Receive free Coronavirus updates
We’ll send you a myFT Daily Digest email rounding up the latest Coronavirus news every morning.
National statistics are sometimes like Hieronymus Bosch paintings: the longer you stare, the more uncomfortable you feel. Or maybe they’re more like advertising board photos: clear from a distance, pixelated up close.
Or maybe, sometimes, they’re just wrong.
How do we categorise Friday’s significant Blue Book 2023 data revision from the Office for National Statistics, which basically destroyed the entire narrative of post-lockdown UK growth? From the mainFT report:
Official statistics on Friday added almost 2 per cent to the size of the UK economy, in a surprise move that showed the country recovered much faster from the pandemic than previously reported . ..
The changes reveal that, by the end of 2021, the country’s economy was 0.6 per cent larger than pre-pandemic levels — instead of 1.2 per cent smaller, as previously estimated.
Here’s the damage (or, we suppose, undamage):
Significantly, this revision moved the UK from being the worst-performing G7 economy since the pandemic kicked off to being middle-of-the-pack.
MainFT’s Chris Giles made a good chart showing this, which we’ve remixed slightly (think Luniz I Got 5 On It vs the Luniz ft. Bay Area All Stars I Got Five On It):
Naturally, this economic correction of the desirable kind has been received with serenity and grace.
Alphaville’s instinct is to get into the weeds, rather than the trenches, but if your response to this improved data was/is:
— “oh no the UK economy has been doing better than I thought, maybe Brexit’s reputation is going to improve and people will dislike the Tories less”, or;
— “ha ha ha yes take THAT stupid mainFT, Remoaners, my estranged father, Alistair Campbell, and everyone else who ever criticised me”
. . . we’d encourage you to briefly go outside and touch grass, and then to think about the following:
— should newspapers not report the headline findings of the independent national statistics agency?
— should the national statistics agency not try to constantly get better at being the national statistics agency?
— is good news not good news when it goes against your priors?
— are GDP figures (especially ones that are this volatile) at all relevant to real life?
There’s obviously an important point to be made about how the media reports on national statistics (and, perhaps more importantly, how those national statistics are used to shape economic decision-making). But that has to be tested against the perfectly possible counterfactual scenario in which the numbers were proved to have already been broadly right.
Still — the ONS’s communications, in Alphaville’s view, could be improved. This does not seem to be, we should add, a pure screw-up, like the one we had back in January.
The ONS has given a few explanations for the revision:
— it has struggled with measuring intermediate consumption (ie the costs facing businesses relative to their turnover) amid rapid changes in inflation
— it has struggled to measure company inventories since the pandemic started
— it made some tweaks to its process and methodology
On the former, it says the input data has improved through use of the Annual Purchases Survey and Annual Business Survey (more detailed, but slower, than other surveys):
These show that in 2021 many businesses in the manufacturing sector actually incurred even more costs than we initially assumed. The picture in the services sector is more complex, though, with these new data showing the price of wholesale and retail services was lower than previously estimated, while the amount of health services provided was higher than previously thought.
This feels reasonable: better data can take a long time to emerge.
But here’s the Newport gang’s thin explanation for the inventory measurement errors:
In 2020, average volume GDP is now estimated to have fallen by 10.4%, revised up by 0.6 percentage points (Figure 3). This upward revision reflects both updated data and methods changes. Measurement of inventories is challenging over this time period; the changes in the inventories component is now estimated to have increased by £2.5bn in 2020 (previously this was a £11.4bn fall).
This tells us nothing, except that the pandemic was a tough time to do certain things. A separate blog post by Craig McLaren, the ONS’s head of national accounts, is hardly more informative:
Separately, we have also updated our estimates for the amount of stock owned by UK companies, known as ‘inventories’ in 2020. This showed that the total stocks held by UK manufacturing and mining companies fell less than we first thought, reducing the scale of the fall in total GDP in 2020.
Producing estimates of GDP in unprecedented times, when much of these data we need to produce complete national accounts are not available for a couple of years after the event is clearly a challenge. We are among the first of our international colleagues to update our initial estimates with more detailed data.
We think they should be clearer about this kind of error. We assume it isn’t a Eurostat-esque, “Germany can’t use a computer properly so we just made up a number for them”-type issue, but . . . what then? Why not just say?
Either way, we can’t wait for Blue Book 2024, which might change the narrative yet again. National statistics, such fun.