It was even worse than that: the paper had multiple problems, of which the Excel bug was one, but they also chose an outrageously dumb sampling method. Basically, if a country had data for five years, it was weighted five times as heavily as a country that had that data for only one. Which, conveniently, lined up perfectly with overweighting countries whose development matched their hypothesis and underweighting countries that contradicted it. Frankly I will never be convinced that it wasn't outright academic fraud in order to generate the conclusion they wanted, and I'm angry that they still have jobs.
The benefactors desired a conclusion, and the authors delivered it for them. They served their purpose. That "academic integrity" was not a purpose reflects the state of affairs in economics.
Those are two possibilities. Another is that the government is the biggest player in the bond market and has unlimited funds at their disposal that they can categorize as Other.
But of course, that's illegal, so it would never happen would it. But at the same time, you and me will never, ever, hear from the bond vigilantes again.
I coming to this topic from utter ignorance, but if longer term data might be considered more reliable, then might it also be reasonable to preferentially weight the more reliable, longer term data, all else being equal?
