England and Wales Court of Appeal (Civil Division) Decisions

IN THE COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM THE HIGH COURT OF JUSTICE
QUEEN'S BENCH DIVISION, ADMINISTRATIVE COURT
Lord Justice Elias, Mr Justice McCombe and Mr Justice Sales
Case Nos CO/3570/2011 and CO/4082/2011

		Royal Courts of Justice Strand, London, WC2A 2LL
		20/03/2012

The Master of the Rolls:

1. This is an appeal from a decision of the Divisional Court (Elias LJ and Sales J, McCombe J dissenting in part and in the result), rejecting a challenge to the Government's alteration to the basis upon which public service pensions are annually adjusted (or 'up-rated') to take account of inflation. Such adjustments are normally made each April by statutory instrument, and, for many years, they have been up-rated in accordance with the increase in the Retail Price Index ('RPI') over the year ending the previous September. However, the Government decided that, from and including April 2011, such adjustments should be made in accordance with the increase in the Consumer Price Index ('CPI'), rather than RPI, over the year ending the previous September.

A summary of the issues

2. Until recently, public service pensions were all based on final salary, but more recent schemes base such pensions on career average salary. Such pensions (together with some state benefits) are statutorily required to be annually up-rated to take into account inflation. The decision made in April 2011 to base any up-rating on CPI rather than on RPI, if effective, would detrimentally affect the value of every pension in payment. And, in the case of career average schemes, it would also detrimentally affect the way in which the career average is calculated. The question raised by these proceedings is whether the decision to up-rate public service pensions by reference to CPI rather than RPI, and the statutory orders implementing that decision, were lawfully taken and made.
3. Before us, as before the Divisional Court, there are a number of applicants, who include a few individuals but are mostly trades unions and other bodies representing large sections of public employees. Some of the applicants are represented by Mr Beloff QC and Mr Westgate QC, and others by Mr Giffin QC and Mr Knight. Before the Divisional Court, the applicants advanced, between them, four main grounds (all of which were rejected) for challenging the decision to change the basis upon which annual inflation adjustments are made. Before us, only two of those grounds are maintained.
4. The first ground it is convenient to consider is advanced by Mr Beloff. It is that the centrally relevant statutory provision, section 150 of the Social Security Administration Act 1992 ('the 1992 Act'), does not permit the annual inflation adjustments to be made by reference to CPI, because of the way in which CPI is compiled. Ultimately, once the relevant feature of CPI is appreciated, this issue turns on the proper interpretation of section 150 of the 1992 Act ('section 150').
5. The second ground has been primarily developed by Mr Giffin, with Mr Beloff making some supportive points. That second ground is that, when making the decision to make the annual inflation adjustments by reference to CPI rather than RPI, the Government (or, more precisely the Secretary of State for Work and Pensions, whose decision it was) took into account an irrelevant consideration; to put what is essentially the same point another way, the Government made the decision for an improper purpose.
6. The Divisional Court unanimously rejected the applicants' case on the first ground – see [2011] EWHC 3175 (Admin), paras 44-51. On the second ground, Elias LJ and Sales J again found for the Government (see [2011] EWHC 3175 (Admin), paras 52-67), while McCombe J agreed with the applicants – see [2011] EWHC 3175 (Admin), paras 118-139.
7. The structure of the remainder of this judgment is as follows. First, I shall discuss RPI and CPI; then I shall describe, in summary form, the evidence relating to the choice of CPI as the index used for the 2011 up-rating; next, I shall set out the relevant statutory provisions. I shall then turn to the first main issue, described in para 4 above. I shall then deal with the second issue, described in para 5 above, by considering three questions, namely (i) whether the effect on the national economy can be taken into account by the Secretary of State when selecting an index for the purposes of up-rating; (ii) if not, whether the 2011 up-rating could, as a matter of principle stand; and (iii) whether, in the light of the answers to (i) and (ii), the decision in 2011 to up-rate by reference to CPI can stand.

