Wednesday, August 21, 2013

Post-Treatment Euvoluntarity Evaluation

An easy trap for a novice applied econometrician to fall into (and take it from me, I'm one myself) is to include variables that capture post-treatment effects in a regression. If you're interested in the effects of X on Y and you want to control for variance introduced by A, B, and C, you should pause for a moment and ask whether X itself influences A, B, and C. For example, if you want to know what influence a bit of crude oil tax policy has on prices at the pump, including reserve stock as a control variable can skew your results since the tax policy is likely to influence how much crude the firm warehouses.

Well, that might be a silly example. Ask Giberson or Kiesling if you want to know more about the actual econometrics of the oil industry. It's not exactly in my wheelhouse.

But euvoluntary exchange is and similar arguments apply to evaluating euvoluntarity apply. Take illicit drugs for example. If you want to discover whether or not a market in marijuana is euvoluntary, you must make your point of analysis free of the confounding influence of existing legislation. Feel free to include whatever other institutional factors you like, but the treatment is the legislation. You can't just say drugs are bad because they're illegal and they're illegal because they're bad. Even small children can recognize that such arguments beg the question.

And this problem is massively widespread when discussing a wide range of topics. Worried about the effect of immigration on public schooling? The way public schools are administered is the treatment, people. Ditto medicine, ditto deposit insurance, ditto lending markets. Ditto, well, take a quick peek at the volume of the Federal Register and make your own list.

Be very cautious about what actually counts in your EE calculus. I think you'll probably find that things aren't always as they might seem and there's quite a lot of exchanges that could be more easily categorized as euvoluntary but for the scritch-scratching of legislators' pens.

No comments:

Post a Comment

Do you have suggestions on where we could find more examples of this phenomenon?