Receive free Economic statistics updates
We’ll send you a myFT Daily Digest email rounding up the latest Economic statistics news every morning.
The writer is the UK’s National Statistician
At a time like this, it’s hard not to reach for the words attributed to John Maynard Keynes on decision making with uncertain evidence: “When the facts change, I change my mind.” Yet in upgrading our official estimates for economic growth in the pandemic years of 2020 and 2021 — and in effect rewriting elements of the UK’s economic narrative in the process — that is the position the Office for National Statistics has found itself in.
Measuring the entire value of a national economy worth about £2.5tn is not a simple task. It depends on gathering huge volumes of data covering every sector of activity — from households, retailers, banks and corporations as well as across government and the public sector. Some of this takes the form of survey returns from organisations, the rest is “hard” data such as tax information.
The UK is one of the fastest countries in the world to do this, but the component parts of gross domestic product still take time to gather fully. Naturally, the users of these statistics can’t wait years to find out how the economy is doing. So we produce preliminary estimates based on the data already to hand. This comes largely in the form of business survey returns, which together amount to only about two-thirds of the eventual, complete information.
In normal times this has provided an accurate early picture of the economy. Revisions in the years immediately before the pandemic tended to be small. But the disruption of the Covid-19 years was clearly not normal. It is hardly surprising, then, the measurement challenges presented by the deepest-ever contraction in our economic history, and the subsequent recovery, have resulted in larger revisions.
Early survey responses — perhaps influenced by the unprecedented nature of events — turned out to be less positive than the scenario the later and more comprehensive data revealed.
It’s understandable that some people should be surprised by the scale of these changes. Not unreasonably, they expect statisticians like us to “get it right” first time. So while I remain confident that our initial estimates for the Covid years were the best possible given the data available — and that our recent estimates are unlikely to be subject to similar uncertainty — I want to doubly underpin faith in our work.
First, we have invited the official watchdog, the Office for Statistics Regulation, to review our analysis of these revisions, how we deal with sharp discontinuities in economic performance and, equally, how they are communicated. This will show if there is anything we could do differently.
Second, we will prioritise the drive to get key data earlier in the production cycle to improve our preliminary estimates. When revisions are necessary they can then be made sooner. This will have implications across government, and both public and private sectors, requiring co-operation.
We have made good progress in getting “fast” data into official statistics over the past few years, including latest retail prices scraped from the internet and real-time payroll data from HMRC to inform our view of the jobs market. I want this work to continue until the advantages of big digital data have been introduced fully.
When data is subject to uncertainty and “confidence intervals” are wide, then we say so. Yet the economy is so often portrayed as either flourishing or slumping when the reality is normally somewhere between those two poles and the true position more nuanced. So we must redouble our efforts to better inform the public discourse.
The UK’s official economic statistics are rightly seen as among the world’s best. While we will not hesitate in future to “change our minds” again when the evidence demands, the continuous improvement of our statistical system is a task to which I and my colleagues are wholly committed.
