27 October 2008

Mankiw splits a double header

Part I

Greg Mankiw lists "Engaging in data mining" as a common sin among economists.

Damn it, Mankiw, no. Data mining is an academic discipline, not a dirty word. Like all disciplines, it can be misapplied. How would he feel if I said "Have you been engaging in [sneer] macroeconomics?!"

What I assume he really means is "Have you been running scores of multivariate regressions and latching onto any statistically significant results, regardless of what your domain knowledge tells you ought to be important?" That is bad. But that does not invalidate all the great results you can get when you properly apply the techniques of data mining. I think Mankiw knows this, as does Robert Shiller, who I have also caught in this inexcusable slander, which makes it all the more annoying that they use such sloppy and mildly insulting language.

Part II

Here Mankiw redeems himself with a great breakdown of his returns to work under McCain and Obama's tax plans.
  • He calculates his incentive to work as being 62% lower under Obama's plan compared to McCain's. (If he earned $1 today and passed it on to his children in 35 years he estimates he would be giving them $28 without any taxes. Under McCain's plan that is reduced to $4.81 and under Obama's it goes down to $1.85.

  • This is a good example of how taxes add up quickly. A couple of percent for payroll taxes, a couple more for capital gains and pretty soon he's looking at a 93% marginal tax rate. (And that's before any of this money is actually spent.) That's one good reason to support a single unified federal tax, be it a VAT, or a "fair tax" or what have you. Of course, that's "transparency" that no one on the hill is actually interested in because the plebs might begin to realize how much is being hoovered out of their wallets.

No comments:

Post a Comment