Best of #econtwitter - Week of February 20, 2022 [2/3]
Welcome readers old and new to this week’s edition of Best of Econtwitter. Thanks to those sharing suggestions, over email or on Twitter @just_economics.
Paper summary threads
New results from a 20-year follow-up on Progresa. Children whose parents received cash transfers in Mexico 20 years ago earn 15% more than those who didn't povertyactionlab.org/sites/default/…
Incredibly striking similarity between US and France in the relationship between parental income and college attendance - and surprising given the massive differences in cost of attending college between the two countries. (new work by & Sebastien Grobon) [1/3]
Also new on https://t.co/sInc4TxNw1: Unequal access to higher education based on parental income: evidence from France, by C. Bonneau & S. Grobon Bottom line: link btw parental income & access to higher education is almost as strong in France as in the US https://t.co/OMRtHuwcGp https://t.co/eXVyabbypg
Thomas Piketty @PikettyLeMonde
A thread on BUE & BLUE🔥 Gauss-Markov condition: 1) y=Xβ+ε 2) E[ε|X]=0 3) Cov(ε|X)=σ^2Σ Standard GM: 4) Σ=I The GM thm shows that OLS/GLS is BL(inear)UE. Hansen (’20) shows it holds for all unbiased est (inc. nonlinear) w/ an elegant proof (tilted density + Cramer-Rao) 1/n
The OLS estimator is BUE, on top of being BLUE, under classic Gauss-Markov conditions — check out @BruceEHansen’s new amazing #econometrica paper. Textbooks require some updating @jmwooldridge? Question: how much a BUE can depart from linearity? https://t.co/Ci4J5S0vuw
Giuseppe Cavaliere @CavaliereGiu
Not exactly. I like Bruce's approach in this paper and it yields nice insights. But in twitter and private exchanges last week, and what I've learned since, it seems that the class of estimators in play in Theorem 5 include only estimators that are linear in Y.
It is time for @jmwooldridge to rewrite his books. https://t.co/ltnejZHyzn
Thanks ! A thread on this short paper 🧵 (no econometric toys are harmed in the writing of this paper)
Very cool new paper by @jiafengchen42 that gives a justification for synthetic control methods He shows that when treatment timing is random and you have many time periods, SC has close to optimal regret when an adversary chooses the potential outcomes https://t.co/lD6rpERhfM https://t.co/0rx8L7dIzb
Jonathan Roth @jondr44
Happy Black History Month- today I want to talk about protests, and specifically whether protests can work to influence economic redistribution using evidence from our newish paper here:researchgate.net/publication/35… 1/n
In a randomised trial with 18k+ participants with suicidal ideation, "training of dialectical behavior therapy skills (mindfulness, mindfulness of current emotion, opposite action, and paced breathing)" raised the risk of self-harm relative to usual care.
Very cool new tinyurl.com/nberp! Losing a job has very different implications across Europe. For Northern Europeans, 5 years after job displacement, earnings are 10% lower than before, while the loss is 3X larger for Southern Europeans (Italy, Portugal, and Spain). 1/3
The consequences of losing a job across countries using a harmonized research design and social security records from seven nations, from Antoine Bertheau, @eacabbi, Cristina Barcelo, Andreas Gulyas, Stefano Lombardi, and Raffaele Saggio https://t.co/aqPElO80Io https://t.co/0esIS2wgzf
47 journals with explicit short paper options where economists publish their research An annotated list: bit.ly/36qNVfn [And a thread]
Excellent new packages for synthetic control estimation and inference in r, python, and stata arxiv.org/abs/2202.05984 nppackages.github.io/scpi/ The nppackages.github.io nppackages group sets the standard for serious econometrics work paired with good software
New harmonized data set tracking ~80m people for 2-3 quarters; 49 countries across five continents, covering wide range of economic development. Looks like more and more data will be provided online, which seems like a public service worth highlighting:
I find animations useful to teach metrics to undergrads. I believe it improves understanding to see what goes on. Here is a thread with my favorite GIFs, most of them are re-makes of stuff from .
This one I have seen several versions of before. It illustrates attenuation bias with classic measurement error in x.
My latest addition to my GIF collection is this one which illustrates fixed effects by demeaning x and y. It also demonstrates Simpson's paradox + shows that demeaning of y does not change the slope coefficient (cf. regression anatomy vs Frisch-Waugh).
^more in thread