basket of different goods and services. Mr Johnson comments that:

"This measure was to evolve in time to become the Retail Prices Index (RPI)."

16. It is not entirely clear when the Index of Retail Prices became the Retail Prices Index, but it does appear that it did so by a process of evolution rather than substitution, in contrast to the way in which the Interim Index of Retail Prices had replaced the Cost of Living Index, and the Index of Retail Prices had replaced the Interim Index of Retail Prices. As the RPI developed certain categories of the population were excluded (e.g. pensioners dependent on state benefits), and the composition of the basket and the relative weightings given to the different goods and services was changed from time to time. During the 1960s variants of RPI were produced to meet specific user needs. Thus state pensioner indices were introduced in 1969. Later, in 1979, a Tax and Prices Index was introduced. Rather than measuring the change in a representative basket of goods and services as the RPI did, the TPI measured the change in the average person's gross income (after income tax and national insurance) needed to buy the basket. In 1981 the Government began to issue index-linked gilts, for the purpose of which increases in prices were measured by reference to the RPI. From 1992 the Government set its inflation target by reference to RPIX (which was a variant of the RPI excluding mortgage interest payments). In the 1980s the Rossi index was introduced, which excluded most housing costs. It was used to uprate state benefits.
17. From time to time the RPI has been rebased; that is to say that the starting point for measuring changes in prices is reset to 100. The process of rebasing was carried out by the authority responsible for publication of the RPI.
18. The origins of the CPI lie in Europe. The Maastricht Treaty of 1992 required EU member states to develop a harmonised measure of consumer price inflation principally for the purpose of assessing whether member states met the criteria for joining the European Monetary Union. This resulted in the Harmonised Index of Consumer Prices (HICP), which the UK first published in 1997. By degrees the HICP has evolved into the CPI. The publisher of the RPI has also changed. In the 1950s responsibility for the index was that of the Ministry of Labour; in the 1960s the Department of Employment and Productivity and latterly the Department of Employment. In the early 1990s responsibility for RPI passed to the Central Statistical Office which became part of the Office for National Statistics in 1996.
19. In recent years statisticians have come to accept that there are flaws in the RPI which no longer meets international standards. HM Government has since 2010 used the CPI rather than the RPI for the uprating of pensions and other benefits. It sets its macro-economic inflation target by reference to the CPI rather than the RPI. In 2013 the UK Statistics Authority stripped the RPI of its status as a National Statistic. However despite its shortcomings the National Statistician recommended its retention; and it has been retained to this day. As Mr Johnson explains in his report:

"The RPI … still attracts widespread attention despite it no longer being accredited as a National Statistic. In part, this reflects the wide range of uses still made of the RPI. A significant proportion of UK government debt is linked to the RPI, and it is used for uprating rail fares, utilities and other contracts, as well as many index-linked private sector pension schemes. It does not, however, meet international standards."

20. The current position, at least at the date of Mr Johnson's report, is that the Office of National Statistics publishes four principal measures of inflation: the CPI, CPIH (which includes owner occupiers' housing costs), RPIJ (in which the "J" signifies use of a Jevons formula that uses a geometric rather than an arithmetic mean) and the RPI. Mr Johnson also tabulates a number of variants on these indices. Three of them are variants of the CPI, and two are variants of the RPI. In addition there is the TPI, the Pensioner Price Indices and the Rossi Index. The CPI is now the headline measure of inflation largely because of EU rules, although Mr Johnson recommended moving towards making CPIH the main measure.
21. Against that background I can now turn to the interpretation of the definition of the RPI in rule 53.
22. Mr Rowley pointed to the difference in language between the 1988 rules and their predecessor 1978 rules. The 1978 rules contained a definition in the following terms:

"INDEX shall mean the Government's Index of Retail Prices or any other official cost of living index published by authority in place of or in substitution for that Index."

23. Although in cases about the interpretation of commercial contracts judges have held that an earlier agreement between the same parties is admissible for the purpose of interpreting a later one, they have rarely found that to be of much help. The 1988 rules are not an amendment of the 1978 rules: they are a completely new set of rules. As the judge said at [30] the structure and wording of the two sets of rules is very different, and to fasten on one particular difference between the old rules and the new is likely to be misleading. Moreover, in a case like this, where the 1988 rules would apply to subsequent joiners who would have had no knowledge of what a previous version of the rules might have said, I do not think that this exercise in forensic archaeology is profitable. As Robert Walker J said in National Grid Company plc v Mayes [1997] Pens LR 157 at [70]:

"… it is often hard enough for trustees and their advisers (and even harder for members or pensioners who may not have easy access to advice) to interpret a pension scheme as it stands, without also having to delve into the archaeology of the scheme."

24. I will, however, deal with the point that Mr Rowley made about the 1978 rules. It was common ground that under this definition any replacement of the RPI would be an index published by authority; and that if there were only one replacement index it would be substituted automatically for the RPI. From this Mr Rowley argued that the definition of "Retail Prices Index" in the 1988 rules departed from this in two significant respects. First, it gave the trustees a discretionary role which was absent in the 1978 rules. Second, it referred to "Approval" which the 1978 rules had not. There is in fact a third difference, which was not touched on during argument. Whereas the 1978 rules referred to "the Government's Index of Retail Prices", the 1988 rules referred to "the General Index of Retail Prices published by the Department of Employment". In addition the definition in the 1978 rules did not cater for the possibility that there might be more than one replacement index. Mr Rowley said that under those rules if there were more than one replacement index there would have to be an amendment to the rules under the power of amendment contained in them. As we know from Mr Johnson's report even in the 1960s there were variants of the RPI, and I do not consider that it is at all fanciful to suppose that the draftsman of the 1988 rules wished to cater for the possibility that any new index would also have variants. The possibility of more than one replacement index is, to my mind, at least a possible explanation for the change from the 1978 definition to the 1988 definition, because in such an event (a) the trustees would have the power of selection but (b) that power could not be exercised so as to prejudice Approval. That there might be more than one index acceptable to HMRC was treated as a possibility by Newey J in Arcadia Group Ltd v Arcadia Group Pension Trust Ltd [2014] EWHC 2683 (Ch), [2014] 067 PBLR (018) at 31 (iii); and that there might be one index that was acceptable to HMRC and another that was not was canvassed by the judge at [55]. I might also add that even where there has been a change in language between one version of a scheme and a later version, the new language may still mean the same as the old, as indeed turned out to be the case in National Grid Company plc v Mayes [2001] UKHL 20, [2001] 1 WLR 864 at [24]. I do not, therefore, consider that this point advances the case either way.
25. Mr Rowley argues that the second sentence of the definition in the 1988 rules plays no active part in the rules (as opposed to the Appendix) because nothing in the rules themselves requires any payment to be adjusted "in line with [RPI]". From this he argues that the second sentence of the definition should simply be ignored. The judge rejected this argument at [32] and [64] and so would I. First, it is trite law that any instrument must be construed as a whole. Where the court is confronted with a difficult question of interpretation it is surely entitled (if not bound) to look for such linguistic clues as there are. The second sentence of the definition is part of the "documentary context" of the first sentence. Second, the first sentence of the definition uses the word "Approval" which is not defined in the rules, but which is defined in the Appendix. Third, clause 3 of the deed governing the scheme provides that:

