UK statistics office to strengthen recession-spotting skills

By David Milliken
Offices are seen at dusk as St. Paul's Cathedral and construction cranes are seen on the skyline in the City of London, Britain November 2, 2015. REUTERS/Toby Melville

By David Milliken

LONDON (Reuters) - Britain's statisticians are trying to become sharper at spotting when the economy is sliding into recession, after missing early signs of a downturn the last time around in 2008, the country's statistics agency said on Thursday.

British gross domestic product data is produced faster than in almost all other major economies, but early versions contain a large chunk of estimated data.

If an initial wave of businesses report a sharp downturn, Britain's Office for National Statistics must judge if that represents a real slowdown, a distortion in the data, or a mix of the two.

"It's bloody difficult," ONS statistician Darren Morgan said at a briefing on Thursday.

The ONS's first preliminary estimate of quarterly British growth data - produced about a month after the end of the quarter - contains less than half the full data used. Even the first 'full' estimate, which comes two months later, is regularly revised, often years later.

In the three months to June 2008 the ONS initially estimated Britain's economy was growing, albeit slowly. But in fact Britain had actually already entered what was to be its deepest recession in decades. Only months later did the Bank of England begin to cut interest rates in earnest.

Part of the problem, Morgan said, was that the ONS was reluctant to take at face value its raw data and models which showed an almost unprecedented slowdown.

In 2009 the ONS was slow to spot Britain was coming out of recession, at a time when many external economists were highlighting a sharp improvement in business surveys.

While the ONS does not formally declare whether Britain has entered a recession, most economists use its figures to see if there are two consecutive quarters of contraction in output - the standard benchmark for a recession in Europe.

The United States takes a different approach, with its National Bureau of Economic Research looking at a wider range of data, including retail sales and unemployment, to date the start of a recession to a specific month.

A year ago, a review by former Bank of England deputy governor Charles Bean concluded that confidence in Britain's Office for National Statistics had fallen, and said statisticians needed to think more about whether their figures made economic sense.

However, the months after last year's vote to leave the European Union offered a boost to the ONS.

While many analysts latched on to a sharp downturn in private-sector business surveys as pointing to an imminent recession, ONS data correctly showed broadly steady growth.

Beyond tweaking forecasting models used to create early GDP estimates, the answer for the ONS lies in using a wider range of data than its monthly business surveys, which form the backbone of early GDP estimates.

The surveys have problems, such as businesses being slow to return them if they are unexpectedly overwhelmed with work - or conversely if they are suddenly struggling to survive.

Later this year the ONS plans to start using other figures, such as monthly government sales tax data, to help show if these effects are swaying survey responses in a given month.

Another goal is to use more BoE data on Britain's large financial services sector, and also to look more at raw figures before any anomalies or early trends get blurred through inflation adjustments.

Chris Hare, a London economist at Investec, said ONS growth data stood up well compared to other countries, despite weak spots in areas like business investment.

But some future misses were inevitable, he added.

"You are not always going to be able to call a turning point perfectly well."

(Editing by Andrew Bolton)