Data vs Common Sense: Fight!

Last week, I posted about the increasingly obvious problems the Office for National Statistics is having in measuring key aspects of the economy. In summary, the UK’s official statistics organisation has postponed the release of UK trade data following the discovery of an error in trade statistics relating to goods and services going all the way back to 2023. The ONS now plans to release the corrected data series on March 28th, along with the full balance of payments and quarterly national accounts data on the same date.

Unfortunately, this isn’t an isolated incident. In recent months, the ONS has admitted problems with several other important datasets, including the labour force survey, earnings data, and its living costs and food surveys—all critical components used to calculate official inflation and GDP figures. Even migration data, though less economically impactful, has faced repeated and politically sensitive revisions.

These repeated inaccuracies reinforce my earlier argument that the ONS is struggling significantly to measure the real state of the UK economy accurately. Issues with trade data appear to be just the tip of the iceberg. The Office for Statistics Regulation itself, effectively the watchdog for national statistics, recognised this late last year when it stated clearly that 14 sets of ONS data could no longer be considered “official statistics.”

Unfortunately, my hunch is that the ONS’s measurement problems run deeper and are more troubling than those now identified in last week’s announcement.

Making a claim like this is hard to prove, not least because I do not have access to a giant bureaucracy, the underlying surveys, or an alternative data series. However, I do have some ability to apply the laws of common sense to this issue, and that is what I have attempted to do here. This time, it is related to the UK’s manufacturing sector.

At the end of last week, the ONS announced January's GDP output data, which showed the economy had contracted by 0.1%. This announcement, not surprisingly, got the usual crowd of gloomy pessimists writing about worrying stagnation in the UK economy and a lack of growth. Maybe, also unsurprisingly, they seemed to forget that according to the ONS, the economy had grown by 0.4% in the previous month. Given that the UK’s long-term monthly average growth rate is about 0.2%, this is what we should have expected.

According to the ONS, although services rose by 0.1% in January, this was more than offset by a slight contraction in construction but a much bigger fall in production output, which fell 0.9%. (This series includes manufacturing output, alongside mining and quarrying, which is much smaller and utility output, which is and should be pretty stable) Nevertheless, the data got me thinking again, and I looked back at how the economy has performed since the start of 2020. What interested me in particular is how two key components of the economy (manufacturing and services) have performed since the pandemic began in early 2020.

This is what the official ONS data shows.

According to the official data, after a collapse during the initial lockdown in early 2020, manufacturing output (yellow line) recovered very rapidly and outstripped its previous peak but then began a rapid decline, such that from Q4 2020 to Q3 2024, it is down an astonishing 12.2% (Q4 2024 data is not out yet). On the other hand, services have grown over this same period, as has the wider economy.

But is this slump in UK manufacturing output believable? I am not so sure. Here’s why.

I had a look at manufacturing employment (easy to measure and very reliable data) over the same period and compared the two data series. Here, they are shown in a combined chart.

This chart shows that over the four years since Q4 2020, manufacturing output has fallen by just over 12%, but astonishingly, manufacturing employment has increased by just over 1%.

In other words, UK manufacturing businesses have employed over 1% more people over the last four years to produce 12% less output. Quite apart from the clear implication that UK manufacturing businesses are run by fools, if this were true, there would have been a coincident collapse in manufacturing profit margins and a huge increase in manufacturing business failures over the same period, neither of which has happened. (In the twelve months to February 2024, manufacturing insolvencies accounted for a stable 7% of total UK insolvencies – construction is up at 17% and hospitality is at 15% - whilst manufacturing contributes 9% of GDP, and goods account for 45% of total exports)

If, like me, you don’t believe that UK manufacturing businesses are uniquely badly run, and you trust the employment data (which I do) and accept that there is no evidence of widespread UK manufacturing business failures over the last four years (which there isn’t), then you have to conclude that the output data is wrong. Very wrong.

Given that a significant proportion of manufacturing output is exported, maybe, and I am guessing here, there is a link between the ONS’s failure to accurately measure the UK’s trade performance correctly and its apparent failure to measure manufacturing output.

Conclusion

In my view, the ONS has not only failed to accurately record the UK’s trade performance in recent years but also has a major problem measuring the output of the UK’s manufacturing sector.

This may seem like a bold and shocking conclusion, and it is one that I cannot prove. However, you don’t need to be a Nobel prize-winning economist to know that the ONS manufacturing output data doesn’t stack up. In the real world, businesses don’t employ more people and pay them more money to produce less output—quite the opposite. The world is a competitive place, and any business that believes it can survive and prosper by doing that would be quickly found out. Another explanation is that the UK has some major measurement problems at its statistics authority, which require a root and branch review of the methodologies adopted by the ONS. In the meantime, I will remain sceptical about official ONS output and trade data and hope that the full range of information policymakers rely on to inform their decisions gives them a better steer about what is really going on.

If I could make one recommendation right away, I’d urge the ONS to abandon its monthly obsession with measuring GDP based on output. Instead, let’s shift focus to measuring the economy by expenditure. Simple, straightforward—and, crucially, more likely to be accurate.