Has labour market data become less reliable?

Falling response rates to the Labour Force Survey mean that labour market data published by the Office for National Statistics (ONS) has become less reliable. In October 2023, the ONS only published some experimental headline statistics for unemployment, employment and inactivity. Less reliable labour market data might makes it harder for the Bank of England to set interest rates appropriately.

This Insight covers why labour market data for Great Britain has become less reliable, what this means for policy makers, and what will happen next.

What’s going on with labour market data?

The Office for National Statistics (ONS) produces official statistics on the labour market using the Labour Force Survey (LFS). In recent years, fewer people have been responding to the LFS, which the ONS has identified as a challenge to data reliability. 

Data for Northern Ireland is collected by the Northern Ireland Statistics and Research Agency (NISRA) in the Northern Ireland Labour Force Survey, which NISRA says continues to have robust response rates.

As shown in the chart below, the response rate for the LFS fell from 47.9% in June to August 2013 to 14.6% in June to August 2023, which means the number of people in the LFS sample in Great Britain fell from 78,994 to 36,526 over this period. Response rates fell particularly quickly during the Covid-19 pandemic, when lockdowns forced a switch from face-to-face interviews to phone calls.

With fewer respondents there is a greater chance that the LFS sample is not representative of the population as a whole, which makes the data less reliable. It is less likely to reflect the true number of people who are employed, unemployed or economically inactive (people not in work and not looking for work), and how these figures change from quarter to quarter.

The ONS already adjusts for the fact that some groups are underrepresented in the LFS: for example, non-UK nationals are less likely to respond to ONS surveys, and during pandemic lockdowns, there were not enough renters in the sample. A smaller sample size makes these problems bigger.

Recently, organisations that use the data for their work, like the Resolution Foundation think tank, the Financial Times and the Institute for Employment Studies think tank, have expressed concerns that LFS data may be too unreliable to use.

Why was the October 2023 data not fully released?

In October 2023 the ONS delayed the publication of LFS data by a week, from 17 to 24 October, because low response rates to the LFS meant the data was too unreliable to report.

Most of the LFS data usually published by the ONS was not updated at all in October: data on full-time and part-time employment, employees and self-employed people, public and private sector workers, men’s and women’s employment, redundancies, and economic inactivity by reason was not published.

Instead, on 24 October the ONS published headline data for unemployment, employment and economic inactivity for the quarter June to August 2023. Rather than using LFS data from that quarter, the ONS used employee data from PAYE records and data on the number of people claiming unemployment benefits to adjust LFS data from previous months.

The ONS has classed these as ‘experimental statistics’, which means they are still in the testing phase and are not yet fully developed. Some think tanks have said this data is also too unreliable to use. For example, the Institute for Employment Studies decided not to publish its monthly labour market statistics briefing note in October.

One of the problems with using PAYE and benefit claimant data to adjust LFS data is that these statistics are quite unreliable themselves and are also classed as experimental statistics.

The Library briefing UK Labour market statistics provides all the most recent available labour market data.

How is the ONS making its data more reliable?

The ONS is in the process of making major changes to the LFS. Part of this is switching to an online survey instead of having researchers ask questions over the phone or knock on doors. The online survey will be supported by follow-ups on the phone and face to face in areas with low response rates.

The ONS also plans to use more administrative data (data used by government departments) like PAYE data to improve the reliability of survey data.

From spring 2024, the ONS plans to publish data from the new ‘transformed’ version of the LFS.

NISRA is also making changes the Northern Ireland Labour Force Survey over the next two years.

How does less reliable data affect policy makers?

The Bank of England uses the unemployment rate as an indicator of how strong the economy is when making decisions about whether to raise or lower interest rates.

Recently, the Bank’s Monetary Policy Committee has raised rates partly because unemployment has been low. If labour market data doesn’t reflect what’s really happening, it makes it harder for the Bank to set interest rates appropriately.

The Government also uses labour market statistics in policy making. For example, in the Spring Budget 2023, the Chancellor announced measures to help disabled people and older working-age people into employment. This was informed by labour market data showing that people in these groups had become less economically active during the coronavirus pandemic.

Uncertainty around the accuracy of labour market statistics makes it more difficult for the Government to target support.

What happens next?

The next ONS labour market data release is due on 14 November and it’s not clear how much will be published then. It’s possible that the ONS will continue to just publish headline statistics using administrative data.

There is hope that the ONS might accelerate its plans and publish new ‘transformed’ LFS data before spring 2024. If not, our picture of what’s happening in the labour market might be blurrier than usual for a while.

About the Author: Brigid Francis-Devine is a researcher at the House of Commons Library, specialising in incomes and the labour market

Image by: (© By IRStone – stock.adobe.com).