A fairly well written article from the Wall Street Journal concerning the literature debate between Hoxby and Rothstein over school competition.

Why can't econometrics textbooks provide such intuitively appealing descriptions of the instrumental variables technique?

Testing a hypothesis in economics isn't as straightforward as, say, testing a drug, where researchers can randomly assign some subjects to receive a placebo. Many economists believe they can approach scientific rigor, however, by taking advantage of random events like draft lotteries and judicial assignments. For Dr. Hoxby, streams offered such an opportunity: Cities with lots of streams had been randomly chosen by nature to have more school districts and more school competition, while cities with few streams were naturally home to fewer districts and less competition.

The current literature spat focuses on Hoxby's particular instrument - the number of small and large streams instruments for the number of school districts in a particular city. Rothstein argues against Hoxby's instrument and ultimately quesitons her conclusion that competition amongst schools improve school performance (which right there says more than anything else, IMHO).

While the criticism of a particular instrument is a common place, after all if there were a perfect instrument then there wouldn't be a need for the instrumental variables technique in the first place, such debates rarely make it outside of the journals or poorly attended conferences. It is interesting to see this particular debate make it into the MSM, and in a fairly well written article.