Measuring GDP is as much an art as a science

GDP estimates are eagerly awaited in the City, and dominate the media headlines. Huge significance is attached to arithmetically trivial differences, whether between market expectations and the announced figure, or to subsequent revisions to the data. But GDP is not something which can be put in a set of scales, say, and measured accurately.

The concept is clear. It is the value of national output at market prices. Market prices? How do we value the public sector, where there are no market prices? A series of plausible conventions has evolved as to how to value such activities. But there is a substantial amount of arbitrary judgment involved.

Even in the market sector of the economy, a vast array of disparate data and estimates have to be taken into account. Measuring GDP is as much an art as a science. National Accounts: Sources and Methods tells us that the potential margin of error around any single estimate is plus or minus 1 per cent. A whole per cent! So revisions of 0.1 or 0.2 per cent are scarcely worth commenting on.

This inherent uncertainty of measurement may offer a clue to a paradox which is receiving a lot of current attention. Employment has been rising in the recent past, and is now effectively back to its pre-recession peak level. In contrast, GDP remains nearly 5 per cent below its previous high point. This contrasts dramatically with the UK’s last big recession, a non-financial one, that of the early 1980s. When output regained its pre-recession level, employment was 9 per cent below its previous peak.

We have been here before. Not in any of the post-war recessions, when each time employment fell more than output. But to the time of the last financial crisis, in the early 1930s. Then, as now, the percentage drop in employment was less than that in output.

The output of the financial services sector is notoriously difficult to estimate. As an excellent but little noticed article in the Bank of England Quarterly Bulletin last autumn notes drily: ‘Users should not have unreasonably high expectations of some of the proxy measures that have to be used to estimate output in the sector’.

In other words, even the Bank has not got much of a clue as to what has been happening to output in financial services. The key point here is that output per worker in this sector is exceptionally high. Even a big fall in output does not have much of an impact on jobs.

Is this what has really been going on? The falls in GDP are broadly correct. But the ONS has got the balance wrong. Financial sector output is down sharply, and the rest of the economy is not doing too badly. This would help explain why employment has held up much better than expected in the financial crises of both the 1930s and now.

Paul Ormerod is an economist and partner at Volterra Partners LLP, and author of Positive Linking: How Networks Can Revolutionise the World