Yagan Hazard @HazardYagan Twitter profile | Pikagi

Pikagi

Yagan Hazard

@HazardYagan

1,023

Followers

2,205

Following

11

Media

910

Statuses

AP @CollegioCA (since Sept. 2023) | PhD @PSEinfo , visited @Brown_Economics | Econometrics & Labor Economics

Paris

Joined September 2019

Don't wanna be here? Send us removal request.

Pinned Tweet

@HazardYagan

Yagan Hazard

2 years

Thanks a lot @instrumenthull ! Sounds like an good time to say a bit more about the paper (written with my amazing co-author @LoeweSimon !). Peter already gave the perfect TL;DR version. Now it's 🧵-time! ⤵️ (headaches free & short, I promise!)

@instrumenthull

Peter Hull

@instrumenthull

2 years

Very interesting JMP from @HazardYagan on improving the precision of IV estimates by predicting and removing non-compliance!

Tweet media one

5

40

182

5

31

101

Last Seen Profiles

@Biffaloe_III

@QemergGlobal

@mdmnn77426743

@Dr_Bonehead

@yncgnr

@Oghene_fegor

@DPAmicrophones

@ABoryas14190

@kitxy29

@PictureHouseUck

@shun_to_saya

@drmsr_dev

@hiroandwife

@radiolewisham

@Isabelltinti

@oliverrr33

@khaleeji2023

@pengen_stw

@ParkJimin_PL

@bokeplokalmalam

@Streamanity

@deflowerdefile

@CardiffWillie

@lissaisalone

@pengen_stw

@codylufccrox

@nawwkcs

@AscensionHelp

@ForumChallenge

@SexyAssVids

@Akizuki_1mat

@CJuntasing

@ShunyuYao12

@CarringtonMich5

@AntClimber

@seven_seven17

@HazardYagan

Yagan Hazard

1 year

Celebrating this here one month late, & still so happy to share that I'll join @CollegioCA next September!🎉 This msg being mainly meant to thank as loudly as I can the awesome pp who supported me and made this happen. @PSEinfo @Brown_Economics @instrumenthull @jondr44 🙏

16

2

122

@HazardYagan

Yagan Hazard

10 months

Looks like I'm PhDone!🥳 So grateful to everybody @PSEinfo for this amazing journey, and ofc to @BehaghelL in particular for his wonderful supervision. (& 🙏 @MGurgand for so many great years since beginning @ENS_ULM ). Looking forward to what's coming next @CollegioCA now! 😁🇮🇹

@PSEinfo

Paris School of Economics

10 months

👨‍🎓 Best wishes to @HazardYagan ( @PSEinfo @ENS_ULM ) who will defend his thesis entitled "Essays on the Econometrics of Treatment Effect Models and Labor Economics" today under the supervision of Luc Behaghel 🍀

Tweet media one

0

1

21

6

0

103

@HazardYagan

Yagan Hazard

10 months

So happy about this! 🥳 Turin is a GREAT city, and @CollegioCA such an amazing place to start this new chapter!

@CollegioCA

Collegio Carlo Alberto

10 months

Excited to welcome a new member in the #CCA Faculty: @HazardYagan has joined CCA as the new Assistant Professor in #economics . Welcome! #CCA #Faculty

0

1

17

2

2

87

@HazardYagan

Yagan Hazard

7 months

🚨We're hiring, and you may have missed our late posting! 🚨 I've been here for only a few months, and yet I can tell you (1) working conditions @CollegioCA are amazing (w/ great colleagues), (2) the offer is *very* competitive in Europe, and (3) Turin is a GREAT city! Join us 😁

@CollegioCA

Collegio Carlo Alberto

7 months

We’re hiring! We’re looking for an Assistant Professor in Economics to join Collegio Carlo Alberto. Applications from female candidates are especially encouraged. Deadline: December 8th, 2023 To apply, please visit: #Economics #cca #EconJobMarket

Tweet media one

1

24

94

1

9

84

@HazardYagan

Yagan Hazard

9 months

Check out @PSEinfo JMC, they're amazing! Don't take my (admittedly biased) word for it ofc, have a look at their really cool research! (And they cover quite a wide range of fields, that's one of the many nice things about PSE!).