The two indices, RPI and CPI

8. The following description is taken in part from the Divisional Court's judgment, [2011] EWHC 3175 (Admin), paras 7-15.
9. RPI and CPI both measure the change in the level of prices, and each is produced and disseminated by the Office for National Statistics ('ONS'). The ONS is an independent public body responsible for producing a range of national statistics, and it is an executive office of the UK Statistics Authority, which now has statutory status and reports directly to Parliament.
10. In a publication dated 2010, 'Implications of the Differences between [CPI] and [RPI]', the ONS describes them as 'the two main measures of consumer price inflation'. In the same publication, the ONS states that 'CPI and RPI both measure the average change in price of a fixed basket of goods and services over time.'
11. In another 2010 publication, 'Consumer Price Indices Technical Manual', the ONS describes CPI as 'the main domestic measure of inflation for macroeconomic purposes' and RPI as 'the most long-standing general purpose measure of inflation in [the UK]'. Later in the Manual there are slightly fuller definitions. RPI 'is defined as an average measure of change in the prices of goods and services bought for the purpose of consumption by the vast majority of households in the UK.' And CPI is described in somewhat less friendly terms as being 'a Laspeyres-type consumer inflation or pure price index measuring the average price change on the basis of changed expenditure of maintaining the consumption pattern of households and the composition of the consumer population in the … reference period'.
12. RPI in its current form dates back to the 1950s. CPI was introduced in 1996, and is governed by EU regulations, as it was designed to enable comparison of inflation levels in different European countries. CPI has been used by the Bank of England as the headline measure for price inflation in December 2003.
13. There are a number of similarities in the ways in which RPI and CPI are determined. They are both calculated by reference to representative goods and services. Each year the ONS identifies a 'shopping basket' of around 700 representative goods and services on which consumers typically spend their money. The items will be changed each year so as to ensure that they reflect changes in the pattern of consumer spending. It is the movement in the price of these goods which is used to measure the relevant price changes. Prices are obtained from many outlets, and an overall inflation rate is worked out by a process of, first, aggregating particular items into defined categories of products and calculating an inflation rate within each category, and then by weighting those categories and the inflation rates within them so as to produce a single overall figure for inflation.
14. There are three main differences between RPI and CPI. First, they are weighted differently in that they reflect different population bases. The population base used in calculating the RPI is narrower than that which is used for determining CPI. The CPI includes all UK private households and foreign visitors to the UK. By contrast, the RPI excludes a number of households including those households where income is in the top 4% nationally. It also excludes pensioner households mainly dependent on state benefits, which constitute some 20% of pensioner households.
15. Secondly, there are certain differences in the goods and services which fill the relevant baskets. So, for example, university accommodation fees are included in CPI but not in RPI, and CPI does not include direct taxes such as TV licences, road tax, or council tax, which are included in RPI. CPI also excludes a number of housing costs, such as mortgage interest payments, building insurance, and depreciation, which again are included in RPI.
16. Thirdly, the basis for aggregation of rates of increase in the prices of items in the basket is different as between the two indices. The RPI uses an arithmetic mean to combine prices within each category of product at the first stage of aggregation whereas the CPI uses that for only around 30% of the categories. For the remaining 70% of categories, CPI uses the geometric mean. It has been estimated that this difference in methodology accounts for rather over 60% of the average difference between the two indices since 1997. This third difference is crucial to the first of the two grounds raised by the applicants, and so it is appropriate to explain it a little more fully.
17. The difference between the two means in terms of calculation may be demonstrated by an example which assumes that there are five types of apple included in an index. The arithmetic mean involves aggregating the price of each of the five apples and dividing the resulting figure by five. The geometric mean involves calculating the product of the five prices (i.e. multiplying them together) and then taking the fifth root of the resulting figure.
18. In practical terms, adopting a geometric mean implies the substitution by consumers of products in a particular category within the year under review whereas the arithmetic mean does not. If, for example, there is an exceptional increase in price of a particular variety of apple, the geometric mean assumes that some consumers will act rationally and switch to a cheaper type of apple. CPI is calculated on the assumption of substitution of products within each category at the lowest level of classification in constructing the index, and not more widely as between different categories. Thus, it does not assume that consumers will switch from apples to some other fruit, but only from one type of apple to another. (The 30% of goods which are subject to the arithmetic mean even for CPI purposes are those goods where the ONS has assessed that there is little opportunity for consumers to switch to a substitute product.)
19. In economic terms, it may be said that, by using the geometric mean, the CPI assumes a perfectly elastic system, or an elasticity of substitution of one, whereas, by using the arithmetic mean, the RPI assumes a wholly inelastic system, or zero elasticity of substitution.
20. The consequence of adopting the geometric mean is, thus, to dampen the effect of larger price increases within each category because it gives items with such price increases less weight when calculating the mean rate of price change than is given to other items in the same category of goods which have smaller price increases. By contrast, the arithmetic approach gives the same weight to the price increase rates for every item in the category. For this reason, the use of the geometric mean usually produces a lower average price rise than the arithmetic mean would do.
21. There is some controversy both as to the extent to which substitution is possible or occurs in practice and as to whether it is in principle appropriate to allow for it at all. Nonetheless the desirability of using a geometric mean in a price index is supported by a significant number of professional economists, who consider it a better method of assessing consumer choice over the requisite period. Indeed, it appears that consideration is currently being given to the possibility that RPI might use the geometric mean, at least with respect to some of its goods.

The factual background

22. Since the 1992 Act came into force, at least until 2010, public sector pensions have been annually up-rated in April by a percentage equal to, or sometimes slightly more than, the percentage increase in RPI over the year to the previous September. In 2010, however, RPI had actually fallen over the previous year, mainly because of the substantial fall in interest rates, and hence in mortgage costs. Presumably not least in the light of the forthcoming general election, the then Labour Government decided that it would be politically unacceptable not to increase public pensions, let alone to decrease them (although decreases do not seem to be contemplated by the applicable statutory provisions).
23. According to the evidence of Dr. James Richardson, Director of Public Spending at the Treasury, once it had been appreciated that RPI inflation to September 2009 would be negative, the Treasury had formed the view that a move to CPI would 'better reflect the experience of those affected by up-rating measures, i.e. benefit and public service pension recipients'. However, that option was not adopted, and, instead, an essentially pragmatic decision was made to up-rate certain social security benefits in 2010 by 1.5%. This was achieved through The Social Security Benefits Up-rating Order 2010, SI 2010/793. It was apparently anticipated by the Treasury that this increase would cost around £300m, and that the 'black hole' thereby created in the finances of the Department of Work and Pensions ('DWP') would be filled by a compensating measure, such as an uplift below the RPI increase, the following year.
24. The new coalition Government elected in May 2010 immediately focussed on ways of reducing the United Kingdom's deficit, and the Treasury put pressure on all other departments to find ways of reducing expenditure. The Secretary of State informed the Chancellor of the Exchequer ('the Chancellor') that one way in which DWP expenditure could be reduced was by switching from RPI to CPI as the index by reference to which the annual up-rating of public service pensions could be effected.
25. Writing on 2 June 2010 to the Chief Secretary to the Treasury, the Secretary of State explained that CPI was 'less volatile' than RPI, and that, unlike RPI, it did not include mortgage repayments 'which are of little or no relevance to the living costs of benefit recipients'. He also referred to the fact that the Bank of England used CPI, not RPI, to measure inflation. He explained that 'if we were to switch indices, we would need, of course, to ensure that we have a legally sustainable justification.' He then expressed the view that 'CPI is … a more appropriate measure of inflation.' On 19 June, the Chancellor recorded an agreement with the Secretary of State that CPI 'provid[ed] a better reflection of recipients' inflation experiences [than RPI] and ensur[ed] consistency with the measure used for monetary policy.'
26. In his budget statement of 22 June 2010, the Chancellor announced that CPI would be used as the basis for the annual indexation of benefits, tax credits and public service pensions from April 2011, in these terms:

'So from next year, with the exception of the State pension and pension credit, we will switch to a system where we up-rate benefits, tax credits, and public service pensions in line with consumer prices rather than retail prices. [CPI] not only reflects everyday prices better, but it is of course now the inflation measure targeted by the Bank of England. This will save over £6 billion a year by the end of the Parliament. I believe that this is a fairer approach than a benefits freeze.'

