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
MR JUSTICE GARNHAM

		Royal Courts of Justice Strand, London, WC2A 2LL
		08/10/2021

Lord Justice Underhill:

INTRODUCTION

1. This appeal concerns an aspect of the benefit cap provisions as they apply to entitlement to Universal Credit ("UC") under the Universal Credit Regulations 2013, which are made by the Secretary of State for Work and Pensions under powers conferred by the Welfare Reform Act 2012. The Claimants, who are the Respondents before us, are a mother, Ms Sharon Pantellerisco, and her three minor children: for convenience I will refer to Ms Pantellerisco as if she were the only Claimant.
2. The Claimant first claimed UC on 4 February 2019, and her claim relates to her entitlement in the period between that date and the issue of proceedings on 12 September 2019. She was at that time employed by Sefton Carers' Centre as a carer for her grandmother, who has dementia. She worked sixteen hours per week (being the period specified in her grandmother's care plan) at the minimum rate prescribed by the National Minimum Wage Regulations 2015 ("the NMWR"), being £8.21 per hour: that rate is described in regulation 4 of the NMWR as "the national living wage rate", and I will refer to it as "the NLW". She was paid on a four-weekly cycle, rather than by the calendar month: as will appear, that fact is central to the issue on this appeal. The Council was what is known as a "real time information" ("RTI") employer, which means that it passed information about her earnings each month to Her Majesty's Revenue and Customs ("HMRC"), who in turn passed it to the Department of Work and Pensions ("the DWP") so that her benefit entitlements could be calculated automatically without her having to make monthly claims.
3. The essential features of the UC regime which give rise to the claim in these proceedings can be summarised as follows:

(1) By regulation 21 of the 2013 Regulations, UC is calculated (subject to some immaterial exceptions) by reference to monthly "assessment periods", running from the date that the claimant makes their first claim. The month in question is a calendar month. Since the Claimant first made her claim on 4 February 2019, her assessment period was from the 4^th of one month to the 3^rd of the next.

(2) Entitlement in respect of any assessment period is arrived at by taking a "maximum amount", calculated by reference to the claimant's particular circumstances, and then deducting any unearned income and a prescribed proportion of any earned income: the paradigm of such earned income – though of course far from the only case – is income from regular employment.

(3) By regulations 79-81 a claimant's entitlement in any assessment period is subject to the so-called "benefit cap", which is set at an amount varying according to the characteristics of the particular claimant, but which in the Claimant's case was £1,666.67 per month.

(4) In order to encourage claimants to work, regulation 82 provides for the disapplication of the benefit cap in cases where their earned income in an assessment period is equal to or exceeds a minimum amount. This is generally referred to as "the earnings-related threshold", though that is arguably not the most apt label – see para. 16 below.

(5) The threshold amount is the amount of earned income that the claimant would receive if they worked for sixteen hours a week during that period at the NLW rate. That reflects the general rule under the Regulations that "earned income" is "based on the actual amounts received in the period" (regulation 54). The fact that the threshold is defined by reference to income received rather than to income earned (in the sense of accrued) is also central to this case.

4. Although permanent employees with regular hours are most typically paid on the basis of the calendar month, it is not at all uncommon for employers to pay instead on a four-weekly, or 28-day, cycle. Where a UC claimant is paid on such a cycle, as in the present case, the inevitable effect is that in eleven of the twelve assessment periods in any given year they will receive four weeks' pay (one-thirteenth of their annual earnings) and that in the twelfth they will receive eight weeks' pay (two-thirteenths).
5. The consequence of that pattern in the Claimant's case was that for eleven months of the year she was unable to take advantage of the disapplication of the benefit cap provided for by regulation 82: although in each reference/assessment period she was working for sixteen hours per week at the NLW rate, she was only receiving payment for her work over 28 days rather than the full calendar month. The result was that the cap applied and her entitlement was some £490 less per month than it would have been otherwise – a reduction of about 20%. Although the cap was disapplied in the single "two-pay-day" month, this could not of course make up for what was lost in the previous months.^[1] It is her case that this effect is arbitrary and irrational, and that the Regulations are unlawful to the extent that they give rise to it.
6. At first instance the effect described above was described as "the lunar month problem", but that phrase is not entirely apt. Quite apart from the fact that the periods with which we are concerned are based on the fact of four-weekly payment rather than on the phases of the moon, the effect of which the Claimant complains would also be suffered by any UC claimant working the same hours at the NLW rate who is paid weekly or fortnightly. Accordingly I will use the labels "pay-cycle effect" or, when referring specifically to the Claimant's case, the "28-day cycle effect".
7. By a judgment handed down on 20 July 2020 Garnham J declared that:

"… [T]he calculation required by regulation 82 (1) (a) read together with regulation 54 of the Universal Credit Regulations 2013 is irrational and unlawful in so far as employees who are paid on a four weekly basis (as opposed to a calendar monthly basis) are treated as having earned income of only 28 days' earnings in 11 out of 12 assessment periods a year."

8. This is an appeal against that order, with permission granted by Dingemans LJ. The Secretary of State is represented by Mr Edward Brown, leading Mr Stephen Donnelly, and the Claimant by Mr Richard Drabble QC, leading Mr Tom Royston. All counsel also appeared before Garnham J. The case was very well argued on both sides.

THE RELEVANT PROVISIONS

9. I need not give a general introduction to Universal Credit. As is well-known, it represents a major reform and simplification of the previous welfare system. It replaces a large number of previous entitlements, but it has been introduced by gradual stages and at the dates material to the claim many of the previous "legacy benefits" remained available for some claimants.
10. The legislative basis of UC is to be found in Part 1 of the 2012 Act. Sections 7 and 8 are headed "Awards" and read (so far as material):

"7. Basis of award

(1) Universal credit is payable in respect of each complete assessment period within a period of entitlement.

(2) In this Part an 'assessment period' is a period of a prescribed duration.

(3) ...

(4) In subsection (1) 'period of entitlement' means a period during which entitlement to universal credit subsists.

8. Calculation of awards

(1) The amount of an award of universal credit is to be the balance of—

(a) the maximum amount (see subsection (2)), less

(b) the amounts to be deducted (see subsection (3)).

(2) The maximum amount is the total of [various elements specified in sections 9-12].

(3) The amounts to be deducted are—

(a) an amount in respect of earned income calculated in the prescribed manner …, and

(b) an amount in respect of unearned income calculated in the prescribed manner ….

(4) …"

"Prescribed" means prescribed by regulations made by the Secretary of State: see sections 40 and 42 (1). The effect of section 43 (3) is that the Regulations in their original form were subject to the affirmative resolution procedure.

11. The benefit cap is also a product of the 2012 Act, but it is not specific to UC. It is provided for under Part 5 of the Act (headed "Social Security: General"). Section 96 reads (so far as material):

"Benefit cap

(1) Regulations may provide for a benefit cap to be applied to the welfare benefits to which a single person or couple is entitled.

(2) For the purposes of this section, applying a benefit cap to welfare benefits means securing that, where a single person's or couple's total entitlement to welfare benefits in respect of the reference period exceeds the relevant amount, their entitlement to welfare benefits in respect of any period of the same duration as the reference period is reduced by an amount up to or equalling the excess.

(3) In subsection (2) the 'reference period' means a period of a prescribed duration.

(4) Regulations under this section may in particular—

(a) make provision as to the manner in which total entitlement to welfare benefits for any period, or the amount of any reduction, is to be determined;

(b) make provision as to the welfare benefit or benefits from which a reduction is to be made;

(c) provide for exceptions to the application of the benefit cap;

(d)-(g) …

(5)-(9) …"

"Welfare benefits" is defined in paragraph (10): it covers not only UC but a wide range of legacy benefits. Paragraph (10) also defines "prescribed" as meaning prescribed by regulations.

12. I have already identified at para. 3 above the particular provisions of the Regulations that give rise to the issue in this case, but I need to say a little more about regulations 21, 54 and 82.
13. Regulation 21 falls under Part 3, which is headed "Awards". Regulation 20 states that Part 3 "contains provisions for the purposes of sections 7 and 8 of the Act about assessment periods and about the calculation of the amount of an award of universal credit". Regulation 21 itself is headed "Assessment periods". Paragraph (1) reads:

"An assessment period is a period of one month beginning with the first date of entitlement and each subsequent period of one month during which entitlement subsists."

Schedule 1 to the Interpretation Act 1978 defines "month" as "calendar month" (though it would in fact be clear in any event from the other provisions of regulation 21 that a calendar month was intended).

14. Regulation 54 falls under Part 6, which contains detailed provision for how to calculate a claimant's capital and income for the purpose of determining the amount of any award. Chapter 2 is concerned with earned income. Regulation 54 is headed "Calculation of earned income - general principles". Paragraph (1) reads:

"The calculation of a person's earned income in respect of an assessment period is, unless otherwise provided in this Chapter, to be based on the actual amounts received in that period."

