How accurate is the first estimate of GDP?

KNOWING how the economy is performing is of critical importance to business and economic policymakers in the Treasury, Bank of England and elsewhere.

But accurately collating huge amounts of data to distill down into a figure for quarterly GDP growth – and doing so within four weeks of the end of the three-month period it assesses – is a mammoth task.

The desire for speed and accuracy puts pressure on the number-crunchers at the Office for National Statistics (ONS). Do their preliminary estimates stand the test of time?

Graph to show how GDP estimates differ from final revisions

That estimate is subject to revisions, in the two months after publication and three years on. As the above graph shows, the ONS’s first estimate, in blue, is sometimes spot on, but it has missed the mark by a long way at times. The collapse in 2008, for example, was missed and only picked up in later estimates – although they weren’t the only ones to be slow to pick up on the huge impact of the emerging recession.

The gap between preliminary estimates and the final figure is highlighted in a second graph, below.

The difference has been as high as 1.5 percentage points, while in 2012 the preliminary estimates said the UK had a double-dip recession, but future revisions showed we hadn’t (albeit, only just).

In tIn the past the data has been criticised for too often underestimating the economy’s growth, but it overestimates nearly as often. In the graph below, positive numbers show when the preliminary estimate had to be revised down.

It is also worth remembering that revisions to the data are not the result of mistakes – it almost always comes in within the statisticians’ margin of error.

GDP graph 2016