"… as from 1 April 1988 the Scheme will be governed by the Rules (including the Appendix) contained in this deed…"

26. So the scheme includes both the rules and the Appendix. The Appendix is referred to again in rule 32 which says that the Revenue limits "are summarised in the Appendix to this Deed." So there is further clear linkage between the rules and the Appendix. As Mr Rowley acknowledges the Appendix is replete with references to payments being adjusted "in line with RPI". Fourth, it is now established that in interpreting a written instrument the court may look at words which the parties have deleted: see Narandas-Girdhar v Bradstock [2016] EWCA Civ 88 [2016] 1 WLR 2366 at [20]. Ex hypothesi deleted words have no operative force in the contract as made, but nevertheless they may be used as an aid to interpretation. Where parties have not deleted the words in question, that seems to me to be a stronger case. Fifth, as Mr Simmonds QC submitted on behalf of the members, if the inverted commas round "in line with the Retail Prices Index" are ignored (rather than the whole of the sentence) the second sentence of the definition can be read as a gloss on rule 30.1.3. (b) so that it does have at least some function within the rules.
27. Starting with the first sentence in the definition, both the word order and the grammatical construction of the sentence lead me to conclude that the natural meaning of the phrase in issue is that replacement precedes adoption by the trustees. It is, to my mind, an unnatural reading to attribute to the phrase the meaning "any other index adopted by the trustees as a replacement" which is the meaning that Mr Rowley urges. This is, perhaps, an intuitive reaction to the use of language, so it is necessary to consider the context in which the disputed phrase is used.
28. What light does the second sentence of the definition in the rules shed on the first sentence? In my judgment quite a lot. That sentence refers to "an appropriate restatement of the later figure if the Retail Prices Index has been replaced or re-based during the period". First, it is plain that the RPI can only be re-based by the authority responsible for publishing it. The natural meaning of the phrase "replaced or re-based" is that the same person is doing the re-basing or replacing. This appeared to be common ground at the start of the appeal although Mr Rowley modified his position during the course of argument. But in my judgment it is an unnatural use of language to specifically contemplate different actors in the same composite phrase. Mr Rowley submits, in the alternative, that the reference to "replacement" in the first sentence of the definition refers to a decision taken by the trustees rather than a decision taken by the authority responsible for publication of the RPI. At [32] the judge said:

"It might be thought that the concepts of 'replacement' and 'replace' were intended to be the same in each sentence."

29. I agree. Not only "might" it be thought, that is the obvious meaning to give to the recurrence of a cognate expression in the same definition. This conclusion also chimes with the definition of "in line with RPI" in the Appendix to the rules which also refers to an "appropriate restatement of the later figure if the Index has been replaced or re-based during the period". As Mr Simmonds points out there is no redundant second sentence in the latter definition, and it would be extraordinary if replacing and re-basing allowed the two activities to be carried out by different persons when each is spoken in the same breath. The way in which the definition in the Appendix is drafted makes it clear that "the replacement of that Index" is replacement by the authority responsible for publishing it and not by the trustees. Again, one would expect the concept to be consistent throughout the document. Since there is clear linkage between the rules and the Appendix, this interpretation makes far better linguistic sense in the documentary context than that for which Mr Rowley contends.
30. In my judgment this also accords with the factual background at the time the rules came into effect. As the history shows official government indices had been replaced from time to time, in each case by the government. Thus the Cost of Living Index was discontinued and replaced by the Interim Index of Retail Prices, and the Interim Index of Retail Prices was discontinued and replaced by the Index of Retail Prices (later the RPI). It is not at all surprising that the draftsman made provision for that eventuality. In addition the definition in the rules refers to the RPI published by the "Department of Employment." The index as thus defined is no longer published. It is now published by the Office for National Statistics. That change of publisher fits comfortably within the definition. We also know from Mr Johnson's report that (contrary to what the judge thought) there was more than one measure of changes in prices published by the government (and there still are). But I do not consider that the judge's error in this respect undermines his conclusion. Rather, it reinforces the possibility that the draftsman foresaw that if the RPI were discontinued there might be several other indices from which to choose.
31. Mr Rowley argued that the requirement in the definition that adoption of an alternative index should not prejudice "Approval" was a strong pointer in favour of his interpretation. If the RPI were to be replaced by another official measure of inflation, it is inconceivable that Approval would be jeopardised. There is some force in that point if the RPI were to be replaced by a single official measure of inflation. But equally it might be replaced by a number of different measures, such as the variety of different measures that now exist and which are used for different purposes. It is by no means certain (and the draftsman of the rules could not have been certain in 1988) that in that event the Inland Revenue would approve all potential measures of inflation. But in any event the draftsman has been at pains throughout the rules and the Appendix to stress the importance of Inland Revenue approval. Necessarily, in the case of this part of the definition he was dealing with a future unknown and unknowable event, and a cautionary approach to drafting is readily understandable.
32. Mr Rowley also placed considerable reliance on IR12 and the possibility that it envisaged of agreeing an index other than the RPI with the Superannuation Funds Office. The first point to make is that although Inland Revenue practice is relevant background to the interpretation of the rules, it is no more than that: background. If the draftsman had wanted to stick to IR12 he would surely have tracked its language (as has happened in other cases). With the precedent of IR12 before him he chose different language. Second, although the requirements of IR12 needed to be satisfied before the scheme could secure approval from the Revenue, there was no mandatory requirement to include provision for a change in the index within the rules, and it is plain from IR12 that a scheme whose rules specified the RPI alone as the chosen index for increases in payment would have achieved approval at least in that respect. There is no reason to suppose that the draftsman intended more than the obtaining of approval under IR12 particularly since other parts of IR12 contained a list of alternatives which are not reflected in the drafting of the rules. Third, Inland Revenue approval to a different index could (and I suspect normally would) be obtained before a particular pension scheme was set up. In other words the agreement of the Inland Revenue to a different index under IR12 could relate to the initial choice of an index rather than to a change from one index to another in an existing pension scheme. Whether such a change would be lawful would depend primarily on the rules of the scheme and not on Inland Revenue practice. But that is the very question we have to decide; and the wording of the rule in this case differs significantly from the wording in IR12 which does not refer to replacement. So reliance on Inland Revenue practice, in my judgment, begs the question. IR12 says nothing in terms about a change from one index to another in an existing pension scheme. Fourth, the Appendix (rather than the rules) contains the draftsman's understanding of Inland Revenue practice. So one would expect IR12 to tie in with the definition of "in line with RPI" in the Appendix. But that definition militates more strongly against Mr Rowley's interpretation than the definition in the rules themselves. The definition in the Appendix does not, in my judgment, contemplate a choice between indices unless and until the RPI is replaced. It may be that the draftsman understood IR12 to be dealing only with the initial choice of index which would explain why the definition in the Appendix is drafted as it is. I do not consider that IR12 militates against the interpretation of the definition that I favour.
33. For much the same reasons I do not consider that an examination of the wording of different pension schemes (or the single published precedent that we were shown) casts any light on the meaning of the definition in this pension scheme. Every instrument must be interpreted in accordance with its own terms. Other decided cases are only of help if they lay down some principle of interpretation.
34. Finally, Mr Rowley argued that the interpretation propounded by the members and adopted by the judge was uncommercial. This seems to me to be a classic case of the invocation of commercial common sense with 20:20 hindsight. There is nothing inherently uncommercial in the specification, in 1988, of a measure of inflation that was the primary measure specified by the Inland Revenue and the principal measure used by HM Government, with a provision that applied in the event that the Government replaced that index as it had done at least twice before. This, to my mind, is the key flaw in Mr Rowley's case. He starts from the end result for which he argues and then seeks to bend the natural meaning of the words to fit that result. From the perspective of the employer (whose interests Mr Rowley advanced so persuasively) that end result is plainly preferable. But as Lord Neuberger emphasised in Arnold v Britton the starting point is the language of the instrument itself and, in particular, its ordinary and natural meaning. It is not legitimate to search for drafting infelicities in order to facilitate a departure from the natural meaning of the words: Arnold v Britton at [18].
35. Mr Rowley also said that adherence to RPI has, as things turned out, been to the members' benefit, but things could have turned out differently with the result that pensions no longer kept pace with inflation. The implications of that submission, if correct, would be that the trustees had power to impose greater financial obligations on the sponsoring employer without obtaining the employer's consent. That is, in my judgment, an unlikely conclusion.
36. For these reasons, which are essentially those of the judge, I would dismiss the appeal. The members raised a point on section 67 of the Pensions Act 1995 by way of cross-appeal. Since we heard full argument on that cross-appeal; and I have reached a clear view, it is right to deal with it, even on an obiter basis, particularly since we do not all agree on the outcome of the appeal itself.
37. The cross-appeal proceeds on the basis that Mr Rowley's interpretation of the definition of the RPI is correct. Thus it is to be interpreted as meaning "the RPI or another index adopted by the trustees as a replacement." I add that the question of "Approval" as defined has fallen away as a result of changes in pensions legislation.
38. Section 67 provides that the exercise of a power to make a regulated modification to an occupational pension scheme is voidable unless certain conditions are satisfied. We are not concerned with the conditions. The point at issue is whether the selection of the CPI or some other index in place of the RPI is a regulated modification. There are two kinds of regulated modification: a protected modification and a detrimental modification. We are concerned with the latter.
39. Section 67A (4) provides:

""Detrimental modification" means a modification of an occupational pension scheme which on taking effect would or might adversely affect any subsisting right of–

(a) any member of the scheme, or

(b) any survivor of a member of the scheme."

40. The key provision is the definition of "any subsisting right" which is contained in section 67A (6). That provides, so far as material:

"Subsisting right" means–

(a) in relation to a member of an occupational pension scheme, at any time–

(i) any right which at that time has accrued to or in respect of him to future benefits under the scheme rules, or

(ii) any entitlement to the present payment of a pension or other benefit which he has at that time, under the scheme rules…"

41. Section 67A (10) (b) provides:

"… a modification would or might adversely affect a person's subsisting right if it would alter the nature or extent of the entitlement or right so that the benefits, or future benefits, to which the entitlement or right relates would or might be less generous."

42. Both Mr Simmonds and Mr Rowley said that the right to an increase in a pension in payment fell within section 67A (6) (ii). I am bound to say that to the uninitiated the right to a future increase falls more naturally within sub-paragraph (i) as being an accrued right to a future benefit rather than an entitlement to "present payment" of a pension. However, I do not think that at least in this case anything turns on that. The real question is what accrued rights or current entitlement the members have.
43. Mr Simmonds submitted, and I agree, that the policy underlying section 67 is that a pension is treated as deferred pay earned by service, and that the purpose of section 67 is to prevent a person's deferred pay from being retrospectively reduced. But that does not answer the question: what is that deferred pay? So once again the real question is what accrued rights or current entitlement the members have.
44. There are three cases that bear on the question. The first is the decision of this court in Aon Trust Corpn v KPMG (a firm) [2005] EWCA Civ 1004, [2006] 1 WLR 97. Rule 7 of the pension scheme under consideration in that case provided for the calculation of a pension by reference to two possible components. The first was the product of contributions made by employer and employee in a relevant year of service. The second was the possible addition of a bonus (under clause 8.4) if the scheme was in actuarial surplus or a reduction in benefits (under clause 8.5) if the scheme was in actuarial deficit. The combination of these two factors dictated the amount of pension which the member would actually receive.
45. There are two important points to be made about the scheme. First, the question whether a bonus would be added or a deduction made was a matter for the discretion of the trustees, subject, in the case of a bonus, to the employer's consent. Second, the application of a bonus or reduction could be made at any time before the pension came to be paid, so that in the case of an active or deferred member it could be made long after the period of service which gave rise to the pension.
46. The argument for the employer was that the power to make adjustments (either by way of bonus or deduction) were "integral to the process of calculating the amount of a member's pension entitlement". The application of the formula to the contributions at stage 1 only gave rise to a provisional figure. Since the power to make adjustments was an integral part of the calculation of the member's pension entitlement the exercise of that power could not be described as the modification of the scheme.
47. The Court of Appeal rejected this argument. Jonathan Parker LJ (with whom Chadwick and Mummery LJJ agreed) said:

