The Labour Force Survey shows 356,000 more people in work than a year ago; payrolls show 111,000 fewer employees. Unemployment fell to 4.7%, but the ONS says to read the sources together.
The official picture of the UK jobs market has split in two, a week before the Budget and a fortnight before the Bank of England sets interest rates. On 20 October the Office for National Statistics (ONS) reported that unemployment fell to 4.7% in June to August, and that the employment rate rose to 75.3%.1 On the same day it reported that the number of payrolled employees, counted from tax records, fell by 111,000 over the year to August.1
The Labour Force Survey, a household survey, puts employment up 356,000 on the year; unemployment down 26,000.1 The two employment measures are about 470,000 apart, in opposite directions.
The gap matters because both the Chancellor and the Bank read the jobs market as a gauge of slack. Noodle News reported on 11 October that there were about 2.5 unemployed people for every vacancy, the most outside the pandemic since 2015. On the latest figures the ratio is about the same: 1.71 million unemployed against 691,000 vacancies.1
Part of the gap is definition. The survey counts people in all kinds of work, including the self-employed. Payrolls count employees in jobs paid through PAYE, and so miss the self-employed.23 The ONS says the rise in part-time work over the year was driven by self-employed people working part-time.1 That cannot explain a fall in payrolls alongside a large rise in the survey on its own, but it points the same way.
The rest is a question of survey quality. The Learning and Work Institute calls the Labour Force Survey “notoriously troubled” because response rates fell after the pandemic.3 The ONS says responses are now close to pre-pandemic levels, at about 80,000 in the first quarter of 2026, but that confidence intervals for unemployment remain wider than before.2 To test a collection problem in May and June this year, the ONS removed about 4,000 affected cases and reweighted; that moved the March to May employment rate up 0.18 points and unemployment down 0.10.2
The ONS does not say which series is right. It advises reading the survey alongside payrolls, the Claimant Count and workforce jobs, and focusing on long-term movements.1 The Claimant Count, an administrative measure of people claiming unemployment-related benefits, was 1.701 million in September and has risen on the year.1
The ONS says vacancy falls come mainly from smaller firms, which may be holding back from recruiting because of higher labour costs.1 That fits the payroll reading better than a jobs boom.
For people in work the picture is of modest gains. Regular pay growth in June to August was 3.7%: 3.0% in the private sector and 6.3% in the public sector, where the ONS says the timing of pay awards distorts the comparison.1 With CPI inflation at 3.1% in August, regular pay rose about 0.8% in real terms.15
For the Bank the question is whether energy costs will spill into wages. In September it judged that slack in the labour market “appears to have stabilised”, though some members thought it premature to rely on that, and it put unemployment at 4.9%.4 A fall to 4.7% supports the Bank’s more hawkish members if it is real. A shrinking payroll count supports those urging patience. Noodle News sets out that debate.
How much of the 470,000 gap is self-employment and how much is survey error, since the ONS has not published a reconciliation. Whether the September payroll estimate, down 66,000 on the year and provisional, will be revised. Whether the fall in unemployment survives the next survey quarter.
Chancellor John Healey delivers the Budget on 28 October, with fresh Office for Budget Responsibility forecasts. The Bank of England’s Monetary Policy Committee decides on 5 November. The ONS publishes its next labour market bulletin on 17 November.1 The ONS plans to move headline statistics to a transformed Labour Force Survey in 2027, which is intended to fix some of these problems.2