15. Part 7 of the Regulations contains provisions relating to the benefit cap. The provisions applying the cap are at regulations 78-81. I need not set them out in full, but I should quote paragraphs (1) and (2) of regulation 79, which read:

"(1) Unless regulation 82 or 83 applies, the benefit cap applies where the welfare benefits to which a single person or couple is entitled during the reference period exceed the relevant amount determined under regulation 80A (relevant amount).

(2) The reference period for the purposes of the benefit cap is the assessment period for an award of universal credit."

16. Regulation 82 provides for exceptions to the application of the cap. I need only set out paragraph (1), which reads (so far as material):

"The benefit cap does not apply to an award of universal credit in relation to an assessment period where –

(a) the claimant's earned income … is equal to or exceeds the amount of earnings that a person would be paid at the hourly rate set out in regulation 4 of the National Minimum Wage Regulations for 16 hours per week, converted to a monthly amount by multiplying by 52 and dividing by 12; or

(b) …"

It is thus paragraph (1) (a) of regulation 82 which provides for the earnings-related threshold and which is the subject of the Claimant's challenge and Garnham J's declaration. At the risk of repetition, I emphasise that because of the definition of "earned income" in regulation 54 (1) the threshold refers to earnings actually received, not earnings accrued. That is why the label "earnings-related threshold" is potentially misleading: it might be more accurate to refer to it as "payments-related".

17. I should note for completeness, though no point arises in relation to it, that regulation 82 (1) as originally drafted provided for an earnings-related threshold in the form of a specified sum calculated on the basis of sixteen hours work per week at the then national minimum wage rate (rounded up). The current version, which was introduced by regulation 2 of the Universal Credit (Benefit Cap Earnings Exception) Amendment Regulations 2017, is substantially similar in effect but is more flexible in as much as it accommodates increases in the NLW without the need to amend the Regulations.
18. The Regulations in their original form were approved by a resolution of each House of Parliament pursuant to the requirement noted at para. 10 above.

THE EVIDENCE

19. The evidence which was before Garnham J at the hearing was mostly not disputed, and most of the essential facts sufficiently appear from paras. 2-5 above. That being so, I can summarise it fairly shortly. There are one or two points which it will make more sense for me to return to in detail when I consider the issues.
20. As for the Claimant's evidence:

(1) She herself made two witness statements explaining her circumstances and the impact on her of having her UC capped, which it is clear was very substantial. She also explains how it was in practice impossible for her, in her particular circumstances, to avoid the 28-day cycle effect either by increasing her hours or her earnings or by persuading her employer to pay her by the calendar month.

(2) Ms Carla Clarke of the Child Poverty Action Group ("CPAG") gave further details of the working of the 28-day cycle effect in the Claimant's case and about the interactions between her and CPAG on the one hand and the DWP on the other when the problem first arose.

(3) Mr Thomas Lee, a policy analyst at CPAG, gave figures showing that in the third quarter of 2019 the hours worked by parents earning at the NLW rate (or less) tend to cluster around certain multiples, such as sixteen and twenty. 124,000 parents earning at that rate work sixteen hours per week. The figures involve an element of estimation, but there is no challenge to their broad accuracy.

(4) Ms Helen Hargreaves, at that time the Associate Director of Policy at the Chartered Institute for Payroll Professionals ("the CIPR"), made a witness statement dealing with "payroll frequency" – that is, the intervals at which employees are paid. Her evidence, based on responses to a survey of employers carried out by the CIPR annually, is to the effect that payment by the calendar month has for many years been the most common pay frequency, being used by no fewer than 96% of employers. Payment on a 28-day cycle is the next most common, being used by 14.5% of respondents to the survey (and 20% among employers in sectors traditionally associated with low pay). (The reason why those figures come to more than 100% is that some employers use both cycles for different employees.)

21. The evidence on behalf of the Secretary of State took the form of a witness statement from Ms Carol Krahé, a civil servant with policy responsibility for the benefit cap. This exhibits a wealth of contemporary documentation evidencing the development of the policy which underlies the impugned provisions of the Regulations. The passages in her statement which are central to the issues before us are as follows.
22. First, at paras. 14-15 she introduces the policy background to the Regulations:

"14. UC was designed as a measure with a far-reaching social purpose. In particular, it was intended to bring about significant behavioural changes, incentivise work and increased earnings and to make the system simpler and fairer. It was designed to address both the complexity and perverse incentives inherent in the legacy benefit system, which had arisen as a result of its piecemeal development over a number of years.

15. Therefore, the policy objectives of UC included:

(a) developing a system that was affordable, rewarding work and personal responsibility;

(b) establishing a fairer relationship between benefit recipients and those who pay for them, particularly between out-of-work benefits and those receiving low pay;

(c) targeting financial support more efficiently, by supporting those invulnerable circumstances;

(d) establishing a simpler system for individuals to understand and for the Government to administer;

(e) designing a system that will operate for all classes of case, straddling in work and out of work cases, which is essential if the system is to have strong work incentives."

23. At paras. 20-28 she gives evidence about the choice of an assessment period of a calendar month. Paras. 21-24 read:

"21. The calculation of UC in each monthly assessment period is a cornerstone of UC policy. All changes that occur in the assessment period are applied to the whole assessment period, and each policy consideration is looked at across the assessment period – such as the inclusion of disability elements, child elements, childcare costs, carer's element, conditionality
arrangements, the treatment of income, capital, deductions, etc.

22. The assessment period is calculated as a calendar month. A calendar monthly basis is used as it is considered to best reflect the most common payment cycles (whether in terms of income, such as salary, or outgoings, such as bill payments).

23. The objective of workability and efficiency requires the same structure to operate for the whole population, notwithstanding that there are, of course, different types of payment cycles (such as irregular pay or weekly or lunar monthly pay).

24. The calendar month structure reflects the general position in modern working life, where individuals, even in more precarious employment, are usually paid monthly. Where claimants are unemployed, monthly assessment and payment of UC creates the discipline of budgeting and managing money on a monthly basis, which is considered to help improve skills which would reduce poverty whether in work or not. The same approach is applied whether a claimant is employed, unemployed or self-employed. This allows UC to be calculated on the same basis whether a person moves in and out of work or whether their earnings are composed of mixed employed and self-employed earnings."

She goes on to quote from the speech of the responsible minister, Lord Freud, in the House of Lords during the passage of the bill that led to the 2012 Act. I need not reproduce the quotation here, but it makes clear that the choice of a calendar monthly assessment period was a matter of deliberate government policy, for the reasons summarised by Ms Krahé in the passage quoted above; and she says that the same point was made on other occasions in the same debate and in the debate on the approval of the Regulations. At para. 28 she says:

"The Secretary of State decided to use the monthly assessment period … as the central component of UC. Accordingly, all entitlements are calculated on this basis, including, for present purposes, the application of the benefit cap. Other periods were considered, but resulted in other issues regarding entitlement which were likely to be detrimental to individuals and to undermine the social policy objectives of UC."

24. Paras. 29-34 explain the approach to the calculation of earnings for the purpose of UC. Para. 30 quotes from a submission to the Minister, as follows:

"Taking earnings into account in the assessment without averaging or attribution – As a first principle we want to reflect the cash flow into the household by taking into account the amount of earnings received in the assessment. For the majority of straightforward cases this will mean there is no need to apply complex averaging or attribution rules."

That recommendation was accepted, and we have seen that it is reflected in the definition of "earned income" in regulation 54. The insistence on basing entitlement on actual receipts in the assessment period – which in the case of RTI employers can be taken straight from their returns – is of course linked to the fact that the entire UC system is, so far as possible, automated, which, as Ms Krahé says at para. 31, removes "error, potential fraud and attribution rules which would result in over and under payments". This point is amplified in para. 77 of the judgment of Rose LJ in the case of Johnson which I discuss below.

25. At paras. 35-58 Ms Krahé explains the policy behind the benefit cap and its disapplication in the case of UC claimants receiving more than a prescribed threshold level of earnings, namely to incentivise claimants to undertake at least a minimum level of paid work. Under the previous (Working Tax Credit) system there was a threshold with a similar purpose, but that was based on a minimum number of hours worked rather than on minimum earnings received. At para. 48 she summarises the reasons for the change to an earnings-based approach as follows:

"(a) It is based upon an assessment period, as is the case for all other aspects of the calculation.

(b) It supports a fundamental principle of UC by reflecting the cash-flow of the household in the assessment period rather than the number of hours that a person is working.

(c) It uses RTI, which allows for an automatic calculation and is therefore efficient. UC is not designed to collect information regarding number of hours worked.

(d) It provides for additional flexibility where, for example, a person might be able to secure employment at a higher rate and therefore choose to work fewer hours.

