Department of Awful Statistics

Chris Hayes points to Dean Baker catching the Washington Post in a rather sizeable error:

T]he Post tells us that "Mexico's gross domestic product, now more than $875 billion, has more than quadrupled since 1987." That's not exactly right. If we pay a trick to the IMF website, we find that Mexico's real GDP in 1987 was $1,130 billion. The IMF projects 2007 GDP as $1,895 billion, an increase of 67.7 percent. That is considerably short of quadrupling.

How on earth does the Post get that Mexico’s GDP quadrupled when it actually only grew by 67.6 percent? Well, they may have taken the growth in nominal GDP. If we don’t adjust for inflation then we can conclude that Mexico’s GDP quadrupled over the last twenty years. Of course, no reasonable person would ever assess growth without first adjusting for inflation since it has no meaning. If we don’t adjust for inflation, Zimbabwe’s economy, wracked by hyperinflation of several thousand percent annually, is the fastest growing economy on the planet. If the Post editorial writers use a consistent measure, we can expect to see warm praise for Zimbabwe’s extraordinary growth on the editorial pages in the near future.

My best guess is that the Washington Post used PPP figures in current international dollars, which give roughly a tripling of GDP since 1987. This is not totally out of line; using constant dollar figures from the Economist Intelligence Unit, I get roughly a doubling since 1987. But quadrupling is just in error according to any data set I can find. And I'm not sure what to do with a current dollar PPP figure.

Unfortunately, Mr Baker has made a few errors too. First of all, the constant-price GDP figures he gives are in pesos, not dollars. While it is true that in Mexico the dollar sign is used to denominate peso prices, the international standard for economics journalism--and certainly, the domestic American standard for same--is to write out "pesos" or abbreviate it (MXP) and then offer the exchange rate equivalent in a common currency, almost always dollars but in a growing number of cases, euros.

I'd also argue that to eliminate this sort of exchange rate confusion, Baker should have used PPP in constant dollars. That figure yields a much more favorable assessment of NAFTA--although still one that makes the Washington Post look wildly off.

Comments

I see more and more stories about mistakes and poor journalism at the Post. Maybe it's always been that way and people only now are getting the tools they need to notice, but it's certainly depressing. Still, it's one of the nation's most respected newspapers, if not quite as blindly trusted as it was.

Posted by liberalrob | December 3, 2007 12:28 PM

A couple of months ago the New Yorker made reference to the quadrillion-dollar US economy.

Posted by Dave | December 3, 2007 1:04 PM

And National Review blamed the well known fact that all Arabs lie for their fabulist reporter.

Posted by Rickm | December 3, 2007 1:30 PM

liberalrob,

I think your first instinct is correct- it was always that way, but now, with the internet, it is far easier to access the information (and to learn how to interpret the information) to verify what is printed in papers. The dispiriting thing is that it should also be easier for journalists to do this as well, but they don't appear to make as much use of it as they should.

Posted by Yancey Ward | December 3, 2007 2:42 PM

Yikes. Depressing.

Yancey, I think part of the problem is that as a corollary of the ease of obtaining information via internet, reporters are now expected to produce vast quantities of material which are actually just too much information for one human being to process intelligently. I'm not sure whether this applies to Baker, but I have a sense that it would have been impossible for a reporter in the'60s or '70s to produce 200 stories a year, which is what a productive reporter at the Post can expect to produce. Or maybe it's that there would have been a separate researcher responsible for hauling up clips for them to get the background info they needed, and that person's responsibility for delivering accurate info would have made this sort of whopper less likely. Or maybe this happened in the old days too. Hm.

Posted by brooksfoe | December 3, 2007 8:37 PM

brooksfor,

If your analysis is correct, then WaPo and others have apparently chosen to pump out masses of unintelligent, error-prone stories rather than some reasonable number of accurate, well researched stories. I suggest that was a bad choice, for them and for their readers.

Posted by Ed Reid | December 3, 2007 9:00 PM

It seems that the apple of Katharine Graham's son Donald fell quite some distance from her tree.

Posted by M. Bouffant | December 3, 2007 9:32 PM

I'm not even sure if its an accident or an intentional manipulation anymore. I spent a few years studying political economy and for every useful statistic there are five to six ways to distort it to fit your political agenda.

It doesn't really matter what message / vision of the economy you want to portray: There's a government (or academic) institution that's created the numbers & ratios that will (appear to) support your thesis.

Posted by underground | December 4, 2007 12:11 PM

This seems less an argument for the terribleness of the Post, and more of an argument for showing your work when writing about numbers.

The confusion stems less from the error
(I've seen worse by more knowledgeable people), than from the fact that there is no way to tell where the number they used actually came from. This would be inexcusable even if the number did accord with a widely used model.

Posted by heedless | December 4, 2007 12:26 PM

heedless,

There is far less "message flexibility" when the source has to be provided and the "work" has to be shown.

Posted by Ed Reid | December 5, 2007 8:52 AM

Out of curiousity, why is 1987 used as the starting point for determining the amount of growth?

Posted by Ken Kellera | December 6, 2007 10:22 AM