Best of #econtwitter - Week of August 27, 2023: paper summaries
Welcome readers old and new to this week’s edition of Best of Econtwitter. Please submit suggestions — very much including your own work! — over email or on Twitter @just_economics.
A lot of Content this week!
^TLDR: “For canonical diff-in-diff, level differences in pre-period are inconsequential. But it is important to think about whether level differences might be due to some unobservables across the two conditions that would respond to treatment differently (non-additively)”
^this paper got a lot of attention (a lot of Georgists on twitter; this is not a complaint) but not full summary 🥲
^(“navel gazing” is not derogatory here!)
^“Interesting to contrast this with industry, where IV is essentially never used, at least to my knowledge. Also see the responses below to my post from a few years back about what causal inference methods are used in data science.”
^my perception is that this is much more the case for the natural sciences. Additional discussion here
^2018 SSC post on this paper; my takeaway not having investigated closely has been that there can be “true placebo effects”, but you have to be careful to disentangle regression from the mean:
Also, claims about power:
More paper summaries
^thread with charts here