@PSEinfo

Paris School of Economics

9 months

Let's meet our candidates for the PSE #JobMarket 2023-24! Each of them will be presented in dedicated posts in the next few weeks. The occasion to discover more about their profiles, research fields and Job Market papers! 🗓 See you this Friday to introduce our first candidate!

Tweet media one

1

14

56

0

0

21

@HazardYagan

Yagan Hazard

3 years

I am really excited (and a bit impressed !) to present our work in such a great session, alongside so many inspiring researchers !

@rolandrathelot

Roland Rathelot

@rolandrathelot

3 years

Interesting in online job search and hiring? Join our #ASSA2022 session Sunday at 10am (EST) *The Automation of Job Search and Matching* with @danielle__li Philippe Kircher @tlebarbanchon Anita Glenny and @hazardyagan ! ,,

Tweet media one

0

5

11

1

0

12

@HazardYagan

Yagan Hazard

1 year

& far too many twitter-less pp to thank. My incredibly kind and supportive advisors Luc Behaghel ( @PSEinfo ) and Xavier D'Haultfoeuille ( @CrestUmr ), the awesome Toru Kitagawa ( @Brown_Economics ), & all my brilliant co-authors/friends/family who've put up w/ me all this time. 🙏

1

0

9

@HazardYagan

Yagan Hazard

1 year

This. And (most importantly) all the answers!!

@danascoot

Dana Scott

1 year

how do people deal with emotional ups + downs of research. like one minute everything is awesome and exciting and you love your job, and the next minute your variation is useless and you're having an existential crisis about your agenda. How do you people live like this!?

83

59

965

1

0

9

@HazardYagan

Yagan Hazard

7 months

Yet another amazing and insightful piece of work by the one and only @amorygethin 🤩. Crazy data work, to answer such an important q°: taking into account (the evolution of) consumption of public goods turns out to be key to understand the long run trends in global poverty!

@amorygethin

Amory Gethin

7 months

Global poverty is commonly measured by counting the number of people whose consumption falls below a given threshold. This approach overlooks an enormous component of people’s economic well-being: public goods. 🧵

21

399

2K

1

0

8

@HazardYagan

Yagan Hazard

7 months

I didn't know about @CollegioCA before getting on the job market last year, and now I couldn't be happier with my choice to come here! And come on, Italy is the real deal 👌

0

0

7

@HazardYagan

Yagan Hazard

2 years

So far so good. But when it comes to your own experiment, you can sometimes get depressed by how weak a shifter Z is for D... bc that increases variance A LOT! (Why can't people do what they're told to?🥲) That's where this paper tries to make you 😀 again: all is not lost!

Tweet media one

1

0

5

@HazardYagan

Yagan Hazard

2 years

Some very interesting & inspiring papers on the subject were written while we were working on this w/ @LoeweSimon e.g., work by @jannspiess @stephencoussens , @borusyak @instrumenthull , @nickchk , @jiafengkevinc , @jmwooldridge . We discuss connection w/ some of these in the paper

1

0

5

@HazardYagan

Yagan Hazard

1 year

So true, congrats @LeroutierM !! 😁🎉

@MortenStostad

Morten N. Støstad

1 year

A great hire by @CrestUmr -- Marion is extremely talented and a great person!

1

1

15

1

0

6

@HazardYagan

Yagan Hazard

2 years

@paulgp My intuition is: focusing on a single treatment alternative (say, arm B) provides 0 insurance against the case when arm A actually dominates (and vice versa). So ex-ante, max regret is higher than if I had "diversified" (Manski's words again). Real-world ex. attached.

Tweet media one

1

0

5

@HazardYagan

Yagan Hazard

2 years

Last but not least, there're (far too many) people to thank for feedback & crucial support on this project. Too many people from @PSEinfo and @CrestUmr who're not on twitter (yet), & @Brown_Economics , among which @jondr44 @instrumenthull @ksoonu (Toru's not on twitter yet!) 🙏

0

1

3

@HazardYagan

Yagan Hazard

2 years

@paulgp Your 1st sentence set the bar too high for me to answer, so here's C. Manski's answer from his 2007 text. I think the intuition for "randomization" rules (fractional rules here) under partial id + minimax regret criterion is the same when we focus on identification alone.

Tweet media one

1

0

5

@HazardYagan

Yagan Hazard

2 years

@toniwhited @instrumenthull Almost incredible how easy it looks (though figuring out the exact form of influence functions might be tricky sometimes I guess...)

3

0

4

@HazardYagan

Yagan Hazard

2 years

@pmichaillat Yet again another great public good, thank you so much ! :)

0

0

4

@HazardYagan

Yagan Hazard

2 years

Should I infer that Corsican people once caused trouble to LSE HR staff? Really? 🤔😅

Tweet media one

0

0

4

@HazardYagan

Yagan Hazard

2 years

@causalinf @thinkific Ok thanks again in any case for all this work! I'll be more reactive for the next @instrumenthull (and other) seminars (hope this will happen again!) ;)

1

0

4

@HazardYagan

Yagan Hazard

2 years

@instrumenthull @pedrohcgs @jmb112485 @paulgp Oh I see. But how will a policy-maker make use of this precisely estimated effect that represents the effect on a "convex weighted average of people" in a way? I'm guess policy-makers would (+/- implicitly) make some restrictions on TE heterogeneity to make use of this info

2

0

4

@HazardYagan

Yagan Hazard

1 year

@BehaghelL on Twitter, very cool vibes! 👌

@BehaghelL

Behaghel Luc

1 year

Launching a blog -- and hoping to make it a useful and lively place. Disclaimer: new to blogs and to twitter, please be patient

0

10

34

0

0

3

@HazardYagan

Yagan Hazard

3 years

@instrumenthull L. Behaghel et al, "Private and Public Provision of Counseling to Job Seekers: Evidence from a Large Controlled Experiment". Two treatment arms (Public/Private counselling), and randomization strata. Data available online I think (or possible to ask to Luc), and large.

0

0

3

@HazardYagan

Yagan Hazard

2 years

@arindube @paulgp @instrumenthull @pedrohcgs @jmb112485 Yes there is a minimax estimation way (but this is not the OLS weighting scheme anymore): @CdeChaisemartin has a paper on this (and this is in line with Tim Armstrong and Michal Kolesar work on ATE estimation under unconfoundedness)

0

0

3

@HazardYagan

Yagan Hazard

2 years

There's a lot more in the paper, such as very nice Monte-Carlo simulations (see pic below) comparing TS estimator w/ TSLS & other estimators proposed in the lit., and two empirical apps. Feel free to drop me an e-mail if you'd like to discuss about applying TS to you own exp.!

Tweet media one

1

0

2

@HazardYagan

Yagan Hazard

4 years

Fascinating summary of recent research on the causes of the (extremely) unequal returns to wealth over the past decades. In french, but based on Bach et al. (2020, AER) for the most curious of us !

@ULojkine

Ulysse Lojkine

4 years

Pourquoi le rendement sur le patrimoine des plus riches est-il près du double de celui de la classe moyenne ? @laurent_bach ( @essec ) mène l'enquête à partir de données suédoises extraordinairement détaillées. À lire dans le @Grand_Continent .

3

37

123

0

0

3

@HazardYagan

Yagan Hazard

2 years

BUT clever guys can come up w/ randomly allocated instrument (Z) that shift D for some indiv. Ex: Z = compulsory schooling laws (AngristKrueger91) Then we can estimate the effect of D on Y for indiv. whose D is shifted by Z ("compliers"). If D and Z binary, good old TSLS gives:

Tweet media one

1

0

2

@HazardYagan

Yagan Hazard

2 years

@toniwhited Neat! Thanks a lot Toni, after seeing your comment on @instrumenthull poll I was trying to figure out exactly what you meant ; now that's crystal clear!

1

0

3

@HazardYagan

Yagan Hazard

2 years

@instrumenthull @YoungStatS2 @borusyak @paulgp @stephencoussens It's been in my calendar for months, looking forward to it! Thanks for putting this together :)

0

0

3

@HazardYagan

Yagan Hazard

2 years

At this point, you might still be suspicious. How am I still estimating the LATE (as plain vanilla 2SLS does) despite this selection step? Couldn't we mistakenly drop groups w/ some compliers (in which case we wouldn't estimate the LATE anymore)? Thanks for asking! 🙃

1

0

2

@HazardYagan

Yagan Hazard

2 years

[Hang on, we're getting to the crux & the end 💪] We'll distinguish two cases to study the asymptotic properties of TS estimator. (1) No (erroneous) dropping of compliant groups asymptotically ("standard" asymp.) (2) Some dropping of such groups even asymp. (weak-IV-like asymp.)

