Lies and Statistics

Much of our public discourse depends on economic statistics -- unemployment is up this much or down that much, GDP is up this much or down this much -- and we throw these numbers around as if they were read off highly accurate meters. But they are not; they are generated by teams of government economists based on surveys of varying degrees of accuracy and formulas that are in some cases very controversial. While I am not sure that statistics really drive government action as much the press seems to claim (I think politicians often use statistics as excuses to do what they wanted to do anyway), they surely have some impact.

It is therefore salutary to be reminded, from time to time, of how questionable our economic statistics really are. Consider the claim that during 2007-2009 recession US productivity grew at the impressive annual rate of 2.4%. I have seen this number dozens of times in the Times, the Post, and elsewhere, usually accompanied by some notion of the "underlying health" of the economy. But take a look at this long analysis of that one number by Michael Mandel. At the very least, some of the numbers that went into the calculation of this value are highly suspect. (For example, who can believe that over this two-year period productivity in US mining grew by 23%?) By Mandel's calculation, there was no US productivity growth at all in this period; any "productivity" gains really came from more efficient offshoring, for example shifting from cheap Taiwanese suppliers to even cheaper Vietnamese suppliers.

Now Mandel is grinding a personal ax here, which is his argument that our economy is doomed to long-term stagnation by a decline in economically useful innovation. His calculation reinforces this view. But after reading this analysis it is hard to take the government's numbers seriously, and the whole public discourse based on them is revealed as so much hot air.