Slightly more technical detail from the paper: Essentially, the paper uses the instrumental variable[1] of "which colleges did you apply to" to create a rough m

Slightly more technical detail from the paper: Essentially, the paper uses the instrumental variable[1] of "which colleges did you apply to" to create a rough metric of applicant-confidence/pedigree. They then find that this indicator better explains an earnings gap between college graduates than simply looking at which college you attended. Namely, the simple fact that you thought you were good enough for MIT (even though you may have been rejected and went to a slightly worse school) is a larger predictor of future success than actually going to MIT. Where this assumption doesn't hold is for minority students though. I won't get into that because its complicated and I'm not sure I fully understand the subtleties there. But focusing on the main point; I think the conclusion would resonate very well with the HN community: that simply applying to top tier schools boosts your expected income because it reflects something innate about you --- some kind of self-confidence that helps you throughout life. This seems similar to a study[2], which I believe (I am going by memory here and this was a while ago) was discussed in Steven Levitt's Freakonomics[3], where the authors found that in places with really crummy schools; where sometimes there were lotteries to get into charter/magnet/better schools; that people who won the lottery didn't have better outcomes than people who lost the lottery. The theory was that the lottery outcome didn't matter so much as entering the lottery: because entering the lottery signified that the student had parents who cared enough about your education to enter you in the lottery... And in the long run this matters far more than going to a better school. All in all, a cool use of instrumental variables for the inner academic-economics nerd in me. [1] http://en.wikipedia.org/wiki/Instrumental_variable http://en.wikipedia.org/wiki/Instrumental_variable [2] http://papers.nber.org/papers/w13443 http://papers.nber.org/papers/w13443 [3] Going off on a tangent here, Levitt really took the study of various previously unmeasurable phenomena via the use of instrumental variables to a new level. Some people argue as to the merits of this technique and whether it actually is very insightful. Critics deride it as statistical sophistry taken to the point of being almost useless in assigning causal inferences. I'm somewhat torn. I spent a while reading tonnes of instrumental variable studies, and while there are a significant set that are confusing and probably not very useful, it does seem on average a legitimate statistical technique, albeit with very subtle problems that need to be handled well. For more on this controversy, there is [4], [5], [6]. While this is a completely irrelevant tangent to the article, I thought it useful to see and think about a side of economics that is unrelated to the recent financial mess that macroeconomists are taking a lot of flack for. [4] A hit piece on Levitt in the New Republic: http://www2.canada.com/components/print.aspx?id=b654079c-3fc5-4e6b-bb03-60ab5f5fd859 http://www2.canada.com/components/print.aspx?id=b654079c-3fc... [5] Greg Mankiw's (Harvard Economist) take on it: http://gregmankiw.blogspot.com/2007/04/is-steve-levitt-ruining-economics.html http://gregmankiw.blogspot.com/2007/04/is-steve-levitt-ruini... [6] Levitt's reply, including mention of Mankiw's analysis: http://freakonomics.blogs.nytimes.com/2007/04/25/am-i-ruining-economics-or-not/ http://freakonomics.blogs.nytimes.com/2007/04/25/am-i-ruinin...