Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
Back to all articles & webinars

Clarity for Irregular Workers: Paid Annual Leave Accrual in the Supreme Court


Joel Wallace provides a review of the recent Supreme Court decision in Harpur Trust v Brazel [2022] UKSC 21. A copy of the judgment can be found here. Topics include: annual leave entitlement, annual leave pay, zero-hour contracts, gig workers and irregular workers.

Please click here for a PDF version of this article.


Most UK employers and employees are cognisant of a standard entitlement to paid annual leave. In the main, workers can take 28 days of paid holiday per year as a minimum. Lawyers and HR professionals will know that this entitlement stems from the Working Time Regulations 1998 (SI 1998/276) (“WTR”), regs 13 and 13A, which frames the entitlement in terms of weeks. According to the WTR, workers are entitled to 4 weeks (reg 13(1)) plus 1.6 weeks (reg 13A(2)(e)) of paid leave per leave year, which gives 5.6 weeks in total.[1] For the 9am-to-5pm, 5-day-per-week worker, the amount of paid annual leave is straightforward to calculate in days:

5.6 weeks x 5 working days = 28 days per year.

An apparent difficulty in calculation arises when working time is irregular: the worker works neither 9 to 5 nor 5-days-per-week and, to add a layer of complexity, there is no way of determining which hours or how many days they will work within a given week, month or year. How should employers calculate statutory annual leave pay for these irregular workers? The Supreme Court has confirmed the answer to this question in Harpur Trust v Brazel [2022] UKSC 21.

This article endeavours to provide a digestible overview of Brazel and some key points in the case. For the time-conscious reader, the main takeaway (how one calculates holiday pay for irregular workers) is found in the section entitled “Competing Methods of Calculation” and the conclusion below.

The Dispute

Ms Lesley Brazel is (and was) a music teacher on a zero-hours contract: she had no entitlement or obligation to work for any particular number of hours in a given week. Ms Brazel was employed at Bedford Girls School, which was run by Harpur Trust (“the Trust”). In the main, Ms Brazel’s hours varied with the number of students that she taught and tasks that she was given. Further, Ms Brazel did not work during school holidays.

A disagreement arose between Ms Brazel and the Trust over holiday pay. Ms Brazel contended that the Trust had miscalculated her holiday pay and that she had been underpaid. The Trust maintained that its calculation was correct. Ms Brazel brought the matter before the Employment Tribunal as an unlawful deduction from wages claim pursuant to the Employment Rights Act 1996 (“ERA 1996”), pt II. The Employment Tribunal sided with the Trust’s calculation and dismissed the claim. Ms Brazel also claimed under the Part-time Workers (Prevention of Less Favourable Treatment) Regulations 2000 but this dismissed claim was not the subject of the subsequent appeals.

The matter then took the usual route to the UK Supreme Court. Ms Brazel appealed the Employment Tribunal’s decision to the EAT and then, with UNISON intervening, the Trust appealed to the Court of Appeal. Both the EAT  and Court of Appeal disagreed with the first-instance decision and favoured Ms Brazel’s method of calculating her annual leave entitlement.

Competing Methods of Calculation

The Trust’s Methods

The Trust forwarded two methods of calculating Ms Brazel’s holiday entitlement before the Supreme Court. The first was called the “Percentage Method”. This method was practiced by many employers and is reflected in a now defunct edition of ACAS’s guidance on annual leave (taken from ACAS Guidance: Holidays and Holiday Pay, p 6):

What leave do casual workers get?

If a member of staff works on a casual basis or very irregular hours, it is often easiest to calculate holiday entitlement that accrues as hours are worked. The holiday entitlement of 5.6 weeks is equivalent to 12.07 per cent of hours worked over a year.

The 12.07 per cent figure is 5.6 weeks’ holiday, divided by 46.4 weeks (being 52 weeks – 5.6 weeks). The 5.6 weeks are excluded from the calculation as the worker would not be at work during those 5.6 weeks in order to accrue annual leave.”

The second method, called the “Worked Year Method”, reduced the leave entitlement of an irregular worker by the proportion of weeks worked in a year when compared to those hours worked by a full-time worker. A full-time worker works 46.4 weeks in a year. Ms Brazel’s 34 weeks worked in the year means that Ms Brazel works 73% of the weeks worked by a full-time worker. According to the Worked Year Method, Ms Brazel’s entitlement is reduced by that same percentage: 5.6 weeks x 73% = 4.09 weeks.

Behind these methodologies lies, on the Trust’s case, the pro-rata principle (dubbed the “conformity principle” before the Supreme Court). It is a principle of holiday accrual which attaches to the time that an irregular worker actually worked as a proportion of a full-timer’s working time. In other words, the principle assumes that 5.6 weeks’ paid leave is a full-time worker’s entitlement and an irregular worker’s entitlement to leave is calculated as a proportion of that entitlement. The more hours the irregular worker works, the greater the entitlement to leave.

