Economic forecasts have become a political hot potato. The Office for Budget Responsibility’s (OBR) predictions, presented as part of the chancellor’s Autumn Statement, have put the government under pressure. The OBR has revised down its forecast for GDP growth over the next four years by 1.4 percentage points.
The real controversy is that their gloomy projections for GDP and government finances have been put down to Brexit. In the simple phrase of the OBR: “Any likely Brexit outcome would lead to lower potential output”. Lower output leads to lower tax receipts, and worse government finances.
To be fair, the OBR does say that “in current circumstances the uncertainty around the forecasts is even greater than it would be in normal times”. But just how great is this uncertainty?
Studies are published from time to time about the accuracy of economic forecasts. The best set of records is kept in America, though less systematic evidence for the UK shows that the track records are very similar in the two countries.
The Survey of Professional Forecasters (SPF) collects the forecasts on variables such as GDP growth and inflation from a wide range of forecasters. Its database goes back almost 50 years to 1968. Just one quarter ahead, the predictions are on average completely accurate. “One quarter ahead” means the next three months, so it would currently refer to the period January to March 2017.
This average accuracy conceals errors in most forecasts for any particular quarter, the errors cancel out over time. For example, the quarter from July to September 2008 marked the onset of the major recession of the financial crisis. At an annual rate, GDP fell by 1.9 per cent compared to the previous quarter. But the SPF predictions made in the April to June period for July to September were for growth of 0.7 per cent.
The SPF predictions account for only 25 per cent of the variability around the average. When we go four quarters ahead – just one year – the predictions are even worse. Negative growth, for example, has never been predicted, even though there have been 26 quarters of negative growth since 1968.
The track record, which has not got any better over time, shows that in relatively calm times, forecasts just one year ahead have a reasonable degree of accuracy. But when major changes are taking place, just when they are really needed, they have none.
The OBR cannot be blamed for producing predictions four years ahead when the track record of the forecasting community shows them to be of no value. That is what George Osborne mandated it to do when he set the independent body up in 2010. But four years ahead, almost any set of predictions is just as good – or bad – as another.
It would be much better to abolish the OBR and restore responsibility to the Treasury and, ultimately, to the politicians. If they get it wrong and are too optimistic, we can at least kick them out.