After our friend Don Luskin posted comments claiming that econometrics was a pseudo-science. I sent off an e-mail asking him to explain why he thinks that econometrics is such a sham and he replied with an update on his blog. Now, the discussion has become a fierce debate with others joining the fray. You can read the debate here.
With that being said, allow me to respond to the latest comments.
I am still not truly sure what Luskin meant by not being able reproduce results. In fact, first semester econometrics students are often given assignments in which they are required to reproduce someone else’s results and then expand on them. Further, Luskin claims that that the data is often inaccurate and poorly defined. Although this may be true in some cases, these studies are quickly proven wrong and dismissed by academics as having “poor data”.
He continues by claiming:
Because it posits cause and effect relationships in data that cannot be proven — and uses these very relationships themselves as “proof” of its claims — it is neither “logical” nor “subject to verification.”
These relationships are proven through statistical methods. Are we to assume that statistics is a pseudo-science as well? Econometricians set up models with a dependent variable and several independent variables. They then use regression techniques such as ordinary least squares, ridge, spatial, tobit, probit, etc. to determine the relationships the independent variables have on the dependent variables holding all others constant. The studies are “subject to verification”. Econometrics has been used to show that consumption, investment, government spending, and net exports determine GDP, which is in line with economic theory.
Finally, the fact that errors can sometimes occur and that data can be manipulated is a problem common to all science. Simply because errors occur is not a reason in and of itself to dismiss a method of study.
UPDATE: The debate continues here