Ryan Bourne of the Centre for Policy Studies exposes the perpetual over-optimism of Treasury growth forecasts

Ryan Bourne is the Economic and Statistical Researcher at the Centre for Policy Studies.

At the CPS growth event last week, Lord Griffiths opened the debate with a remarkable statistic. “Between 1975 and 2008,” he said, “on 31 occasions the initial estimate of GDP on a quarterly basis was negative. The subsequent average upward revision was 0.95%.”

This shows just how difficult it is to measure GDP growth accurately. And if measuring GDP is difficult, then it is even tougher to forecast. Until last year, the unenviable role of assessing growth prospects was undertaken solely by the Treasury. But there were growing suspicions during Gordon Brown’s premiership that growth forecasts emanating from HMT tended to be over optimistic.

Archived Budget and pre-Budget reports allow us to examine whether this suggested over optimism has statistical justification. The Budgets and PBRs forecast growth up to three years ahead, and so allow us to examine the accuracy of forecasts at different time intervals from the reported Treasury outturn.

Given the difficulties in forecasting outlined above, we would expect there to be quite substantial deviations from eventual outturn for individual calendar years, especially when forecast up to three years ahead. But averaging over the whole time period, we would expect these deviations to be small.

Instead, what we do see is a relatively strong upward bias to forecasts for the period 2000-2010:

Indeed, overall results for the period show that Treasury forecasts of years three years into the future have been over optimistic by 1.3 percentage points on average; with forecasts two years into the future over optimistic by 1.3 percentage points; and 0.75 percentage points over optimistic when forecasting the next year’s growth.

Unsurprisingly, the lack of foresight for the financial crisis and subsequent recession meant that forecasting accuracy was particularly poor for the period between 2007 and 2010. But even for the ‘good years’ of 2000-2006, there was still an average upward bias in growth forecasts of 0.5 and 0.3 percentage points when forecasting two years and one year ahead.

If growth figures have been massaged in the past, George Osborne’s new independent Office for Budget Responsibility should restore forecasting honesty. But if unchanged models with an in-built upward bias are still being used by HMT and OBR, there’s a high probability that this government will not meet its structural deficit target.