Ms Brazel’s Method: the Calendar Week Method

Ms Brazel’s contention (labelled the “Calendar Week Method” by the Supreme Court), on the other hand, was arguably far simpler: 5.6 weeks of leave per leave year under the WTR means 5.6 weeks of leave per leave year. On this reading of WTR, regs 13 and 13A, leave entitlement is fixed regardless of the number of hours, days or weeks that one actually works. If Ms Brazel ended her employment having worked half of the leave year, then her entitlement up to that point would simply be half of 5.6 weeks. The amount of pay that she would receive for each week of leave was a different question and one which would be resolved in accordance with WTR, reg 16 and the Employment Rights Act 1996, ss 221-224 ie by calculating Ms Brazel’s average weekly pay in the 12 weeks of work prior to the first day of leave.[2] The reader should note that those 12 weeks (or 52 weeks as is currently the case) are weeks on which remuneration was payable by the employer; non-remunerable weeks do not count.

The Trust’s Arguments before the Supreme Court

The conformity principle underlying the Trust’s calculation was based on a “sophisticated” rationale.[3] The Trust’s primary submission was that the conformity principal must be read into the WTR because EU law prescribes the principle. The Working Time Directive 2003/88/EC (“WTD”), art 7 naturally formed the legislative pillar for the Trust’s contention since the WTR is the domestic progeny of the WTD:

“1. Member States shall take the measures necessary to ensure that every worker is entitled to paid annual leave of at least four weeks in accordance with the conditions for entitlement to, and granting of, such leave laid down by national legislation and/or practice.

2. The minimum period of paid annual leave may not be replaced by an allowance in lieu, except where the employment relationship is terminated.”

The Trust then turned to the recent case of QH v Varhoven kasatsionen sad na Republika Bulgaria (C-762/18) [2020] EU:C:2020:504. This case was concerned with whether national law could prevent the accrual of paid annual leave for the period between a worker’s unlawful dismissal and subsequent reinstatement. The CJEU responded in its judgment that, as with instances of sickness, if the worker was prevented from working then paid annual leave would still accrue notwithstanding the fact that no work had been performed; national law could not prevent paid annual leave from accruing during the period in question. However, it was noted by the CJEU that the situation in QH was an exception to the general rule that,

Entitlement to paid annual leave must, in principle, be determined by reference to period of actual work completed under the employment contract”[4]

According to the Trust, this general rule was in fact an iteration of the conformity principle. In fact, for the Trust, this iteration merely confirmed a feature of CJEU jurisprudence which had been in existence for some time. One noteworthy case is Hein v Albert Holzkamm GmbH & Co. KG C-385/17. Mr Hein was a concrete worker who alleged that he had been underpaid for annual leave. The alleged underpayment arose from the method of calculation which took into account a period in which Mr Hein worked less than normal hours. A question referred to the CJEU can be crudely simplified as follows: Does EU law permit a calculation of holiday pay where that calculation leads to a reduction in the amount that the worker would have received had the worker been working normal hours?  The CJEU answered no. Courts must interpret domestic law, so far as possible, so that remuneration for annual leave is not less than the average of normal remuneration during periods actually worked.[5] For the Trust, this case and its conclusion made it clear that entitlement to leave and remuneration must be calculated with reference to normal periods of work under the employment contract. However, this contention is difficult to maintain given that the CJEU in Hein expressly separated the issue of entitlement (how much annual leave can be taken) and remuneration (how much should be paid whilst on annual leave). Whilst entitlement, as a minimum, had to accrue for periods actually worked (although it is open for states to legislate more generously), it was only the remuneration that had to reflect normal work periods.[6]

Another important case in the Trust’s argument is Greenfield v The Care Bureau Ltd C-219/14, [2016] ICR 161. Here, the CJEU was tasked with answering the question of whether the WTD and Framework Agreement oblige or prohibit member states from recalculating annual leave in accordance with a worker’s new work pattern, which included increased hours. The CJEU answered by holding that retrospective recalculation of annual leave was not required. Rather, a new calculation must be performed for the period of time in which the hours were increased.[7] Accordingly, so the Trust submitted, the CJEU confirms that pay for annual leave must follow time worked. Interestingly, in Greenfield the CJEU also held that the principles in operation for calculating annual leave must be consistent regardless of whether the calculation is for the purposes of determining outstanding leave during employment or determining pay in lieu of leave on termination of the employment contract.[8] This point must surely have weakened the Trusts argument since the calculation for pay in lieu of leave is clearly defined under the WTR, reg 14 and does not resemble the Trust’s proposed methodologies.