Tweet media one

Tweet media two

1

0

2

@HazardYagan

Yagan Hazard

1 year

Looks very useful to get started with moment inequality models!

@IllanesGaston

Gaston Illanes

1 year

New paper! @IvanCanay , @amilkant and I have written a practitioner’s guide for models defined by moment inequalities, link below. We hope that it’ll be useful for those of you navigating the literature on how to estimate these models. Comments welcome!

1

29

155

0

0

3

@HazardYagan

Yagan Hazard

2 years

IV is VERY popular in applied work. A good reason is that it's sometimes the only tool we have to estimate a policy impact when said policy can't be randomly allocated. (+ Peter's lobbying 👌) Take the effect of schooling (D) on income (Y). We'll never allocate D randomly...

1

0

2

@HazardYagan

Yagan Hazard

2 years

Thank you so much @CdeChaisemartin ! I wrote a bit more on the content of the paper here:

@CdeChaisemartin

Clément de Chaisemartin

@CdeChaisemartin

2 years

I agree! Departments interested in hiring an applied econometrician that also does applied work, and conversely, should check out @HazardYagan

0

3

13

0

0

2

@HazardYagan

Yagan Hazard

2 years

@pedrohcgs @jmb112485 @paulgp @instrumenthull Ah, I thought the point of this paper was about what causal parameter I can estimate with the lowest variance. But (at least without any restriction on TE heterogeneity), it cannot be used to say anything about bias-variance trade-off for a given parameter (say ATE), right?

1

0

3

@HazardYagan

Yagan Hazard

2 years

@instrumenthull @restatjournal That's great news ! 🎉 (Congrats!)

0

0

2

@HazardYagan

Yagan Hazard

1 year

And I forgot to say how grateful I am to all the brilliant researchers I met over the past few months. The kind of support and kindness they brought to the (exhausting) market experience made it so much easier to enjoy!

0

0

2

@HazardYagan

Yagan Hazard

2 years

@paulgp I think that when planners have absolutely no idea on which treatment dominates, they end up randomizing instead of assigning in some deterministic way (e.g., true to extent in France for the allocation of caseworkers to job seekers, instead of trying to match based on obs.).

2

0

2

@HazardYagan

Yagan Hazard

1 year

@deaneckles @MITSloan Very nice syllabus, thanks for sharing! Do you have any class material as well? (Ofc it looks like the reading list is already providing far enough material though)

1

0

2

@HazardYagan

Yagan Hazard

3 years

@pmichaillat Thank you so much, that's amazing ! :)

0

0

2

@HazardYagan

Yagan Hazard

9 months

@LukasLehner_ @Unibocconi @DondenaCentre @ZParolin Congrats Lukas!! And I'm really happy to hear you'll be in Northern Italy now, we should meet again soon then :)

1

0

2

@HazardYagan

Yagan Hazard

9 months

@ULojkine Bravo !! 😁

0

0

1

@HazardYagan

Yagan Hazard

2 years

@instrumenthull @kylefbutts Oh yes @kylefbutts your beamer template is 🔥. I am using it all the time now (and was wondering how to appropriately pay tribute to you, apart from recommending it to everyone!)

0

0

2

@HazardYagan

Yagan Hazard

1 year

Now if that's half as good as Clément's lecture notes are (and of course it is!)... What a treat!! :-)

@CdeChaisemartin

Clément de Chaisemartin

@CdeChaisemartin

1 year

If you are an economist and French, you have to write a book. So we started writing one on, guess what, TWFE and DID :-) First chapters available here: . Comments welcome! Also, we are looking for a publisher: reach out if would be a good fit for your press

5

56

345

0

0

2

@HazardYagan

Yagan Hazard

1 year

@CyrilVerluise @a_bergeaud Congrats!! Really cool to see this kind of projets advertised more :)

0

0

2

@HazardYagan

Yagan Hazard

2 years

@paulgp Oh yes, I find it a bit crazy too. I find the maximin criterion (which is what MMR eventually is if we restrict ourselves to "singleton"/non-randomized rules) more realistic. Is it what you have in mind as well?