27. 'Budget 2010', a document printed by order of the House of Commons on 22 June 2010, stated that 'the Government will use the CPI for the price indexation of benefits and tax credits from April 2011'. The document then described CPI as 'a more appropriate measure of benefit and pension recipients' inflation experiences than RPI', on the ground that CPI 'excludes the majority of housing costs faced by houseowners …, and differences in calculation mean that it may be considered a better representation of the way in which consumers change their consumption patterns in response to price changes.' It then stated that '[t]his change will also apply to public service pensions …'.
28. There was, of course, a significant gap between the date of the Chancellor's statement in June 2010 and the decision of the Secretary of State, concurred in by the Chancellor, to increase pensions and other payments in accordance with CPI, as contained in a statutory instrument laid before Parliament in February 2011. During that period, it is clear that significant further consideration was given in the DWP and the Treasury, and some consultation took place, in relation to the possibility of switching from RPI to CPI for section 150 purposes.
29. In that connection, the contemporary documents and witness statements in these proceedings clearly show that (i) the relatively short term financial requirement to fill the 'black hole' which resulted from the 2010 1.5% up-rating, and (ii) the longer term desire to reduce Government expenditure were significant factors driving the DWP in deciding to base up-rating on CPI rather than RPI. Indeed, it is clear that they were the driving force (or to use Dr Richardson's expression, 'the policy imperative') for the decision. However, it also appears that the question whether to switch to CPI for the purpose of section 150 was also addressed more narrowly.
30. In his full witness statement prepared for the purpose of these proceedings, Mr Harry Cunniffe, a senior policy advisor to the DWP said this:

'As well as the considerable savings that could be made by the switch from the RPI to the CPI, the Secretary of State considered that the switch would have other advantages: (a) the fact that CPI was already the headline measure for price inflation, used by the Bank of England, and so a more widely recognised measure of the general level of prices; and (b) our view that the CPI was more suitable than the RPI as a measure of inflation for benefit claimants, as it excludes mortgage interest payments but includes all pensioner households.'

31. Mr Cunniffe also said that 'the Secretary of State considered that the fact that RPI moved into negative growth in 2009' because it included mortgage interest 'which had no bearing on living costs for most benefit and pension recipients' showed 'that it was not a reasonable benchmark for price inflation.' Accordingly, for this reason also, according to Mr Cunniffe, 'the Secretary of State concluded that CPI, which excludes mortgage interest, was a more appropriate index for up-rating purposes'. Mr Cunniffe added that the use of the geometric mean in CPI 'better reflected substitution behaviour' in appropriate categories of goods. He acknowledged that many pensioners paid council tax which was not in the CPI basket, but overall he concluded:

'it was recognised that no index can perfectly capture everyone's experience of inflation,.. it was considered [by the Secretary of State] that a single index should be used for up-rating and the CPI was considered to be the most appropriate.'

32. Dr Richardson, who also provided a full witness statement, referred to the Treasury's view 'as to the superiority of … CPI as a measure of the general level of prices in the UK in comparison to … RPI' for the reasons given by Mr Cunniffe for the purpose of up-rating all benefits, tax credits and public service pensions. The Treasury, he said, was of the view that 'CPI provides a fairer reflection of inflation experience than RPI over the longer term, including, for example, because it used the geometric mean'. He also said this:

'The policy imperative of a move to indexation by the CPI such as we had considered in 2009 became particularly urgent in early to mid 2010 given … the need to ensure long-term fiscal savings ...'.

33. Dr Richardson gave a number of reasons why the selection of CPI was better for the overall economy and 'would substantially increase the long-term sustainability of benefits and public service pensions expenditure'. However, the final reason which he gave for the change to CPI was that it 'would ensure that up-rating was carried out in a way which better reflected the inflation experience of pensioners and so would best protect the purchasing power of public service pensions against inflation'.
34. Meanwhile on 12 October 2010, the ONS released the September 2010 inflation figures, which formed the basis of the following April's uplift for benefits and pensions. The annual rate of inflation according to CPI was 3.1%, whereas according to RPI it was 4.6%. The Secretary of State approved a draft order implementing up-rating by reference to CPI increase, and it was laid before both Houses of Parliament on 3 February 2011. It was approved by the House of Commons on 17 February, and by the House of Lords on 14 March, 2011, in each case pursuant to the affirmative resolution procedure. There were debates in both Houses. The order was finally made on 16 March 2011 as the Social Security Benefits Up-rating Order 2011 SI 2011/821 ('the 2011 Up-rating Order'). The Treasury then made a consequential Pensions Increase (Review) Order 2011, SI 2011/827 – ('the 2011 Pensions Order') increasing pensions by the same amount.
35. Lord Freud, the Parliamentary Under-Secretary of State at DWP, said this during debate on the draft 2011 Up-rating Order in the House of Lords on 14 March 2011:

' … [T]here is a body of empirical evidence that people do substitute and that the geometric mean is an appropriate reflection of that. In Australia in 2009 a study … found that, in the overwhelming majority of cases, elasticity of substitution was much closer to one than to zero and therefore that the geometric mean was a more appropriate reflection of consumer behaviour. One of their key findings was that consumers are very responsive to price changes at the elementary aggregate level, the level on which the geometric mean operates. However, the study went further, finding that even the geometric mean might not fully capture substitution, with some elasticities exceeding one. There is separate evidence, for example, that brand-level elasticity is often more in the one and a half to two range.

Closer to home, also in 2009, the Scottish Government published an overview of evidence on food prices. … I hope that this reassures noble Lords that consumers do substitute when prices rise; not necessarily that they substitute all the time, for the geometric mean does not demand that; simply that some people will substitute when an item has risen sharply in price and there is a good substitute.

