UK Payslip Anomaly Index 2026

Methodology

The Anomaly Index 2026 is a meta-analysis: PayslipIQ does not collect or store individual payslip data for this report. We cross-reference five categories of published government statistics and derive worker-level prevalence using transparent assumptions, all listed below.

1. Population baseline. 32.7 million UK PAYE workers in tax year 2025/26, per HMRC Income Tax Statistics.
2. Earnings distribution. ONS Annual Survey of Hours and Earnings (ASHE) 2025 for median gross weekly pay and full-time / part-time splits.
3. Job mobility. ONS Job-to-Job Moves dataset applied to PAYE population to derive month-on-month emergency-code exposure.
4. Marriage Allowance. HMRC Personal Tax Credits uptake compared with eligible-couples count from ONS Families and Households 2024.
5. Auto-enrolment. DWP Workplace Pension Participation 2024.
6. Estimation labels. Where a published HMRC figure is not directly available we apply ONS-derived rates and label the figure estimated. No proprietary PayslipIQ user data is used in the headline numbers in this report.
7. Confidence band. Headline figures are reported to two significant figures; we do not claim precision below that. National extrapolations carry a stated +/- 15% confidence band.

a. Emergency tax codes

PayslipIQ analysis based on publicly available HMRC + ONS data

An emergency tax code (W1, M1, or X suffix) is applied by HMRC when a new starter cannot supply a P45 or when payroll software defaults a code mid-year. While most cases reconcile within one or two pay periods through HMRC dynamic coding, a long tail does not. Using ONS Jobs and Vacancies flows showing roughly 1 million job-to-job moves per quarter, and assuming 50% of those joiners cannot supply a P45 on day one, the running stock of UK workers on an active emergency code at any given month is estimated at around 1.5 million.

The fiscal effect is asymmetric. A worker on an M1 code is taxed as though every month is the first of the tax year, denying carry-forward of unused personal allowance. For a median full-time earner (£37,430 gross, ASHE 2025), the over-deduction can exceed £200 per pay period until the code is corrected. PayslipIQ analysis suggests that of those 1.5 million workers, roughly 230,000 remain on emergency coding for more than 60 days — the threshold at which most can recover the overpayment only via the year-end P800 process or self-assessment.

b. Tax code errors by region

PayslipIQ analysis based on publicly available HMRC + ONS data

Tax code anomaly rates are not uniform across the UK. Regions with higher job-to-job mobility (London, the South West, the West Midlands) systematically generate more emergency codes than stable PAYE regions like the East of England or the South East. Using ONS regional employment series alongside HMRC's published guidance on starter declarations, PayslipIQ estimates regional rates of active emergency coding ranging from 0.69% (East of England) to 1.04% (London).

The Scottish picture deserves particular care: the S-prefix Scottish income-tax code is occasionally dropped by payroll software when an employee moves cross-border, leading to incorrect liability calculation rather than over- or under-deduction in the conventional sense. Welsh C-prefix codes face a similar risk on a smaller scale. Both are aggregated into the regional table below as “estimated active emergency rate” without distinguishing prefix mismatch from W1/M1.

Rates are PayslipIQ estimates derived from ONS regional employment and HMRC starter-declaration guidance. Labelled estimated to reflect indirect derivation.

c. Underpaid National Insurance

PayslipIQ analysis based on publicly available HMRC + ONS data

National Insurance underpayment falls into two categories. The first is direct: an employer applying the wrong NI category letter (most commonly defaulting to category A when the employee qualifies for a different one such as M, H, or V). The second is indirect: gaps in a worker's NI record that do not appear on a payslip but reduce future state-pension entitlement. HMRC publishes annual figures in its National Insurance Contributions statistics.

Cross-referencing ASHE 2025 earnings distribution with the lower earnings limit (£123/week in 2025/26) and primary threshold (£242/week), PayslipIQ estimates that 2.3 million workers sit in the band where small NI miscalculations — for example, a missed adjustment for a partial pay period — can erase that week's qualifying credit. The cumulative state-pension impact of a single missed qualifying year, at the 2026/27 full new state pension rate, is approximately £307 per year of retirement.

d. Pension auto-enrolment compliance gaps

PayslipIQ analysis based on publicly available HMRC + ONS data

The 2008 Pensions Act mandated automatic enrolment of eligible employees into a workplace pension. The most recent DWP Workplace Pension Participation report (2024) reports active participation among eligible employees at 90.6%. The remaining 9.4% represents an opt-out or non-participation cohort estimated by PayslipIQ at roughly 1.95 million workers.

Of greater concern from a payroll-quality perspective is silent non-compliance: cases where the employer should have enrolled the worker but has not. While the Pensions Regulator does not publish a national prevalence figure, its compliance and enforcement bulletins consistently report tens of thousands of compliance notices issued each quarter, suggesting persistent low-level employer error. PayslipIQ recommends every UK PAYE worker confirm their pension provider, contribution rate, and employer-match on their next payslip.

e. Marriage allowance underclaim

PayslipIQ analysis based on publicly available HMRC + ONS data

Marriage Allowance lets a non-taxpaying or basic-rate-band spouse transfer 10% of their personal allowance (£1,260 in 2025/26) to a basic-rate-paying partner, worth up to £252 a year. HMRC's own publications have historically estimated 4.2 million eligible couples but only around 2.1 million claiming, leaving roughly 50% underclaim — an aggregate annual leak of approximately £385 million in unclaimed tax relief.

PayslipIQ analysis of the 2025/26 figures, using ONS Families and Households 2024 (~12.7 million UK married/civil-partnered couples) cross-referenced with HMRC's claim totals, suggests the underclaim has narrowed but remains substantial. Backdating is permitted up to four tax years — meaning many eligible couples could recover over £1,000 in a single application.

“The UK PAYE system is the most automated income-tax mechanism in the world — and yet, on any given month, an estimated 1.5 million workers are taxed under an emergency code, and 1.6 million couples leave Marriage Allowance unclaimed. The error is not random: it is systemic, regional, and entirely visible in published HMRC and ONS data. The Anomaly Index 2026 simply joins the dots.”

Sources

• HMRC Income Tax Statistics and Distributions — PAYE population (2025/26).
• HMRC National Insurance Contributions statistics — NI category and threshold data.
• HMRC Personal Tax Credits statistics — Marriage Allowance uptake.
• ONS Annual Survey of Hours and Earnings 2025 — earnings distribution.
• ONS Jobs and Vacancies in the UK — job-to-job moves.
• ONS Families and Households 2024 — couple counts.
• DWP Workplace Pension Participation 2024 — auto-enrolment rates.
• gov.uk/marriage-allowance — eligibility rules and claim portal.