0

0

2

@HazardYagan

Yagan Hazard

2 years

@causalinf That's really great, thank you so much for this! I was wondering : suppose I fail to register for a session before it is sold out, but I'd still be happy to pay for access to the (very high quality) content. Will you have some plan for this? :)

1

0

2

@HazardYagan

Yagan Hazard

1 year

@DEF_TorVergata @Caro_cly Yeees let's go @Caro_cly !! 💪

0

0

2

@HazardYagan

Yagan Hazard

2 years

What are the general properties of this TS estimator? ✅ Consistent for the LATE under "standard" asymptotics, w/ asymptotic variance ≤ TSLS ✅ These properties remain under "weak-IV-like" asymptotics IF we restrict treatment effect heterogeneity (as desc. in pic).

Tweet media one

1

0

1

@HazardYagan

Yagan Hazard

2 years

@NLonguetMarx @amorygethin My thoughts exactly, I'd be curious to get a sense of how much Zemmour affected the distribution of far-right voting along the income distribution.

0

0

1

@HazardYagan

Yagan Hazard

2 years

@instrumenthull @pedrohcgs @jmb112485 @paulgp Exactly, that's precisely what I would love to see!

0

0

2

@HazardYagan

Yagan Hazard

2 years

@jondr44 @jiafengkevinc Given @jiafengkevinc activity on Twitter, tricking him into solving problems doesn't sound extremly hard 😅

0

0

2

@HazardYagan

Yagan Hazard

11 months

@borusyak Probably a bit off, as it seems you're more about a good example of a theoretical justification behind the formulation of a Roy model... But what about the Generalized Roy Model (Eisenhauer et al 2015, JPE) and it allowing for identification of cost and benefit separately ?

1

0

2

@HazardYagan

Yagan Hazard

2 years

But hopefully we're on the right track! We can actually fix this selection procedure to get a valid one (using data-splitting and cross-fitting)! That's the Test&Select (TS) procedure we propose and study in this paper. ✅ TS yields valid inference where its naïve version failed

Tweet media one

Tweet media two

1

0

1

@HazardYagan

Yagan Hazard

2 years

No pb: you could TEST in your sample whether the share of "compliers" in each group is significantly ≠ from 0, and decide which group to drop accordingly. ⛔️ This kind of naïvely implemented pre-test will introduce too much bias, leading to invalid inference.

Tweet media one

Tweet media two

1

0

1

@HazardYagan

Yagan Hazard

2 years

@paulgp I hope I'll have the opportunity to teach a course that great in the future (in which case your material would be a huge source of inspiration -- it already is as a TA!). I can't imagine how much I'd get to learn along the way! Really inspiring!

1

0

2

@HazardYagan

Yagan Hazard

1 year

@BunelSimon @CollegioCA @PSEinfo @Brown_Economics @instrumenthull @jondr44 Merci Simon ! Les cafés offerts à la banque ont joué à coup sûr ;)

0

0

1

@HazardYagan

Yagan Hazard

2 years

Calling G=1 white indiv., G=0 ethnic minorities, the pic below gives intuition for why getting rid of G=0 obs. sounds like a good idea. The tricky part is: most of the time, you wouldn't know ex-ante which group to drop and which group to keep, right?

Tweet media one

1

0

0

@HazardYagan

Yagan Hazard

9 months

@jondr44 Well, Omar Sy's Lupin Season 3 is out today (if you're after some French vibes) :). Wishing you a speed recovery!

0

0

1

@HazardYagan

Yagan Hazard

1 year

@josueawonon @CollegioCA @PSEinfo @Brown_Economics @instrumenthull @jondr44 Merci Josué !! J'espère que tout se passe bien pour toi, un petit bout de chemin parcouru depuis le moment où tu accueillais ma cohorte de master à PSE ahah ;)

0

0

1

@HazardYagan

Yagan Hazard

9 months

@SimonBriole @CEEMMontpellier Félicitations Simon!!

