"Words ought to be a little wild, for they are the assault of thoughts on the unthinking" - JMK
- Tyler Cowen asks why India's PISA rankings are so low. A better question is whether we should even be using PISA rankings. Hal Salzman and Lindsay Lowell provide a dose of skepticism about these metrics here.
- This J. Scott Armstrong paper on regressions has been making the rounds. I have not read it yet, but people seem to think it's good. The basic message seems to be a correlation is not causation point, and then some ideas about restricting the number of variables in a forecast. I can't speak to the latter, but I'm generally skeptical of economic forecasting anyway. The former point about correlation and the role of regression is something that I think econometricians are well aware of anyway. I think critics get a little too zealous on this. The point is that causal assertions (1.) are not proven by regressions, but they are framed by regressions, and (2.) clearly any causal claim has to originate from an underlying theory - not from a regression coefficient. People clearly shouldn't be confusing causation with correlation. I don't think they are, generally. But people also should not be causation agnostics. Don't be afraid to make causal statements, just know that it's not given to you by your regression. Your grounds to make a causal statement are given to you by underlying theory, and confirmed by an empirical analysis that is cognizant of problems of endogeneity and spuriousness.
- Brad DeLong points out that Rick Perry is not a fan of states' rights when the state in question is my own state of Virginia.
1 hour ago