(e) A fixed amount can be changed in future with minor adjustments.

(f) The original fixed amount (described below) was a clear comparison with the level at which lone parents could access Working Tax Credit. It was recognised that the benefit cap would apply to lone parents to a greater extent and as such this approach would assist with smooth transition between the old and new systems."

At para. 51 she exhibits a submission to the Minister identifying various options for defining the earnings threshold. The option chosen, as we have seen, was defined by reference to the amount that would be earned by a claimant working sixteen hours per week at the NLW rate.^[2]

26. It is convenient at this point to note that at para. 52 of his judgment Garnham J describes the Claimant as being "treated as if she were not working enough, when in fact she is". With respect, that way of putting it blurs the important difference between the old regime and the new. The threshold is deliberately defined not in terms of "working enough" in the assessment period but of earning – or, rather, receiving – enough.
27. At paras. 59-65 Ms Krahé accepts that the Regulations as they stand result in the pay-cycle effect. At para. 61 she notes that the issue had been raised by the Work and Pensions Select Committee in a report published on 12 March 2019, as a result of evidence given to the Committee by CPAG. She refers to the Department's response and says, at para. 63:

"Accordingly, the DWP's position is that the current calculations are correct and in compliance with the law. The DWP is however always considering possible changes (consistent with the 'test and learn' philosophy) and this is one issue under consideration."

28. I make two general observations arising from the parties' evidence as summarised thus far.
29. First, it is clear from the Claimant's evidence both that it is a common pattern for employees to work sixteen hours per week at the NLW rate and that it is common for employees, particularly those who are less well-paid, to be paid on a 28-day cycle (though still much less common than payment by the calendar month). But it was accepted before us that it was impossible to establish from that evidence how many UC claimants conformed to both patterns so that the 28-day cycle effect would impact on them as it does on the Claimant; and the Secretary of State's evidence does not fill that gap. All that can be said is that the number must be substantial.
30. Second, claimants can of course try to escape the pay-cycle effect, at least when and if they become aware of it. They can in principle do so by working an hour or two more per week, either for the same employer or for another; likewise if their employer is prepared to pay them at slightly above the NLW rate or to change to calendar-monthly payment. We were told that since the hearing at first instance the Claimant herself has managed to achieve this by changing jobs: she still works sixteen hours per week at the NLW rate, but her new employer pays her on a calendar-monthly basis, so that she is no longer subject to the 28-day cycle effect and crosses the threshold for the disapplication of the cap. But such a change will in many or most cases not be available for the asking: employers may have good reasons for fixing the number of hours at sixteen (for example, in the Claimant's case that was the figure specified in her grandmother's care plan); and they may not be prepared to pay more than the NLW rate or to change their payment systems.
31. During the hearing Garnham J invited further evidence on the specific question of whether the returns completed by RTI employers included data about pay cycles. Evidence was filed accordingly, which I summarise below. It was not, however, the subject of further submissions from the parties: with the benefit of hindsight I think this may have been a mistake – see para. 81 below.
32. The Claimant filed a further statement from Ms Hargreaves. This explained the workings of the RTI system, but the essential point is that she identified a particular box (no. 42) on the relevant form which required the provision of information about pay cycles.
33. The Secretary of State filed a further witness statement from Ms Krahé. This is not directly responsive to Ms Hargreaves' statement^[3], but she acknowledges that pay-cycle data does appear in the RTI returns (albeit that it is used by DWP only for policy purposes and is not included in the information processed for the purpose of calculating UC entitlement). She goes on, however, to explain at paras. 12-13 why the fact that that data is available is not a complete answer to the problem. She says:

"12. It is not therefore possible with the current arrangements to process the 'pay cycle' data as part of the UC award. As I indicated in my first statement, the DWP is considering reforms in relation to pay cycles. However, this requires considerable input from policy, legal and technical stakeholders. It is not realistic to rely on the fact that HMRC is able to capture pay cycle data to conclude that UC decision makers can approach the calculations in the way suggested by the Claimants (i.e. by assuming that any salary payment which is paid with a 28-day 'indicator' will inevitably be paid in the same way on an ongoing basis).

13. The fundamental problem remains the need to maintain the integrity of the monthly Assessment Period structure which underpins UC and the need to base calculations upon actual data rather than predictions as to future earnings (which, in this context, are notoriously uncertain). Any reform needs to have regard to the complications of fluctuating pay patterns, multiple employment arrangements (with different patterns), and other forms of earnings and income."

34. That passage is addressed to what Ms Krahé evidently understood to be the Claimant's case as to how the 28-day cycle effect^[4] could be eliminated, as summarised in the brackets at the end of para. 12. I discuss this later, but her essential point is that any solution to the problem has to take into account the fact that UC entitlement itself was based on receipts in the calendar month and the crucial principle that payments must be based on actual data rather than assumptions or predictions; and that this would not be straightforward for the kinds of reason identified (albeit only in headline terms) in her final sentence. It is convenient to note here that Mr Drabble did not advance any submissions before us addressing the specific points made by Ms Krahé in this passage.

JOHNSON

35. Following the hearing before Garnham J, but before his decision, this Court gave judgment in a case raising a different, but arguably analogous, challenge to an aspect of the Universal Credit Regulations – R (Johnson) v Secretary of State for Work and Pensions [2020] EWCA Civ 778,  [2020] PTSR 1872. The parties filed further written submissions about the extent to which the reasoning in Johnson applied to the circumstances of the present case. Since, as will appear, it was central to Garnham J's eventual reasoning, I should at this stage summarise what Johnson decided and why. I should note that in Johnson also the Secretary of State was represented by Mr Brown. The leading judgment was delivered by Rose LJ.
36. The issue in Johnson arose out of the practice of employers who pay salary by the calendar month of advancing payment by one or two days when the contractual payment date falls on a weekend or bank holiday – so-called "non-banking day salary shift". Where non-banking day salary shift occurs at the beginning or end of an assessment period – which depends on what Rose LJ described (see para. 39 below) as the "happenstance" of at what point in the employer's payment month the claimant made their first claim – it will have the effect of doubling their receipts from the employment in question in one assessment period and reducing it to zero in another. This led not only to artificial fluctuations in payment of UC but (overall) to substantially lower entitlements, specifically as regards entitlement to "work allowance": the effect is summarised in paras. 2 and 3 of Rose LJ's judgment and explained in detail at paras. 51-67.
37. At first instance the Divisional Court held that on their true construction the Regulations did not have the effect complained of. This Court overturned that conclusion, but it went on to hold that the Regulations were irrational to the extent that they failed to incorporate an appropriate adjustment to correct the effect of non-banking day salary shift. Rose LJ's reasoning can be summarised as follows.
38. At paras. 46-50 she considers the overall nature of the claim. As part of that section, at paras. 48-49, she discusses the correct approach to an issue of irrationality, concluding at para. 50 that:

"We need to consider [A] what are the disadvantages of deciding not to 'fine-tune' the Regulations thereby allowing the non-banking day salary shift problem to persist unresolved; [B] what are the disadvantages of adopting a solution to the non-banking day salary shift problem; [C] would a solution be consistent or inconsistent with the nature of the universal credit regime; and [D] has a reasonable balance been struck by the SSWP - or rather [D (1)] is it possible to say that no reasonable Secretary of State would have struck the balance in the way the SSWP has done in this case?"

(I have inserted the letters in square brackets for ease of reference.) Rose LJ structures her dispositive reasoning broadly by reference to those questions, though her headings depart from them to some extent.

39. As to question [A], this is discussed at paras. 51-67, to which I have already referred: for our purposes the summary that I have already given will suffice. But I will quote a short passage from para. 56 which amplifies a point made by Ms Krahé at para. 22 of her witness statement in this case:

"… [T]he choice of monthly assessment periods was based in part on the increasing prevalence of that salary cycle amongst the working population and because many household bills are payable monthly. Indeed, the alignment of the duration of the assessment period with household outgoings is intended to encourage responsible budgeting by low income claimants."

I should also note that at para. 46 she observed that "the happenstance of the date on which [the claimants] applied for universal credit results in them losing, several months each year, the entitlement to the work allowance which the Regulations clearly intend to confer on them".

40. At paras. 68-83 Rose LJ addresses question [B], under the heading "the disadvantages of resolving the problem". Three such disadvantages were relied on by the Secretary of State, namely

"(i) that the supposed irrationality is based on a misconception because it aligns the assessment period with the period of a calendar month; (ii) there is a need for bright lines to ensure that the universal credit system operates in a coherent way; and (iii) it is important to ensure that the calculation of the monthly award of universal credit can take place in an automated way …".

