The Office for National Statistics has published early research suggesting that payroll data could help explain why some employment figures have recently told different stories. The work does not revise official statistics today, but it reinforces the ONS view that users should consider administrative Pay As You Earn data alongside survey results.
The issue matters because businesses, policymakers and investors use labor-market data to judge hiring conditions, wage pressure and the strength of the economy. Those conclusions are harder to draw when surveys and tax records do not move in the same direction.
Two data sources measure different things
The Labour Force Survey, or LFS, is a household survey. It asks people about their work, job search and economic status. Pay As You Earn Real Time Information, known as PAYE RTI, is administrative data sent by employers to HM Revenue and Customs when they run payroll.
Neither source is a simple substitute for the other. PAYE RTI is particularly useful for counting employees on payroll, while the LFS provides information payroll records do not, including self-employment, unemployment and people outside the labor force.
The ONS said it linked the two datasets to examine whether changes in survey response patterns had affected estimates of employment, unemployment and inactivity. This is called non-response bias: if people who do not answer a survey differ systematically from those who do, the survey result can be distorted.
What the research found
The ONS found evidence that employment-related non-response bias in the LFS changed over time. Its proposed PAYE RTI weighting method would have produced a lower employment rate during the pandemic peak and a higher rate between 2022 and 2024 than the published LFS series.
At the beginning of 2024, when the achieved LFS sample had only recently begun recovering from a low point of about 14,000 households in Great Britain, the research indicated an employment rate 0.8 percentage points higher under the proposed method. It did not identify a substantial increase or decrease in the employment rate during 2025.
That is an illustration from a research method, not a new official employment figure. The ONS said the work is proof-of-concept research and does not establish the true level of employment, unemployment or inactivity.
Why the distinction matters
The LFS is especially valuable because it shows more than payroll employment. Someone may be unemployed, self-employed, working informally, studying or inactive for reasons that payroll data cannot fully capture.
At the same time, PAYE RTI has the advantage of coming from records created when employers pay staff. The ONS now says it currently provides the most reliable measure of employees while survey response patterns remain an issue.
This does not mean the survey is being abandoned. The ONS said that improving the LFS collection process and response rates had helped narrow the difference between survey and administrative measures during 2025. It will continue testing whether the new method could be used with its Transformed Labour Force Survey, the online replacement for the LFS, once that survey is sufficiently mature.
Businesses still need to read several signals
A payroll count alone cannot show every part of the labor market. A company deciding whether to hire will still want to consider vacancies, pay growth, unemployment, participation and sector conditions.
That is why the research is useful even before it changes any published series. It explains why one number should not carry the entire burden of describing a complicated labor market.
MBN’s earlier report that UK business activity strengthened while services employment still fell showed how company surveys can also give a timely but partial view. The ONS work is a reminder that statistics become more useful when their different coverage and limits are understood.
