Christopher Walters
@c_r_walt
Followers
3K
Following
136
Statuses
42
Professor @berkeleyecon studying labor/education. Affiliate @nberpubs @JPAL_NA @BlueprintMIT. Coeditor AEJ:Applied @AEAjournals.
Berkeley, CA
Joined November 2017
RT @aganimian: Our study w/@karthik_econ @c_r_walt on how to expand state capacity to improve child development in India is now out on @JPo…
0
22
0
It's not too late to register for this week's Mixtape session on empirical Bayes methods -- sign up at the link below to join! Lectures and live coding labs MWF, 6-9pm eastern
@c_r_walt's Empirical Bayes course is live!! Including a lab using labor market discrimination data from their QJE article Last chance to sign-up for tonight: Course material:
0
1
11
Topics will include teacher/school value-added, employer-level labor market discrimination, individualized treatment effect predictions, connections to machine learning and multiple testing, and more
0
0
6
RT @causalinf: I’ll be posting from time to time about this coming workshops. This one we just finalized and it’s ready. It’s on empirical…
0
1
0
RT @instrumenthull: Our new handbook chapter w/ @metrics52 & @c_r_walt is now available online! Check out the latest in estimating institut…
0
63
0
@mister_kessler @nberpubs This result is in lemma 1 of the first version of my paper on job-level discrimination with Pat Kline:
1
0
1
RT @DNealEconUofC: I am teaching a labor economics course 7/3 to 7/7 at the Barcelona School of Economics. The course should be beneficial…
0
14
0
RT @instrumenthull: Want to learn the latest in estimating school quality w/ quasi-experimental data? Or the various IV tips & tricks @Blue…
0
104
0
@jiafengchen42 @instrumenthull I think that would do it, but this seems like a reductio ad absurdum against the negative weight criterion. This loss says implement the policy as long as *someone* potentially gains at least X — that’s all you know for sure if you have a conv avg TE of X w/arbitrary weights
2
0
1
@jiafengchen42 Agree. I have never seen an explicit defense of such a loss function, so I’m perplexed that convexity/positive weights has become such a focus recently
1
0
6
@jiafengchen42 I agree in general, though I think (negative weights -> bad worst case risk) is only true if your loss function cares specifically about sign errors for some reason. Otherwise it’s not clear why negative weights are such a big deal
1
0
6
RT @instrumenthull: Michal Kolesár, Chris Walters (@c_r_walt), and I were very happy to have been invited by the Scandinavian Journal of Ec…
0
64
0