"[163] So I conclude that, on the true construction of clause 8.4 and 8.5, the existence of a surplus or a deficit (as the case may be) does no more than set the scene for the possible exercise of the powers of adjustment of benefits conferred by those clauses. The relevant power may not be exercised at all; or it may be exercised in a way which leaves some part of the surplus or deficit still in existence. There is, therefore, no question of any automatic adjustment of benefits following an actuarial valuation which reveals the existence of a surplus or a deficit; still less of an adjustment which will have the effect of extinguishing the entirety of that surplus or deficit.

[164] There is also the time factor to be considered. Even if the trustee, with the employer's consent and acting on actuarial advice, decides to exercise the appropriate power of adjustment (according to whether the actuarial valuation has revealed a surplus or a deficit) there will inevitably be an interval of time (which may be substantial) between the date when the surplus or deficit first arose and the date when the power is exercised. Yet during that time rights will inevitably have accrued under the scheme and benefits will inevitably have been paid out, notwithstanding the existence (by definition) of a continuing mismatch between assets and liabilities. Moreover, the effect of any adjustment of benefits (whether upwards or downwards) will usually take effect over time, by gradually eroding the surplus or deficit. Thus even in a situation where the adjustment is designed entirely to extinguish the surplus of deficit, the mismatch between assets and liabilities may continue for a substantial period of time after the adjustment has been made.

[165] In the light of the above analysis, I conclude that [the employer's] basic submission must be rejected. So far as the clause 8.4 power to increase benefits is concerned, the declaration of a bonus will give the member the right to an increased pension. But it does not follow that the member has no right to a pension under rule 7 until the trustee has considered whether or not to exercise that power (and, it may be, decided not to exercise it, or to exercise it not by declaring a bonus but by reducing contributions). The same consideration applies, in my judgment, to the clause 8.5 power. In my judgment it does not follow from the existence of that power that a member has no right to a pension under rule 7 until the trustee has taken a decision as to whether the power should be exercised, and if so how.

[166] The correct analysis in law, in my judgment, is that on the true construction of the scheme a member has an accrued right to a pension under rule 7 in the (unadjusted) amount calculated by aggregating the total amounts referred to in rule 7.2(1), but subject to any adjustments made under clause 8.4 or clause 8.5. I therefore reject the notion that that calculation produces only a "provisional" sum… In my judgment, to read the scheme in that way is to attempt to force a square peg into a round hole."

48. Having decided that as a matter of construction of the scheme a member had a right to a pension by reference to the unadjusted application of the contributions to the formula, it was a relatively simple step to conclude that a downward adjustment was caught by section 67 (in its then form). The mere fact that the adjustment would be made in the exercise of a power contained in the rules of the scheme did not prevent it from being a "modification" for the purposes of the then section 67.
49. There are two subsequent cases which have considered Aon. The first is Danks v QinetiQ Holdings Ltd [2012] EWHC 570 (Ch), [2012] Pens LR 131, a decision of the Chancellor (when sitting as Vos J); and the second is the decision of Newey J in Arcadia Group Ltd v Arcadia Group Pension Trust Ltd [2014] EWHC 2683 (Ch). Both these decisions held that where an increase in pension was to be calculated by reference to an index whose definition admitted of alternatives, there was no accrued right to payment until the index had been selected. Because there were two first instance decisions directly in point, the arguments on the cross-appeal were not ventilated before the judge, who would in practice have been bound to follow those two decisions.
50. In Danks v Qinetiq a member was entitled to an increase calculated in accordance with the "Index". "Index" was a defined term which meant:

"the Index of Retail Prices published by the Office of National Statistics or any other suitable cost-of-living index selected by the Trustees".

51. The issue was whether the trustees were entitled to select the CPI in place of the RPI without engaging section 67. That in turn required the judge to decide what their entitlement (or accrued right) was. At [55] Vos J held that the right of a member to an increase in pension was not a right to an increase at any particular rate, but was only a right to an increase "since it is only a right to have an increase each April 'by an amount equal to the percentage increase in the Index …', and the Index is defined as being RPI or 'or any other suitable cost of living index selected by the Trustees'." He continued:

"Thus, the point is really one of timing. A member with a pension in payment, who has had an increase under Rule 49.1 at RPI on 1 April 2011 (for example) could not have that increase reduced without there being a detrimental modification. But, in advance of the next Rule 49.1 increase date (1 April 2012), the member has no entitlement to an increase at any specific rate, since the Trustees always retain a power to change the Index by which the increases are to be calculated. To repeat the point, the member has only a right to a future increase at RPI 'or any other suitable cost of living index selected by the Trustees'. The difference in Aon was that the members with pensions in payment were entitled to pensions calculated in accordance with rule 7, and the exercise of the Clause 8(5) reduction would obviously have been a detrimental modification. Here the entitlement is only to a future increase at a rate that the Trustees have power to change."

52. Newey J took the same approach in Arcadia in which the "Retail Prices Index" was defined (with immaterial differences in some scheme documents) as:

"the Government's Index of Retail Prices or any similar index satisfactory for the purposes of HM Revenue and Customs"

53. Having considered both Aon and Danks v QinetiQ Newey J said at [58]:

"On balance, it seems to me that Vos J's analysis in Danks v QinetiQ Holdings Ltd must have been correct. Far from being persuaded that that case was wrongly decided, I respectfully agree with the decision. The better view, I think, is that members have a "subsisting right" to increases and revaluation at rates consistent with the definitions of "Retail Prices Index", but not to increases and revaluation specifically by reference to RPI."

