First draft of UK economic history
The preliminary estimate for a fall of 0.4% in the third quarter is clearly below expectations - and formally makes this the longest period of consecutive falls in national income since comparable records began.
Many economists have said they expect this recovery to be slower than usual. But, if right, these data suggest it has not yet even arrived.
The very weak state of bank lending - and low state of household and business confidence in the UK and many other economies - are still taking their toll.
News of further falls in manufacturing and industrial output earlier this month had suggested that the figures might be weaker than initially hoped. But most expected the service sector to show modest growth.
However, the Office for National Statistics (ONS) estimates that output in this sector - which accounts for by far the largest share of the economy - showed another slight fall in the past three months, of 0.2%.
I suspect that the further decline in the service industries in the past quarter is a large part of the reason why so many forecasters have been caught short.
But it is worth remembering that the ONS does not have complete data for that part of the economy at this early stage - far from it. It is quite possible that the number will be revised.
Historically, official output numbers have tended to be revised upwards, over time. But this has been less true in recent times. The estimate for the first quarter of this year was eventually revised - downwards - by 0.7%.
The bottom line is that we should take this as very much a first draft of UK economic history - but clearly a disappointing one.
Update, 11:34: There's been a rapid response to today's figures from Ben Broadbent, of Goldman Sachs, who has long argued that the ONS was understating the strength of the economy.
To say he thinks the figures should be treated with a grain of salt would be an understatement:
"Do today's data tell us anything about what is really going on in the economy? Probably not. In the past, the ONS's early GDP estimates have contained no statistically useful information about growth. In the 10 years to 2007Q4 (the latest quarter for which we have fully reconciled UK GDP data), the correlation between quarterly GDP growth according to the first release and quarterly GDP growth based on the most recent data is only 0.10. In a regression of the latest quarterly growth on the first quarterly estimate, the coefficient on the early official estimate is not significant. The correlation between the UK's latest data and the Euro-zone's first release is much higher, with a correlation coefficient of 0.34. Amazingly, if one wants to know in real time what is happening in the UK economy, it has been better to follow the Euro-zone's early GDP estimates than the UK's GDP estimates.
"The data also run contrary to pretty much every other indicator of UK growth. The Composite PMI - which has historically been a more reliable indicator of UK growth than the ONS's preliminary estimate - is consistent with growth of +0.7%qoq." "
Broadbent notes that it's not just the PMI data - car sales, retail sales, net exports, housing starts are all growing at a pretty good clip. And it's curious, to say the least, that employment seems to be stabilising even as output (officially) continues to fall.
For the non-statisticians amongst you, the shorthand way to express those correlation numbers would be that you'd not be much worse off tossing a coin to find out where the economy was headed. I look forward to hearing the ONS response.
Update, 17:30: I promised you a response: this just in from ONS.
"ONS produces the earliest estimate of GDP of any of the major economies, around 25 days after the reference quarter - on 23 October for July-September 2009. This provides policy makers with an early estimate of the real growth in economic activity, but this is inevitably revised as the underlying data sources mature. However, we are open about how much our preliminary estimate later gets revised - over the last five years, the average absolute revision (that is, without regard to the plus or minus sign) has been only 0.03 percentage points between the first estimate and the one a month later, and 0.08 percentage points between that estimate and the third estimate a month thereafter. We are confident that today's figure is our best, central, estimate at this point in time."
I suspect the Goldman boys would say the big revisions happen after the first three estimates referred to here, perhaps a year or two after the event, when the ONS reconciles the output data with more complete information on the amount of spending in the economy, and people's incomes. But I'll let you know what they say.