More personal news!
Last week our second born turned 18. The he got his voter registration. Now, today, he committed to
@VassarBaseball
where he will spend the next four years of his life in pursuit of his academic and athletic dreams! ๐
#proudpapa
Your periodic reminder to not ignore endogenous controls because they are not "of interest". The may bias may affect the estimates of the parameter "of interest" even if that covariate is exogenous.
Covering a common theme I like to harp on in econometrics in class today. Related to IV. Say the model is
y = a + b1*x1 + b2*x2 + e,
where x1 and x2 are endogenous, b1 is the "parameter of interest", and z is a valid IV for x1 in the usual sense. You do not have an IV for x2.
Today a current PhD student Iโm advising threw in the towel. I rejoiced!
No, itโs not what you think. He came to the realization that he can get his dissertation done faster in
@Stata
and he has abandoned R.
Sometimes you gotta let โem fail so theyโll come to the light.
Zillowโs home buying business lost them $500,000,000, 25% of their stock value, and 25% of their workforce.
How did this happen to a company with so much data on housing prices?
Bad model evaluation.
Hereโs the fatal error they made that you must avoid when deploying models๐งต
@Catherineoscopy
@HowardAdelman2
@louieeckhardt
Lemme say that obv disc sucks. But the wage gap alone is not evidence something is wrong. Presumably VC should pay what the market dictates. Different disciplines have their own market. You are worth what it costs to replace you if you quit. Economists have many outside options.
For those who don't want to spend the time teaching themselves, there are 6 critical things to know about measurement error.
1. Classical ME in Y leads only to imprecision. Classical ME in X leads to attenuation bias.
2. The vast vast majority of ME is not classical.
Protip for those teaching applied econometrics (as well as both complementing and offending another field) ... poli sci journals have lots of great metrics method articles that are "dumbed down" in the sense that they explain the basics in great detail w/o getting sidetracked w/
Professional news:
I am thrilled to be named the Robert H. and Nancy Dedman Trustee Professor
@SMU
!
Very thankful to all my wonderful colleagues in the
@SMUEconDept
and
@SMU
, as well as the many mentors who helped me over the (gulp!) decades.
Time to keep paying it forward!
๐
I am beyond excited that people seem interested in my econometric factoids! So, here is another episode of "How the Sausage is Made" on IV, following up on my previous thread. This thread contains another reason why (i) IV is finicky and (ii) weak instruments make it worse.
Thanks to all who have followed, replied, and reached out. I love my Twitter community. Donโt know anything more other than the hostages are physically unharmed. I donโt know the congregants, but Rabbi Charlie has been our rabbi for a long, long time and is a great person.
#EconTwitter
After a long hiatus, I decided I cannot stop thinking about econometrics and am announcing the resumption of my blog
#EconometricSausage
. Motivated today by
@jmorenocruz
โค๏ธ!!!
New post is called "Dig a Little Deeper".
Thanks for reading!
#econtwitter
An offer to anyone, but esp PhD students planning on being on the market next year or juniors ... if anyone wants to send me paper drafts for comments, please do! I donโt care how unpolished. Canโt guarantee they will be good comments, but it will help motivate me!
So, I've decided to call my psuedo-blog on
#EconTwitter
about random econometric factoids "How the Sausage is Made". It will be me sharing (hopefully correct) tidbits that I think most applied types are unaware and may be of use. A bit of insight into "how the sausage"
Had a great day yesterday, receiving a wonderful surprise. Flattered to receive this award! Definitely appreciate all the support of my colleagues and
@SMU
during my time here. It has been a wonderful institution to spend my entire career at (not that I'm retiring anytime soon!).
Congrats to our chair, Prof.
@dlmillimet
, who was named the
@SMUDedman
2022 Dedman Family Distinguished Professor! The award recognizes his contributions over the past 2+ decades to
@SMU
in terms of research, teaching, and service.
#PonyUp
We are currently conducting SWAT operations around the 6100 block of Pleasant Run Rd. All residents in the immediate area are being evacuated. Please avoid the area.
The delta method is awesome. It is also used in most
@stata
nonlinear test commands. However, most people - I suspect - donโt understand how it works. Let me tell youโฆ
I agree. I tried for a while when I was young to get into NBER. Went every summer. Presented. Discussant. Never a peep. Iโm not much of a networker. Was very bitter. I let it go eventually. Sort of. Donโt go any more. Donโt really care to. Iโve learned to be be comfortable in my
As NBER recently finished its summer institute and BREAD is announcing its upcoming conference, let's discuss how "insane" it is that these two important institutions in economics do not have open and transparent criteria for membership/affiliation.
I'm afraid that many applied researchers will tune this point out bc they haven't learned DAGs and/or do operator. But the point here is HUGE and can be explained easily to those who use OLS. DO NOT IGNORE THE ENDOGENEITY OF CONTROLS JUST BC THEY ARE NOT YOUR OBJECT OF INTEREST.
