How GDP Is Compiled Really Matters
Updated: 8:11am UK, Thursday 25 October 2012
By Ed Conway, Economics Editor
I've covered economics for a decade or so, but I confess that until very recently I didn't really know what GDP really is.
I mean, like most of you I knew it was the broadest and most widely-used measure of our economy's health - that it determines whether we're officially in recession or not (two or more quarters of shrinking GDP equals a recession).
I knew it was the sum of everything spent, earned or made in Britain.
What I didn't know was how it's actually put together.
I guess I vaguely assumed - and I don't think I'm entirely alone - that the Office for National Statistics had some kind of electronic hotline into British business, some privileged access to their numbers, which in turn became the Gross Domestic Product number.
Turns out I was monumentally wrong.
For it transpires that GDP - that big number we're all so focused on, the figure that tells us whether we're in a recession or booming, that can end a political career and swing an election - is actually a big, big survey.
I know this because earlier this month I spent some time in the ONS headquarters in Newport with the team who put together this most significant of all numbers.
For the first time, they allowed cameras into their offices to show how GDP really comes into being - and the genesis might well surprise you.
At this point it might be worth explaining why this matters so much: there is arguably no other number out there that can swing the financial markets quite so much, that can influence Britain's feelgood factor, that dominates the headlines and strikes fear into politicians.
And yet there are many people who question whether we can really rely on the numbers.
Some economists argue that the GDP figures in recent months have painted a far more negative picture of the UK economy than is actually the case.
Some argue that Britain never really experienced a double-dip recession - but that this reality will only ever be confirmed many years into the future when the ONS revises those initial estimates.
So how GDP is put together really matters. And it all starts with the pounds in your pockets.
For the first estimate of GDP - the one today - is created from data collected in surveys of tens of thousands of surveys from businesses around the country - whether they're manufacturers, construction firms, retailers or others.
Each month a large sample of them is asked by the ONS to tell them their turnover (how much money is going through the till), along with a few other industry-specific questions which form part of the retail sales, manufacturing output and other releases.
The turnover number is what matters from the perspective of GDP. They fill the relevant questionnaire in and post it to the ONS (they can also submit the data through an automated telephone system).
When those envelopes arrive there the questionnaires are scanned and the numbers go into the ONS' systems.
The problem is that by the time that first estimate needs to be produced, the ONS only has 44% of the relevant data (the rest arrives in dribs and drabs over the following months, hence the revisions). In particular, the ONS only has early responses for the final month of the quarter.
So there are some pretty big gaps to be filled, and the ONS has to make some estimates about what the other data will eventually say when it comes in.
It relies for this on computer models, backed up by assumptions and calculations from the ONS staff themselves. After they make these calls they meet and discuss them in so-called "balancing meetings": the statisticians ask each other whether the data are reliable and their assumptions have foundation.
During this entire period, those GDP assumptions and the ultimate figure are kept locked up (quite literally - there are safes into which they are put) such that only a dozen or so statisticians actually know the number before it comes out.
So far as anyone knows, there has never been a leak of a number as sensitive as this from the ONS. But 24 hours before the figures are published, selected ministers and officials also get a look.
The figures are revised again a month after that initial release, and then again a month later. During that period, more information has come in from quarterly surveys which measure families' and businesses' incomes, and other spending data.
As I said, GDP can be measured in terms of what we spend, what we earn and what we make - they should all add up to the same number, since what one person buys another person sells. And the extra data furnishes that initial estimate and, occasionally, contradicts it.
The ONS maintains that its record of revisions is acceptable by international standards. It points out that its surveys have far more respondents than those put together by independent competitors.
But some, most notably Kevin Daly of Goldman Sachs, argue that it has a tendency to revise the more distant history so substantially that often periods we thought at the time were slumps were actually booms.
A case in point is the early 1990s: at the time, the ONS said the UK was suffering a double-dip recession.
But by the end of the millennium it had revised its assessment: far from slumping, the UK was actually bouncing back forcefully at that point. When Norman Lamont referred to "green shoots", it turns out he was absolutely right.
Today, the GDP figures have been telling an altogether different story to the unemployment figures, which seem to suggest there never was a double-dip. Based on precedent, we are unlikely to know the definitive story for years to come.
Which implies that the ONS, and the way it puts together this most important of all numbers, will remain in the spotlight for the foreseeable future.