Best of #econtwitter - Week of January 17, 2021
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.
A lot of new papers this week!
Paper summary threads
New paper: How much do cities really differ in productivity?
@zivoren & I find: up to 75% of the naive between-city (within-industry) variance of average TFP reflect plant-level idiosyncracies ("granularity")---rather than systematic place effects.
PDF:
eml.berkeley.edu/~schoefer/scho…
^see also: on granular spatial economics
^paper here. File under, “measuring the correct concept of wealth is hard”
Results show that trade generates about 11% higher gains for those at the 90th percentile than those at the median
But openness also increased the fall in inequality caused by domestic shocks. Without trade, the 90-10 ratio of earnings would have decreased by 17% instead of 32%.
Striking chart showing population shifts away from densest areas - paper by @AlthoffLukas et al. (h/t @alex_doser ) fpeckert.me/AEGW.pdf
Our new paper on judicial bias in India’s lower courts, using new data from 6 million criminal court cases. ⬇️
Do Muslims or women get better outcomes when their cases are heard by Muslim / female judges?
Surprise: No evidence on bias! 🧵 1/11
devdatalab.org/judicial-bias
2/ We know people make mistakes in health insurance choice. We know less about *why* and what policies matter for the “behavioral”. We think default rules matter a lot, just as in retirement savings and organ donation. Our focus: Low-income subsidy program in Medicare Part D.
^thread from another coauthor here
@SunMengyi , Jainabou Danfa, and myself took a close look at the effects of this policy change. We found that review scores given to the most prestigious 1/3 of authors decreased after transition to double-blind
3/ Authors show that marginal ADHD diagnoses cause additional diagnoses of ADHD to be made in younger siblings of patient and even cousins. This could be family history effect in doctors office or parents seeking care bc of the index marginal diagnosis. Either path interesting.
"We conclude that outsourcing leads to some, though modest, positive productivity effects, and that these gains benefit firm owners and deteriorate workers’ labor market prospects."
Adrien Bilal @AdrienBilal
. @nico_de_roux, @meslava_twit, @FrancoTabaresSa, and I have a new working paper on production-function estimation. Might seem a little inside baseball, so let me try to explain what we do and why we think it matters. 1/ nber.org/papers/w28323
NBER @nberpubs
Job market paper from @katempenn — with new construction, rich people move in without poorer people moving out, a happy outcome for everyone.
(I call it “gentrification without displacement” in TRITDH)
dropbox.com/s/oplls6utgf7z…
More (a lot this week): effect of AI on job postings; school management and teaching effectiveness; math olympiad medalists and immigration; effects of GDPR; coal in India; history dependence of city location; household deleveraging and slow recoveries; international trade and earnings inequality; training vs. outsourcing; education in Africa; lit review on macro wage compression; polio eradication in Africa; effect of voting on beliefs
Public goods
(Trying this again with direct links that work!)
The professional development sessions organized at #SEA2020 by @AEACSWEP @AEACSMGEP & @LGBTQ_Econ are now online!
#EconTwitter
Links follow:
Interesting discussions
But there may be a publication bias concern with these prior studies. In the absence of publication bias, we should see no correlation between effect sizes and standard errors.
Here is a graph of the effect magnitudes and errors in prior studies: 8/11
^ie:
Interesting: in this literature, there is a linear relationship between regression coefficients and standard errors. The slope is close to 2, because coefficient/standard error>2 is required to get published. Hence, the smaller and noiser the dataset, the bigger the findings
Paul Novosad @paulnovosad
Awesome set of suggestions to make online teaching more lively, engaging and restoring human interaction somewhat
Alice Evans @_alice_evans
Professional navel-gazing
I often say to grad students: a reasonable thing to aim for in grad school is ~2-3 essentially complete and posted papers, alongside your JMP. This often looks like your JMP, plus your 2nd-yr paper, a paper from RA'ing for a faculty, or from working with classmates
^long advice thread — summarized by Shosh Vasserman as “work on writing multiple papers” — that provoked a lot of reactions. (Including the usual kind of backlash.) Context from Zhang on his objective function. Some followups:
This thread has led to some disagreement. Does saying this stress some students out? Probably, but it does students no favors to obscure the standard. Don’t hide the hidden curriculum. https://t.co/Bguy6PCiLq
Anthony Lee Zhang @AnthonyLeeZhang
@ashdgandhi I’ve found the opposite to be true in practice- typically candidates with “worse” pedigree have more papers, potentially to make up the skepticism gap. It’s not unreasonable to think candidates can produce more than one paper. Aren’t dissertations ~3 papers on average?
Good thread, which links to some data on number of papers typically completed by grad students.
healy.econ.ohio-state.edu/papers/PJsAdvi…
Shengwu Li @ShengwuLi
^Healy summarizes his linked slides/advice here
