At 9.30 am on Wednesday all eyes in the City will be pinned to their screens as the ONS releases its latest estimate of the GDP in the third quarter of 2009.

The preliminary figures, which came out a month ago, caused consternation. They showed the economy was still in recession in the third quarter, shrinking by 0.4 per cent. The figures signalled the longest recession since records began in 1955, and confounded economists, the City, and the Bank of England, which had all convinced themselves that the economy had resumed growth in the third quarter.
 
Many were disbelieving. Surveys, including the influential Purchasing Managers Indices for services and manufacturing, had given clear signals of modest growth. Chris Williamson, the chief economist at Markit, which produces the PMI surveys, told The Times: “We think the numbers are wrong. A whole host of indicators other than ours show that the economy grew in the third quarter, powered by a stronger services sector. There is a risk that these figures could lead to disastrous policy mistakes.”
 
The Bank of England’s Monetary Policy Committee admitted puzzlement at its November meeting.  “Initial estimates of GDP were prone to revision as more complete data became available, providing grounds for interpreting the release cautiously” the minutes of the meeting say. “In addition, some business surveys for the third quarter pointed to a somewhat stronger number than the initial ONS estimate.”
 
Two analysts at Goldman Sachs, Kevin Daly and Adrian Paul, went even further. In a paper published as the preliminary estimates appeared, they claimed:  “There has been no statistically useful information about GDP growth in the ONS’s early GDP estimates ... amazingly, if one wants to know what is happening in the UK economy, it has been better to follow the Euro-zone’s GDP estimates than the UK’s GDP estimates.” The correlation between the preliminary estimates and the final figures was a miserable 10 per cent, they claimed. (Not everybody agrees with this, as will become apparent.)
 
So a lot hangs on Wednesday’s figures. If the ONS makes a large revision to its preliminary estimate, its reputation will suffer a blow, and the economists will crow that they know more about the economy than the statisticians. Seldom has a single figure carried such a heavy weight of expectation.
 
GDP figures are often revised. The ONS produces its first estimate early – barely three weeks after the end of the quarter to which it refers – and it is not invariably right. Revisions of 0.1 percentage points between the first estimate and the second, a month later, are not unusual. As time passes and more data are collected the revisions can become bigger.
 
Wednesday’s figures are very unlikely to show revisions great enough to change last month’s judgement that the economy was still in recession. But in time, a revision that large is certainly possible. The first quarter of 2009 has already been revised sharply, from the preliminary figure of -1.9 per cent issued in April to -2.4 per cent in figures issued at the end of June and -2.5 per cent in September’s Quarterly National Accounts.  The second quarter has been revised in the opposite direction, from  -0.8 per cent to -0.6 per cent.
 
The ONS argues that revisions this large are exceptional. The average over the past five years between the preliminary figure and the next month’s is 0.03 percentage points, and between that and the quarterly figures 0.08 percentage points. Revisions between first publication and estimates three years later, when all the data are in, are larger: an average of 0.21 percentage points.
 
Changes in the figures are hardly surprising. The earliest estimates are based on output; the second set includes expenditure and income measures; and the third includes the national accounts reconciliation. Later estimates still include tax data, and may be changed retrospectively if the methodology is changed or improved.
 
So are the preliminary estimates useless, as Goldman Sachs suggested? A counter argument published by Erik Britton and Danny Gabay of the Fathom consulting group reaches the opposite conclusion. They say the Goldman Sachs team is wrong to search for correlations between the first estimates (M1) and the final published figures of GDP because the final figures include the effect of retrospective methodological changes and M1 does not.
 
It is pointless, they say, to compare a first estimate made at the time with a final figure that incorporates changes retrospectively introduced. When they compare like with like, they find a 60 per cent correlation between M1 and the final figure, not the 10 per cent found by Goldman Sachs.
 
Fathom thus backs the ONS and confirms the value of its preliminary GDP estimates. That doesn’t rule out a revision in the GDP data on Wednesday – and as a rough rule of thumb, when the economy is in decline, the revisions are downwards (as in the first quarter) and when it is recovering, they are up.
 
So we could possibly see the -0.4 per cent figure shaded to -0.3 per cent. That’s roughly what the MPC is betting on. Any more than that would be a surprise. Revisions may not attract much attention in normal times, but when everybody is betting on a recovery from recession, they loom very large indeed.