Jonathan Rothwell offers the opinion that the new MI report on segregation is not Glaeser and my best work. (Note that when Ed and I authored a very similar report after the 2000 Census, it was published by Brookings and to my knowledge did not receive any criticism from fellows there). It is of course difficult to read a phrase like "Glaeser, and to a lesser extent Vigdor," but I'll try to stay out of the muck here.
Rothwell wishes we had emphasized that segregation is high. He notes that many blacks live in areas with segregation indices above 60. That's true. And it is true that 60 is a sort of "magic number" that many people -- myself included -- have used as a dividing line between "high" and "low" segregation. But there isn't a compelling scientific rationale for using 60 rather than 61 or anything else, other than that it is a round number. If we set the bar at 70 -- an equally arbitrary round number -- we learn that the number of African-Americans living in "over 70" metro areas has declined from 14.6 million to 3.9 million since 1970, and the number of metro areas meeting that definition shrank from 134 to 19. Is that not a change we should notice and care about?
Rothwell also cites the cognitive dissonance between the Cutler/Glaeser QJE paper of 1997 and the current report. I worked as an RA on the Cutler/Glaeser paper and know it well. At the time it was written, it was the best non-experimental evidence we had on the impact of segregation. Now we have experimental evidence which is both more reliable and paints a starkly different picture. I'll also note that I published an article back in 2002 noting some fragility in the Cutler/Glaeser results, and Bill Collins and Bob Margo found that the results that C&G found in the 1990 Census could not be replicated with earlier enumerations.
I appreciate Rothwell's link to Rucker Johnson's work. But it should be noted that Rucker is looking at the impacts of school desegregation, not neighborhood desegregation per se. School segregation has not been trending downward over time, and there are definite reasons to be concerned about that. Of course, there's also very strong experimental evidence that black students perform better when assigned to black teachers -- a pattern which we can't take advantage of when schools are completely integrated. So even school segregation is a subject about which we have much to learn.
Comments