A number of commentators have suggested that the news on growth could not be as bad as the official data suggests. No worries, it couldn’t possibly be true that the government’s policies are failing. All will be revised upwards and all will be well. But the data tells another story.
Kevin Daly of Goldman Sachs, for example, has noted that the “on-the-day surprise to the preliminary release has averaged 0.5 per cent in the past two years”. These claims have drawn some attention from government supporters who are desperately in need of some better news. It should be said that Daly has been an overly optimistic forecaster, so if the data were to be revised upwards that would get him out of a really bad hole.
Data available on the ONS website — called its revisions triangle — for GDP growth rates does suggest that Daly’s claims, along with those of others such as David Smith of the Sunday Times, who has also been overly bullish and wrong on the data, are unlikely to be true. Gavyn Davies in the FT also joined in the chorus that things are likely to be better than they seem. Maybe not.
The data since 2007 is presented below. The first column is the month-one estimate, followed by the latest estimate available today and the difference.
A brief look at the facts reveals the following:
1) Since Q1 1993, the average revision from the first estimate to the latest estimate we are using today is +0.11 per cent, not +0.5 per cent, as Daly has claimed.
2) Data revisions before 1993 are unlikely to be much of a guide to what happens today, given technological advances and computerisation. The likelihood is that data revisions today will be smaller than in the past.
3) The average revision since Q1 2007 has actually been a reduction of 0.1 per cent.
4) The most striking data revision has been to the data for Q2 2008, which had an initial estimate of 0.2 per cent, which has now been revised to a reduction of 0.3 per cent. The consequence of this revision, of course, has been to move the starting point for the onset of recession that most commentators failed to spot.
5) The three quarters at the depths of the recession, from Q3 2008 to Q1 2009, which were initially estimated to show output drops of 0.5 per cent, 1.5 per cent and 1.9 per cent respectively, have had the biggest revisions downwards. They are now estimated as -0.9 per cent; -2.1 per cent and -2.2 per cent.
6) It is perfectly plausible to expect to see a growth number revised upwards during a boom, perhaps because factors that, at first, appear to be seasonal are actually a trend. In a slump, the reverse occurs and numbers are more likely to be revised downwards. We haven’t had negative quarters for a very long time, as under Labour every quarter from 1997-2008 Q1 had positive growth.
7) Many forecasters, including the MPC, seem to believe that past results will be revised upwards. This is unlikely. Interestingly, the MPC held a similar view in August 2008, when it failed to spot the recession. At the time, it forecast that data for the years 2006 and 2007 in particular would be revised up. In reality, it has been revised down. Below, I present the data that was current in August 2008 and the latest estimate and the differences, which in seven out of nine quarters, are down.
Based on the history of past revisions, there is a better than 50-50 chance that the data will be revised down, rather than up.