Haven’t seen much discussion of this paper https://scholar.harvard.edu/files/jroth/files/roth_jmp_honestparalleltrends_main.pdf#page49 coauthored with Rambachan(whose 3rd year PhD at Stanford I think?).

In one sense this paper seems really important for applied researchers finding another way to to estimate the true cost of not having the ideal parallel trends. On the other hand it is so uninspiring, these are supposed to be the future of metric theory in our field and the scope seems so small.

I’ve also been going through the supplementary paper (https://scholar.harvard.edu/files/jroth/files/roth_jmp_honestparalleltrends_supp.pdf) tonight and the proofs are so long and dry with basically 0 elegance which for me always raises alarm bells as that’s what theory ppl normally use to avoid too much scrutiny. I haven’t found any mistakes yet but there is more restrictiveness nested into the estimator than just glancing at the paper would have you assume...

So yh am I reading this wrong, is this paper brilliant and I’m too stupid or old to realise, what do ppl think of it?