54. Both Mr Simmonds and Mr Rowley agreed that the question whether section 67 applies to a change in index must turn on the substance of the right given to members under the rules of the scheme, rather than on the details of drafting. There is a lot to be said for that approach, although drafting may well have a significant role to play, as examples canvassed in the course of argument demonstrated. It is, I think, better to stick with the actual drafting than to attempt to imagine other forms of words.
55. Mr Simmonds argued that the definition of "Retail Prices Index" in rule 53 was analogous to a power of appointment over trust property. Unless and until the trustees made a positive decision to exercise their discretionary power to select another index, members had the right to an increase in pensions in line with the RPI. The RPI, he argued, was the "default rule" unless and until something happened to change it.
56. This, too, is at bottom a short point of construction. It seems to me that if a person has a right to "A or B" one cannot say that he has an accrued right to A. He has a right to one or other of them. As Newey J put it a member has the right to an increase consistent with the definition; or as Vos J put it the member has a right to a future increase at RPI or any other suitable cost of living index selected by the Trustees. I agree with both of them. I do not consider that there is a default rule in the way that Mr Simmonds suggested. On the basis of Mr Rowley's construction of the definition, the trustees have a choice; and until that choice has been exercised, it is not possible to say that the member has a right to an increase measured in any particular way.
57. I would dismiss the cross-appeal.

Lord Justice McFarlane:

58. Whilst I am, of course, most grateful to Lord Justice Lewison and to The Chancellor, Sir Geoffrey Vos, whose judgments I have read in draft, for the clarity of reasoning by which they have each reached opposing conclusions upon the construction of the 1988 Rules of the Barnardo's Pension Scheme, the fact that they have been able to do so goes to demonstrate that there is no clear and unambiguous answer to the issue at the centre of this appeal. It now falls to me to express my own conclusion on that issue and, as I do so, where I am forced to differ with either of My Lords, it is with the greatest respect to their superior experience in matters of this nature.
59. It is not necessary for me to rehearse the details of the scheme or the various important points in the submissions of each side which have already been described in full. I agree that the two conflicting interpretations can be formulated in the manner suggested by The Chancellor at paragraph 74 (i) and (ii) below.
60. At the centre of the dispute on construction there is a difference of opinion as to the significance of the 'second sentence' in the definition of "Retail Prices Index" and the significance of The Appendix and the definition of "in line with RPI" which appears there. Neither of My Lords take Mr Rowley's extreme position, which is to hold that these provisions should effectively be ignored. Lewison LJ, in common with Warren J at first instance, holds that the second sentence and the Appendix fall to be construed as part of the scheme. The Chancellor takes a more nuanced position by accepting that the second sentence cannot be ignored, but holding that it may not help a great deal with the construction of the first sentence. With respect to the Appendix, Sir Geoffrey considers that it is relatively clear that any replacement of the "General Index of Retail Prices published by the Department of Employment" would be one published by the Department of Employment and not another different index adopted by the Trustees for the reasons that he gives at paragraph 79. However he goes on to distinguish the identification of the Index in the Appendix from the terms of the Rules of the pension scheme themselves because the Appendix is intended to be no more than a summary of the existing Revenue Limits in place at the time.
61. Having now considered the competing arguments, with the assistance that each of my lords has provided by the clarity of their judgments, I am persuaded that Lewison LJ's construction of the key passages is to be preferred for the reasons that he has given. In expressing this conclusion I would wish to highlight the following points which I regard as being of particular weight (I need only refer to these in headline form):

i) The document should be construed as a whole and it is artificial to afford no, or very little weight, to the second (longer) sentence of a two sentence definition that appears in the Rules when interpreting the first sentence;

ii) The pension scheme includes both the Rules and the Appendix. The Appendix refers to adjustments "in line with RPI" which phrase, in turn, falls to be interpreted in line with the definition of that term in the Appendix;

iii) The reference to "a replacement" in the definition of "in line with RPI" must, as The Chancellor explains, be a reference to a replacement published by the Department for Employment, and not another index chosen by the Trustees;

iv) Interpreting the document as a whole, the construction of "replacement" in that context in the Appendix should weigh significantly in the interpretation of "any replacement" in the definition of "Retail Prices Index" in the Rules. This must be particularly so where the phrasing of both definitions is, as Lewison LJ observes, in very similar terms;

v) The natural meaning of the first sentence in the definition is that replacement of the index precedes adoption by the trustees, and not the other way about;

vi) In the second sentence there is some welcome solid ground, which is accepted by all parties and in both of my lords' judgments, namely that the reference to an index that has been "re-based" in the context of the Retail Prices Index can only be to action by the authority responsible for publishing it;

vii) From that solid ground, I agree entirely with Lewison LJ's analysis at paragraph 28 to the effect that it would be an unnatural use of language to bisect that short phrase and hive off "replaced" from its neighbour "re-based" so as to contemplate the trustees undertaking replacement, in contrast to the authorities who are responsible for re-basing, where there is no indication at all that the two elements of this composite phrase fall to be understood in such a manner;

viii) I do not agree that it is clear that something has gone wrong with the drafting of the second sentence or that it has no real place in the Rules. Whilst the phrase "in line with RPI" does not appear in the Rules, it does appear in the Appendix. The inclusion of a definition of the phrase within the Rules is not so perverse or out of kilter with the document as a whole as to justify holding that it should be afforded very little weight;

ix) Construing the scheme as a whole, and starting from a position that the two sentences in the definition of "Retail Prices Index" should be read as a whole, I regard the interpretation of "replaced" in the second sentence as being of significance in any understanding of the meaning of "any replacement" in the first sentence. For the reasons given by Lewison LJ at paragraph 29, looking at these key words overall, and taking account of the Appendix as part of the scheme, the interpretation given by the judge makes far better linguistic sense than the alternative argued for by Mr Rowley.

62. Although it does appear that the judge fell into error in his understanding that at the time when the 1988 Rules were introduced the RPI was the only official index published, the position before this court is clear. There were four additional relevant indices in existence at the time. Any error by the judge has not, therefore, compromised our attempts on appeal to construe this scheme.
63. Secondly, in terms of any potential error by the judge, whilst it may be unusual for the exercise of interpretation to start with consideration of the Appendix, it is clear that, by the time he had concluded his analysis, the judge had considered all of the relevant provisions before reaching a decision. Again, in this court, the exercise has been undertaken afresh and any apparent error, if error there be, on the part of the judge must be of little relevance.
64. Finally, although the authority of Arnold v Britton was introduced into the process of submissions by the court rather than counsel, attempts to consider whether one interpretation or the other is, or is not, in accord with commercial common sense do not seem to be of any great assistance given that the exercise has to be looked at in terms of the world as it was in 1988. Mr Rowley's argument clearly, but wrongly, invited the court to look at the matter from the perspective of 2016. There is, however, no evidence that we have seen that suggests that, if the question of commerciality is considered against the backdrop of 1988, the interpretation argued for the appellants should be preferred.
65. I therefore conclude, for the reasons that I have summarised and on the basis that I am in full agreement with the analysis contained in the judgment of Lewison LJ, and for that matter the judge, that this appeal should be dismissed. I also agree that the cross-appeal should be dismissed for the reasons given by Lewison LJ.

