Provisional ONS estimates point to 145,000 fewer employees on UK payrolls than a year ago — a 0.5% fall that adds to a run of signals of a cooling jobs market.

The number of people on UK company payrolls was around 145,000 lower in August 2026 than a year earlier, according to early estimates from the Office for National Statistics (ONS) published on 15 September. That amounts to a fall of 0.5% over the year, leaving roughly 30.2 million payrolled employees on the latest count.

The figure is drawn from HM Revenue and Customs Pay As You Earn (PAYE) records and gives one of the timeliest available reads on the state of employment. It is an important measure precisely because it captures actual Payroll records rather than survey responses, but the most recent month is provisional and the ONS cautions that it is likely to be revised as more data arrive.

A year of gradual erosion

A single month can be noisy, which is why the annual comparison carries weight. Comparing August with the same month a year earlier smooths out short-term wobbles and shows the underlying direction: over twelve months, the payrolled workforce has shrunk rather than grown. A 0.5% decline may sound small as a percentage, but 145,000 is a substantial number of individual jobs no longer showing up on payrolls.

The annual fall is consistent with other measures pointing the same way. Over the three months to July, the ONS's comparable estimate showed payrolls down 84,000, or 0.3%, on the year, and down 39,000 on the previous quarter. Taken together, the monthly, quarterly and annual figures describe a labour market that has been losing employees gradually rather than through any sudden shock.

What the payroll measure captures

The PAYE payroll series counts employees paid through the PAYE system. It does not include the self-employed, who are captured in separate survey-based statistics, so it is a measure of employee jobs rather than of total employment across every category of worker. That makes it a strong indicator of trends among employers, but not a complete map of everyone in work.

It is also essential not to read a payroll decline as a headcount of confirmed redundancies. A payroll can fall for many reasons that the data do not itemise: fewer people being hired to replace those who leave, retirements, people moving into self-employment, or workers leaving the labour force altogether. The ONS figures record that the total has fallen; they do not, on their own, tell us how much of that reflects active Job losses as opposed to slower hiring.

The wider labour-market backdrop

The payroll numbers sit within a broader set of labour-market indicators that have softened in tandem. The Unemployment rate stood at 4.9% in the three months to July 2026, and the employment rate for people aged 16 to 64 was 75.1%, with an economic inactivity rate of 20.9%. None of these is at an alarming level by historical standards, but the movement in recent releases has generally been towards a looser market.

Demand for new staff has eased alongside it. The number of Job Vacancies fell to 702,000 in the three months to August, down 8,000 on the previous quarter. Fewer vacancies alongside a shrinking payroll suggests employers have become more cautious about adding to their workforces, which is one plausible route through which a payroll total can drift lower over a year even without large-scale redundancies.

Pay is still rising

A cooling in the quantity of jobs has not, so far, been matched by a collapse in the price of labour. Average regular pay, excluding bonuses, grew by 3.5% over the year to July, and total pay including bonuses rose by 3.9%. After adjusting for inflation, regular pay was still up in real terms, by 0.6% on the CPIH measure. In other words, those in work have generally continued to see modest real-terms pay gains even as the overall number of payrolled employees has fallen.

That combination — fewer employees but continued pay growth — is a familiar feature of a labour market that is loosening gradually rather than seizing up. It does not resolve the question of where the jobs market heads next, but it does temper any reading of the payroll fall as evidence of acute distress among those still employed.

Why the provisional label matters

The ONS is explicit that August's payroll estimate should be treated as provisional and is likely to be revised when more data are received. Recent history shows these early figures can move, sometimes materially, between first publication and later revisions. The safest reading is therefore to treat the 145,000 annual fall as a firm signal of direction rather than a number fixed to the last employee.

Even with that caveat, the message is reasonably clear. Across monthly, quarterly and annual horizons, and alongside falling vacancies and a slightly higher unemployment rate, the payroll data point to a UK jobs market that has been contracting modestly. Whether that softening deepens or stabilises will depend on data still to come — and on how far these early estimates are revised in the months ahead.

A timely but partial gauge

The payroll series has become one of the most closely watched labour-market indicators precisely because it is so timely, arriving weeks ahead of the survey-based measures that cover the same period. Its speed, however, comes with the trade-off of provisionality: the newest month is always an early estimate. That is why analysts tend to focus on the run of recent months together rather than fixating on the latest single figure, which carries the greatest risk of later revision.

It is also worth setting the annual fall in proportion. A decline of 145,000 against a payroll of around 30.2 million is a fall of roughly one in every 208 employees. That is enough to register clearly as a change in direction after years in which the payroll count generally rose, but it is far from the kind of sharp drop associated with a recession. The significance lies in the Reversal of trend as much as in the size of the number itself. Comparing August with the previous August also strips out much of the seasonal pattern in employment, leaving a cleaner view of the underlying change over the year. For now, that underlying trend runs gently but clearly downward.