Statisticians grapple with inflation impact of Covid-19

Collecting reliable inflation figures during lockdown is not straightforward

Statistical authorities are scrambling to adapt their processes for gathering inflation data, as standard approaches have been severely impacted by lockdown measures in response to the Covid-19 pandemic.

The standard approach to gathering inflation data has not changed much over the years. Statisticians design a basket of representative goods and services and then go out and gather their prices. The advent of e-commerce has shifted some of this work online but, before lockdown, a large number of prices were still gathered from physical shops.

The UK’s Office for National Statistics released March’s figure for consumer price inflation on April 22, showing a slight drop from 1.7% in February to 1.5% in March.

The ONS says March’s figures were only marginally affected by the UK’s lockdown measures, which were applied across the whole population on March 23. Some elements of the figures were based on fewer observations than usual, it adds.

Statisticians are still figuring out how to deal with April’s figures, which will fall completely within the lockdown period. A spokesperson says: “This is a rapidly changing situation and we intend to publish a methods paper in the coming weeks to outline the decisions we are making.”

Around 45% of CPIH, the CPI measure plus owner-occupied housing costs, is gathered from physical locations, the ONS estimates. The ONS is working with other statistical authorities and says it will communicate any changes to methods or data sources to the Bank of England before implementation.

The European Union’s Eurostat is facing similar problems with its headline harmonised index of consumer prices (HICP). “In a number of countries, field price collection has become impossible either because the statistical office stopped this or because shops are closed,” a spokesperson says.

Statistical authorities in Europe are turning to alternative sources of data to try and bridge the gap. High-frequency data that does not require an individual to visit a physical shop can provide some of the same information – examples include scanner data and prices “scraped” from web pages.

One useful source of data is shops such as supermarkets that have similar prices online and offline. Pricestats, a firm founded by economists Alberto Cavallo and Roberto Rigobon to gather high-frequency data, is offering some item-level data free to statistical authorities during the Covid-19 crisis.

The firm is also publishing data on stock levels, to give a quick picture of when supermarkets sell out of key goods due to unusual patterns of demand.

This is a rapidly changing situation and we intend to publish a methods paper in the coming weeks to outline the decisions we are making

ONS spokesperson

The ONS is not currently using high-frequency data in its headline inflation figures but has not ruled out the option for the future. It is publishing the data in parallel, however, and finds prices of “high-demand products (HDPs)” have moved up sharply. Prices for these items were up 1.5% week on week in the period from March 30 to April 5.

Andy King, head of CPI production and user engagement at the ONS, says the web-scraping is part of a broader move by the authority to adopt alternative data sources by 2023.

But he cautions that there are drawbacks. “The prices are collected in a rigorous approach, however the HDPs do not necessary match the items collected (there are additional HDPs which are not included in the CPIH/CPI basket) and, rather than daily prices, for the consumer price indices we rely on prices collected at or around index day each month.”

It is hard to guess whether inflation will rise or fall in response to the virus. The rapid rise in prices of some goods has been matched by dramatic falls in other areas. The recent plunge in the oil price – to below zero for the WTI benchmark – is likely to have knock-on effects to a wide range of goods.

The coronavirus outbreak caused a rapid shift in the sorts of goods and services consumers are buying and where they are buying from. As well as shops shutting their doors, some goods and services cannot be produced, either due to supply disruptions or virus-containment measures. For example, holidays have been cancelled and the number of passenger flights has dropped sharply.

Where web scraping, scanner data, or telephone enquiries fail to turn up reliable data, authorities may turn to imputed prices. As Eurostat explains in a guidance document, this involves multiplying the previously obtained price “with an appropriate price change”, for example a change observed in similar goods or at a higher level of aggregation.

Eurostat admits that for gauging the price of goods and services where demand has dropped to zero, such as hairdressing, “there is no optimal method”. It recommends authorities take imputation decisions “on a case-by-case basis”.

“Eurostat and the statistical offices in the EU are committed to continue the dissemination of HICP data to the best of their abilities,” the spokesperson says.