In this post I talked about the problem of confusing different margins of comparison. This is what I mean w.r.t. the public school issue:
In the United States we are living in the left columns, with a mix of public and private schools where private school students generally have better outcomes than public school students (btw - I went to public schools through high school, William and Mary is public, and George Washington University and American University are both private). We often make this comparison between public and private school students - either formally and rigorously or informally - when we discuss these issues. Those comparisons have some value, of course. From a parents' perspective it helps them think about where to send their kids, and it also helps us think about the structure of public education and the wisdom of things like charter schools and vouchers. But when we pontificate on the role of government in education its exactly the wrong comparison to think about. When we talk about the role of government we have to compare the left columns with the right columns. That's an entirely different question, and it's harder because distributional issues start to enter the equation. Not to mention the problem that we have no data on that question!
Another place this pops up is the question of wage cyclicality. If you plot out the raw data, real wages seem to be high in recessions - it looks like wages are counter-cyclical. This bolsters both New Keynesians who think that sticky wages matter a lot and anti-intervention types who want to blame the minimum wage, labor laws, etc. The problem is its a big example of margin confusion again. If high wages were the culprit in recessions, what we would want to look at is the spot market for labor - the intensive margin. We would want to look at trends in the cost of one more unit of labor over the business cycle. Easy - that's the hourly wage data series that the Bureau of Labor Statistics produces, right? Wrong. Aggregate hourly wage statistics are influenced by fluctuations on the intensive and the extensive margin, and as such they can be very misleading in discussions of wage cyclicality. The problem is that different sorts of workers are employed at different points in the business cycle, so wage data is picking up fluctuations on the extensive margin - the decisions to hire or fire different types of workers, rather than fluctuations on the intensive margin for a constant pool of workers. Brad DeLong summarizes the literature on this problem in his excellent critique of Vedder and Gallaway here. The empirical evidence on this is very consistent - wages are not highly counter-cyclical. At most they are acyclical, although some find a modest pro-cyclicality. Abraham and Haltiwanger (1995) provide a survey of this literature, and I suggest anyone who thinks empirical economics is all over the map read it. The findings are very consistent and all point to the same conclusion. Your margin of comparison matters.
I don't have time to go over it now, but this also matters a lot for the economics of discrimination and the work that Becker did (which a lot of people get confused about and draw the wrong lessons from). I'll try to post on that in the near future. But out of curiosity I want to ask - what do people think Gary Becker concluded about discrimination in the labor market?