Best of #econtwitter - Week of May 1, 2022 [3/3]
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.
This is part three of three.
Paper summary threads
Intuition: * All admissible decision rules, under a constraint on identification strength, turns out to be (Lipschitz) continuous in the GMM objective function * The argmin is not continuous in the GMM objective function * Hence GMM is not admissible □
How much value-added tax revenue is lost to non-compliance? This paper estimates gaps, exploiting differences in VAT at international borders. Gaps are highest in countries with poor governance. Financial development and tougher administration help improve compliance.
. provides best causal evidence I know for cognitive impairment from noise: • Worker productivity down in randomized noise experiment • It's cognitive, not just effort • Noise actually isn't salient to workers; their WTP is low joshuatdean.com/wp-content/upl…
@arpitrage This study out of Kenya shows results about the effects of noise even without spatial sorting? https://t.co/Yu3Hldqq7T
Alon (they/them) @alon_levy
New paper argues that Zoom conversations might be worse for coming up with new ideas, but equally good for deciding b/t ideas. Interestingly, they track eye movement w/ in-person vs Zoom meetings and find that gazing off abstractly into space was correlated with idea generation.
Delighted to share the first peer-reviewed article from my PhD research, where I estimate the effect of university on political values, just published in . sciencedirect.com/science/articl… Quick🧵to tell you all about it!
Many people in our field are debating the recently-released results of READI Chicago, a well-known anti-violence program that provides cognitive behavioral therapy and subsidized employment to those at the highest risk for violence. For my take, read on.
Thoughtful post on READI Chicago findings & how they got reported in the press. I'm one of the READI evaluators, and I don't agree with everything here. I see REALLY hard trade offs in how fast one should report results, and how to help policymakers weigh ambiguous results. BUT
News reports on the READI Chicago study are misleading and are overly positive. As I argue in this post, the study itself makes it easy for casual readers to come to the same conclusions, part of a broader trend of what I call Press Release Research. https://t.co/fs3R6Ygqi5
Jacob Kaplan @JacobKaplanCrim
🚨 Public good alert 🚨 I made a website covering all of the resources I used when I was applying to pre-docs + some additional advice from going through the process. Hope this can help people considering these jobs in the future! zahrathabet.com/resources.html…
Academics often forget that titles are very precious space. I keep seeing papers where it feels like the title is an afterthought, thus wasting a huge opportunity for communication with potential readers. Making your title maximally informative is hard, but worth it.
The more generalizable version: Use notation that tells the reader what's a function of what.
@DerekRury Off the top of my head: - Outcome subscripts should give the max unique identifier of obs. E.g. y_it for a panel of individuals over time - All regressor subscripts should be in terms of the outcome subscript. E.g. x_i, w_it, q_s(i) where s is a f'n giving i's state
Peter Hull @instrumenthull
You're invited to a job talk and have to give a mock lecture? A thread 👇of do's and don'ts based on three hiring committees I was part of in the last months. Caveat: Business/economics at a small German university. Not sure how generally valid. 1/8
Another year of my data science class has finished, and so now it's time to talk about this year's run of "the Hunt for the Worst Data Visualization." This year had some great submissions from the class: