Payrolling benefits from April 2027: predicting the spec HMRC hasn't published yet

From 6 April 2027 company cars, vans, fuel and medical benefits get taxed through payroll in real time, with employer Class 1A NI reported on the Full Payment Submission (FPS) every pay period. The final spec isn’t out until autumn 2026, about five months before go-live. But most of the pieces are already published, so you can predict most of it now. That’s what this post does, then it goes through where I think it’s going to hurt.

I’m a software developer, not an accountant. Everything here is from published material (linked at the end). The “Things I worked out myself” section is my own arithmetic, so check it.

The short version#

Today the tax on a company car or medical cover is sorted out after the year ends: a P11D in July, a tax code adjustment, one annual Class 1A payment. From April 2027 the benefit value is added to taxable pay every payday as “notional pay” (taxed, never paid out) and Class 1A is worked out and reported every payday too.

How we got here#

The policy was announced, delayed, then split into phases, mostly because the industry told HMRC the timetable didn’t leave room to build and test. ICAEW asked for a delay, a phased rollout and a penalty soft landing. HMRC gave them all three.

So 2027/28 is a dual-running year. Cars go through payroll, a gym membership can still land on a P11D in July 2028. The P11D code path stays for years.

Predicting the spec#

A draft spec already exists. HMRC sent developers over 100 new FPS data items in November 2025 covering full mandation. The June 2026 phasing pulled 94 of them out. What’s left for phase 1 is about 32 fields, 14 already in RTI, so roughly 18 new ones to build.

Annex 3 of the interim guidance lists the likely phase 1 fields by P11D section letter:

My base prediction is Annex 3 plus validation rules, delivered as the usual RIM artefact and business rules. The open questions:

1. Validation. Can a YTD cash equivalent go down? Are negative period values allowed? The correction model below needs both.
2. End-of-year update transport. Plain FPS resubmission, an Earlier Year Update style thing, or something new? I think revised final FPS, but it’s not in writing.
3. P46(car). Not mentioned once. Probably redundant for payrolled cars.
4. The Class 1A adjustment box. Up to nine pending sub-fields says HMRC knows per-period Class 1A won’t reconcile cleanly. Expect this to change latest.

One surprise: voluntary payrolling of non-mandated benefits won’t use the new FPS fields. So from April 2027 an engine might run three regimes at once: mandated with FPS fields, voluntary without, and P11D. Worth knowing before you commit to a data model.

The mechanics HMRC has explained#

Spreading#

Annual cash equivalent ÷ pay periods, added to taxable pay each period as notional pay, rounded down to 2dp. Taxed through the tax code, no employee Class 1 NIC, never touches net pay. Try £2,000 (a £40,000 electric car at 5%): twelve rounded-down periods come to £1,999.92 and the guidance doesn’t say whether the last period trues up.

Also confirmed: 53-week years divide by 53, annual schemes payroll the lot in one period, and the notional amount is expected on the payslip.

Recalculate forward, never backward#

When a value changes mid-year you don’t amend earlier periods. New full-year figure, minus what’s already payrolled, spread over the periods left. HMRC’s example: 4-weekly, £2,600 becomes £2,800 after six periods, £1,200 already through, so £1,600 over the last seven at £228.57.

Everything funnels through this one algorithm: corrections, car swaps, premium changes, estimates trued up. Your benefit object needs annual value, payrolled to date, periods remaining, and a recalc that runs on every change. There’s a simulator in “Things I worked out myself”, because it breaks on one input.

Estimates, the 50% limit, and the year-end tail#

• Estimates are expected. If the value isn’t known, use a reasonable estimate (last year’s premium is fine), fix it in-year. Zero for a benefit you know exists is not acceptable. No penalty for a reasonable estimate. So estimate-vs-actual is a first-class state.
• The 50% limit stops being rare. Tax in a period can’t exceed 50% of cash pay. Someone on maternity pay with a company car goes straight through it. The excess must carry forward within the year, and anything left at year end HMRC collects from the employee (P800, Simple Assessment, SA). You also keep filing an FPS for benefits-only employees with no cash pay. That’s a per-employee carry-forward ledger most engines don’t have.
• Making good. Payroll the value net of expected reimbursement. Employee pays by 6 July, employer revises by 22 July if they don’t.
• Leavers. Benefits continuing after termination move to the FPS, reportable by 22 July. A value change after the P45 (premium refund in month 11) isn’t described anywhere. ICAEW has asked.
• Employee statement by 1 June. No format, nothing on the P60 or P45. A document generator for your backlog.
• Corrections after 19 April become end-of-year corrections, expected to be a revised final FPS with amounts due 19/22 July and P60s reissued. Final deadline still “expected”, so TBC.