The CPI deals only with substitution on the elementary aggregate level, the lower level. …

Suffice to say that the theory and evidence for consumer substitution is compelling, that the geometric mean is an appropriate method of capturing that behaviour and therefore that the CPI's method of aggregation is superior. That is why the geometric mean is used in the consumer prices index of the United States, Canada, Australia, Denmark, Finland, Ireland, Italy, Luxembourg, … France, Portugal, Spain, Sweden and Austria.'

36. Similar statements were made in the House of Commons by the Minister of State, Steve Webb MP. One of them was on 8 December 2010 and is quoted in part at [2011] EWHC 3175 (Admin), para 27; another was on 8 December 2010, and it included the following:

'[W]hen we looked at this issue as a new Government, we were prompted particularly by the context of a year in which the RPI had been negative. … In April 2010, up-rating had been nil for the state earnings-related pension scheme, public sector pensions and all the connected pensions. That is not because inflation for pensioners had been nil … but because that is what the RPI said. The RPI was clearly not doing its job then, and that focused our mind on whether it was the right thing. It is true that, on average, the CPI tends to be lower – not always, but generally. I have looked at the past 20 years, and in five of those the RPI has been lower than the CPI. That improves the situation in a difficult financial position; I would not pretend that it does not. However, our job is to have an appropriate, stable measure of inflation, and that is what the CPI achieves. Indeed, it is much less volatile. ...'

The relevant statutory provisions

37. The statutory background to this case was admirably summarised by the Divisional Court at [2011] EWHC 3175 (Admin), paras 28-36, in the following terms:

'28. Public service pensions, including those for the civil service, police, the NHS and local government, may be increased in accordance with the rules established under the Pensions (Increase) Act 1971 [("the 1971 Act")]. That Act creates a link between public sector pensions and certain state benefits. The effect is that when benefits are increased to take account of the rise in prices that same rate is used to increase public service pensions.

29. The mechanism works as follows. Section 150(1) … obliges the Secretary of State [for Work and Pensions] to review certain sums annually

"in order to determine whether they have retained their value in relation to the general level of prices obtaining in Great Britain estimated in such manner as the Secretary of State thinks fit."

30. Section 150(2) then sets out what the Secretary of State must do if there has been a rise in the general level of prices:

"Where it appears to the Secretary of State that the general level of prices is greater at the end of the period under review than it was at the beginning of that period, he shall lay before Parliament the draft of an uprating order –

(a) which increases each of the sums to which sub-section (3) below applies by a percentage not less than the percentage by which the general level of prices is greater at the end of the period than it was at the beginning;

(b) if he considers it appropriate, having regard to the national economic situation and any other matters which he considers relevant, which also increases by such a percentage or percentages as he thinks fit any of the sums mentioned in subsection (1) above, but to which subsection (3) below does not apply; and

(c) stating the amount of any sums which are mentioned in subsection (1) above but which the order does not increase."

31. Section 150(3) then sets out certain benefits in social security legislation, such as the additional state pension. The effect, therefore, is that certain benefits are automatically up-rated in line with the percentage price increase whereas in the case of other benefits there is a discretion whether to give effect to that increase or not, and one of the factors the Secretary of State is required to consider in the latter case is the national economic situation.

32. Section 150(9) provides that the Secretary of State shall make an order in the form of the draft if it is approved by a resolution of each House.

33. Section 189(8) of the 1992 Act provides that an order under section 150 "shall not be made by the Secretary of State without the consent of the Treasury."

34. Where an up-rating order is made under section 150 …, section 59(1) of the Social Security Pensions Act 1975 [("the 1975 Act")] then requires the Treasury to make an order applying the same up-rating percentage used for the additional state pension (which is listed at section 150(1)(c) …) to what are described as official state pensions, as defined in the [1971] Act, which include the relevant pension schemes in issue in this case. So far as relevant, section 59(1) states:

"Where by virtue of section 150(1) … a direction is given that the sums mentioned in section 150(1)(c) …are to be increased by a specified percentage the Minister for the Civil Service shall by order provide that the annual rate of an official pension may if a qualifying condition is satisfied or the pension is a derivative or substituted pension or a relevant injury pension, be increased … by the same percentage as that specified in the direction."

35. It is no longer the Minister for the Civil Service who exercises that power, but the Treasury, pursuant to the Transfer of Functions (Minister for the Civil Service & Treasury) Order 1981.

36. Section 59(6) of the 1975 Act provides that an order made under this section has to be made by statutory instrument and shall be laid before both Houses of Parliament after being made.'

