Using economic statistics to explain recent UK inflation
By Josh Martin
Economic data are a vital foundation for the Bank of England’s analysis, forecasting and research. I used to work at the Office for National Statistics (ONS) as a producer of economic statistics, but since moving across the statistical aisle and becoming a user, my perspectives have shifted somewhat. One key difference is that I now rely on a wider set of statistics, often bringing together data on different economic concepts to get a more complete picture.
In an ideal world, these datasets would be consistent, cover extended time periods, and be available in one place. In practice, we often have to combine data sources and vintages (the version of a dataset available at a certain point in time) and make further adjustments before the data are ready to use.
In a new ESCoE Discussion Paper with Jonathan Haskel and Lennart Brandt, we use a range of economic statistics to shed light on the recent UK inflation experience. This links with an ESCoE talk on UK inflation given by Jonathan Haskel in July 2024.
There have been various suggested stories for the increase in inflation in 2021-22 and subsequent fall in 2023-24. These include energy prices, supply chain disruptions and a tight labour market, among others. One way to assess the various explanations is to estimate a model that accounts for different factors at the same time, and then test which ones really matter in the data.
One of these models was developed by Olivier Blanchard and Ben Bernanke in their influential paper on the US inflation experience. We estimated this model with UK data; since the UK economy differs from the US, the estimated equations and coefficients of the model might differ too.
Of course, estimating an economic model requires data. In this case, we needed data for only a handful of variables, but on a quarterly basis and over a fairly long period of time – since 1990. The ONS publishes a wide range of economic data and was the main source for most of our variables. However, some of the currently published data series have a shorter time series than we needed, or required adjustments before we could use them.
Starting with inflation data, the Consumer Price Index (CPI) is available since 1988, which was long enough for our purposes. We used the headline CPI index, as well as the series for energy products (both household electricity and gas; and fuels for vehicles) and for food and non-alcoholic beverages. However, official CPI statistics are only available non-seasonally adjusted. Since our model is estimated using quarter-on-quarter changes, we had to seasonally adjust the data. Recent research has also taken this approach to CPI.
For data on wages, we used the Average Weekly Earnings (AWE) series for private sector regular pay. Since this series only begins in 2000, we needed to look for an alternative. ONS publish a historical AWE series, but not for private sector regular pay (only for total pay, including bonuses, which makes the data noisier). So, we used a historical Bank of England series instead. More recently during the pandemic, wage growth figures were distorted by compositional affects and the furlough scheme. To account for this, we used a Bank of England estimate of “underlying” wage growth.
One key variable is the vacancies-to-unemployment ratio (V/U) – a measure of the tightness of the labour market. The official data on vacancies (from the Vacancy Survey) starts in 2000, so we needed to find a historical source. The Bank of England Millenium of Macroeconomic Data contains a historic series of vacancies, based on Job Centre postings, and a factor linking it to the current vacancies series. We used this linked vacancy series alongside the long-running unemployment figures from the Labour Force Survey.
For productivity, we focused on output per hour worked in the private (or market) sector. Currently published series start in 1997, while some historic series go back to 1994 on a quarterly basis. To take it back further, we used the series from the ONS growth accounting estimates, which contains annual estimates back to 1970.
To measure supply chain disruption, we used the Global Supply Chain Pressure Index (GSCPI) published by the New York Fed. This isn’t a UK-specific measure, which is a slight limitation in this context. We previously used the Google trends series for “shortages” which is UK-specific, but we found that it was less robust.
Finally, we needed data on inflation expectations: both one-year, and long-run (5-10 year) expectations. There are many different types of inflation expectations from different economic actors – households, financial markets, experts, and firms. To capture a consistent picture, we used a series from the Millennium dataset, which extends a composite measure designed to highlight the common signal across these different sources.
The Bernanke-Blanchard model combines inflation, wage growth and inflation expectations, allowing for lags and feedback dynamics. The estimated parameters of the model are similar to those for the US. However, the UK appears to have stickier wage and price inflation and more persistent effects of food price shocks. The initial increase in inflation in 2021 can be explained by supply chain disruptions and energy price shocks. In 2022 and 2023 there was also an important role for food price shocks and labour market tightness. It was only by using the data together that we could paint this more nuanced picture. More detail on the model and findings can be found within the paper.
Teams at several other central banks also applied the Bernanke-Blanchard model to their economies, as summarised in Bernanke and Blanchard (2024). The UK contribution to that paper is documented in our ESCoE Discussion Paper.
This work was made possible by the availability of broadly consistent data on a range of economic variables across many advanced economies.
Having comparable, high-quality statistics allows researchers and policymakers to analyse complex economic relationships, test theories, and draw meaningful conclusions. It is a clear example of how reliable economic data underpins important policy analysis, helping central banks and governments to make better-informed decisions.
This blog explores how economic statistics and modelling help explain recent UK inflation. Using key data on prices, wages, productivity, and labour market tightness, the work shows how combining multiple datasets reveals the main drivers of 2021–2024 inflation, including energy and food price shocks and supply chain disruptions. The analysis highlights how reliable economic statistics underpin informed policy decisions and a deeper understanding of UK inflation.
ESCoE blogs are published to further debate. Any views expressed are solely those of the author(s) and so cannot be taken to represent those of ESCoE, its partner institutions or the Office for National Statistics.