Reconciling estimates of employment rates

The Office for National Statistics (ONS) has faced challenges with the quality of UK labour market statistics, particularly from the Labour Force Survey (LFS), following a decline in survey response rates and the implementation of recovery measures. These changes have made labour market trends more difficult to interpret, as movements in headline indicators reflect both genuine changes in the labour market and changes resulting from survey improvements. As a result, there has been reduced consistency between similar labour market measures across different data sources.

This project will investigate the potential to statistically reconcile multiple data sources to produce robust estimates of key labour market indicators over time. By combining information from existing and emerging data sources, the research aims to improve estimates of labour market developments while accounting for the fact that new data contain both meaningful signals (“news”) and measurement error (“noise”). While statistical reconciliation techniques have been widely applied to different measures of GDP (from the expenditure, income and output approaches), they have not yet been developed for key labour market indicators.

The project builds on existing statistical methods developed to reconcile different measures of GDP and adapts them to the labour market. Unlike GDP, labour market statistics come from multiple sources that differ in their concepts (for example, measuring jobs rather than people in employment), publication frequency, revision patterns, and the length of their historical time series. These differences present both challenges and opportunities for developing a single, coherent estimate of labour market trends.

The research will combine evidence from the Labour Force Survey (LFS), HMRC PAYE Real Time Information (RTI), and Workforce Jobs (WFJ) to produce a reconciled time series for UK employment as a proof of concept. The project will initially focus on quarterly estimates before exploring whether the approach can be extended to produce monthly estimates using mixed-frequency methods. The findings will provide a foundation for future work to incorporate additional labour market indicators, account for data revisions, and make use of more granular data.

The project will provide a methodological foundation for improving the coherence of labour market statistics by demonstrating how multiple data sources can be reconciled into a single, robust estimate of employment.

The approach has the potential to be extended to other key labour market indicators, such as unemployment and inactivity, supporting the development of integrated labour market tables.

Beyond the UK, the methods could be shared with other National Statistical Institutes (NSIs) facing similar challenges, and, in the longer term, adapted to other areas of official statistics where survey and administrative data provide complementary measures of the same underlying concepts, such as migration or household expenditure.