At ASSA: Interviewer asked me โWhy should we care about the effects of kids on female labor market outcomes? My wife stayed home with our kids and was happy.โ ๐คฆ๐ปโโ๏ธ
I do remember the other interviewer was equally mortified.
This is all excellent advice. I will reemphasize FINISH PAPERS. Being smart is necessary but not sufficient for tenure. Unfortunately you need to prove it with pubs and you cannot pub w/o submitting. Many smart people have not succeeded in academia (if thatโs your goal) bc
I have been writing academic papers for nearly 25 years. Today I spent 3 hours and revised 3 paragraphs in the introduction of a paper. And I still don't really like it. Writing takes a LOT of time. Don't underestimate it. Don't skimp on it.
@causalinf
@christopherruhm
This is what Manski has been teaching and writing about for decades.
@JohnMullahy
can confirm. In his book he talks about how politicians need to be able design policy based on partially identified parameters but they donโt; they prefer certainty even if certainly wrong.
#EconTwitter
This week's blog post on
#EconometricSausage
is now available.
It is called "Horseshoes and Hand Grenades"
Plus, a little something extra for
@PHuenermund
Enjoy!
Peace. Justice. Health. Some leadership (and elections)! โค
@mcuban
Respectfully disagree on the racist part. Fox can distort views on policies, etc., but cannot distort the words that come directly from Trump. He reveals his racism directly. And if you are ok with it, then you too are racist.
#EconTwitter
I have no idea if this site still works, and if anyone is listening, but ...
@SMUEconDept
will be hiring THREE tenure-track positions this year. Ad coming soon! Reach out to myself or any others to find out how cool we are (well, not in the summer)!
Happy to see
@jondr44
's paper finally out. It's an important reminder that, in any context, if you condition your choice of estimator based on the result of a prior spec test, you end up with a new estimator:
ฮธ_3=I(fail test)*ฮธ_1+I(pass test)*ฮธ_2
with different properties! Poof
#EconTwitter
#Econjobmarket
For transparency, we received 465 job applications for our (one) junior position. We are in the process of contacting individuals chosen for first round interviews. As always, it is fascinating being on this side.
Iโm am pleased to report that I was notified of an editorโs decision on a paper I recently referred for JPE. The letter notes that that JPE is making a concerted effort โto raise the barโ. So, all you jokers that got your papers published there before now dodged a bullet! ๐
Thought of the day: universities feel compelled to makes profs go through hrs of training to teach online but no one ensures profs know how to teach in person.
I know ๐คซ
I have been publishing for a long time. I have been on editorial boards for a long time. I have tenure. There is a pandemic. All that said, I am VERY annoyed (as are my coauthors) how slow some journals are in making a first decision. Just brutal. Do better.
Imputation = I didnโt have any data so I made it up.
Multiple imputation = I made it up lots.
Data augmentation = I didnโt have any data so I made it. Bayesian style.
More personal news!
Last week our second born turned 18. The he got his voter registration. Now, today, he committed to
@VassarBaseball
where he will spend the next four years of his life in pursuit of his academic and athletic dreams! ๐
#proudpapa
Propensity score matching is like "I know a spot," but it only looks like a spot
Instrumental variables is like "I know a spot," but it's always rainy
Partial identification is like "I know a spot," and it's somewhere around here
#EconTwitter
I have a new
@iza_bonn
WP with my colleague Hao Dong. Some might find it shocking, but I really like the paper. A ๐งต
"Embrace the Noise: It Is OK to Ignore Measurement Error in a Covariate, Sometimes"
But, a VERY imp PSA to JMCs IMHO (lots of acronyms!) ... If your JMP uses DID, you best be tooled up on the latest developments, which have come fast and furious. To me, it is not a good signal if you are using DID and there is no mention of all the new work in the last 5 yrs.
Someone told me that we shouldnโt care about the effects of children on female labor market outcomes (my JMP) bc his wife stayed home to raise their kids and she was happy. ๐คท
My Stata-to-Python translation page has been updated to include a few more concepts, like 'tab' and 'bysort'.
Thanks to everyone who provided feedback.
The Department is pleased to announce that Prof. Daniel Millimet (
@dlmillimet
) will serve as the next department chair beginning August 1, 2020. At that time, Prof. Santanu Roy will conclude 6 years at the helm. We owe him a great thanks for his exemplary leadership.
#PonyUp
#EconTwitter
Guess what? It's summer. In an effort to avoid work, I started my econometrics blog, instead of just doing twitter threads!
It is now available (I think!) at
And I posted my twitter threads from earlier in the semester. I hope it's useful!
@jmwooldridge
Exactly. I tell students not to let Stata drop something for you. If it does, figure out why, drop something yourself, and then reestimate the model.
Without inviting snarky comments on IV, does anyone else view the credibility of instruments as non-monotonic in their strength? If the IV is weak, the results are suspect. If the IV is very strong, then it is hard to imagine that itโs not corr w/ the source of endogeneity.