Why you should take economic statistics with a bucket of salt

TODAY is GDP day, when the Office for National Statistics releases its first, early and ultra-preliminary estimate of the growth of the UK economy. As ever, the media, the markets and the political classes will massively over-react to whatever gross domestic product (GDP) number is released, good or bad. I would urge everybody not to pay too much attention.

It is clear that the UK is performing poorly, with real wages falling, retail sales stagnating and the budget deficit increasing again. But it is hard to accurately measure the size of the overall economy and GDP figures are not only spuriously accurate but also subject to shockingly large revisions.

Vital research by Sam Williamson of the University of Illinois, published on econobrowser.com, shows that the extent of these revisions can be extraordinarily significant.

Until about 18 months ago, the Office for National Statistics (ONS) thought UK GDP (in real terms, measured in 2009 prices) had grown by 2.43 per cent per year over the past 62 years. Per person, GDP had grown by 2.06 per cent a year, from £6,428 in 1948 to £22,410 in 2010. Then came the Blue Book 2011 and 2012, the key annual documents about the economy, and everything changed, thanks seemingly to a decision to use a different measure of inflation.

The ONS now believes the economy has been growing by 2.68 per cent a year over the past 62 years – much faster. GDP per person is now deemed to have grown by 2.31 per cent a year, from £5,559 in 1948 to £22,920 in 2010. Our statistical masters have decreed that the economy was actually 14 per cent smaller than previously thought in 1948 – astonishingly, figures first out more than six decades ago are still being revised, which suggests that today’s data will probably look very different by 2075.

There are numerous other almost unbelievable changes. The biggest revisions are to our performance in the 1950s (up almost 0.4 per cent a year) but every decade has been improved, at the stroke of a statistician’s pen. Thousands of economic studies, including hugely important work, have become worthless – assuming, of course that the new data is more accurate than the old. Either way, it’s a disaster for our understanding of economic history.

The new numbers suggest the economy grew by 3.03 per cent a year in the 1950s, peaking at 3.18 per cent a year in the 1960s before slowing to 2.07 per cent in the 1970s, accelerating back to 3.09 per cent in the 1980s, before expanding by 2.77 per cent in the 1990s and by a miserable 1.77 per cent in the 2000s, the worst of the post-war decades. These figures would be fascinating if they hadn’t emerged from such a breathtaking re-rewriting of history.

Over the past decade, I have witnessed and written about other such massive revisions. There were huge changes when new European accounts were introduced and when supposed quality improvements in the public sector (such as better A-Level grades) started to be treated as increased output. Then there are all the short-term changes, the endless revisions going back a few years that sometimes turn what we thought was a terrible, negative quarter into a positive one. The inevitable problem is that the ONS relies heavily on surveys of the activity of existing firms, thus missing some new ones.

The fact that statisticians can simply rewrite the past is deeply depressing; what seems like objective reality is in fact little more than a social construct (as an over-caffeinated French philosopher might put it). The economy is doing poorly – but don’t take today’s figures too seriously.

Follow me on Twitter: @allisterheath


In Other News

April 17, 2014, 1:02pm
April 17, 2014, 1:41am
April 16, 2014, 7:50am