Sir Geoffrey Vos, Chancellor of the High Court

66. I shall not repeat the facts and background so clearly explained by Lord Justice Lewison. I shall also use the same defined terms.
67. It is useful, however, to set out the main clauses contained in the 1988 Rules that are to be construed as follows:-

i) Rule 53 defines "Retail Prices Index" as meaning "the General Index of Retail Prices published by the Department of Employment or any replacement adopted by the Trustees without prejudicing Approval. Where an amount is to be increased "in line with the Retail Prices Index" over a period, the increase as a percentage of the original amount will be equal to the percentage increase between the figures in the Retail Prices Index published immediately prior to the dates when the period began and ended, with an appropriate restatement of the later figure if the Retail Prices Index has been replaced or re-based during the period".

ii) Rule 30 provides for pensions and deferred pensions over the Basic GMP (as defined) to be increased by the "prescribed rate", which is defined by Rule 30.1.3 as meaning "an increase at the rate of the lesser of:- (a) 5%, and (b) "the percentage rise in the Retail Prices Index (if any) over the year ending on the previous 31 December".

iii) Rule 32 provides that "Benefits payable under the Scheme must not be of amounts or paid on conditions or in a manner or in circumstances which would exceed Revenue Limits or otherwise endanger Tax Approval. The Revenue Limits are summarised in the Appendix to this Deed. The Trustees must observe the terms of any undertaking they have given to the Board of Inland Revenue. If a combination of payments would prejudice Tax Approval or cause the Trustees to be in breach of such an undertaking, the Trustees will reduce all or any of them in such a manner as they think fit but to the extent only as may be necessary to prevent Tax Approval being prejudiced or a breach of the undertaking".

iv) The Appendix to the Rules summarises the Revenue Limits. The Appendix uses the term "in line with RPI" on 5 occasions. Paragraph 2(3) limits the increase of deferred pensions between leaving pensionable service and normal retirement date to "5% p.a. or in line with RPI (whichever is greater)". Paragraph 6 limits increases in pensions in payment to "3% p.a. compound of (if greater) in line with RPI".

v) The term "in line with RPI" is defined in paragraph 10 of the Appendix to the Rules as follows: ""in line with RPI" over a period means in proportion to increases between figures in the General Index of Retail Prices published by the Department of Employment (or a replacement of that Index not prejudicing Approval), immediately prior to the dates when the period began and ended, with an appropriate restatement of the later figure if the Index has been replaced or re-based during the period".

68. I would start by endorsing what Warren J said about the proper approach to the construction of pension scheme documentation in paragraphs 10 and 11 of his judgment. I agree, however that there is also a useful summary of the principles applicable to pension schemes in paragraphs 26-32 of Arden LJ's judgment British Airways Pension Trustees Ltd & others v. British Airways plc & others [2002] PLR 247. I also agree that the principles of statutory construction most recently summarised by Lord Neuberger in Arnold v. Britton [2015] AC 1619 at paragraphs 15-23 are applicable, although they do not deal specifically either with pension schemes or with a situation where there is obvious ambiguity.
69. Neither party has contested the judge's summary of the applicable law. It has, however, been submitted for the employer that the judge failed properly to apply his own summary. What is said, in effect, is that the definition of RPI was ambiguous (i.e. it admitted of two possible constructions) so that the court should have preferred the construction that was consistent with business common sense and rejected the other. For my part, I do not think the judge departed from his summary of the applicable law, since he implicitly recognised that there were two possible constructions, and held that neither had more commercial reality than the other (paragraph 68). He chose the meaning he did on the basis of the language used in the Rules and his view of the context.
70. I do, however, think that the judge adopted an over-complex approach, and that he fell into error in three important respects. His first error was to say at paragraph 20 that the evidence before him established that "at the time when the 1988 Rules were introduced, RPI was the only official index published" (see also paragraph 41, which is similarly in error). It is and was common ground that that is not the case. The following other "official" indices were published in and after 1988:-

i) A Pensioner Price Index introduced in 1969.

ii) RPIX (RPI excluding mortgage interest payments) introduced in 1975 RPIX was adopted as the UK inflation target between 1992 and 2003, and was therefore the precursor of CPI in one regard at least.

iii) The Tax and Price Index introduced in 1979.

iv) The Rossi Index used at times to uprate income-related state benefits, introduced in 1981.

71. The judge's second error in my judgment was in relation to his approach to IR12. At paragraphs 51-55 and 62 he explained why he derived no assistance either from the 1978 Rules or from the propositions (i) that the change in wording was intended to provide the Trustees with a new discretionary power and (ii) that this change was reflective of IR 12. But in my judgment, each of the previous definition of "INDEX" in the 1978 Rules, the new discretionary power accorded to the Trustees by the Rule 53 definition and IR12 were of importance to the construction process. Arden LJ made this clear in British Airways supra when she said at paragraph 30 that the factual situation at the time included the "practice and requirements of the Inland Revenue at the time" (see also National Grid Co plc v. Mayes [2001] 1 WLR 864 where Lord Hoffmann said at paragraphs 18-21 that pension schemes must be construed against their fiscal background).
72. The judge's third error was, in my judgment, one of approach in that it seems to me that it is appropriate to start the process of construction of the relevant rules by looking at the Rules themselves rather than the Appendix, which is expressly included only to summarise the applicable Revenue Limits. Moreover, it is noticeable that the Appendix uses the term "RPI" whilst the Rules use the term "Retail Prices Index". The judge started his approach to construction with the Appendix and worked backwards. That approach was, I think, a mistake.
73. In construing then the definition of "Retail Prices Index" in Rule 53, one should also recall that, whilst words are normally used in the same sense in the same document, that is not always the case. The context will play an important part (see, for example, Lord Hofmann at paragraph 27 in Oxfordshire County Council v. Oxford City Council [2006] UKHL 25, where he construed the word "locality" used twice in the same sub-section as meaning "a single locality" when first used and "locality or localities" when used for the second time).
74. Taking the words of the first sentence of Rule 53 first, it seems to me that at least two possible meanings are available. I have interposed words in square brackets to make those meanings more apparent in the following formulations:-

i) the General Index of Retail Prices published by the Department of Employment or any replacement [for that General Index of Retail Prices published by the Department of Employment which is then] adopted by the Trustees without prejudicing Approval.

ii) the General Index of Retail Prices published by the Department of Employment or any replacement [index of prices which is] adopted by the Trustees without prejudicing Approval.

