The Quality and Usefulness of ONS Statistics

The panel expressed the highest confidence in consumer price statistics and the lowest confidence in labour market statistics. Around three-fifths of the panel reported high confidence in consumer price statistics, while two-fifths reported low confidence in labour market outputs and a further tenth reported very low confidence.
The panel widely viewed improvements to labour market statistics as the most important for strengthening the conduct of economic policy. Three-fifths of the panel identified labour market statistics as the most important output to improve for policy. Regional outputs, the digital economy and national accounts statistics each received a tenth of the votes, while one panellist prioritised statistics on the financial sector.

Question 1: How much confidence do you have in each of the following sets of ONS statistics as a basis for economic policy analysis and decisions?

This question received eighteen responses.

The panel expressed the strongest confidence in consumer price statistics, with three-fifths reporting high confidence and none reporting low confidence, although one panellist reported very low confidence. By contrast, confidence was weakest in labour market statistics, with over half of the panel reporting low or very low confidence and just one tenth reporting high confidence. Confidence in national accounts data was more mixed, with similar proportions of the panel reporting high and low confidence. Confidence in trade and public sector finance statistics was moderate to high, while financial services statistics attracted the largest share of moderate confidence responses.

Question 2: In which of these areas would an improvement in the quality of ONS statistics do most to improve the conduct of economic policy?

This question received eighteen responses.

Three-fifths of the panel selected labour market statistics as the area where improvements in quality would do most to improve policymaking. National accounts, the digital economy, and regional and subnational statistics each accounted for around a tenth of responses, while financial services statistics were selected by one panellist. No panellists identified consumer prices, trade statistics or public sector finance statistics as the area where improvements would have the greatest impact on policymaking.

Question 3: What do you see as the most important outstanding issues to resolve with the quality and resilience of ONS surveys and economic statistics?

This question received six responses.

Productivity and labour market statistics were highlighted as areas of concern by panellists who commented on this question. Jonathon Hazell (LSE) argued: “Obviously the worst important product put out by the ONS is the labour force survey, which has many well documented issues. Alongside the better known problems, the hours worked series in the ONS appears to be suspect. For instance, the GDP per hour worked and GDP per worker series are diverging in puzzling ways, alongside severe attrition bias in the hours worked series. I suspect that measurement of hours worked during the new age of remote and hybrid work is challenging, but ever more important. One less well known but equally concerning area is industry productivity statistics. As AI starts to become important, it is imperative to have timely industry information on output, output per worker, productivity and so on—in order to understand which industries are benefitting or not. Moreover one can often learn from cross industry patterns what are the aggregate drags on growth—for instance, Brexit or housing would harm some industries more than others. My sense is that the revisions to industry accounts are too large to draw any firm conclusions right now.”

To access the full panel responses, including the free text comments where respondents expanded on their answers, please download this MS Excel sheet.