She assesses those disadvantages in turn and (broadly) does not find them to be cogent. I need not summarise her detailed analysis. I should, however, quote para. 77, where she explains the importance of the calculation of UC being automatable:

"The witnesses on behalf of the SSWP emphasise the importance of the fact that benefit awards made each month are not calculated manually by DWP officers but are generated by a computer programmed to take the many and varied inputs about the claimant's family and financial circumstances and work out the award each month. The automation of the process is an important advantage of the universal credit regime over the legacy benefits system for a number of reasons. First, it generates substantial savings in the costs of administration thereby releasing more money to be paid out in benefits. Secondly, it enables the amount awarded to respond immediately to changes in circumstances; a payment made seven days after the end of the assessment period can take into account any changes reported by the claimant during that assessment period. Thirdly, the overall move to digitisation prompts claimants to develop skills which enable them to access other online services which they might previously have been reluctant to use. This may itself be a way of reducing poverty, for example enabling access to cheaper online services offered by utilities. In order to achieve that objective it requires simple and automatic processes."

It is clear that she accepts that evidence, although she goes on to find that adjustments could be made to accommodate non-banking-day salary shift within the automatic processes.

41. Paras. 84-107 are headed "Is it irrational for the SSWP not to enact a solution to this problem?", i.e. questions [D] and [D (1)]. Rose LJ considers first whether the Secretary of State had when making the Regulations made a deliberate policy decision not to accommodate the problem on non-banking day salary shift and concludes that there was no evidence that she had. She then turns to consider, under the heading "Other Relevant Factors", four factors – "(a) the size of the cohort affected, (b) the duration of the impact on them, (c) the arbitrary occurrence of the effect and (d) the inconsistency between the effect of the problem and the aims of the universal credit regime". At paras. 93-106 she reviews those factors. Again, I need not set out her analysis.
42. Rose LJ concludes, at para. 107:

"The threshold for establishing irrationality is very high, but it is not insuperable. This case is, in my judgment, one of the rare instances where the SSWP's refusal to put in place a solution to this very specific problem is so irrational that I have concluded that the threshold is met because no reasonable SSWP would have struck the balance in that way."

It will be noted that what was held to be unlawful was not the making of the Regulations themselves but the Secretary of State's refusal to put in place a solution to the problem caused by non-banking day salary shift once it was appreciated.

43. Irwin LJ agreed with Rose LJ's judgment. I gave a short concurring judgment. I should quote paras. 113-114:

"113. I start by saying that I recognise, as does Rose LJ, the extraordinary complexity of designing a system such as universal credit, and that it necessarily involves a range of practical and political assessments of a kind which the Court is not equipped to judge. I also accept that in order to be workable any such system may have to incorporate bright-line rules and criteria which do not discriminate fully between the circumstances of different individuals. … For those reasons I fully accept that a Court should avoid the temptation to find that some particular feature of such a system is 'irrational' merely because it produces hard, even very hard, results in some individual cases.

114. However, for the reasons which Rose LJ gives, that is not a sufficient answer in this case. Non-banking day salary shift is common and entirely predictable, and its arbitrary effect on entitlement to universal credit is now well-recognised, whether or not it was actually predicted when the scheme was being designed. That effect has a severely harmful impact, which they can do nothing to avoid, on very large numbers of vulnerable claimants. … That impact is not the inevitable consequence of the application of some fundamental principle of the legislation. I of course understand that it is a fundamental principle that entitlement should be based on monthly receipts, however much a claimant's income may vary from month to month. But an adjustment specifically to address the non-banking day salary problem would not in any real sense undermine that principle: indeed it could be said to vindicate it, since the receipt of salary in the 'wrong' month because of the mechanics of bank payment is purely factitious (unlike other kinds of irregular payment discussed in argument). … If anything, the present operation of the scheme runs positively counter to its declared purpose, as Rose LJ points out at paras. 100-101. It follows that I cannot accept that there is no way in which an appropriate adjustment can be made without prejudicing the overall statutory purpose; and I agree with Rose LJ that there is nothing in the evidence to justify the conclusion that no solution can be devised without causing unacceptable cost or problems elsewhere in the system."

44. The claimants had advanced a separate claim that the Regulations discriminated against them, contrary to article 14 of the European Convention on Human Rights, in conjunction with article 1 of the First Protocol. In view of its decision on irrationality the Court found it unnecessary to decide that issue: see paras. 108-109 of Rose LJ's judgment.
45. The Secretary of State did not seek permission to appeal to the Supreme Court.

THE JUDGMENT OF GARNHAM J

46. In his judgment in this case Garnham J explicitly approaches the issues by reference to the four questions identified in para. 50 of Rose LJ's judgment in Johnson, though his headings too depart from them a little. I take them in turn.
47. Paras. 51-56 are headed "The disadvantages of allowing the lunar month problem to persist unresolved". Garnham J identifies four problems about the 28-day cycle effect, as follows:

(1) The first is the basic effect which I have outlined at para. 5 above. As he puts it at para. 52:

"… [T]he First Claimant is in continuous and regular employment, earning regular amounts of money throughout every assessment period. However, in 11 out of 12 assessment periods the Regulations treat her (and others in her position) as having earned less income for that period than is in fact the case, because of the dates on which she was paid. The result is that she receives substantially less UC, perhaps some £400 per month or 20%, less, than would be the case if she was paid monthly. That is, self-evidently, a very significant reduction for somebody of modest means. As Mr Drabble correctly puts it, the First Claimant is treated as if she were not working enough, when in fact she is. Then, once a year, the Regulations treat her as having almost double the income she has actually earned in that period."

(2) The second is that, contrary to the avowed legislative policy of encouraging work, the result of the 28-day cycle effect "discourages work when the work available is paid on a lunar month basis". He says at para. 53:

"The scheme is said to be designed to be responsive to changes in earned income, and to make work pay to the fullest possible extent. But in these circumstances, it is neither."

(3) The third is that the result of the 28-day cycle effect is that it "causes the First Claimant's household income to fluctuate dramatically once a year, making it difficult to budget": see para. 54. He notes that what he describes as "a similar phenomenon" was caused by the non-banking day salary shift and was part of the basis of the decision in Johnson.

(4) The fourth is that it will in the great majority of cases be impossible for UC claimants to persuade employers who pay on a 28-day cycle to shift to payment by the calendar month. Garnham J says, at para. 56:

"… [T]he idea of having to choose employment based, not on the nature of the work, but on the particular pay cycle operated by the employer seems to me absurd. The consequence is to give the UC scheme an appearance of arbitrariness."

He quotes a statement by Rose LJ, at para. 57 of her judgment in Johnson, that:

"It is … no part of the policy underlying universal credit to encourage claimants to base their employment choices on the salary payment date offered by a prospective employer. Yet that is what is happening for these Respondents."

48. Paras. 57-65 are headed "The disadvantages of adopting a solution to the lunar month problem". At para. 57 Garnham J says that the disadvantages in question are the same as those relied on by the Secretary of State in Johnson – see para. 68 of Rose LJ's judgment. He says at para. 58 (in effect) that he adopts Rose LJ's comments about those disadvantages and proposes only to "pick up points of particular importance on the facts of the present case". As to that:

(1) As regards the first disadvantage, at para. 59 he notes that payment on a 28-day cycle cannot possibly be described as "irregular".

(2) As regards the second, he quotes passages from paras. 73 and 75 of Rose LJ's judgment to the effect that although legislation often legitimately requires the drawing of "bright lines" that could not reasonably justify the "significant, predictable but arbitrary effects" resulting from non-banking day salary shift; and he says, at para. 62, that the same effects occur in the present case.

(3) As regards the need for systems to be automatic, he refers at para. 64 to a finding by Rose LJ that the computer programme employed for UC was sophisticated enough to be adjusted to accommodate non-banking day salary shift. At para. 65 he says:

"On the evidence in the present case, it is plain that some, at least, of the necessary computer software is in place and can readily be utilised. As noted above …, the data provided by employers to HMRC includes pay frequency. Presently, the DWP routinely receives only a subset of that data which does not include pay frequency; it is said that that information 'is not routinely accessible to operational delivery staff'. However, it is not suggested that other staff do not have access to that data or that it could not be made available."

49. Garnham J's third heading is "Would a solution be consistent or inconsistent with the nature of the universal credit regime?". He begins, at para. 66, by quoting para. 14 of Ms Krahé's first witness statement and continues:

"67. One important element of the UC regime was to align the assessment period with monthly payment and charging cycles. Rose LJ said at [56]:

'… It is no part of the policy underlying universal credit to encourage claimants or employers to adopt a non-monthly salary cycle; on the contrary the choice of monthly assessment periods was based in part on the increasing prevalence of that salary cycle amongst the working population and because many household bills are payable monthly. Indeed, the alignment of the duration of the assessment period with household outgoings is intended to encourage responsible budgeting by low income claimants.'

