Sixty-seven ways to guess GDP

The private value of predicting official statistics before their release is large but its social utility is zero, which is why procuring it is at once the best paid and most futile form of economic research.

At 8am last Friday, the Bloomberg financial information service carried 67 different predictions of US gross domestic product growth in the third quarter. The median was 3.6 per cent. At 8.30am, the Bureau of Economic Analysis issued its official estimate of 3.8 per cent. This modest difference was well received by the market: the Dow Jones index rose immediately by 50 points and was up 170 points – almost 2 per cent – on the day.

As the figure was announced there were cheers in some dealing rooms, commiserations in others. Wall Street economists who get these numbers right earn large bonuses: those who miss the mark soon miss their salary cheques as well.

Although these 67 figures are described as estimates of GDP growth, none of the 67 houses had really made an independent assessment of trends in American national income. The figures compiled by the BEA are based on extensive data collection using the legal powers, moral authority and financial resources of the US government under assurances of confidentiality. Such an exercise could not easily be replicated by any research firm or investment bank and is not.

Moreover, if anyone did try to measure GDP independently of the federal government, it is likely that they would come up with a substantially different answer. The official statistics are based on samples, cover only a fraction of economic activity and are open to considerable revision. The BEA’s own assessment is that this figure will be revised by up to 1-2 per cent on two-thirds of occasions. It is therefore more likely than not that the number – itself still an estimate – will be outside the range of between 3 per cent and 4 per cent which included almost all analysts’ judgments and the BEA’s own provisional figure. If an infinitely knowledgeable analyst actually established the correct answer, his prediction would probably be off the chart: not just in this quarter, but every quarter. He would soon lose his job.

How is it that all estimates are so close together when the underlying uncertainties are so large? The 67 teams of analysts are not trying to guess what actually happened to American national income in the third quarter. The traders who use their figures do not want an estimate of what is really going on: they want an estimate of what the BEA will announce. And so the economists who service them are trying to guess the number in the press release. Their main source is other statistics that the BEA has already issued. That is why their judgments, although not necessarily right about the economy, are close to the actual BEA figure. Their analysis is also influenced by the numbers posted by the other 66 analysts. It is dangerous to be right, but safe to be conventional.

The lazy equity analyst focuses on trying to anticipate quarterly earnings announcements, forming a symbiotic relationship with company finance officers and investor relations people who are trying to massage market expectations. A better observer can sometimes add value by penetrating corporate public relations and obtaining his or her own information.

But the BEA is trying to find the truth and is better placed to do so than any Wall Street economist. The bureau is not concerned to please the market and, unlike companies, is serious about confidentiality. Financial economists might as well be lazy since they have little to add to the work of the bureau. Correctly predicting the official estimate 30 minutes before its release may be profitable but contributes nothing to our understanding of the economy. The private value of such information is large but its social utility is zero, which is why procuring it is at once the best-paid and most futile form of economic research.

Keynes likened professional investment to a beauty contest, in which “it is not a case of choosing those which are really the prettiest, nor even those which average opinion genuinely thinks the prettiest. We have reached the third degree where we devote our intelligence to anticipating what average opinion expects the average opinion to be”. And it is so, to the power of 67.

Shopping Basket
Scroll to Top