Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
professional age forecaster Profile picture
professional age forecaster
@agoodmanbacon
Black Lives Matter. Senior Economist, Federal Reserve Bank of Minneapolis he/him/his
Aug 23, 2022 • 9 tweets • 5 min read
So @DrJPCunningham and I updated our paper on Legal Services, switching the whole thing from TWFE to Callaway/Sant’Anna (@pedrohcgs)

Learned a ton while doing this and I think our methods section turned out great and help some folks:

goodman-bacon.com/pdfs/cgb_lsp.p… The key thing is that we need to use control strategies to find a good comparison group. Our (staggered) treatment went mainly to big cities and we chose to use an *untreated* comparison group.

So we faced all the control options available in Sant’Anna/Zhao (@ZhaoBean).
Apr 5, 2022 • 5 tweets • 3 min read
Macroeconomists are doing some wicked data entry recently. Three examples I’ve seen probably in the last month alone: @A_Fieldhouse, Howard, Koch, and Munro digitize UI claims by state and month to extend state level u/e measures back to 1947!

econstor.eu/bitstream/1041… Image
Nov 1, 2021 • 12 tweets • 5 min read
I recently dug into @TymonSloczynski's awesome and mind-bending paper, "Interpreting OLS Estimands When Treatment Effects Are Heterogeneous: Smaller Groups Get Larger Weights

people.brandeis.edu/~tslocz/Sloczy…

I had to make some graphs to figure it out and I do love to share a good graph! the main result is that if you run this:
y = AX + BD + e
you get average of the Average Treatment Effect on the Treated (ATT) and on the Untreated (ATU)

BUT

The weight on ATT is *inversely* proportional to the size of the treatment group. Wait WHAT?
Oct 26, 2020 • 14 tweets • 8 min read
Excited to *finally* share an old paper w/ @seth_freedman and @nehamm!

goodman-bacon.com/pdfs/fgbh_medi…

Some folks say "Medicaid kills people" b/c they compare recipients to non-recipients while controlling for "a lot" of things. This is a bad idea.

Medicaid does not kill you. @Avik, for example, made this argument forcefully in 2011 (and it got lots of political traction) based on a paper in @AnnalsofSurgery. That paper is called "Primary Payer Status *Affects* Mortality for Major Surgical Operations" (emphasis mine):
ncbi.nlm.nih.gov/pmc/articles/P…
Sep 9, 2019 • 19 tweets • 10 min read
Excited to share my paper with @DrJPCunningham on welfare participation and family structure in the 1960s (topics that arouse virtually no controversy whatsoever…)

We document an important role for the War on Poverty's Legal Services Program (LSP):

nber.org/papers/w26238 Family structure was pretty stable from at least 1880-1960. About 10% of mothers were unmarried, usually b/c they were widows.

Then an enormous change. Maybe the biggest demographic phenomenon of the 20th century aside from the baby boom. Single parenthood has almost quadrupled.
Aug 25, 2019 • 12 tweets • 6 min read
In light of this question, I thought I'd do a little thread on purely practical event-study stuff.

No theory, just a bunch of pictures of how I make event-studies and what you get when you make different mistakes/choices.

I do my event-studies totally by hand. I choose the omitted category, I deal with endpoints myself, and I make the dummies ahead of time.

Here's how:
Apr 29, 2019 • 8 tweets • 2 min read
I want to talk about a simple but illuminating thing: "distribution regression".

You have a design and a continuous outcome, but get noisy junk for the average?

Make dummies, 1{y>x}, move x through support of y, estimate effects on each dummy. Traces out effects on F(y). 1/4 Chernozhukov, Fernandez-Val, and Melly (2013) discuss it in all the metricsy detail, but I have found actually using it to be very straightforward.

Here it is for my long-run Medicaid paper. I get a noisy-ish thing for avg earnings, but really striking effects for the dist. 2/4
Sep 10, 2018 • 30 tweets • 14 min read
Does your diff-in-diff treatment variable turn on at different times? Of course! Most of them do!

nber.org/papers/w25018

Very happy to share my paper showing how DD with variation in treatment timing works and what that means in theory and practice. 1/29 So…

Maybe there are untreated units and units treated at different times.

Maybe you ONLY have variation in timing (ie. everyone winds up treated).

Essentially your data look like this: 2/29
Aug 14, 2018 • 14 tweets • 7 min read
OK, a thread of weird graphics from historical reports that I like. Not (Medicare) sure (could) what ("cover") the (us) subtext (all) here (you know) could (like) be (a blanket?!).
Jul 10, 2018 • 6 tweets • 2 min read
I'm so excited to have ordered a bunch of Frederick Wiseman documentaries today (for work!).

The most immediately relevant one is "Welfare": 3 hours of unadorned footage from a NYC welfare office in 1975. It's not for everyone, but it is totally for me

cinetrafic.fr/video/extrait-… "Hospital" (1970) gives the same treatment to Metropolitan Center Hospital in East Harlem:

Jul 1, 2018 • 4 tweets • 2 min read
pilot: welcome to seattle, keep your seatbelts fastened as we pull into the gate.
my 3 yo: daddy, I'm having a pee accident. it was so fun to take kids out there, though. per @dynarski, we just spent three straight days prowling around in tide pools and it was perfect and engaging for everyone.
Jun 18, 2018 • 5 tweets • 2 min read
This paper provides really good evidence on a really terrible phenomenon: ICE activity (stretching back to 2008) even reduces hispanic citizen participation in transfer programs out of fear that their non-citizen family will be deported.

nber.org/papers/w24731.… The roll-out of Secure Communities, which massively increased sharing of fingerprint information between local law enforcement, the FBI, and Homeland Security. It led to sharp reductions in food stamp participation for hispanic adults, but not white or black adults:
Jun 9, 2018 • 5 tweets • 3 min read
The other #ASHEcon18 session in which I'm involved as more than just a discussant is, surprise, about diff-in-diff.
ashecon.confex.com/ashecon/2018/w… I'll talk about my work on DD with variation in timing, which I'm eager to share very soon. I do a lot of stuff and am really excited to get comments from Michael Anderson, whose work continually introduces me to technical things I'd otherwise never tackle.
Jun 6, 2018 • 4 tweets • 3 min read
I'm happy to have put together this ASHEcon session with Seth Freedman.

The papers are methods-related, but neither totally new topics nor totally new methods so we settled on "research practices". First, @ajhollingsworth tells us that one does not, but more importantly CANNOT, detect effects of the ACA on mortality. We simply don't have the power. @smilleralert discusses.
May 16, 2018 • 4 tweets • 1 min read
I agree. Talking about blue-->red transfers serves a political purpose, but redistributing to poor *people* will probably involve a lot of that.
Apr 17, 2018 • 5 tweets • 3 min read
Even John Snow (1855) had a reviewer number 2:

academic.oup.com/ije/issue/42/6 Here are some agitated zingers
Feb 7, 2018 • 9 tweets • 4 min read
So Seema Verma took to the WaPo to defend work requirements in Medicaid and some argument seemed...off

washingtonpost.com/opinions/makin… No one gets left behind (unless they miss a nominal payment)