There are other more refined possibilities but these two will do, I think, for the purposes of this discussion.

75. It is also important to note that, assuming the Retail Prices Index or RPI are the same, the provisions for pension increases as defined in Rule 30 and in paragraph 6 of the Appendix do not appear to be in conflict. If RPI is below 5%, the lower RPI will be the prescribed rate, which will not contravene the limit in paragraph 6 of the Appendix (which is 3% or RPI if greater). If RPI is above 5%, the higher RPI will be the prescribed rate, which will also not contravene the limit in paragraph 6 of the Appendix.
76. Taking the Rule 53 definition alone, I cannot agree with Mr Keith Rowley QC, leading counsel for the employer, that one should wholly disregard the second sentence. It is true that the second sentence makes a reference to the term "in line with RPI" used in the Appendix, even if the term is "in line with Retail Prices Index" in the second sentence of the Rule 53 definition. But I am not sure that the second sentence helps a great deal with the construction of the first. In the first place, the second sentence is, as Mr Rowley pointed out, strictly redundant since it attempts to define the term "in line with the Retail Prices Index" that is used nowhere else in the document, not even in the Appendix. It is true that it uses the words "with an appropriate restatement of the later figure if the Retail Prices Index has been replaced or re-based during the period" to qualify the figure which should be taken when comparing Retail Price Indices. But the words "replaced" and "re-based" are not themselves qualified. An index would be "re-based" by the authority publishing it, but an index might in this context be "replaced" either by the trustees or the authority responsible for it. The use of the two alternatives does not obviously imply that they are both to occur as a result of the actions of the same entity.
77. Mr Andrew Simmonds QC, leading counsel for the members, argued that, because the word "replaced" must mean "replaced by the authority responsible for it" in the second sentence, it must mean the same in the first sentence. I do not agree for the reasons I have already given, and because the second sentence, whilst not being completely disregarded, must be given less weight when something has clearly gone wrong with the drafting. On any analysis, the second sentence has no real place in the Rules and should be, if anywhere, in the Appendix.
78. In these circumstances, the proper meaning of the first sentence of the Rule 53 definition has, in my judgment to be informed by the factual matrix, proper context and business common sense.
79. The first thing to consider is, of course, the Appendix itself even if, for the reasons I have given, the use of the terms "RPI" and "replacement" in the Appendix will not be conclusive as to the meaning of the terms "Retail Prices Index" and "replacement" in the main body of the Rules. It is, however, relatively clear in my view that "in line with RPI" in the Appendix is defined in a way that contemplates that any replacement of the "General Index of Retail Prices published by the Department of Employment" would be one published by the Department of Employment and not another different index adopted by the Trustees. That is because of the use of the capital "I" for "Index" and the word "that" in the bracketed words "or a replacement of that Index not prejudicing Approval" (emphasis added), and the use of the capital "I" for "Index" in the words "if the Index has been replaced or re-based during the period" (emphasis added). There is no indication in the definition of "in line with RPI" to suggest that the RPI of which it speaks can be changed by the Trustees. That is, however, hardly surprising since the Appendix is defining the existing Revenue Limits, not seeking to define what pension increases the Rules might provide for. There is no reason in principle (as Mr Simmonds ultimately accepted) why the summary of Revenue Limits in the Appendix should not refer to RPI and any replacement provided for it by the Department of Employment, whilst the Rules refer to Retail Prices Index and any replacement index adopted by the Trustees. If that were the case, it would, of course, cease to be so clear that the pension increases would not breach Revenue Limits. But Rule 32 provides that they must not breach Revenue Limits, so the Trustees would not be able to increase pensions in such a way that they did so.
80. The first element of factual matrix that I regard as indicative is the fact that the definition of "INDEX" in the 1978 Rules was in a form that made very clear that only the official Retail Prices Index or a substitution for the Retail Prices Index published by the authority responsible for it could be employed. The new definition was not just a matter of drafting style (though that was quite different). The Rule 53 definition introduced two new aspects entirely missing from the 1978 definition. First, it introduced the concept of any replacement being "adopted by the Trustees", which plainly gave the Trustees some new discretion, and secondly, it provided that whatever was adopted must not prejudice "Approval". It is true that the term "Approval" is only defined in the Appendix and not in the Rules, but Mr Simmonds fairly accepted that not much could be drawn from that apparent drafting anomaly. It seems to me that it may properly be said that the introduction of these two new aspects of the definition point towards a substantive change having been intended. It would be surprising, but I accept not impossible, for the draftsman to have introduced these elements whilst intending to achieve the same result as was clearly the meaning of the 1978 definition.
81. The terms of IR12 are also, as I have said, admissible context. The version of IR12 at the relevant time made it clear that pension increases could be made by reference to RPI or "by any other suitable index agreed for the particular scheme by the Superannuation Funds Office".
82. Mr Simmonds accepted that there were at the time two forms of Revenue approval. The first was mandatory approval where the terms of the scheme satisfied statutory requirements later reflected in the Income and Corporation Taxes Act 1988 (which Mr Rowley told us few schemes actually achieved). The second form was discretionary approval, which required the draftsman to show that the terms of the scheme satisfied the provisions of IR12. This is reflected in the Foreword to IR12 itself. In these circumstances, it seems to me that there was substance in Mr Rowley's submission that the draftsman might have been expected to have been trying to satisfy the provisions of IR12. Even though Mr Simmonds was right to say that the provision of IR12 concerning appropriate indices introduced a flexibility that the draftsman had no obligation to adopt, I think the existence of that approach in IR12 points towards the likelihood that a draftsman trying to satisfy the provisions of IR12 would want to reproduce it in the rules he was drafting. This is supported by the fact that it seems likely that a similar kind of definition having the effect contended for by the employer was included in the Encyclopaedia of Forms and Precedents, at least by 1999. We were asked to infer that it was there by 1991, but that has not been confirmed.
83. The next element of relevant factual matrix is the Pension Schemes Office Manual, which contains guidance prepared for the staff of the Inland Revenue in considering the approval of pension schemes. Although the copy we have been shown was published in December 1995, we were asked to infer that a similar manual would have been available to Inland Revenue staff at the time the Rules were drafted. In relation to the indices to be used to increase pensions, the manual says that "[t]he general principle to be followed is that increases should protect the real value of the pension's purchasing power". It explained that the most commonly used index for calculating cost of living increases was RPI, and that if any other index was proposed, the practitioner should be asked why it was considered more appropriate than RPI, and the answer should be referred to the Section Manager for a decision. It was common ground that the purpose of the Rule 53 definition was accurately summarised by the judge in paragraph 78 of his judgment as follows:-

