Join our community of smart investors

On spurious precision

The answer is: very little. The OECD says that GDP in Q2 2012 will be 0.2 per cent down from 2011 Q3. This is within measurement error of “flat”; the ONS says the average revision to its initial estimate of GDP since 2003 has been 0.23 percentage points. Sir Mervyn and the OECD are saying pretty much the same thing – don’t expect much growth between now and the summer. It’s just that the OECD is forecasting a recession – in the technical sense of two successive quarters of negative growth – whilst Sir Mervyn, in that statement, isn’t (quite).

What we have, then, are similar messages, but dissimilar approaches. The difference between them illustrates the difference between accuracy and precision. The plumber who tells you he’ll arrive at 10.27 is being precise but inaccurate (like heck he will). The one who tells you he’ll turn up in the morning is being accurate (we hope) but imprecise. The OECD is like our first plumber; economic forecasts being what they are, it is highly likely to be wrong. Sir Mervyn is like our second. Like Keynes, he would rather be roughly right than precisely wrong.

I much prefer Sir Mervyn’s approach. Spurious precision is a dangerous thing, not least because it breeds overconfidence. It was the belief that they could measure and control risk precisely that got so many banks into trouble in 2008: Gaussian copula models were precise but inaccurate. It is insufficiently appreciated that banks survived for decades by employing economists, but collapsed soon after they hired physicists.

I fear some stock-pickers fall into a similar error. It’s easy to worry about precise numbers in company accounts, whilst underweighting important but imprecise influences upon stock returns such as investors’ cognitive biases or the firms’ SWOTs.

Sadly, however, I fear I am in a minority of journalists here. Journalists love precision. And demand calls forth supply – hence headlines such as “public sector strike may cost economy £500m”, “50p tax rate could cost the country £350bn”, “Obesity and chronic disease cost UK PLC £20bn a year.” These numbers are inaccurate. But, being numbers, they look precise. And journalists, more than others, mistake precision for accuracy. Their attitude to statistics is like their attitude to sausages; they love them, but they’re not interested in how they’re made.