Saturday, December 29, 2012

Glaeser... Glaeser... Where have I heard that name before?

Joseph's last post has got me thinking that it might be a good time for a quick Edward Glaeser retrospective.

Glaeser is, unquestionably, a smart guy with a lot of interesting ideas. Unfortunately, those ideas come with a heavy dose of confirmation bias, a bias made worse by a strong tendency to see the world through a conservative/libertarian filter, a provincial attitude toward much of the country and a less than diligent approach to data. The result is often some truly bizarre bits of punditry.

The provincialism and cavalier approach are notably on display in this piece of analysis involving Houston, a city Glaeser has written about extensively.
Why is housing supply so generous in Georgia and Texas? It isn’t land. Harris County, Tex., which surrounds Houston, has a higher population density than Westchester County, N.Y.
The trouble is Houston is IN Harris County (technically, the town does spill over into a couple of other counties -- Texas counties are on the small side -- but it's mainly Harris).




Keep in mind that Glaeser is one of the leading authorities on cities and Houston is one of his favorite examples.

[Update: Glaeser has correctly placed Houston is Harris County in the past, though the Harris/Westchester comparison still raises questions.]

Glaeser's flawed example was part of a larger argument that "Red State growth is that Republican states have grown more quickly because building is easier in those states, primarily because of housing regulations. Republican states are less prone to restrict construction than places like California and Massachusetts, and as a result, high-quality housing is much cheaper."

Like so much of Glaeser, it's an interesting idea with some important implications but as presented it doesn't really fit the data.

This confirmation bias can lead to some other truly strange examples:
But there was a crucial difference between Seattle and Detroit. Unlike Ford and General Motors, Boeing employed highly educated workers. Almost since its inception, Seattle has been committed to education and has benefited from the University of Washington, which is based there. Skills are the source of Seattle’s strength.
The University of Michigan is essentially in a suburb of Detroit. UM and UW are both major schools with similar standings. Washington is better in some areas, Michigan is better in others, but overall they are remarkably close. When you add in Wayne State (another fine school), the argument that Seattle is doing better than Detroit because of the respective quality of their universities is, well, strange.

Glaeser's confirmation bias has led him to make a number of other easily refuted arguments. His predictions about the auto bailout aren't looking good. Joseph pointed out numerous problems with his statements about food stamps. Dominik LukeŇ° demolished his school/restaurant analogy. His claims about Spain's meltdown are simply factually wrong.

To make matters worse, Glaeser doesn't seem to show much interest in engaging his critics. As far as I know, none of these points have ever been addressed.



4 comments:

  1. Glaeser knows where Houston is and its relation to Harris County; his *point* is that the relatively high growth region (Harris County/Houston) had higher density than the lower growth region (Westchester County). Hence, he argues, higher growth cannot be attributed to lower density. Here's Glaeser elsewhere, even more clearly revealing he does actually know where Houston is:

    "If our Houston family’s income is lower, however, its housing costs are much lower. In 2006, residents of Harris County, the 4-million-person area that includes Houston, told the census that the average owner-occupied housing unit was worth $126,000."

    In contrast to your previous criticism (http://observationalepidemiology.blogspot.ca/2011/01/curious-case-of-dr-glaeser.html), Glaeser does not ignore population density in his analyses. See for example:

    http://www.economics.harvard.edu/pub/hier/2005/HIER2062.pdf

    Further, regarding the `ice cream causes murder' allegation, he is quite aware of the dangers of unobserved confounders and consequently uses econometric methods which eliminate or minimize resulting issues.

    In general, it seems inappropriate and unfair to focus on nitpicking his pop writing rather than seriously considering his academic work. Glaeser focuses on casual case studies in pop writing to illustrate theoretical arguments, but the research on which he's basing his conclusions is in the form of a long sequence of highly influential statistical studies published in top peer-reviewed journals. He's not basing conclusions on anecdotes and confirmation bias.

    ReplyDelete
    Replies
    1. "Glaeser knows where Houston is and its relation to Harris County; his *point* is that the relatively high growth region (Harris County/Houston) had higher density than the lower growth region (Westchester County). Hence, he argues, higher growth cannot be attributed to lower density"

      This argument depends on the two counties being analogous, which would be arguable if Harris really did surround, rather than include Houston. When you look at comparable regions, you'll see that density in and around Houston is much lower than those in and around New York.

      "In general, it seems inappropriate and unfair to focus on nitpicking his pop writing rather than seriously considering his academic work."

      Glaeser's popular work is better known and probably more influential than his academic work. It has to stand on its own.

      Delete
  2. Glaeser has an insurmountable edge over you: he is a better writer.
    Your hole thing about harris county etc is gibberish; you have left out so much of the argument that onlya congnoscenti could understand what you are trying to say

    go back and re read galbraith on book reviewing

    ReplyDelete
  3. Sorry to spite the critics, but population density does have play a role in housing prices. It is also quite ignorant to be making an apples to oranges comparison.

    Local housing and zoning regulations might be a factor in the migration, but Glauser's article in the NY Times proves nothing.

    ReplyDelete