1

0

1

@HazardYagan

Yagan Hazard

2 years

@bonneau_cecile @Caro_cly Merci pour le fil, c'est... Assez effrayant !!

0

0

1

@HazardYagan

Yagan Hazard

2 years

@PossebomVitor @paulgp @instrumenthull @pedrohcgs @jmb112485 @bruno_ferman I like split sample, but then part of the gains in variance from the estimation strategy would be altered by the use of only half of the data for estimation, right? (Well, no free lunch, so that might be a good way still of course!)

0

0

1

@HazardYagan

Yagan Hazard

10 months

@economeager @Vassar That's insane...

0

0

1

@HazardYagan

Yagan Hazard

1 year

@Apoorva__Lal Congrats!!

0

0

1

@HazardYagan

Yagan Hazard

1 year

@cottet_sophie @PSEinfo @KatarzynaBilic1 @giulia_giupponi @jarkko_harju @attilalindner @AntoineBozio Bravo Sophie !! 🎉🥳

0

0

1

@HazardYagan

Yagan Hazard

3 years

@paulgp Thanks a lot, super cool deck !

0

0

1

@HazardYagan

Yagan Hazard

1 year

@aletondini @CollegioCA @PSEinfo Thanks Alessandro, I hope we can see each other again soon! :)

0

0

1

@HazardYagan

Yagan Hazard

4 years

@jasonmlindo Though you would usually include CI in your RD graphs, so that even "not so compelling graphs" in this gif would look much better if they correspond to an RD with t-stat > 1.96. Right ?

0

0

1

@HazardYagan

Yagan Hazard

2 years

@edwardhkennedy Oh that's gold, combined with your previous notes and slides! Thank you so much for all of this !! 🙏

0

0

1

@HazardYagan

Yagan Hazard

10 months

@LukasLehner_ @PSEinfo @BehaghelL @MGurgand @ENS_ULM @CollegioCA Thank you Lukas! 😊 Hope everything is going well in Boston!

0

0

1

@HazardYagan

Yagan Hazard

2 years

@LukasLehner_ @HorowitzFdn @maxkasy @INETOxford @DSPI_Oxford Congrats Lukas, very well deserved!!

1

0

1

@HazardYagan

Yagan Hazard

10 months

@jondr44 @CrestUmr @rolandrathelot @Brown_Economics @ksoonu Thank you Jon!!

0

0

1

@HazardYagan

Yagan Hazard

3 years

@paulgp (I should say this syllabus is already an very cool public good of course 😄)

0

0

1

@HazardYagan

Yagan Hazard

1 year

@steve_tadelis Thanks Steve! :)

0

0

1

@HazardYagan

Yagan Hazard

8 months

@paulgp Aaaaah that's what I'm waiting for 🤩 Will that be posted online as well? Definitely yea

0

0

1

@HazardYagan

Yagan Hazard

1 year

@instrumenthull @CollegioCA @PSEinfo @Brown_Economics @jondr44 Thank you so much Peter!

0

0

1

@HazardYagan

Yagan Hazard

1 year

@WeidnerEcon Thanks for the nice and wise words Martin, really cool to have people like you sharing their 2c here :)

0

0

1

@HazardYagan

Yagan Hazard

1 year

@EmilySilcock1 @CollegioCA @PSEinfo @Brown_Economics @instrumenthull @jondr44 😁🎉 looking forward to celebrating this with you and @LoeweSimon !

0

0

1

@HazardYagan

Yagan Hazard

1 year

@instrumenthull Yes indeed! :)

0

0

1

@HazardYagan

Yagan Hazard

2 years

@paulgp @karlrohe Sorry to pop up here: what would be the best reference paper on Puffer transformation? I know about the application of the method in the "Smart pooling and pruning to select policies", but I am not sure where I should go for an analysis of Puffer transformation per se. Thanks!

1

0

1

@HazardYagan

Yagan Hazard

2 years

@_criminologist @instrumenthull Hi Alex! Great q°, as it's a key ≠ btw. our paper and other alt. Here we're trying to get a very specific weighting scheme in which groups w/ 0 compliance get 0 weight, and groups w/ ≠0 compliance get standard weight.