Where it’s going to hurt#

1. Class 1A goes per-employee, per-period. Today it’s one annual sum, one P11D(b), one July payment. From April 2027 it’s on each period’s cash equivalents in its own FPS fields, “paid in real time”. HMRC hasn’t said how it’s paid: same monthly remittance or a separate stream. Promised “in a future update”. I think it’ll be the same reference and the same 19th/22nd deadlines, but that’s a guess.
2. July 2027 is a double-payment month. The last annual Class 1A for 2026/27 is due 19/22 July 2027 while monthly 2027/28 payments are already running. HMRC flagged this itself. Calculator below.
3. Tax code squeeze. HMRC strips mandated benefits out of codes before April 2027 but will not remove prior-year underpayment adjustments. So someone pays real-time tax on this year’s car and arrears on last year’s. Correct, looks like double taxation. Expect “my tax is wrong” tickets April to June 2027, and put an explanation on the payslip.
4. Medical data timing. Car data lives in fleet systems and rarely changes. Medical data lives with an insurer, renews mid-year, and arrives after cut-off from a different team. My prediction: medical, not cars, generates most in-year corrections. Make “update value, respread, log it” one click.
5. Odd payrolls. 53-week years (small statutory mismatches ICAEW has queried), annual schemes (whole benefit in one period, straight into the 50% limit), benefit-only payrolls, and modified PAYE expats where HMRC is “considering retaining the P11D”. Nobody knows yet. Plan for both.

Things I worked out myself#

None of this is published anywhere as far as I can tell. It’s what you get by taking the rules above and doing the arithmetic.

Two years of Class 1A get paid in one#

Scale it up: Class 1A raises about £1.4bn a year (£9.5bn of benefits at 15%), and phase 1 is most of it, so the transition year is a one-off timing gain for the Exchequer in the high hundreds of millions. I haven’t seen anyone put a number on it. And permanently, employers pay the same money about eight months earlier than before. At 7% borrowing that’s roughly £1,400 a year for the £30,000 employer, every year.

Recalculate forward can go negative#

Nothing published says what happens then. Either the spec allows negatives and falling YTDs (and HMRC, DWP and your reports cope), or it rejects them and pushes every over-estimate to year end, which contradicts its own “fix it in-year” rule. Leavers with post-P45 refunds hit this first.

The simulator shows the other property too: a correction’s impact depends on its date, not its size. £1,200 in month 1 is £100 a month. In month 12 it’s £1,200 on one payslip, right where the 50% limit kicks in. So it now matters when a medical scheme renews. A January renewal squeezes the correction into three periods, a May one spreads it over eleven. Bureaux could usefully record every client’s renewal month today.

Nobody has mentioned the P11D(b)#

An employer with a company car (mandated) and a gym membership (not until 2028) pays Class 1A through two channels in 2027/28. Today’s P11D(b) asks for Class 1A on all benefits. Unless it’s redesigned, that employer pays part of it twice. No published document mentions a P11D(b) change. It isn’t an FPS field, it’s a form owned by a different bit of HMRC, so I’d expect it to be announced late.

The real crunch is summer 2028#

April 2027 is a clean-slate go-live with 18 new fields. Look at what stacks into April to July 2028, all for the first time:

If you’re planning capacity, it’s an April 2027 go-live followed by a bigger peak twelve to fifteen months later. It’s also why I think phase 2 slips.

Universal Credit#

1. UC shrinks as earnings rise, 55p per £1.
2. DWP reads earnings straight from the FPS.
3. UC regulation 55(2) says benefits in kind do not count as earnings.
4. Payrolling puts benefits inside taxable pay on the FPS.

So somewhere the benefit has to be subtracted back out, and no DWP-side change has been announced. It hasn’t mattered because company car drivers are rarely on UC. Mandation sweeps in every van driver and everyone with employer medical, and 3.1 million UC claimants are in work. A flat-rate van benefit is about £4,000 a year, £335 a month of notional pay, which if it leaks costs the claimant about £184 a month. They’d have no way to trace it, and the question lands on the employer’s payroll team first.

The predictions#

And the 2027/28 penalty easement (no inaccuracy penalties except deliberate non-compliance; late filing, late payment and interest still apply) is the soft landing everyone’s first year will lean on.

What to build before the spec arrives#

1. A period-level benefit model: annual value with estimate-vs-actual state, payrolled to date, periods remaining, recalculate-forward as the one path for every change.
2. The 50% limit carry-forward ledger with a year-end residual.
3. Class 1A period and YTD accumulators per benefit type, plus an adjustment mechanism you expect to rework.
4. The boring bits: notional pay on payslips in plain words, the 1 June statement, dual-running P11D, fleet and insurer imports. The tax calc is the easy part compared with getting a premium in before cut-off.

And get on SDST’s mailing list (SDSTeam@hmrc.gov.uk). The June 2026 phasing news reached developers by email before GOV.UK caught up.

Sources#

HMRC / GOV.UK:

• Changes to reporting of benefits in kind from April 2027 (L-Day policy paper + draft legislation, July 2026)
• Interim draft guidance: reporting benefits in kind in real time - especially Annex 3 (FPS fields) and the phased introduction
• Technical note: mandating the reporting of benefits in kind through payroll software - an update (April 2025)
• Employer Bulletin, June 2026
• RTI support for software developers (spec collection)
• Benefit in kind statistics commentary, July 2026 - source of the ~£9.5bn total taxable value used above
• The Universal Credit Regulations 2013, regulation 55 - the benefits-in-kind exclusion from UC employed earnings
• DWP: Universal Credit quarterly statistics, February 2026 - the 3.1 million in-work claimants figure

Industry:

• ICAEW REP 19/25: Mandatory payrolling of benefits in kind (February 2025)
• ATT: HMRC announce phased implementation of mandatory payrolling
• PAYadvice: Payrolling benefits April 2027 - phasing (SDST communication)
• CIPP: payrolling to be phased from April 2027

This describes announced policy, draft guidance and draft legislation as at September 2026, plus predictions that are clearly marked as mine. It isn’t tax advice, and plenty of it will probably change before April 2027.