Have you ever been working with a spreadsheet and you get a message saying ‘Catastophic error’ followed by a sinister hex number? (&H8000FFFF for all you geeks) No, Microsoft; that’s not catastrophic. This is a catastrophic error, now that Excel has been accused of tanking the global economy. This is a bit more serious than a corrupted file on your disk. Sure, there have been some spectacular disasters before, as when a first-year associate at Cleary Gottlieb Steen & Hamilton made an Excel reformatting error that mistakenly added a massive 179 contracts to an agreement to buy Lehman Brothers assets, or when JPMorgan lost $6 billion in 2012 on a credit-derivative bet due to an incorrect risk calculation on an excel spreadheet that drastically underestimated the risk, or when the same company admitted misreporting their ‘Value at Risk’ risk exposure for years due to an Excel error. But so far, there have been no Excel catastrophes to compare with the one that has misled global economists since 2010 about the growth of companies in a time of debt, and their reasoning and evidence has become a foundation of the arguments for spending austerity and debt reduction in the developed world, especially in the EU. Would the Greek debt crisis have been handled differently without it?
A paper, written by Carmen Reinhart and Kenneth Rogoff, and published in 2010, showed with seeming authority that growth rates for countries where debt was over 90% of the GDP was 1% slower than it would be otherwise. It has now been discovered that it is riddled with methodological errors and inconsistencies, one of which is a fairly bad error in an Excel spreadsheet used for part of the calculation.
What’s strange is that it’s repeatedly been reported in this way – “Excel Error”, “Flaw in Excel Spreadsheet” rather than the more accurate version of the issue, which is a combination of some peculiar mathematics, some strange and poorly-explained methodology, and flat-out calculation errors. Excel gave the right answer: the error was in the question. It’s hard to imagine the same headlines if all the calculations were performed in R.
It might be that there are solid journalistic reasons for the repeat of the “Excel error” line, such as laziness. Far be it for me to suggest that this story was reposted endlessly with little analysis or original thought applied each time. It could be that (heaven forfend) enough of us use Excel every day that mentioning it in a headline marks its use here as a genuine case of human interest. It could be that the reporters have no understanding of the broader analytical issues. Regardless, it is disingenuous, like blaming a word processor for a poorly thought-out law. It reframes the story as the original authors being brought low by the shoddy tools they are forced – forced! – to employ, rather than a matter of their own error.
So is Excel evil? Clearly not, no matter what hyperbolic news reports would have you believe. It’s a tool, and like any tool can be used well or used badly. We suffer the same problem with the misuse of relational databases. When things go seriously wrong, the reflex reaction is to say that relational databases can’t cope with our big data. This may seem like a small matter, but when large and trusted media organisations blame the tools rather than the person, it perpetuates the notion. Even ambiguous language (“Excel error” could of course be an error made in or by Excel) doesn’t help here.
Of course, it could just be the final revenge of a disgruntled former Microsoft employee...