The Trust then presented arguments for how this conformity principle could or should be accommodated by domestic law. The Trust asserted that the WTR must be interpreted so as to conform with EU legislation, per the well-established rationale in Marleasing SA v La Comercial Internacional de Alimentatión SA [1990] ECR I-4135. Further, the Trust argued that the Calendar Week Method gave rise to an absurdity in the WTR because irregular workers might be paid more for annual leave proportionate to their annual pay than their regular-hour counterparts. The Trust also argued that the WTR prescribes a gateway by which the conformity principle and its methods of calculating annual leave could be given effect. A “week’s pay” under the WTR is defined with reference to the ERA 1996, ss 221-224. Those sections are supplemented by ERA 1996, s 229(2) which provides that “other payments” may be apportioned “as is just” where those payments do not “coincide with the periods for which [it is] calculated“. The Trust argued that the conformity principle could be read into section 229(2)’s “apportionment”.

The Supreme Court’s Decision

The Supreme Court dismissed the appeal, with Lady Rose and Lady Arden jointly authoring the rationale for the court’s decision. The Supreme Court agreed with the Trust that the conformity principle was a feature of EU law and that the principle is not found within the WTR.[9] However, the Supreme Court held that the Trust’s methods of calculation were wholly new and were not envisaged by the WTR. Further, the court held that it need not be read into the scheme. Although the WTR should be read so that it is consistent with the WTD per Marleasing, there is nothing which prevents states from conferring additional benefits and rights to workers over and above those required by EU law.[10] In Ms Brazel’s case, the WTR’s scheme meant that she would be paid more for annual leave that she would have receive under the conformity principle and the Trust’s methods of calculation. Accordingly, the Calendar Week Method espoused by the WTR could not be said to derogate from Ms Brazel’s rights under EU law; on the contrary it enhanced the right.[11]

Whilst the absurdity argument was “one of the strongest arguments”[12] for interpreting the WTR in the manner suggested by the Trust, the Supreme Court did not,

[R]egard any slight favouring of workers with a highly atypical work pattern as being so absurd as to justify the wholesale revision of the statutory scheme which the Harpur Trust’s alternative methods require. We agree with Underhill LJ’s observation as to the odd results produced in extreme cases… General rules sometimes provide anomalies when applied in untypical cases…”[13]

Further, the Supreme Court went on to hold that ERA 1996, s 229(2) is not a provision which entitles a court (or tribunal) to dispense with the methods of calculating a week’s pay under sections 221-224. Rather, section 229(2) is designed to deal with payments which fall outside of the usual remuneration, such as bonus pay.


Ms Brazel’s method of calculating annual leave entitlement, the Calendar Week Method, has now received Supreme Court backing. Whilst this method is at odds with old employment practices and ACAS guidance in respect of irregular workers, this author hopes that irregular workers and their employers have not been caught off guard; the EAT in 2018 and the Court of Appeal in 2019 have already endorsed the Calendar Week Method.

Those employers who have yet to implement the Calendar Week Method may find themselves in some difficulty. The recent gig-worker decision in Smith v Pimlico Plumbers Ltd [2022] EWCA Civ 70[14] highlights the risk that an employer faces where they fail to provide workers with their paid annual leave entitlement in accordance with EU law: such employers may remain liable for the outstanding pay, particularly if the worker ceases employment before any correction is made.

Finally, the fact that Calendar Week Method paid Ms Brazel more than she would have received under the Trust’s calculations gave the Supreme Court reason to avoid Marleasing because Ms Brazel’s rights under EU law remained intact; in fact, Ms Brazel received more pay under the Calendar Week Method. However, this evasive manoeuvre leaves the door ajar for further argument. If an irregular worker finds themselves in a worse position, with less pay under the Calendar Week Method, tribunals and courts will have to address Marleasing and any inconsistency between the WTR and the conformity principle head-on. Such a case will undoubtedly be rare and may also give rise to discriminatory treatment against part-time workers in any event.

[1] Note that this basic entitlement is capped at 28 days (WTR, reg 13A(3)). Of course, it is always open to employers to give workers more than the basic entitlement.

[2] Note that the 12-week average was extended to a 52-week average as of April 2020 per Employment Rights (Employment Particulars and Paid Annual Leave) (Amendment) Regulations 2018 (SI 2018/1378), reg 10(3)(b).

[3] Using Underhill LJ’s terminology in the Court of Appeal, see [2019] EWCA Civ 1402 [36].

[4] QH [58].

[5] Hein [53].

[6] Hein [24], [26]-[27] and [32]-[34].

[7] Greenfield [42]-[44].

[8] Greenfield [57].

[9] [2022] UKSC 21 [42]-[52].

[10] See R (Z) v Hackney London Borough Council [2020] UKSC 40; [2020] 1 WLR 4327 [114]

[11] [2022] UKSC 21 [71].

[12] [2022] UKSC 21 [43].

[13] [2022] UKSC 21 [72].

[14] A summary of this case can be found here.

Related Members
Joel Wallace
Shortlist Updated