Monday, November 16, 2009

I realize today, that, as a "researcher", I've changed fundamentally in many ways. I also realize how much the faculty here has helped shape my thinking.

This is a good feeling.

And, I semi-demonstrated today that the results of a 20 year old seminal paper in econometrics do not hold up. Now, I cannot access the data that paper used - the Dept of Commerce changed the definition of the variable which is trying to be understood -

but one would imagine a change in definition should not affect the qualitative nature of an important macroeconomic indicator (implicit price deflators), right?

Wrong! It does, and it points to the fragility of at least one statistical result in the face of seemingly minor changes in the definition of a variable.

Although somewhat depressing, doing this sort of work makes you feel very intimately connected with economics.

It may not be sexy like Steven Levitt and Co's approach to "research" (take an interesting social phenomenon, subject it to a barrage of statistical tests, leaving aside issues of generalizability or usefulness) but it shows the very fragile nature of applied statistics when dealing with economic outcomes.

You'd imagine price deflators were among the more "solid" variables out there. Forget the sex, drugs and rock n roll economics, this is basic stuff one would take for granted.

The implications are troubling.