The legitimacy of using CPI for up-rating under section 150 of the 1992 Act

38. The argument that it is not open to the Government to use the annual change in CPI as the basis for annually up-rating public service pensions is put in two ways. First, it is said that, rather than, or at least in addition to, measuring changes in the general level of prices as envisaged by section 150, CPI measures consumer responses to changes in price. Secondly, it is said that, contrary to what is envisaged by section 150, up-rating by reference to the annual change in CPI does not involve comparing like with like, because CPI does not involve the general level of prices being assessed according to the same criteria at the start and the end of the year in respect of which the exercise is being carried out.
39. Both these arguments, which ultimately appear to me to be different ways of making the same point (though none the worse for that), turn on the feature of CPI discussed in paras 16-21 above, and in particular on the practical effect of using the geometric mean mentioned in para 18 above. In the end, the point resolves itself into the question whether an index compiled on the basis of the geometric mean, rather than the arithmetic mean, is a permissible basis for determining whether, and if so to what extent, state pensions and other annual payments have lost 'their value in relation to the general level of prices', within section 150.
40. In agreement with all members of the Divisional Court, and despite the elegantly expressed arguments of Mr Beloff to the contrary, I am of the view that the answer to that question is in the affirmative.
41. As a matter of ordinary language, the change in CPI over a given period can fairly be said to be a measure of the change in 'general prices' over the period in question. It is obvious, both as a matter of common sense and in the light of the closing words of section 150(1) ('estimated in such manner as the Secretary of State thinks fit'), that there will be more than one way of measuring a change in general prices over a particular period. Questions such as which particular goods and services one takes, precisely how one determines the price of each good and service, and how one weights each good and service, must, within the limits of rationality, be a matter of opinion and judgment. So too, it seems to me, must the question how one calculates the change in general level of prices, from the raw data of each specific increase in price, be a matter of opinion and judgment. In particular, whether to use the geometric mean or the arithmetic mean.
42. Support for this conclusion can be found from the publications issued by the ONS referred to in paras 10 and 11 above. In particular, I have in mind the description of CPI (as well as RPI) as 'measur[ing] the average change in price of a fixed basket of goods and services over time', and the definition of CPI as being a 'pure price index measuring the average price change'. It may well be that CPI can be described in different terms; so too it may well be that the descriptions I have just quoted may, at least in some people's views, justify adding a qualification (e.g. with the addition of the words 'on the basis of changed expenditure …' in the case of the second quotation). However, the fact that a thing may be fairly described as A as well as being fairly described as B, or that the description of a thing as B requires qualification, cannot, as a matter of logic, detract from the fact that the thing is fairly described as B.
43. The fact that the ONS describes CPI in terms which plainly assume that it may be used for the purposes of a provision such as section 150(1) is, at least in my view, of importance. When it comes to accounting procedures and practices for the purposes of satisfying statutory requirements, such as for assessing liability to tax, it is clear that well established accountancy practice (e.g. for assessing profit) is relied on by the court. So, too, it seems to me, when asking whether a particular index is appropriate for assessing whether there has been any change, and if so what change, in 'general prices', it is appropriate to consider whether the index sought to be used for the purpose of that assessment is regarded by the relevant professionals, in this case economic statisticians, as having the appropriate characteristics.
44. Also relevant in this connection is the fact that CPI is used for the purpose of comparing inflation rates in different EU member states, as is the fact that the Bank of England has used CPI to assess the level of inflation since the end of 2003.
45. It is true that the Government did not compile or publish an index calculated by reference to a geometric mean when the 1992 Act was enacted. However, in my judgment, it is highly relevant that CPI is aimed at assessing changes in prices of goods and services nationally, and is compiled by a method which has substantial support nationally (by the Bank of England and ONS), internationally (by the EU and by the individual countries mentioned by Lord Freud), and from relevant experts (as shown by the evidence in this case and by what was said by Lord Freud).
46. It would, in my view, have required clear exclusionary or limiting words in section 150(1) before one could fairly conclude that Parliament had intended that such an index could not be invoked by the Secretary of State for the purposes of the section. There are no such exclusionary or limiting words. Indeed, the closing words of section 150(1) seem to me to emphasise the width, rather than the narrowness, of the Secretary of State's field of choice.
47. In the end, it appears to me that the position is this. Section 150(1) leaves it to the Secretary of State to select the method by which he estimates whether, and if so to what extent, certain benefits and pensions have lost 'their value in relation to the general level of prices obtaining in Great Britain' during a particular year. The obvious way, or at least an obvious way, of making such an estimate is the use of an official, professionally compiled index, whose function is to measure the extent to which prices of consumer goods and services have increased in sterling terms over the period in question, and CPI is such an index. Inevitably, there will be different opinions as to the best way of compiling such an index, including the identification, pricing and weighting of its constituent items. Provided that the Secretary of State acts rationally and takes all appropriate (and no inappropriate) matters into account, it is a matter for him which such index he chooses. The instant criticism of his choice of CPI is based on the way in which the weighting of items within a category of items is effected. It is quite impossible to say, and Mr Beloff specifically eschews saying, that the weighting method adopted by CPI is irrational, and that, as I see it, is the end of the matter.

Taking account of the national economy when selecting an index under section 150(1)

48. Turning to the second ground relied on by the applicants, the first question it is convenient to consider is whether the Secretary of State can take into account the effect on the national exchequer when selecting the index for determining whether, and if so by how much, the purchasing power of certain benefits and pensions has declined for the purpose of section 150.
49. There is force in the view that the answer is no, for the reasons expressed by McCombe J, in his powerful dissenting judgment below, which, like the argument of Mr Giffin, has considerable attraction. However, I have come to the conclusion that, when selecting the appropriate index or other method for making his estimate under section 150(1), the Secretary of State is not precluded from taking into account the effect of his selection on the national economy, although it can only play a relatively attenuated role in limited circumstances.
50. In summary terms, the applicants' case is that, on a fair reading of subsections (1) and (2)(a) of section 150, what the Secretary of State has to do is first, under subsection (1) to assess how far the purchasing power of the pension or other payment has fallen during the year in question, and then, under subsection (2)(a), to compensate recipients of the payment by making an up-rating order which accurately reflects, or compensates for, that fall. Such an exercise, said Mr Giffin, leaves no room for taking into account the impact of the up-rating order on the national exchequer. He further contended that this argument is reinforced by the express reference to the Secretary of State being required to 'hav[e] regard to the national economic situation' in section 150(2)(b).
51. As Lord Reid put it in Padfield v Minister of Agriculture, Fisheries and Food [1968] AC 997, 1030:

'Parliament must have conferred the discretion with the intention that it should be used to promote the policy and objects of the Act; the policy and objects of the Act must be determined by construing the Act as a whole, and construction is always a matter of law for the court.'

To similar effect, in Tower Hamlets LBC v Chetnik Developments Ltd [1988] AC 858, 873, Lord Bridge of Harwich said:

'. . . [B]efore deciding whether a discretion has been exercised for good or bad reasons, the court must first construe the enactment by which the discretion is conferred. Some statutory discretions may be so wide that they can, for practical purposes, only be challenged if shown to have been exercised irrationally or in bad faith. But if the purpose which the discretion is intended to serve is clear, the discretion can only be validly exercised for reasons relevant to the achievement of that purpose.'

