MNI INTERVIEW2: UK Covid Data Distortions Less Than Feared

Distortions of UK statistics due to changes in behaviour during the Covid pandemic have been smaller than feared, the deputy national statistician and a former senior Bank of England official told MNI, adding that headline figures remain meaningful even if some new biases must be taken into account in inflation and labour force data.

At the onset of the pandemic it appeared possible that lockdowns and the public's health fears might throw consumption patterns out of line with weights in inflation baskets, throwing the data's usefulness into question, said Jonathan Athow, Director General of Economic Statistics at the Office for National Statistics. But the fear that the prices of some goods and services would tumble due to plummeting consumption as people stayed home, whilst others which saw rising demand would be squeezed higher has proved unfounded, he said.

"I was really nervous we would see very large changes in prices across the basket in such a way that it would mean, actually, it wasn't meaningful," Athow said in a joint interview with Martin Weale, a former Bank of England Monetary Policy Committee member and ONS adviser, following the annual conference of the Economic Statistics Centre Of Excellence.

"All the work we have done so far, and we have done attempts at re-weighting baskets of different degrees of sophistication, suggest that those effects were pretty small," Athow said.

An ONS study published last month indicated that the difference between headline and estimated actual inflation was just 0.2 percentage point.

The ONS will press ahead with the development of alternative inflation baskets, even as it continues to publish continuity CPI and CPIH measures, while Weale said being ready to change weights based on real-time spending patterns may be appropriate given evidence of a second Covid wave.

"Why shouldn't we think of applying forecasting and nowcasting techniques to produce up-to-date weights of consumption shares?" Weale said.

"So far it has simply been seen as a measurement problem and I am just wondering whether there is some gain in seeing it as a nowcasting or forecasting problem. That is an issue to think about over the next few months," he added.

LABOUR FORCE DISTORTIONS

Another risk of data distortion comes from the ONS's forced switch to telephone interviews for the international standard Labour Force Survey, as the pandemic meant its staff were no longer able to visit households. This skews the survey to homeowners rather than renters.

"The data collection issues around the Labour Force Survey they were one of my top risks early on," said Athow, who has argued for checking the data against alternative surveys.

"Our view at the moment is that there is more uncertainty than normal, but the headline numbers are pretty reasonable," he said, although but he acknowledged ongoing problems including that of failing to fully capture immigrant labour.

"If we are getting more homeowners and fewer renters does that mean, for example, that we are getting fewer non-UK citizens answering it as they are more likely to be renters?"

One way the problem could be addressed was by re-weighting the labour data to better capture the rental sector, said Weale.

"If you know the total number of renters and you know the total number of homeowners, then you can increase the weight on the renters," he said. But he added the caveat that this could only go so far before it risked creating additional distortions by placing too much weight on small samples.