68. I accept that accommodating a four-weekly salary cycle would not advance the behavioural change of encouraging people to plan their working lives around a monthly work and payment pattern, and that is a consideration in favour of the Secretary of State's present stance. But first, the weight to be attached to that consideration is somewhat reduced by the evidence discussed at [64] above to the effect that the data needed to manage four-weekly payments is already collected by HMRC. And second, that is not the only behavioural change in prospect. A solution to the lunar month problem would encourage people for whom the only employment available, or likely to become available, was paid four-weekly to take and keep such work, a central element in the UC regime.

69. I accept too that introducing a solution to the lunar month problem would make the UC process technically more complicated, although for the reasons discussed, it seems to me unlikely to be unmanageable. Furthermore, in my view, such a solution would make UC conceptually much simpler.

70. For the reasons discussed above, a solution to the lunar month problem would reduce disincentives to work, make the system fairer and reduce perverse incentives. In those respects, a solution of the lunar month problem would be consistent with the nature of the UC regime.

71. Ms Krahé says in her witness statement that the fundamental problem remains the need to maintain the integrity of the monthly assessment period structure which underpins UC and the need to base calculations upon actual data. But in my judgment the collection and deployment of the 'actual data' provided to HMRC would enable calculations to be made that not only respect the integrity of the monthly assessment period structure but enhance it by ensuring the calculation of income in each assessment period accurately represents actual receipts."

50. The fourth heading is "Other relevant factors". The factors considered are those addressed by Rose LJ at paras. 84-107 in Johnson (i.e. the extent to which the problem had been considered by the Secretary of State at the time that the Regulations were made, together with the four other factors that she enumerates). Taking them in turn:

(1) After reviewing the evidence Garnham J concludes, at para. 74, that "there was no evidence that specific consideration was given to solving the lunar month problem as it is identified in the present proceedings".

(2) As to "size of the cohort" he finds that "payment in four-weekly periods is not uncommon", with "something like 13% of new UC claimants [being] in their last job paid 4-weekly (with a further 58% paid on other non-monthly but regular patterns which could be vulnerable to the same problem)": see para. 75.

(3) As to "the duration of the impact on the Claimants", he makes the point that the 28-day cycle effect arises in eleven out of every twelve months and will persist throughout a claimant's period of entitlement; see para. 76.

(4) As to the arbitrariness of the 28-day cycle effect, he says at para. 77 that "there is no evidence that the First Claimant was warned at the time she made her claim for UC of the potential effect of the fact that her employers paid her on a four-weekly basis" and goes on to quote a passage from her witness statement expressing how dismayed she was to discover that she was subject to the benefit cap despite working sixteen hours at national minimum wage and what a devastating impact that had and continued to have. (I note in passing that I am not sure that that point goes to arbitrariness.)

(5) As to the issue of disincentivising work, Garnham J notes at para. 78 that he has already addressed that.

51. Garnham J's final heading derived from Rose LJ's analysis is "Can it fairly be said that no reasonable Secretary of State would have struck the balance in the way the SSWP has done in this case?". Since that is the ultimate question in the case I should quote what he says in full:

"79. The consequence of the lunar month problem is that for 11 months out of 12 the First Claimant's earned income is treated as being her earnings for just 28 days. The result of that is that the benefit cap is applied, and her UC is reduced, by perhaps as much as 20%. As discussed above, the disadvantages of allowing the lunar month problem to persist are manifest and serious. By contrast, the principle [sic] suggested disadvantages cannot survive the analysis of the Court of Appeal in Johnson.

80. The importance of ensuring that the payment system can be automated is clear and not in dispute. During the hearing, much the most powerful consideration in favour of maintaining the status quo was the suggested difficulty in collecting and deploying the data necessary to enable the calculation of earned income in relevant assessment periods to be carried out automatically when payment had been made on a four-weekly basis. But that difficulty substantially disappeared when the further evidence was obtained from Ms Hargreaves and Ms Krahé [this is a reference to their second witness statements]. There was little evidence that the SSWP ever focused on the lunar month problem, as opposed to the general benefit of a universally applicable monthly assessment period, and nothing to suggest the possibility of solving that problem was ever considered and rejected.

81. In those circumstances, it seems to me that the outcome of the balance is obvious and irresistible. I cannot see how any reasonable Secretary of State could have struck the balance in the way the SSWP has done in this case."

52. Finally, Garnham J addresses a submission from Mr Brown that a finding in the Claimants' favour would represent "an objectionable extension of Johnson". He says, at paras. 83-87:

"83. It is right to say that the Court of Appeal's conclusion in Johnson was expressly and deliberately confined to the specific problem of 'non-banking day salary shift'. Rose LJ said at [107] that 'the SSWP's refusal to put in place a solution to this very specific problem' was unreasonable. That conclusion was confined to the 'very specific problem' of 'non-banking day salary shift'. At [86] Rose LJ said, 'We are not concerned here with making an exception for people who are paid at frequencies other than monthly.'

84. In his concurring judgment Underhill LJ added, at [116],

'…I regard this as a case which turns on its own very particular circumstances. It has no impact on the lawfulness of the universal credit system more generally.'

85. Mr Brown argues, against that background, that any extension of the Johnson approach to other pay cycles would reintroduce precisely the uncertainty which the Court of Appeal, in overturning the Divisional Court, were concerned to remove.

86. It is plain that the Court of Appeal was anxious, whilst correcting the unfairness and irrationality in cases such as Ms Johnson's, to preserve intact the structure and tenets of the UC scheme as approved by Parliament. But in my view, the principles the Court identified, and the essential logic of the argument they accepted, apply with equal force to cases of claimants paid on a four-weekly basis.

87. In fact, it can fairly be said that the logic applies with even greater force. First, four-weekly payments are genuinely and consistently regular; they do not incorporate the inevitable, if occasional, irregularity that comes with monthly payments as described by the Court of Appeal. Second, in monthly payment cases the difficulty arises in a few months each year; with lunar monthly cases such as the First Claimant's, it arises in 11 months out of 12. In those circumstances, it seems to me that the case for the Regulations making an exception for such claimants is even stronger than it was in Johnson. The one substantial ground on which a Secretary of State might reasonably decline to make such an exception is if the availability of data by RTI threatened the integrity of the automated processing of claims. And such evidence as there is points in the opposite direction on that issue."

53. Having reached that decision Garnham J said at para. 89 that the grounds of challenge based on article 14 of the Convention did not arise for consideration and he made no finding on it.

THE TEST OF IRRATIONALITY

54. In Johnson Rose LJ noted that the Court had not received detailed submissions on the test of irrationality: see para. 48 of her judgment. The claimant had relied squarely on "the Wednesbury unreasonableness that has been a ground for a public law challenge since the early days of the modern jurisprudence on judicial review". Rose LJ referred to para. 90 of the judgment of Leggatt LJ and Carr J, sitting as a Divisional Court, in R (Law Society) v Lord Chancellor [2018] EWHC 2094 (Admin), [2019] 1 WLR 1649. This reads (so far as relevant):

"The second ground on which the Lord Chancellor's Decision is challenged encompasses a number of arguments falling under the general head of 'irrationality' or, as it is more accurately described, unreasonableness. This legal basis for judicial review has two aspects. The first is concerned with whether the decision under review is capable of being justified or whether in the classic Wednesbury formulation it is 'so unreasonable that no reasonable authority could ever have come to it': see Associated Picture Houses Ltd v Wednesbury Corp [1948] 1 KB 223, 233-4. Another, simpler formulation of the test which avoids tautology is whether the decision is outside the range of reasonable decisions open to the decision-maker: see e.g. Boddington v British Transport Police [1998] UKHL 13; [1999] 2 AC 143, 175 (Lord Steyn). The second aspect of irrationality/unreasonableness is concerned with the process by which the decision was reached. …"

Rose LJ observes that the challenge in Johnson was essentially of the first kind, and the same is true in this case.

55. No doubt taking their lead from Johnson, counsel before us did not feel the need to advance any detailed submissions on the test of irrationality. That being so, this is not the case in which to attempt any wide-ranging analysis.^[5] I am broadly content to adopt the very general formulation derived from Boddington which appears in the Law Society case: it is clearly not intended to be essentially different from the time-honoured Wednesbury language, but, as the Divisional Court there says, the Boddington formulation is simpler and less tautologous.
56. It is now well-recognised that the degree of intensity with which the Court will review the reasonableness of a public law or act or decision (including a provision of secondary legislation) varies according to the nature of the decision in question. There are many authoritative statements to this effect, but I need only quote from para. 51 of the judgment of Lord Mance in Kennedy v The Charity Commission [2014] UKSC 20, [2015] AC 435, where he says:

"The common law no longer insists on the uniform application of the rigid test of irrationality once thought applicable under the so-called Wednesbury principle. The nature of judicial review in every case depends upon the context."