As Mr Giffin points out, the House of Lords in that case rejected the proposition that, in exercising a discretionary power to refund a mistaken overpayment of rates, the rating authority was entitled to have regard to its financial circumstances or to those of the ratepayer.

52. On the issue of interpretation, it is perhaps convenient to deal first with the point that the reference to 'the national economic situation' in section 150(2)(b) indicates that Parliament did not intend the Secretary of State to have regard to that factor when carrying out his functions under sections 150(1) and 150(2)(a). In my view, that is not a telling point.
53. By referring to the national economic situation in section 150(2)(b), Parliament was making it clear that, before the Secretary of State could make any up-rating, in his discretion under subsection (2)(b), he was required to have regard to that situation. It does not follow, as a matter of logic or natural implication, that he was not intended to be allowed to have regard to that situation when performing his functions under subsections (1) and (2)(a). In other words, the furthest this point goes is to suggest that the Secretary of State cannot always be required to consider the national economic situation when carrying out his duty under section 150(1) and (2)(a).
54. Mr Beloff argued that it was obvious that the Secretary of State would have regard to the national economic situation when according an up-rating under section 150(2)(b), and that, even without the express reference to it, any decision to accord an up-rating under section 150(2)(b) without referring to the economic situation would be judicially reviewable. I am rather inclined to agree, but, in my view, it is unnecessary to decide the point, as I do not think that it impinges on the point made in the immediately preceding paragraph.
55. That leads me back to the question whether the Secretary of State is entitled to take into account the effect on the public purse when making his decision as to which index, or other measure, to use for the purpose of estimating whether, and to what extent, public service pensions, and other payments, have failed to 'retain… their value in relation to the general level of prices obtaining in Great Britain' under section 150(1). The determination of that question raises, argued Mr Giffin, a purely objective assessment, as to which the consequences, whether for the recipients or for the Government, are irrelevant.
56. I accept that the estimation exercise to be carried out under section 150(1) must involve an honest and rational assessment. However, it is not, as Mr Giffin was initially disposed to suggest, an assessment to which there is, even in theory, a single right answer, as there is, in the example he gave, to the level of concentration of a solid dissolved in a liquid. As I have already said when considering the first point, there can no doubt be different opinions as to the best way of compiling such an index, including the identification, pricing and weighting of its constituent items. And provided that the Secretary of State selects an index which is a rational choice, his selection cannot be criticised, unless of course it is for impermissible reasons, which is what the applicants contend in this case.
57. As the oral argument developed, the parties' respective positions became clear. Take a case, which really reflects the position in the present proceedings, where there are two possible indices which the Secretary of State could use for making his estimate for section 150 purposes. The applicants' contention is that, when making that choice, the consequences for the national exchequer are simply irrelevant. (There was an arguable concession made below that such consequences may be a relevant factor if there is a choice between two equally good indices. Such a concession was not expressly made by Mr Giffin, and was effectively withdrawn by Mr Beloff before us; in these circumstances, it is best ignored.) The duty of the Secretary of State, on this argument, is simply to select the index which, in his view, best reflects the change in the level of prices in Great Britain for section 150(1) purposes. Mr Eadie, on behalf of the Secretary of State, contended that, once the Secretary of State has identified two indices which could properly be used for the purposes of his section 150(1) estimate, he is entitled to use that which is, in his opinion, the best from the viewpoint of the national exchequer.
58. When considering this issue, I do not find much assistance from the wording of section 150(1). It does not require, at least in terms, the Secretary of State to use what he considers to be the best index or method when making the estimate required of him. Indeed, the closing words of section 150(1) suggest that he is intended to have a relatively wide discretion. There is no doubt that he must seek to identify a rational basis for making his estimate, and one which he honestly believes achieves the aim of the section, namely to assess the decline in the purchasing power of benefits and pensions (and other payments). However, that does not rule out his taking into account the wider consequences, subject to proportionality.
59. An argument relied on by the applicants, and which weighed with McCombe J, is based on the structure of section 150. The argument is that subsection (1) simply requires the Secretary of State to estimate the decline in the purchasing power of benefits and pensions and other payments, and it is only after he has made that estimate, that up-rating under subsection (2)(a) comes into play. Accordingly, runs the argument, it is putting the cart before the horse for the Secretary of State to take into account factors which are only even arguably relevant to the subsection (2)(a) up-rating when making the prior subsection (1) estimate.
60. Seductive though it is, I do not accept that argument. Like all statutory provisions, section 150 must be read as a whole, and it is unreal to treat each subsection, or each step mandated by the section, as somehow insulated from, or blind to, any subsequent subsection or step. I consider that it is not an appropriate way to read a statutory provision to say that, simply because of the way in which section 150 is structured, the exercise to be carried out under subsection (1) must ignore the consequences under subsection (2).
61. Viewing the matter more broadly, the applicants' contention that, whatever the circumstances, the Secretary of State should, as a matter of course, be required wholly to put out of his mind the effect on the national economic situation when carrying out his functions under section 150(1) and (2)(a), seems to me unreal. The exercise required by section 150 is macro-economic in nature, unlike the micro-economic exercise involved in Chetnik Developments [1988] AC 858, and it has the obvious potential of having a significant effect on the country's finances. It therefore seems to me unrealistic to say that the Secretary of State is required to ignore the wider economic realities, irrespective of the circumstances, when carrying out his functions under section 150.
62. I cannot, however, accept Mr Eadie's argument without qualification. Thus, I do not consider that the Secretary of State could opt for an index which was clearly less good, and more detrimental to the recipients of pensions, than another index, simply because the former index was beneficial to the national exchequer. Indeed, if the Secretary of State thought that one index was significantly less reliable or less accurate than another, I find it very hard to conceive of any circumstances where he could select the former index merely because he thought it was just about acceptable for the estimating exercise required by section 150(1).
63. While I am not seeking to lay down a firm standard, it seems to me that, before the Secretary of State could invoke the benefit to the national exchequer by selecting an index he considered less good, three requirements would normally have to be met. Those requirements are (i) there would, in the Secretary of State's view have to be little to choose between the indices in terms of reliability and aptness, (ii) the benefit to the national exchequer of choosing the less good index would have to be significant, and (iii) the need to benefit the national exchequer, in terms of the national economy and demands on the public purse, would have to be clear.
64. In other words, the Secretary of State could only select the less good index if it was proportionate to do so, and, bearing in mind the purpose of the up-rating exercise, the circumstances would normally have to be unusual before it could be proportionate to select an index, or other method, which the Secretary of State considered was less good than another.
65. It may be said that this approach represents a dangerous concession by the courts to the executive, as it would permit the wider economic consequences to be taken into account by any decision-maker in any circumstances. I do not agree. First, as cases such as Chetnik [1988] AC 858 make clear, the question has to be determined by reference to the specific language, context and purpose of the statutory provision concerned. Secondly, this case concerns a macro-economic decision affecting millions of people, and therefore it will inevitably have very wide-ranging consequences. Thirdly, at least in my view, the ability of the decision-maker in this case to take into account the interests of the national exchequer is pretty circumscribed, as described in para 63 above. Fourthly, it seems to me unsurprising to suggest that the wider public economic interest can have a proportionate and limited bearing on an annual decision which will cost the national exchequer many hundreds of millions of pounds in most years.

