Goldman Sachs has put out a note on the GDP figures headed "Unbelievable. Literally", which points out the that the early estimates of GDP in the UK have a correlation of just 0.1 with the final data. Bizarrely, there is a closer correlation between first estimates of Eurozone GDP and the final UK numbers than the Office for National Statistics' first stab.
Looking at the data, one big shock came from services and the ONS's guesstimate of a 0.2% drop in the sector, which has a weight of 759 in 1,000 in the output-based GDP figures. The ONS thinks the index of services averaged 104.5 in the third quarter and has published figures showing monthly estimates of 104.9 and 104.6 for July and August.
That means a hefty 0.6% drop between August and September, at a time when the surveys were suggesting a return to strong growth for the sector. So something odd is happening, particularly as the ONS has little service-sector information of its own for September. If Q3 service sector output turned out to be merely flat, that would halve (subject to rounding) the estimated GDP fall in the quarter. If it bounced back only a little, the fall would be eliminated. A lot of ink has been spilt over some odd looking first estimates.