Cooking the books can lead to a half-baked result
‘Greece has brought criminal charges against the official responsible for measuring the country’s debt, thereby calling into question the validity of its €172bn second bailout by the EU and International Monetary Fund.’
Financial Times, January 23
What?
Could be worse. When Andreas Georgiou, who was head of Greece’s independent statistical agency Elstat last time I looked, was first placed under investigation in late 2011, there was talk of him facing life in prison for a kind of statistical treason. As it is, he and two of his senior staff are facing charges of corruption and making false statements, with a mere five to 10 years if found guilty.
But what’s behind all this?
Mr Georgiou was brought into Elstat in 2010, and although he’s a Greek citizen he worked for the IMF for a few decades. He thus represents the global forces of technocracy. The case against Mr Georgiou is that, as Greece was negotiating a bailout from the international community, he deliberately exaggerated the country’s deficit statistics in a way that weakened its negotiating position. His motive, presumably, would have been to strengthen technocrats in the IMF and Eurostat, the EU’s statistical agency.
Not much of a motive.
Mr Georgiou says he is being prosecuted not for cooking the books but for failing to cook them. Greece’s budget deficit doubled overnight in late 2009 shortly after a new government was elected and announced that the previous deficit had been massively underreported. So there was a lot of statistical horseplay going on before Mr Georgiou showed up. It would certainly be comforting for some Greeks to believe that it was Mr Georgiou, not his predecessors, who were producing inaccurate numbers; Eurostat, however, has blessed Mr Georgiou’s figures as accurate.
Is that why he’s being prosecuted, then?
He seems to have been embroiled in full-contact office politics. One of the people accusing him, Zoe Georganta, was an Elstat board member who was sacked (along with most of the board) in 2011. Another, Nikos Logothetis, was accused by Mr Georgiou of hacking his email and has been criminally investigated. Elstat’s union has weighed in against Mr Georgiou, too. As to the merits of these accusations, your guess is as good as mine.
Gosh. Who would be a Greek statistician?
Not just Greek. The most recent issue of Significance (the Hello! magazine of the statistical community) included an interview with Graciela Bevacqua, formerly head of Argentina’s statistics institute, who complained that she had been sacked, fined and threatened with imprisonment after publishing inflation statistics that were insufficiently flattering to Argentina’s government. The same issue included a discussion of the L’Aquila earthquake verdict, in which six scientists and an Italian official were sentenced to six years imprisonment for providing “inaccurate, incomplete and contradictory information” regarding the risk of a quake that, when it occurred, killed 309 people.
But this is outrageous. Statistics should be above politics. It reminds me of the time Alabama changed the value of pi to three.
I’m afraid life will never be that simple. (For one thing, Alabama never did change the value of pi – that’s a myth.) Statistical judgments are just that: judgments. There is right and wrong in statistical practice, but typically there are also shades of grey. Walter Radermacher, Eurostat’s boss, told The New York Times that “the truth is not my business . . . statistics is about measuring against convention”. He has a point. The UK’s Office for National Statistics did chose not to correct a significant flaw in the way the retail price index was calculated, a decision that seems to have been motivated by a preference to overcompensate pensioners, index-linked bond investors and others for inflation.
Was the ONS right?
No, it should have corrected the RPI and published an alternative series, RPIAFU.
RPI all fudged up?
Something like that. RPIAFU would have been calculated using the old method and if the government wished to continue to benefit pensioners and bondholders at the expense of taxpayers and the purchasers of price-regulated products such as railway tickets, then fine. But the truth is that governments will often prefer statistical obfuscation as a way of achieving their goals indirectly, and statisticians will find themselves in political debates with no unambiguous, correct answer.
Also published at ft.com.