Erroneous data? not a spreadsheet kind of guy (part I)
My mind doesn’t think in terms of little boxes and straight lines. I have an open-ended, curvy type brain. It is a good thing that I have Excel fans on my team, I just don’t happen to be one of them. I think I get more than a little guff about this; the sorting, classifying and counting of information and data is what info-professionals do, after all. Well, not I — and I have come to find that the difference between those column-loving fans and me is the magic that makes this library sing, albeit a little offkey.
However, when you are out-numbered by people who are very, very good at counting and analyzing data, imagine my delight to read how an Excel coding error almost wrecked the world-wide economy. Yes, you read that correctly — an Excel coding error! I rub my hands doing my best Simon LeGree, but realize that doesn’t work unless you are on the other side of the bean counter equation.
But! Let me not digress without explaining the error and how it crippled the thinking of many bright folks who were ruling the world once upon a time (and even now). I love this story.
As Paul Krugman explains in his thoughtful opinion piece in the New York Times, The Excel Depression — in 2010 two Harvard economists, Carmen Reinhart and Kenneth Rogoff, circulated a paper, “Growth in a Time of Debt,” that supposedly identified a critical “threshold” (aka a tipping point) for government indebtedness: when debt exceeds 90 percent of gross domestic product, they claimed, economic growth drops off sharply. The paper came out just after Greece went into crisis and as a result, it instantly became famous, becoming what Krugman says was “the most influential economic analysis of recent years.” Only problem was, it was based on faulty data. Other researchers, using what seemed like comparable data on debt and growth, couldn’t replicate the Reinhart-Rogoff results. Hmmmm.
Researchers looking at the original spreadsheet found a vital clue to why they couldn’t replicate this data — R&R had omitted some data; they used some questionable statistical procedures; and finally, yes, they made an Excel coding error. Did I say they made an Excel coding error? I just had to make sure that you understood that part.
Correcting for these errors, and you get what other researchers have found: there is no sign at all of that 90 percent “threshold.” The basis for some of the most conclusive “proof” that is behind the austerity economic theory went blip.
Did I say they made an Excel coding error? I am a little bit OCD sometimes.