The late 1990s saw aspiring eurozone members carrying out the economic equivalent of the crash diet, as they tried to meet the rules for joining the new club. Inflation, deficits and debt all needed
to be low — a tough target for
a country such as Greece. The Greeks nevertheless appeared to comply. Over time, though, it became clear that all was not as it had seemed. The European statistics agency repeatedly complained about the misreporting of Greek data. (In 2006, Greek GDP jumped 25 per cent overnight when the authorities sought to incorporate illegal industries such as money laundering and prostitution.)
Had Greek economic statistics been more accurate, the country may well have been excluded from the eurozone, which — with hindsight — would probably have been the best thing all round. But could the problems have been spotted at the time? Perhaps. Loyal readers of this column will recall me writing about Benford's Law — a statistical regularity that often occurs in 'real' data but not in manipulated numbers. Now four researchers have published a paper in the German Economic Review using Benford's Law to examine Greek macroeconomic data.
Benford's Law, named after physicist Frank Benford, is a little odd. It predicts the frequency of the first digits of a collection of numbers. Measure the populations of cities, lengths of rivers, or just the numbers mentioned in a copy of The Economist, and see how many of the digits begin with 'one' (164 metres; 1,351 miles) versus 'four' (4,980,000) or 'seven' (£745m). Benford found that, 30 per cent of the time, the first digit is a 'one', and this result holds true for a vast range of different sources of number. Nobody is absolutely sure why this might be true. What is clear is that the law does not hold for artificially assigned numbers, such as telephone numbers.
Benford's Law also often fails to hold for — ahem — fraudulent numbers. It is easy to get an idea of why not. For instance,
a salesman who must submit receipts for expenses over £10 may end up filing claims for lots of £8 and £9 expenses — and the data will then contain too many eights and nines. A forensic accountant can check for Benford's Law very quickly and, while it is not an infallible check (Bernard Madoff filed Benford-compatible monthly returns), it's an indicator of possible trouble.
Which brings us back to Greece. According to the German Economic Review paper, Greek macroeconomic data is further from the Benford distribution than that of any other EU member state. Perhaps Benford analysis could have kept Greece out of the eurozone; although it seems likely that politics would have won out over statistics. That's unfortunate: by the yardstick of Benford's Law, Greece's data was particularly odd in 2000. That was the year before it joined the euro.
Tim Harford is a Financial Times columnist and author of Adept: Why Success Always Starts With Failure (Little Brown).
blog comments powered by
