This discussion is fascinating! How would a modern empirical economist, student of Duflo and Piketty answer this? The fundamental empirical question is it worth going to Harvard when US State school university would do? A Thread. (1/n) https://twitter.com/nadeemhaque/status/1282928723235348480

A similar question is what worried Dale and Krueger (QJE, 2002). Suppose Steve Jobs went to Harvard not Reed College in Portland as he actually did, would it matter? (2/n)

The problem with comparing students who go to Harvard and those than go to some US state school is they are not apple to apple comparisons. Suppose we compare Jamal and Ahmad, where Jamal got into Harvard because his parents invested into his SAT private tuitions and Ahmad .(3/n)

could not get into Harvard because he had lower SAT scores and no daddy money to invest so Jamal and Ahmad are not apples-to-applies comparison. If you compare their income of Jamal and Ahmad, Jamal would earn higher. (4/n)

In fact when we look at Dale and Krueger data, yes indeed, as @nadeemhaque notes, what they observe is students that go to Harvard indeed earn much more. It seems indeed there is this huge college premium. But is it really the college premium of Harvard/elite schools? (5/n)

The problem is that students that go to elite schools also differ in their family background, socioeconomic status, childhood nutrition and so forth as noted by @nidkirm. We are not comparing apples to oranges! (6/n)

Suppose Jane got in Harvard but decided to stay close to home when her mother got into an accident and attend a local State School instead, while Jennifer got into Harvard and was lucky enough to not have this misfortune. (7/n)

Now comparing hundreds of Janes with Jennifers are apples to apples comparison. They are both similar yet one did not attend Harvard due to chance. (8/n)

What happens when we make applies-to-apples comparison in Dale and Krueger, the huge college premium of elite schools, disappears! (9/n)

Reference: Dale, S.B. and Krueger, A.B., 2002. Estimating the payoff to attending a more selective college: An application of selection on observables and unobservables. The Quarterly Journal of Economics, 117(4), pp.1491-1527. End Thread (n/n)