What if the effect on the national economy cannot be taken into accout?

66. If the effect on the national economy cannot be taken into account by the Secretary of State when selecting the index for the purpose of up-rating, then the fact that the effect was taken into account in 2011, when selecting CPI for the purpose of up-rating pensions and other payments under section 150, would not automatically invalidate the 2011 Up-rating Order.
67. Where a decision-maker has taken a legally irrelevant factor into account when making his decision, the normal principle is that the decision is liable to be held to be invalid unless the factor played no significant part in the decision-making exercise. Thus, in Simplex GE (Holdings) Ltd v Secretary of State for the Environment (1989) 57 P&CR 306, 325-6, Purchas LJ (with whom the other two members of the Court of Appeal agreed) approved an observation of Forbes J in R v Rochdale Borough Council ex p Cromer Ring Mill Ltd [1982] 3 All ER 761, 766-7, explaining that a decision would not be set aside where the irrelevant factor was 'insignificant or insubstantial', as opposed to a case where the irrelevant factor's 'influence was substantial'.
68. Even where the irrelevant factor played a significant or substantial part in the decision-maker's thinking, the decision may, exceptionally, still be upheld, provided that the court is satisfied that it is clear that, even without the irrelevant factor, the decision-maker would have reached the same conclusion. Thus, in Simplex GE (1989) 57 P&CR 306, 326, Purchas LJ approved the following passage in the judgment of May LJ in R v Broadcasting Complaints Commission ex p Owen [1985] 1 QB 1153, 1177:

'Where the reasons given by a statutory body for taking ... a particular course of action are not mixed and can clearly be disentangled, but where the court is quite satisfied that even though one reason may be bad in law, nevertheless the statutory body would have reached precisely the same decision on the other valid reasons, then this court will not interfere by way of judicial review.'

In Smith v North East Derbyshire PCT [2006] 1 WLR 3315, para 10, (a different) May LJ said this:

'Probability is not enough. The defendants would have to show that the decision would inevitably have been the same and the court must not unconsciously stray from its proper province of reviewing the propriety of the decision making process into the forbidden territory of evaluating the substantial merits of the decision.'

(See also per Keene LJ at [2006] 1 WLR 3315, para 16, as well as Simplex 57 P&CR 306, 327 and 329, and R v Secretary of State for the Environment ex p Brent LBC [1982] QB 593, 646.)

69. There is, in theory at least, a possibility that, even if the court concludes that it ought otherwise to set aside a decision on the ground that a legally irrelevant factor was taken into account, it can nonetheless uphold the decision, if it is satisfied that it would be pointless to require the decision-maker to reconsider the question afresh, because he would reach the same answer. It appears to me that that is a theoretical point, at least in this case, because, if the Secretary of State cannot succeed in showing that the irrelevant factor was not a significant factor in his thinking or that he would have selected CPI as the relevant index anyway, it is hard to see how he could hope to persuade the court that there would be no point in setting aside the decision and requiring it to be reconsidered. The evidence all relates to the period in 2010 and 2011 when the decision was being made, so there is no reason to believe that the thinking of the Secretary of State, or indeed the Treasury, is any different now from what it was then.

Was the 2011 decision to up-rate by reference to CPI lawful?

70. Having considered the two questions of principle in relation to the second issue before us (as described in para 5 above), I must now turn to that issue: was the Secretary of State's decision to make the estimate in April 2011 under section 150(1), and his decision to make the subsequent 2011 Up-rating Order and the 2011 Pensions Order, valid? In order to answer this crucial question, it is appropriate to summarise the effect of the evidence set out in paras 22-36 above.
71. In that connection, it seems to me that certain matters are clear. First, (i) the need to fill the 'black hole' created by the 1.5% increase in 2010, and (ii) concerns about the future effect on the national economy provided the driving force, or spur, for the Secretary of State's selection of CPI rather than RPI as the index by reference to which up-rating was to be effected in 2011. That was realistically conceded by Mr Eadie. Secondly, even if its effect on the national economy could not have been taken into account, the selection of CPI was a rational decision for a Secretary of State to have made when effecting the up-rating in 2011. That was realistically conceded by Mr Giffin and Mr Beloff.
72. Thirdly, and perhaps more controversially, CPI was selected not merely because of the wider economic consequences, but also because it was believed to be an inherently more satisfactory index than RPI for up-rating purposes. That was for a number of reasons, which one can glean from the evidence of Mr Cunniffe and Dr Richardson, the contemporary documentation, and the statements of Lord Freud and Mr Webb MP.
73. These reasons are based on the various differences between CPI and RPI, and can be summarised as follows:

(i) CPI did not include mortgage interest and was therefore (a) less volatile than RPI and (b) more representative of many pension-receiving households than RPI;

(ii) CPI did not, unlike RPI, exclude some pension-receiving households, and hence was a better measure for up-rating pensions;

(iii) CPI was used by the Bank of England to measure inflation, and therefore was a better index for section 150 purposes than RPI;

(iv) The equivalent of CPI was used as an inflation measure in many other European countries, and so it was a better measure of inflation than RPI;

(v) CPI was compiled largely by reference to the geometric mean, and therefore was a better measure of changes in prices than RPI, which only used the arithmetic mean;

(vi) CPI was less volatile than RPI, and, unlike RPI, it would not have produced an unrealistic negative inflation rate for the year to September 2009.