"The Rules provide for increases to pensions in payment and in deferment: the purpose of such a provision is to protect the members to some extent at least from the effects of price inflation. The definition of "Retail Prices Index" is there to provide a measure by which those increases are to be awarded. A power to switch the index, if such a power exists, ought properly to be exercised only to ensure that the index in use best reflects the policy of providing protection from inflation."

84. Both sides pointed to the competing consequences produced by the different constructions of the Rule 53 definition. Mr Simmonds said that it was inconsistent and inappropriate to give such a power to the Trustees alone, when all other powers of amendment were vested by Rule 46 in the employer and required the consent of the Trustees. Mr Rowley, on the other hand, pointed to the inconvenient inflexibility of having no power to change the appropriate index absent the abolition or formal replacement of RPI, particularly when Rule 46 could not be used to change to a less generous index even in the absence of section 67 of the Pensions Act 1995, because of the first proviso to that Rule. The employer submitted also that business common sense pointed clearly towards the Trustees having a discretionary power to change the index in case it ceased properly to reflect the level of inflation that the pension increases were intended to protect members of the Scheme against.
85. Both parties accepted that the construction exercise should be undertaken as at the time that the Rules were agreed, which was actually some time after 1988 in April 1991. It would not be appropriate to consider economic factors that arose only after that time. In particular, I accept Mr Simmonds' criticisms of the submissions Mr Rowley made about how CPI has become the more appropriate index and it would be uncommercial if the Trustees were unable to change to it. It is quite clear that CPI was not in existence in 1988 or 1991 and that RPI was the main accepted index at these times.
86. Against this background, it is necessary to apply the principles the judges stated and those helpfully summarised in Arden LJ's judgment in British Airways supra to determine which of the two possible meanings of the Rule 53 definition is the correct one.
87. Ultimately, I have taken the clear view that the judge was wrong in the conclusion he reached and that the correct construction of the Rule 53 definition is that Retail Prices Index as used in the Rules means "The General Index of Retail Prices published by the Department of Employment or any replacement [index of prices which is] adopted by the Trustees without prejudicing Approval". As regards the language of the first sentence of the Rule 53 definition itself, I regard it as equally consistent with either of the two possibilities that I have adumbrated. For the reasons, I have given I do not gain much assistance from either the second sentence or the Appendix, though I accept that they point marginally towards the judge's construction.
88. I do, however, find the context that I have already summarised as indicating clearly that the draftsman would have been likely to have wanted to provide a meaningful discretion to the Trustees to choose another index other than RPI. Other such indices were known about and available at the time. IR12 contemplated a choice between RPI and other indices, as did the Pension Schemes Office Manual. Moreover, the introduction of a discretion in the Trustees and of the need for the exercise of that discretion not to prejudice Revenue approval points to an intentional change in the effect of the definition from that contained in the 1978 Rules. On the judge's construction, there was no substantive change, because the previous Rules would have allowed the index to be changed if RPI were replaced by another index, and that is the only situation in which the judge envisaged that the Trustees could exercise their discretion. It is, I think, far-fetched to think that the drafting changes were made just in case RPI was replaced by more than one index. That was hardly a likely possibility.
89. It is, in my judgment, far more likely that the draftsman using his own drafting style was seeking to follow the suggestion made in IR12 to the effect that the Trustees might wish to choose a replacement index other than RPI and to submit it for approval to the Pension Schemes Office. I do not think that the fact that this construction would create a potential difference between the possible increases in pensions under the Rules and the Revenue Limits summarised in the Appendix is a good reason why the power to switch to another more appropriate index would not have been included. If there were a problematic conflict, it could always be remedied by the use of the amendment power in Rule 46.
90. Both sides accept that a resort to business common sense is available on the basis of an accepted ambiguity in the Rule 53 definition (see Lord Clarke's well-known dictum in Rainy Sky SA v. Kookmin Bank [2011] 1 WLR 2900 at paragraph 21 cited by the judge at paragraph 10). There is obvious value in flexibility for the Scheme if the Trustees could choose a more appropriate index reflecting inflation to protect members' pensions. As Millett J said in Re Courage Pension Schemes [1987] 1 WLR 495 at page 505G in relation to powers of amendment: "[i]t is important to avoid unduly fettering the power to amend the provisions of the scheme, thereby preventing the parties from making those changes which may be required by the exigencies of commercial life". I accept, therefore, that there would have been business common sense, even in 1988, in including a discretionary power in the Trustees (even without an employer's veto) to change the index in case it ceased properly to reflect the level of inflation. I do not, however, think that the resort to business common sense would by itself be sufficient. Rather I think it is one factor which, when taken alongside the words used and the contextual matters I have mentioned, leads to the clear conclusion that the Rule 53 definition must be taken to have been intended to give the Trustees a discretion to choose another index to replace RPI.
91. I would therefore hold that the Rule 53 definition should be construed as meaning the "General Index of Retail Prices published by the Department of Employment or any replacement [index of prices which is] adopted by the Trustees without prejudicing Approval". I have had the opportunity of seeing the judgments of Lewison and McFarlane LJJ in draft, and the appeal on this point will, therefore be dismissed by a majority.
92. On the section 67 cross-appeal, I am in complete agreement with Lewison LJ. In these circumstances, I would dismiss the members' cross-appeal.