by Jeff Sovern
The story is headlined 'Plague on both your houses': Cordray, Noreika get scolding from statisticians. Excerpt:
" * * * There isn’t strong evidence either way,” said Bruce Meyer, a professor at the University of Chicago.
“There is weak evidence that says more than likely there is a positive effect on the cost of credit,” Meyer said of the OCC analysis. “But the accepted approach is you don’t crow about something unless you have stronger evidence.”
As for Cordray’s rebuttal, “it was wrong, too,” Meyer said. “You can say that there’s an 88 percent probability if you believe that the study that’s being quoted. There are always potential problems with research studies.”
* * *
Campbell Harvey, a Duke University professor and a past president of the American Finance Association, scolded both the bureau and OCC for fighting over things like p-values, which can be hacked or manipulated. He called the OCC’s probability calculations “vague” and said agency researchers “left themselves open to criticism that they don’t understand basic statistics.”
But he also said both sides should stop bickering and rethink what’s important. In that light, he noted that the CFPB couldn’t rule out the chance, however remote, that an arbitration ban could increase the cost of consumer borrowing by more than 3 percentage points.
“By standard methods we wouldn’t consider this significant. That doesn’t mean that you ignore it,” Harvey said. “While it’s not significant at traditional levels, so what? It could be economically important.”
There's more in the article. Two statisticians is better than we had, but still only a small (you should forgive the expression) sample. I wonder what others would say. I would also like to know more about what Harvey meant when he said that the agency researchers left themselves open to the criticism that they don't understand basic statistics. The agency in question seems to be the OCC rather than the Bureau, judging by the earlier reference in the sentence to the OCC.