57. It is also well-recognised that in the context of governmental decisions in the field of social and economic policy, which covers social security benefits, "the administrative law test of unreasonableness is generally applied … with considerable care and caution" and the approach of the courts should "in general … [accord] a high level of respect to the judgment of public authorities" in that field. I take those words from para. 146 of the judgment of Lord Reed (with which the other members of the Court agreed) in R (SC) v Secretary of State for Work and Pensions [2021] UKSC 26, [2021] 3 WLR 428: see para. 146. In that case the Supreme Court was concerned, as here, with a challenge to the legislation relating to welfare benefits (sections 13 and 14 of the Welfare Reform and Work Act 2016). The claimants' case was that the impugned provisions contravened article 14 of the Convention, but in the part of the judgment from which I quote Lord Reed is making the point that the Strasbourg jurisprudence is in line with the approach taken by the common law, and it is the latter which he is describing. He explains the reasons for adopting a less intensive standard of review in this area, including the need for the courts "to respect the separation of powers between the judiciary and the elected branches of government" (see para. 144).
58. Although the decision in SC is very recent (indeed it post-dates the argument before us), Lord Reed emphasises that the approach which he sets out is well-established in domestic law. I should note in particular a statement which he quotes from the speech of Lord Bridge in R v Secretary of State for the Environment, ex p Hammersmith and Fulham London Borough Council [1991] 1 AC 521 to the effect that

"[where a] … statute has conferred a power on the Secretary of State which involves the formulation and the implementation of national economic policy and which can only take effect with the approval of the House of Commons [my emphasis], it is not open to challenge on the grounds of irrationality short of the extremes of bad faith, improper motive or manifest absurdity".

As is evident from the italicised words, the ministerial orders which were in issue in that case were required to be approved by resolution of the House of Commons; and Lord Bridge evidently attached weight to that fact when identifying the appropriate standard of review. Lord Sumption made the same point at para. 44 of his judgment in Bank Mellat v Her Majesty's Treasury (no. 2) [2013] UKSC 39, [2014] AC 700, where he said:

"When a statutory instrument has been reviewed by Parliament, respect for Parliament's constitutional function calls for considerable caution before the courts will hold it to be unlawful on some ground (such as irrationality) which is within the ambit of Parliament's review. This applies with special force to legislative instruments founded on considerations of general policy."

Those observations were endorsed by Lord Reed in R (SG) v Secretary of State for Work and Pensions [2015] UKSC 16, [2015] 1 WLR 1449, at para. 94.

59. Finally, I would repeat what I said at para. 113 of my judgment in Johnson, as follows:

"I recognise, as does Rose LJ, the extraordinary complexity of designing a system such as universal credit, and that it necessarily involves a range of practical and political assessments of a kind which the Court is not equipped to judge. I also accept that in order to be workable any such system may have to incorporate bright-line rules and criteria which do not discriminate fully between the circumstances of different individuals. … I fully accept that a Court should avoid the temptation to find that some particular feature of such a system is 'irrational' merely because it produces hard, even very hard, results in some individual cases."

I would add that the very complexity and difficulty of the exercise is bound to mean that following the implementation of the scheme it may become clear with the benefit of experience that some choices could have been made better. But it does not follow that the legislation was in the respect in question irrational as made, or that it would be irrational not to correct the imperfections once identified: the court cannot judge the lawfulness of such schemes by the standard of perfection. Whether any errors or imperfections are of such a nature or degree as to impugn the lawfulness of the relevant regulations must depend on the circumstances of the particular case, having regard to the appropriate intensity of review.

THE APPEAL

60. The Secretary of State's pleaded grounds of appeal start with the general proposition that "the Judge erred in holding that the calculation of earned income under reg 54 of the Universal Credit Regulations 2013 was irrational and unlawful in respect of employees paid on a four-weekly basis" and then proceed to plead eight specific criticisms of his reasoning. I need not set those criticisms out: I refer to them so far as necessary in the course of my discussion below.
61. Permission to appeal to this Court was given by Dingemans LJ, who observed:

"There is a compelling reason to her this appeal, namely to determine whether Graham J.'s careful analysis in his judgment has carried the effect of R (Johnson) v Secretary of State for Work and Pensions [2020] EWCA Civ 778; [2020] PTSR 1872 beyond its intended scope, see paragraphs 83 to 86 of his judgment. I grant permission on all grounds."

62. The Claimant has not filed a Respondent's Notice and has accordingly not sought to revive her claim based on article 14 of the Convention.

THE PARTIES' CASES

63. As I have already noted, the pleaded grounds of appeal take the form of a number of detailed criticisms of Garnham J's reasoning. But the submissions before us primarily addressed the substance of the challenge to the lawfulness of the earnings-related threshold in regulation 82 (1) (a), and I prefer to take the same approach. One of the problems about viewing the issue through the lens of Garnham J's reasoning is that it is so closely based on Rose LJ's reasoning in Johnson. Although it will indeed be necessary to consider the extent to which the present case is analogous to Johnson, I do not think that it is satisfactory to make that the framework for the analysis: the issues must be addressed in their own right.
64. The essence of the case advanced by Mr Brown can be summarised as follows. It is a fundamental feature of the UC system that entitlement to benefit should be assessed by reference to actual receipts in a calendar month. That was a policy choice by the Minister, deliberately departing from the approach taken in the previous legislation: see the evidence of Ms Krahé quoted at para. 23 above. It was also a deliberate, and explicit, policy choice that the earnings-related threshold for the disapplication of the benefit cap in the case of UC should be defined by reference to the same period: see para. 28 of Ms Krahé's first witness statement. It was both natural and desirable that the two provisions (and those governing the cap itself – regulation 79 (2)) should employ the same measure: they needed to "tessellate". It was self-evident that those choices would mean that where earnings are paid otherwise than by reference to the calendar month the figures for receipts would vary from assessment period to assessment period. It was also self-evident that such variations would produce fluctuations in entitlement from month to month. That might be regarded as a disadvantage of the choice of the calendar month as the assessment period, and of the associated choice of the earnings received in a calendar month as the criterion for the earnings-related threshold. But any choice would have disadvantages as well as advantages; and the Secretary of State made the choice that she^[6] did for the reasons given by Ms Krahé. It was inevitable in the case of legislative schemes of this kind that there would be cases where people are disadvantaged by falling just on the wrong side of a "bright line", as the Claimant and others in her position could be said to do; but that did not mean that the line is arbitrary in the sense of being illegitimate. It was in any event also important to bear in mind that the phenomenon of which the Claimant complained was not the result simply of how you define a "month", which turns on an apparently trivial difference of two or three days. Rather, it was the result of claimants being paid on any cycle other than the calendar month, which the Secretary of State had deliberately chosen as the cornerstone of the system for reasons explained by Ms Krahé.
65. Mr Brown emphasised that the threshold for establishing irrationality is, as Rose LJ said at para. 107 of her judgment in Johnson, very high; and it could not be said that no reasonable Secretary of State would have struck the balance between the advantages and disadvantages of treating receipts in the assessment period as the yardstick for all purposes in the way that was done here. It was axiomatic that the Court should be very slow to interfere with decisions of this kind, based as they are on policy choices which are, as a matter both of constitutional propriety and of institutional competence, the responsibility of Government. That was all the more so where, as here, the statute provides for the rules in which they are embodied to be approved by affirmative resolution of both Houses of Parliament.
66. Mr Drabble emphasised that the Claimant was not challenging the choice of the calendar month as the assessment period for the purpose of quantifying an award of UC. Her challenge was confined to the use of earnings received in the calendar month as the basis of the definition of the earnings-related threshold. Those two contexts were different, and there was no reason why the same method should be used in both. The purpose of the earnings-related threshold was to reward claimants who were undertaking paid work, and from that point of view it was irrelevant at what intervals they received their payment for that work: the position of people earning the NLW and working sixteen hours per week was identical whether they were paid by the calendar month or four-weekly (or indeed weekly or fortnightly). It was irrational to treat people who are in the same position differently.
67. In the course of the hearing Mr Drabble was asked by the Court how, on his case, the Secretary of State should have defined the threshold for the disapplication of the benefit cap. His response was that the criterion should be whether claimants were working a sixteen-hour week (at at least the NLW rate) – that is, that there should have been an hours-related threshold, as under the previous regulations. (Other answers might be possible: I return to this below.)
68. That was Mr Drabble's core submission, but he reinforced it by a number of points. It was clear from the evidence that the 28-day cycle effect had a substantial adverse impact – we have seen that in the Claimant's own case it meant that her entitlement was reduced by almost £500 per month – on a substantial number of claimants and that it could not readily be avoided. He said that it was clear from the second witness statements of Ms Hargreaves and Ms Krahé that the DWP had the data which would enable it to operate an hours-related threshold. As regards the respect to be paid to the Minister's policy choice, it was clear from the DWP's response to the report of the Work and Pensions Select Committee in 2019 (see para. 27 above) that it had not when devising the scheme appreciated that the definition of the earnings threshold by reference to receipts in the calendar month would have the impact that it did on claimants on a 28-day cycle; so no considered choice could be said to have been made.