74. There is no basis for suggesting that these reasons were not genuinely held by those advising the Secretary of State, and by the Secretary of State himself. Nor is there any basis for suggesting that these reasons do not represent rational grounds for choosing CPI rather than RPI, although, on their own, some of them, such as (iii), may appear to be something of a makeweight. Additionally, there is evidence from Dr Richardson that, in 2009, the Treasury thought that CPI would 'better reflect the experience of those affected by up-rating measures, i.e. benefit and public service pension recipients'. Furthermore, there is no evidence which suggests that, during 2010 or 2011, the Secretary of State or his advisers suggested, or even considered, that RPI had any advantages over CPI. Further, the June 2010 correspondence referred to in para 25 above shows that the Secretary of State (and the Chancellor) thought that CPI was more appropriate than RPI.
75. In all these circumstances, it seems to me that, irrespective of whether I am right about the Secretary's right to take into account the effect of his selection of an index on the national economy, the Secretary of State's decision to select CPI as the index by reference to which to up-rate under section 150 was valid.
76. As mentioned above, it was, in my view, open to him to take into account the effect on the national economy, provided that, in his rational view, (i) the index which he selected was not significantly less suitable for section 150 purposes than the alternative, (ii) the choice of index would have a significant effect on the national economy, and (iii) the state of the national economy justified it being taken into account. It seems to me that those three requirements were plainly satisfied here. The fact that the factor which initially drove the selection of CPI was the effect on the national economy does not alter the fact that CPI was considered on its merits to be an appropriate index for making the section 150(1) estimate for 2011.
77. As for the three requirements, the position appears to have been this in April 2011. (i) To put the point at its lowest, CPI was thought by the Secretary of State, by Lord Freud and Mr Webb, as well as by Mr Cunniffe and Dr Richardson, to be no worse than RPI. (ii) So far as the effect on the national economy was concerned, the effect of choosing CPI rather than RPI was significant. (iii) The Government clearly believed that the state of the national economy was grave, and that any savings which could properly be made should be made – and made as soon as possible; if that were not well known, it is obvious from the Chancellor's statement of 22 June 2010.
78. I turn now to the position if I am wrong in my view that it was open to the Secretary of State to take into account the effect on the national economy when selecting the index for up-rating purposes. On that basis, I would still hold that the up-rating decision in 2011 was lawful. On the assumption that the effect on the national economy could not be taken into account, it appears to me that this is one of those relatively rare cases when the court could properly (and therefore should) conclude that, even though the decision-maker took a legally irrelevant factor into account, and it was a 'substantial' (as opposed to an 'insignificant') factor, it is a factor which can be 'disentangled' from the other, valid, reasons for the decision, and, in the light of those reasons, the decision should stand as it plainly would have been the same if the factor had not been taken into account.
79. I have already referred in this section of the judgment to what Mr Cunniffe, at the DWP, and Dr Richardson, at the Treasury, said in their evidence, to what the DWP Ministers said to Parliament, and to what the Secretary of State wrote in June 2010. Even ignoring the effect on the national economy, all of the differences between RPI and CPI which were considered by those people, and are listed in para 74 above, favoured the selection of CPI, rather than RPI, as the appropriate index upon which to base the 2011 up-rating.
80. Putting on one side the benefit to the public purse of adopting CPI, as opposed to RPI, for the 2011 up-rating, it seems to me that the evidence in this case establishes the following points:

(i) The Secretary of State's senior policy adviser on up-rating issues considered that CPI was the more appropriate index to use for up-rating for reasons which were purely related to the inherent advantages of CPI as set out in para 74 above;

(ii) The Director of Public Spending at the Treasury took the same view for the same reasons;

(iii) The June 2010 correspondence shows that the Secretary of State also took that view for at least some of those reasons;

(iv) Perhaps of particular significance, given that the decision was ultimately that of the Secretary of State and these were contemporaneous public statements, the DWP Ministers in the Lords and the Commons also were of that view;

(v) none of the reasons which supported this view (as set out in para 74) could be said to be illogical or unreasonable;

(vi) the Secretary of State was not provided with a single reason to reject CPI and stick with RPI, save (presumably) that RPI had consistently been the index by reference to which up-rating had taken place over the previous twenty years or more.

81. In all these circumstances, while acknowledging the high hurdle which has to be crossed by a decision-maker before he can persuade the court that his decision would have been the same if he had ignored a factor which he illegitimately had taken into account, I am satisfied that that hurdle would be crossed in the present case. It seems to me that the decision to use CPI as the index by reference to which the 2011 up-rating should be effected would certainly have been made by the Secretary of State even if he had put out of his mind any consideration of the benefit to the national economy of that decision.
82. For the same reasons, it appears to me that, if we were to quash the decision to up-rate in 2011 by reference to CPI, and remit it to the Secretary of State on the basis that he should not have taken the national economic situation into account, he would inevitably reach the same conclusion. As explained in para 70 above, at least in this case, that is not an additional reason for reaching this conclusion, but another way of putting the same point.

Disposal

83. I would therefore dismiss this appeal.

Lord Justice Maurice Kay:

84. I agree.

Lord Justice Sullivan:

85. I also agree.