What's in a number? How economic data (should) drive decisions

The ancient Greeks knew a thing or two about democracy. So when a former government statistician from Greece recently made a plea to preserve something essential for “the functioning of democracy itself,” heads turned.

The subject of his juicy take? The very dry business of publishing trustworthy economic data.

The normally low-profile public servants who churn out tables and charts on a regular basis, and dutifully revise as needed, are illuminating lighthouses that prevent policy-makers from blindly crashing an economy on rocky shoals.

Should we lower interest rates? Better have solid inflation figures on hand, or the decision may not deliver the desired results. How about investing in a new factory? Hopefully you have industrious clerks at the ready to provide reliable numbers on the local labour market and energy costs.

When the ancient Greeks sought economic counsel, they turned to a guy strumming a lyre and reciting poems; Hesiod’s verse jibed with the labour theory of value, and was probably more enjoyable than a press release. It was also more accurate than the official budget-deficit figures modern Greece relied on in the first years of this century, which eventually turned out to have been under-reported. A worsened debt crisis and punishing austerity followed.

Other historical examples of bogus information inflicting real damage include agricultural statistics in the Soviet Union often intended solely to hit arbitrary quotas. Political factions implemented a system liable to overestimate grain harvests, and contribute to famine.

In the US, a combination of outdated models and problematic ratings on the safety of mortgage-backed securities stoked the 2008 global financial crisis. More recently, that country’s jobs statistics became news – when a dramatic, downward revision to a pair of monthly figures aroused presidential ire.

As economists argue for maintaining the sanctity of data no matter whom it upsets, an international framework designed to help do just that is being revamped.

An updated System of National Accounts was approved earlier this year, and is wending its way around the world. It’s an international standard for measuring a country’s economic fundamentals: production, consumption, investment, and wealth.

The lofty benefits promised by technology made the need for a global measurement refresh more pressing. Nearly $100 billion in added economic well-being was enjoyed in the US alone last year thanks to artificial intelligence, according to one estimate – but none of that registered in traditional metrics.

The British economist William Petty’s Political Arithmetick was published in 1690. It popularized the fledgling concept of applying numerical analysis to economics, and included its own sort of jobs report: “The Husbandman of England earns but about 4 s. per Week, but the Seamen have as good as 12 s.” People have been bickering about the accuracy of these kinds of figures ever since.

The same US labour-statistics agency that recently incurred the wrath of the White House also found itself in political crosshairs back in 2012. That time, it was an unemployment rate that seemed too good to be true.

322 years earlier, Political Arithmetick sought to pre-empt any 17^th-century political backlash with a preface carefully explaining that its overriding aim was “to shew the weight and importance of the English Crown.”

In the 21^st century, some say it might be best to entirely remove the production of economic figures and other sensitive information from direct political supervision. The ex-Greek government statistician whose recent appeal for data integrity drew widespread attention certainly thinks so. He still faces legal action for his role in correcting the official record in his country more than a decade ago. Meanwhile the pain of subsequent austerity measures there has left a lasting imprint.

Other suggestions for ways to bolster faith in economic data include better explaining the process behind it. US jobs reports provide a good example. Revisions to these numbers are nothing new, which is something most people don't know.

Each month’s new-jobs figure is based on surveys distributed to hundreds of thousands of businesses. It takes a while for many of them to respond.

In the meantime the initial data gets published even if it's premature – but is then refined once the bulk of responses are in hand a few months later. A particularly big revision may signal a significant swing in the economy. It’s a bit like a journalist rewriting months-old copy at least a couple of times to get it right. Or, more right.

Agencies may be able to better flesh out data like this for the public, but other aspects of the job sit further outside of their direct control. Like response rates.

A few years ago, responses compiled for the UK’s benchmark jobs report plummeted so dramatically that its publication was disrupted. That left the country’s central bank in the dark as it sought to set interest rates, with billions of pounds for the economy hanging in the balance. Officials have since considered making responses legally mandatory.

Falling short like that is one thing, fudging is another. Both are equally represented in the annals of history.

Economic data has been likened to vital infrastructure, and to a free information service for investors. The erstwhile Greek statistician who recently sounded an alarm chose to compare it to a mirror. Dim the lamp and preen all you want, but you’ll still look the same in the cold light of day.