After telling Claude it was the Thanksgiving Day game between the Lions and the Bears, I asked it which team it thought had made the gaffe (based on its general knowledge):

I was tipped of by "have you made good profits?"

Seems to do a good job of prioritizing comments into "major" and "minor" etc.

You do need to activate the "image PDFs" feature in Claude which may not be enabled by default.

Surprised to not see @lexfridman.bsky.social on this list

That's a good idea! I didn't think I ended up actually using any proposed sentences wholesale, but it would be best to make 100% sure.

This is probably the most ambitious paper I will ever be involved in. It requires multiple recent econometric innovations to even enable estimation. Coding was done in
Stata, Matlab, and Julia.

We welcome your feedback and comments. Thank you for your attention!

Full information also decreases the proportion of people who attend college, as well as narrowing the gap in college graduation rates between students from above/below-median income households.

We simulate a counterfactual scenario where students have full information about their abilities by the end of high school.

full info increases:
• college graduation rates
• the college wage premium
• ability sorting

full info decreases dropout by about 20% (6pp)

We find that students learn from both school and work outcomes, and that abilities are indeed correlated across different college options and labor market sectors.

Learning affects college enrollment, dropout, re-entry ("stopout"), and sorting in the workforce.

You can think of the model as an extension of Keane & Wolpin (1997), where there are now more schooling choices and agents have uncertain abilities in each of five sectors. Moreover, the abilities are allowed to be correlated.

We estimate the model on men in the NLSY97.

We develop and estimate a dynamic discrete choice model where students are uncertain about their abilities in college majors (2yr/science/humanities) and labor market occupations (blue/white collar).

Students update their beliefs as new information arrives in the form of college grades / wages