Alexander Agnew & others v Chief Constable of the PSNI – Recent Decision of the Employment Tribunal in respect of the calculation of holiday pay.
16/11/2026
The Belfast Employment Tribunal recently handed down it’s remedy decision in Alexander Agnew & Others v Chief Constable of the PSNI (commonly called Agnew II.)
This case will be familiar to Members as the Decision in respect of liability made by Employment Judge Kelly in 2018, established the core principles in respect of the calculation of holiday pay for workers in Northern Ireland. That Decision was appealed to the NI Court of Appeal in 2019 and subsequently to the UK Supreme Court in 2022. The landmark Supreme Court judgment held:
- A gap of more than 3 months will not necessarily break the chain in a series of deductions where there is a common fault or unifying vice factually linking each deduction;
- There is no rule of law that holidays are taken in a particular order; the default position is that each day of holiday is made up of a fraction of each type of leave, i.e. the “composite whole” approach;
- How the appropriate reference period for calculating leave is identified is a question of fact.
Agnew II does not change any of the above, nor is it a new Claim as liability against the PSNI had already been established. What Agnew II does is address the rules governing recovery. It is therefore an important Decision, particularly for employers who have ongoing holiday pay claims.
The questions for the Employment Tribunal in Agnew II were:
- How far back can claims potentially go; 1996 or 1998? (“The Backdating Issue”)
- Can a worker claim interest or an inflationary uplift to any award to reflect the delay in receiving the pay? (“Interest Like Compensation Issue”)
- How should loss be quantified where no records exist? (“The Records Issue”)
- What is the correct divisor to retrospectively calculate holiday pay? (“The Divisor Issue”)
How far back can claims potentially go? (“The Backdating Issue”)
The Employment Tribunal held that Claimants, who have service extending to 1996 and who can establish unauthorised deductions from wages as part of a series of deductions, are entitled to claim back to 23 November 1996, the date by which EU Member states were required to transpose the original Working Time Directive.
It is notable that the Claimants are employed by a public sector body and as such can rely on the Directive by reason of the doctrine of ‘direct effect’ which does not apply to the private sector. For private-sector employers, 1998 may be the earliest potential starting point for a claim (the Tribunal did not however address the private sector position.)
For employers based in GB, arguments around alleged historical underpayments are largely unheard of due to the introduction of the 2 year back-stop in 2014 (which does not apply in NI).
Can a worker claim interest or an inflationary uplift to any award? (“Interest Like Compensation Issue”)
No. The Employment Tribunal held that neither Regulation 43(5) of the Working Time Regulations 2016 nor Article 56(2) of the Employment Rights (Northern Ireland) Order 1996 confer a power on the Tribunal to award interest or ‘interest like compensation’ as a percentage uplift in the Claimants’ claims of underpayment of annual leave.
Calculating loss where records do not exist (“The Records Issue”)
In our experience this is a major hurdle in dealing with historic holiday pay claims, as whilst workers in theory can bring a claim stretching back to 1996/1998, documents rarely exist for the entirety of the period claimed.
The question for the Tribunal was, in the absence of records from either party, should it apply a “broad brush” or “broad axe” approach to calculate loss or should each individual claimant be required to prove their loss?
The Tribunal confirmed the burden of proving a factual allegation, as a general rule, lies with the party making it. The standard of proof is the balance of probabilities. The Respondent argued the Claimants must prove that their earnings during periods of annual leave were less than their lawful entitlement and that “it is for the claimants, not the respondents to produce evidence of the amount of loss.”
The Tribunal determined that it is not appropriate in the circumstances of these
claims, to apply such a ‘broad brush’ or ‘broad axe’ approach. However, it held that the Claimants were not prevented from relying on ‘extrapolated data’ or ‘sampling’ in evidence to establish quantum. It will be a question of fact to be considered in individual cases.
The Tribunal stated:
“The tribunal accepts the argument on behalf of the claimants that there is nothing in principle which prevents a claimant in tribunal proceedings from relying on ‘extrapolated data’ or ‘sampling’ evidence in any individual claim…The Industrial Tribunals are not bound by strict rules of evidence and are entitled to consider a wide range of evidential material including evidence from ‘extrapolated data’ or ‘sampling’ to show what is due…There is no obligation on a tribunal to accept ‘extrapolated data’ or ‘sampling’ as proof of loss in claims where records do not exist. The extent to which ‘extrapolated data’ or ‘sampling’ can be relied upon is a question of fact and can only be determined by a tribunal after due consideration of all relevant evidence in individual cases.”
Agnew II does not mean that an absence of records will defeat a claim, but neither does it permit an assumption to be made that proven loss in one year will automatically mean loss can be implied for other periods where no records exist. Employers would undoubtedly have preferred the Tribunal to state that each individual Claimant must prove evidence of their specific loss.
Notably (and we refer to this again below), the Tribunal referred to the possibility of settlement negotiations stating “A ‘broad brush’ or ‘extrapolation of sampling’ may be entirely appropriate and pragmatic as a means of quantifying loss for the purposes of settlement negotiations to enable resolution of claims. This may be of considerable benefit to the parties. This tribunal does not discourage such an approach.”
The correct divisor (“The Divisor Issue”)
The parties agreed with the earlier decisions of the NI Court of Appeal and the Supreme Court that calendar days (365) is the wrong divisor and that the calculation will vary for each individual Claimant by reason of their specific working patterns/practices.
The Respondent sought to rely on a divisor of 240 working days expressed as a percentage of 8.33%:
Overtime and allowances received in one year are divided by 240, being the number of working days in the year, in order to calculate a daily rate for overtime and allowances. This is then multiplied by 20, being the total number of days of annual leave under the Working Time Directive. Expressed as a percentage, the figure is 8.33%.
The Tribunal held there is no single divisor that can be applied for the purposes of retrospectively calculating holiday pay in these claims. It confirmed that working days during the relevant reference period is the proper approach and that once a worker’s normal remuneration over the agreed reference period (12 months or such suitably representative period) has been identified, the amount of a week’s pay can be identified.
The Tribunal did not conclude that the 20/240 divisor could never be appropriate, but stated a fixed divisor for all Claimants would not be appropriate.
The Decision separately noted that the current estimate of Claimants in this multiple exceed 7,000; in light of the numbers, the likely variation in roles, working arrangements and patterns etc are, in our view, likely to be relevant to the reluctance to adopt a single divisor approach.
Practical points
Whilst there is the potential for claims to go back as far as 1996, the main issue we see in practice is dealing with the absence of records. We wait to see how the decision in Agnew II around ‘extrapolated data’ or ‘sampling’ evidence will apply in practice.
Finally, the Tribunal quoted the earlier Decision of former President (Employment Judge Kelly) where he stated that due to the significant number of individual Claimants “if each claim resulted in a remedies hearing which took two days to hear and to determine, several decades would have to be spent by a single tribunal.” The Tribunal in Agnew II stated it was unfortunate that agreement had not been reached on remedy. The Tribunal emphasised that the parties must strive to resolve the outstanding issues in compliance with the overriding objective to avoid hearings for individual Claimants which will inevitably result in considerable expense and further delay. We expect this position to be reflected across the case management of all ongoing historical holiday pay claims.