DISCUSSION AND CONCLUSION

69. The pay-cycle effect arises from the interaction of three important elements of the scheme governing UC, namely:

(a) the adoption of a threshold for disapplication of the benefit cap by reference to a minimum level of earned income rather than a minimum number of hours worked (regulation 82 (1) (a));

(b) the adoption of a calendar month as the period for the calculation of that earnings-related threshold (again, regulation 82 (1) (a)); and

(c) the definition of earned income in terms of actual receipts (regulation 54).

70. The elimination (or at least substantial mitigation) of the pay-cycle effect – what in Johnson was referred to as a "solution" – would require a modification of, or a partial exception to, one or more of those principles. In order to decide whether the Secretary of State is required, as a matter of rationality, to make such a modification, it is necessary to understand the extent of the departure involved and (again, in the terminology of Johnson) the "disadvantages", so as to be able to consider what form such a modification would take. This is not entirely straightforward. It is not easy to detect in the Claimant's pleadings, evidence or written submissions any clear identification of her case in this regard. Likewise, although Garnham J discusses the advantages and disadvantages of "the solution" to the lunar month problem, he does not spell out what form that solution might take. It is for that reason that the Court put the question that it did to Mr Drabble: see para. 67 above. I will consider each of the three elements identified above, though not in the same order.
71. As to element (b), it was not positively submitted by the Claimant that the Secretary of State should depart from the assessment period of a calendar month in calculating the threshold for the disapplication of the cap under regulation 82 (1) (a). That would involve using different assessment/reference periods for different parts of the calculation, since a calendar month is the relevant period both for the calculation of the primary entitlement and for the definition of the cap itself (see regulation 79 (2)). As Mr Brown submitted, the various elements in the calculation need to fit together; and in my view any solution of this kind would be incoherent and unworkable. Since it is not being advanced, I say no more about it.
72. I turn to element (a). As we have seen, it was Mr Drabble's submission that the Secretary of State should have abandoned the earnings-related criterion and reverted to the previous criterion based on hours worked. That would involve a wholesale departure from a fundamental policy choice of precisely the kind with which the Court should be very slow to interfere: see paras. 57-59 above. At para. 48 of her first witness statement (para. 25 above) Ms Krahé advances specific reasons for that choice. Those reasons are in my view legitimate and cogent. A departure from this principle would be inherently more damaging to the coherence of the UC scheme than a limited departure from element (c) – to which I accordingly turn.
73. A solution focusing on element (c) would require some modification of the principle of defining earned income by reference to actual receipts. This requires rather fuller consideration. It is unsatisfactory that no specific proposal has been formulated, but the broad lines of such a solution are sufficiently clear from the evidence. What would be required is that a claimant who is paid on a regular pay cycle other than a calendar month should be deemed to have actually received in that month not only the amount received on their actual pay day(s)^[7] but also what they have accrued in the part of the calendar month for which they have not received payment. Two versions of such a solution are referred to in the evidence.
74. First, at para. 87 of its report the Work and Pensions Select Committee records that CPAG told it that it had proposed to DWP a solution to the pay cycle effect. The passage reads:

"[CPAG] recommended using existing arrangements for averaging earnings, which are currently used to determine whether claimants are earning enough to exceed the threshold for in work conditionality – if they are in-work but below the threshold they may be required to look for more work in order to receive UC and face sanctions if they do not continue to seek additional work. The averaging, which is done to make sure the system does not unfairly penalise people, currently has to be done manually. CPAG recommended the system be applied immediately due to the 'serious financial losses for affected claimants'."

The Committee noted at para. 88 that the Department's response was that it was aware of the problem and was looking into it. That is confirmed by Ms Krahé in para. 63 of her first witness statement: see para. 27 above. The proposal as summarised by the Committee is decidedly opaque, at least to the uninitiated. It was referenced in the Claimant's skeleton argument in the High Court but not there developed in any way. It is not referred to in Garnham J's judgment nor in the Claimant's skeleton argument in this Court or Mr Drabble's submissions before us. But, whatever its details, the reference to "averaging" suggests some form of deemed receipt of the kind discussed above.

75. Second, there is Ms Krahé's reference in para. 12 of her second witness statement (see para. 33 above) to a suggestion attributed to the Claimant based on "assuming that any salary payment which is paid with a 28-day 'indicator' will inevitably be paid in the same way on an ongoing basis". That does not seem to be the same as the suggestion attributed to CPAG, though it too is not easy to understand as summarised. Again, however, it is clear from the reference to "assuming" that it involves a deeming exercise.
76. A solution of this kind would seem to be the most natural way of eliminating the pay cycle effect. Ms Krahé has confirmed that the DWP "is considering reforms in relation to pay cycles", and I think it can be inferred from the context that her reference is to reforms of broadly the kind proposed by CPAG and/or which she understood the Claimant to be suggesting as identified above. The question is whether it is irrational for the Secretary of State not to have implemented such a solution. In my view that has not been established. My reasons are as follows.
77. I make three general points by way of preliminary:

(1) The threshold of irrationality in this case is high. I have identified the proper approach of the Court at paras. 54-59 above. In the present case, the features of the scheme which result in the pay cycle effect reflect important policy decisions made by the Secretary of State. Those choices are explicit on the face of the Regulations, which were approved by both Houses of Parliament.

(2) Mr Brown is clearly right in principle to point out that it is often necessary in a complex scheme of this kind to apply general rules or principles which will sometimes produce harsh results in particular cases ("bright lines", in the jargon): both Rose LJ and I made this point in Johnson – see paras. 72-73 and 113. However the threshold is defined, there will inevitably be UC claimants who miss out by a narrow margin.

(3) The effect is suffered only by UC claimants with the very specific characteristics identified – working exactly sixteen hours per week, at exactly the NLW rate, and paid on a regular pay cycle other than the calendar month; and who cannot realistically avoid it in one of the ways noted at para. 30 above.

78. I should make it clear that I do not regard those points as decisive by themselves. Even if the pay-cycle effect affects only a very specific class, the numbers are substantial and the impact serious. At first sight it seems anomalous that so much should turn on the pay-cycle operated by the employer, and it is not difficult to see why the Work and Pensions Select Committee believed that a solution should be found. The fact that the DWP is considering what might be done is a tacit acknowledgment that the effect on this class of claimants is (to put it no higher) not ideal. Accordingly I believe that it is necessary, notwithstanding those preliminary points, to consider whether the principle of basing entitlement on actual receipts can be modified without doing unacceptable damage to the integrity of the system.
79. As to that, the crucial evidence is that of Ms Krahé in para. 13 of her second witness statement. It is clear that the Department has serious concerns about introducing any element of deeming into a system based so deliberately – and clearly for legitimate reasons – on actual receipts. Those concerns are not developed in any detail, no doubt because of the way in which the issues developed before the Judge. But they are not difficult to appreciate in general terms and are entitled to our respect. In my view what Ms Krahé says is sufficient to establish that there is a real risk that any departure from the "actual receipts principle" will seriously impair the workability and reliability of the assessment of entitlement for this group of claimants. That being so, I do not think that it is open to the Court to hold that it is irrational for the Secretary of State not to have modified the effect of regulation 54 so as to eliminate or mitigate the pay cycle effect.
80. It is true of course that Ms Krahé does not say that there is no way in which the concerns which she expresses can be met: on the contrary, it is implicit in her evidence that the DWP hopes to be able to identify a limited departure from the principle which will be workable and produce accurate results. For the reasons given in the previous paragraph, it is very much to be hoped that it succeeds. But deciding whether or to what extent to derogate from a general principle of this kind in order to address the interests of a particular group is quintessentially a question for the Secretary of State (with the assistance, of course, of her civil servants) and not the Court. It requires a detailed understanding of a highly complex scheme, and the technicalities of its administration, which the Court does not have, so as to be able to assess the advantages and disadvantages of implementing any particular solution. It will also ultimately require the striking of a balance between those advantages and disadvantages, which is an exercise of judgment that is the province of the legislator. If it were established that there was a straightforward solution which it was irrational for the Secretary of State not to have pursued the Court could and should nevertheless intervene; but that is not the case.
81. At para. 71 of his judgment Garnham J proceeds on the basis that Ms Krahé's concern that benefit should be calculated on the basis of "actual data rather than predictions" is answered by the fact that the RTI data includes information about pay cycles: using "pay cycle data" would ensure that the calculation "accurately represents actual receipts". (Paras. 63-65 are to essentially the same effect.) I believe, with respect, that that is a misunderstanding of her evidence. The effect of paras. 12-13 of her second witness statement is that simply knowing that a claimant is paid on a cycle other than the calendar month is not enough. That only shows that they may have earned (in the sense of accrued) more in the calendar month than they have received, but her point relates to the conceptual and administrative difficulties of treating that additional amount as if it had been received. Garnham J does not address that aspect. I suspect that if the post-hearing evidence had been the subject of further submissions this misunderstanding would have been avoided.
82. It should be clear from the foregoing that the similarity between this case and Johnson is superficial. In the first place, the claimants' challenge in Johnson was not directed at any fundamental feature of the scheme of the Regulations. They were entitled to be paid by the calendar month, which is the assessment period prescribed by the Regulations and which it is the policy of the Secretary of State to encourage. The problem only arose because of the quirk of their periodically, as I put it at para. 114 of my judgment, "[receiving] salary in the 'wrong' month because of the mechanics of bank payment". That was, in Rose LJ's words, a "a very specific problem" (see para. 107). It is true that in a strictly literal sense the solution (i.e. treating the payment as received on the date that it would have been received if it was a banking day) involved a departure from the principle of assessment by reference to actual receipts. But that could fairly be described, as Rose LJ put it at para. 50 of her judgment, as "fine-tuning", which allowed payments to be treated in accordance with the common-sense reality. The problem in the present case, by contrast, arises from the fact the Claimant is paid by reference to a period which does not correspond to the assessment period which is a cornerstone of the Regulations. That is a real and fundamental mismatch.
83. Two further points about Johnson arise from that distinction:

(1) At para. 46 of her judgment Rose LJ said that the claimants were losing a benefit "which the Regulations clearly intend to confer on them". That was a fair observation in the circumstances of Johnson but it has no application here. The Regulations clearly do not intend that the disapplication of the benefit cap should be calculated by reference to sums received in a period other than the calendar month. The Claimant's argument has to be that that is, as a matter of rationality, what they should have intended; but that is another matter.

(2) A central part of the Secretary of State's case in Johnson was that the non-banking-day salary shift problem could only be solved by manual intervention on a case-by-case basis, which was contrary to the important principle that the calculation of entitlement should be automated. The Court rejected that contention not because it did not recognise the importance of automation but because it found as a matter of fact that the necessary adjustments could (at least in due course) be incorporated in the relevant software, given that the only relevant variable, i.e. the incidence of non-banking days, was wholly predictable – see paras. 81 and 82 of Rose LJ's judgment. Any revisions to the software to address the pay-cycle effect would have to be of a different character and would almost certainly be a good deal less straightforward. But the real point is that the problems identified by Ms Krahé go beyond difficulty of automation and are ultimately based on the difficulties of departing from the straightforward and fundamental principle of working on the basis of actual receipts.

84. It follows that I cannot, with respect, accept Garnham J's conclusion at paras. 82-88 of his judgment that the issue in this case can be answered by simply applying the reasoning in Johnson; and I believe that his own reasoning is at several points vitiated by applying a read-across that is not legitimately available. That can be illustrated in what I have already said, but I will mention one further example. At para. 54 he describes the "dramatic fluctuation" in income in the twelfth month caused by the cap being disapplied as being similar to the budgetary difficulties experienced by the claimants in Johnson. But in truth the two situations are quite different. In Johnson a central feature of the claimants' complaint was that they suffered variations in benefit on an irregular basis depending on the variable incidence of their pay-days falling on non-banking days. In this case the Claimant's real complaint is simply about the reduction in her income in eleven months of the year: the fact that she does not suffer it in the twelfth month is not sensibly described as a "fluctuation".
85. For those reasons I would reject the Claimant's challenge. It should largely be clear why I respectfully disagree with Garnham J's thorough and careful reasoning. There are, however, three points which have not featured in the foregoing discussion and which I should address.
86. First, at para. 53 of his judgment Garnham J says that the 28-day cycle effect "discourages work when the work available is paid on a lunar month basis", contrary to the declared objectives of the scheme: the same point is made in para. 70. It is no doubt correct that a UC claimant is less likely to accept an offer of work which does not qualify them to have the benefit cap disapplied than one which does (assuming that they aware how the criterion operates). But in principle that will incentivise them to find work which does meet the necessary threshold. That is the whole point of having a benefit cap which can be avoided by earning a prescribed minimum. The Claimant's argument is that the way that the definition of the particular threshold applies to a case like hers is irrational; but that argument is not advanced by pointing out that there is no incentive to take work which fails to cross that threshold.
87. Second, at para. 59 he attaches weight to the fact that the Claimant was paid regularly, pointing out that that was a factor which this Court emphasised in Johnson. The reason why the point featured in Johnson was that the Secretary of State contended that the phenomenon of non-banking-day salary shift meant that payments were indeed "irregular". The Court rejected that characterisation, but regularity of payment is not the issue here. Rather, the problem is that the Claimant is paid on a cycle that does not correspond to the prescribed assessment period.
88. Third, he also attaches weight to the fact it did not appear from the materials exhibited by Ms Krahé that the DWP had when initially designing the scheme appreciated that it would give rise to the pay-cycle effect: see paras. 74 and 80 of his judgment. As already noted, this was also a point made by Mr Drabble.
89. As to that, I agree that it seems clear that, although the DWP certainly appreciated that where a claimant was paid on a cycle other than the calendar month their receipts would vary from month to month, the impact of that variation on the threshold for the disapplication of the benefit cap in cases of the present kind was only really focused on as a result of CPAG's post-implementation representations. That means that the Secretary of State is not able to point to a considered policy decision, made at the time, that there was no workable way of avoiding the pay-cycle effect without unacceptably compromising the objects of the scheme. If she were able do so, that would of course reinforce her case that the Court should not interfere. However, the absence of a contemporary decision of that kind is by no means fatal to her case. It is well-established that an interference with (qualified) Convention rights may in principle be justified by reference to considerations which were not present to the mind of the decision-maker at the time, and I can see no reason why the same approach should not be taken in the context of a rationality review.
90. But there is a further point. There is a limit to the extent that it is possible to anticipate every particular effect of a highly complex scheme of this kind, and it may emerge following implementation that there are aspects which are (at least) sub-optimal and where changes should be made: I note what Ms Krahé says at para. 63 of her first witness statement about the DWP's "test and learn" philosophy. Where that happens, it does not in my view automatically follow that the legislation was irrational in the respects in question: it cannot be the case that whenever imperfections in a legislative scheme are corrected by amendment in the light of experience the original version is to be characterised as irrational on the basis that it should have been got right first time round. Whether that is a fair conclusion will depend on the circumstances of the particular case, including the nature of the defect. It may, separately, be irrational for steps not to be taken to resolve the problem once identified: this was the basis of the finding in Johnson (see para. 42 above). But, again, whether that is so will depend on the particular case. For the reasons which I have given I do not believe that irrationality on either basis has been established here.

DISPOSAL

91. For those reasons I would allow this appeal and dismiss the claim.

Haddon-Cave LJ:

92. I agree.

Simler LJ:

93. I also agree.

Note 1    When this judgment was circulated in draft, counsel for the Claimant suggested that the Court might wish to record “that in comparison with monthly-paid workers, 4-weekly paid workers are likely also to experience reduced UC in the 12th month (albeit in that month because of artificially high income, rather than the benefit cap)”. Counsel for the Secretary of State did not accept that that was an accurate characterisation of the position. I need not explore the issue because it was common ground that the point does not affect the basis on which we have decided the case.    [Back]

Note 2    Ms Krahé does not directly address the question of why sixteen hours was chosen as the figure on which that calculation is based. Mr Brown told us that the same threshold was used for the hours-related threshold under Working Tax Credit, but he was unable to follow the trail further back (though he referred us to a rather opaque reference in a report from the Social Security Advisory Committee to “the design of labour market contracts”). Mr Drabble suggested that the sixteen-hour threshold might have its origins in the threshold between “full-time” and “part-time” work historically used for entitlement to employment protection rights. Fortunately, it was common ground that the explanation for the choice was not central to the issues before us.    [Back]

Note 3    Garnham J describes the statement as being “in response” to Ms Hargreaves’ statement; but I think this is a slip, since it was filed only a day later and she does not refer to it.     [Back]

Note 4    Although she refers specifically to a 28-day cycle, the issue is the same in the case of claimants paid on a weekly or fortnightly cycle.    [Back]

Note 5    For the same reason I will not enter the debate about whether “unreasonableness” is a better label for the ground of judicial review with which we are concerned than “irrationality”, as has been suggested in some recent authorities. In truth, neither term is without its problems.     [Back]

Note 6    For convenience I refer to the Secretary of State by the gender of the current incumbent.    [Back]

Note 7    There would be a single pay day if they were paid four-weekly, but four or two if they were paid weekly or fortnightly. Theoretically there could be regular pay cycles by reference to periods other than a week, but they will be too rare to justify consideration.    [Back]