GDP "special events" get pushed into the background

Reading between the lines, last Friday’s release of GDP data for the second quarter of 2011 showed some interesting revisions from the preliminary estimate published in July. Not in the figures – GDP growth remained the same as in the first estimate, a discouraging 0.2 per cent – but in the way they were presented.

The UK Statistics Authority, which has direct responsibility for the Office for National Statistics, is keen on statistical commentary but felt that in July it didn’t quite strike the right note. Read by suspicious minds, and there are plenty out there. the presentation of the preliminary data could suggest that the ONS was acting as an apologist for the Government.

It cited a multitude of special circumstances – the Royal Wedding, the Japanese earthquake and tsunami, the sale of Olympic tickets, and warm weather  – that may have affected economic growth in the quarter. Without these, it estimated, GDP growth might have been half a percentage point higher,

Was this special pleading, or simply a desire to explain as clearly as possible what the underlying issues were? Whichever it was, the result was that the media focussed more on the exceptional circumstances than they did on the data.

The revised estimates do not resile from the explanations offered previously. But they tone them down and chunks of text that appeared in the first release have been deleted from the second. The special events are listed, as before, but the estimates of their likely impact – 0.4 per cent in the services sector and 0.1 per cent in the production sector – are deleted. The second release simply says: “These special events may have had a net overall impact on GDP, as noted in the Q2 preliminary release”. A link to the preliminary release is provided. 

There is also, accessible through the Background Notes, a new analysis of how the effect of special events is measured. This involves making a short-term forecast (sometimes called a “nowcast”) and comparing it with observed data. The difference is the sum of various errors that may affect the accuracy of the nowcast, plus the effects of the special events. In this case, the nowcast produced an estimate of 0.7 per cent, the observed data 0.2 per cent. The difference was attributed to the special events.

In addition, and if you try quite hard to find it, there is an Economic Review article that gives the economic background to the second release, but does not mention any special factors at all.

So all the same data and explanations are provided, but are not so prominent and take some digging out. We can assume that this will be the pattern for the immediate future.

But in November an advisory committee will review the arrangement for handling the effect of one-off events on economic time series, and next spring the ONS will publish an overview of the issue. So this may not be the final word.