2

0

1

@HazardYagan

Yagan Hazard

10 months

@bleistein_linus @PSEinfo Thank you so much for coming and for the nice words @bleistein_linus ! 🙏

0

0

1

@HazardYagan

Yagan Hazard

1 year

@myoungbrun @UniLeipzig @IWH_Halle @_MartinHaensel Congrats Marie !! 🎉

0

0

1

@HazardYagan

Yagan Hazard

8 months

@LukasLehner_ @ESRC @DSPI_Oxford @INETOxford @oxsocsci @UniofOxford @StEdmundHall @sehlibrary @DondenaCentre @MIT_IWER Congrats Lukas, that's so well deserved !!

1

0

1

@HazardYagan

Yagan Hazard

1 year

@ElioNimier @CornellEcon @ChicagoBooth @rolandrathelot @DSraer @dthesmar @CrestUmr Congrats !! 🎉🎉

1

0

1

@HazardYagan

Yagan Hazard

2 years

@_criminologist @instrumenthull So... if we make no mistakes in identifying 0 compliance groups, we're in the clear! But if we do make mistakes, then you might consider that we're doing some kind of "trimming" based on compliance, and changing the estimand. We consider this case (see )

@HazardYagan

Yagan Hazard

2 years

[Hang on, we're getting to the crux & the end 💪] We'll distinguish two cases to study the asymptotic properties of TS estimator. (1) No (erroneous) dropping of compliant groups asymptotically ("standard" asymp.) (2) Some dropping of such groups even asymp. (weak-IV-like asymp.)

Tweet media one

Tweet media two

1

0

2

1

0

1

@HazardYagan

Yagan Hazard

8 months

@borusyak At first I thought that was a tweet about how YOU managed to be so productive and was like "wow, that's something I want to know" 😂

0

0

1

@HazardYagan

Yagan Hazard

10 months

Last but not least, so many people to thank from @CrestUmr (in particular twitter-less Xavier D'Haultfoeuille for his supervision and @rolandrathelot ) and @Brown_Economics (twitter-less Toru Kitagawa, @jondr44 , @ksoonu ) for welcoming and supporting me so generously!!

1

0

1

@HazardYagan

Yagan Hazard

3 years

@ccanonne_ @dzackgarza Sounds like this one could be added IS a very light footnote... In this case, why not including it ? :) Anyway, nice and easy way of proving it indeed !

0

0

0

@HazardYagan

Yagan Hazard

2 years

@jiafengchen42 Reading this is encouraging (though you are most likely an incredibly fast learner 😂) ! Thanks for the good references on the topic ! :)

1

0

1

@HazardYagan

Yagan Hazard

1 year

@bleistein_linus That's what our beers are for Linus, come back soon!!

1

0

1

@HazardYagan

Yagan Hazard

1 year

@Lea_Bousleiman @LKYSch @NUSingapore @nberpubs Félicitations !! 👏🎉

0

0

1

@HazardYagan

Yagan Hazard

2 years

@jmwooldridge @Brown_Economics @instrumenthull @jondr44 It was so nice getting the opportunity to chat with you today, thanks again!!

0

0

1

@HazardYagan

Yagan Hazard

2 years

@toniwhited @instrumenthull Thank you so much Toni!

0

0

1

@HazardYagan

Yagan Hazard

3 years

@SaraSgnr @UvA_Amsterdam @Sandra_McNally @dany_bahar Bravo, c'est génial ! 👏😄

0

0

1

@HazardYagan

Yagan Hazard

2 years

@Basucally @instrumenthull Sounds very interesting indeed! Thanks for highlighting it, I'll look into it :)

0

0

1

@HazardYagan

Yagan Hazard

2 years

@cameron_pfiffer Thank you! So cool that you made all of this freely available! Sounds great, I'll definitely start here when I find the time to actually learn Julia... (You Julia folks are doing a great job at making it it look even more attractive day after day!)

0

0

1

@HazardYagan

Yagan Hazard

1 year

@steve_tadelis That's really cool to read this when about to start a tenure-track. I did not realize how much I needed to hear that from people as successful as you are, thanks so much for sharing this :)

1

0

1