So happy to see in print!🧵Main q: how does adjusting the size of local gov't units affect service delivery? Ex-ante, not obvious! Smaller population=easier to cater to local prefs, monitor leaders, etc but may be more susceptible to elite capture, miss out on econ of scale, etc
I was curious about how “top 5” pubs are distributed across US econ depts so scraped some data for the last 10 years (but note data caveats below). Thought others might also be interested- some was expected, some surprises. Fuller data-
We (
@USC_Econ
) will be on the job market this year with at least 2 open positions: we encourage candidates in all fields (incl seasoned juniors), although labor/econ history/pol econ are prioritized. Link to apply:
#econtwitter
@JOE_listings
We (
@USC_Econ
) are hiring this year for 3 positions: focuses are metrics+applied micro. We are also interested in current APs--feel free to message me if you're interested in moving/want to learn more. Nice dept + living in LA is pretty great, esp as winter arrives...🌞🏄♀️🌴🧗♀️
We're hiring! Here are the ads for two of our open positions that we're hoping to fill this year: join us!
1) econometrics
2) development/political economy - all applied micro welcome
We (
@USC_Econ
) are a little late in posting, but have a post-doc position open in any applied micro field. Please pass it on to any interested JMCs! Living in LA feels better and better as winter arrives back east... 🌞🌴🏖️
#econtwitter
@JOE_listings
Excited to see in print! Reposting our old explainer thread. Mortality for those involved in the criminal-justice system is much higher than for the general population. Is that because incarceration causes poor health, or because this population faces so much disadvantage?
Mortality risk is lower while incarcerated, primarily due to lower risk of overdose or murder. In the July issue, by Sam Norris (
@sambnorris
), Matt Pecenco (
@mattpecenco
) and Jeff Weaver (Jeff_Weaver_)
Call for papers: PacDev conference is at
@USC_Econ
in LA on March 16 2019 and deadline to submit is in less than a month (Dec 14) - please spread the word and submit yourself! Link to submit is open now:
Call for papers: PacDev development economics conference at
@USC_Econ
in LA on Mar 16 2019 - please spread the word and submit! Open for submissions now, deadline to submit is Dec 14 –
#pacdev
This is the key figure, plotting survival likelihood by year since sentencing across these groups. The incarcerated are *less* likely to die in the years after sentencing than the non-incarcerated, and the longer the sentence, the higher the likelihood of survival.
Thanks
@jenniferdoleac
for the chance to talk about some of the ongoing research on parental incarceration, including our paper (with
@sambnorris
& Matt Pecenco)...
tbd whether I'll be able to listen to myself talk for an hour without wincing myself into oblivion.
Episode 7 is up:
@jenniferdoleac
talks with
@Jeff_Weaver_
about the spillover effects of incarceration on the children and siblings of those incarcerated.
To JMCs stressing about the lack of JOE postings, we only got hiring authorization in the last week, so other schools may similarly be late in posting ads 🤞. Good luck, and let me know if you have good candidates we should be looking at!
Amazing fix to the awful display of dropbox pdfs: change "?dl=0" to "?raw=1" at the end of the link. Fixed on my site, and it looks great on both mobile and PC displays. May all the world (of academic economists) know of its glory!
PSA: If you share link to a PDF file on Dropbox, delete the "dl=0" (after the "?" sign) and type "raw=1". This way it opens like a normal PDF link (without the Dropbox crap around it) and is easier to read/scroll/skip.
You're welcome.
Really important piece on cross-state occupational licensing and how it limits the potential of teletherapy. I'm biased since it is by my wife, but would think it is great regardless! May be some good econ papers that could be written on this, PhD students should check it out!
🚨🚨Call for development papers: only a few days left to submit to PacDev, deadline is Dec 14 at 5PM PST! The conference will be held at
@USC_Econ
in LA on Mar 16 2019. Please spread the word, the submission link is here:
Main surprise was that the differences are WAY bigger than I thought, especially at the very top: going down by 10 schools -> # of pubs drops by ~half or more. Would this be similar for non-econ journals? If no, what frac of this is explained by more movement in econ?
If someone wants to fix some of these issues (esp the department roster issue), feel free to take my code (in R) to get the articles/author names… thanks to my RA, ChatGPT, for writing most of it!
@USC_Econ
is rapidly expanding with 15(!) new TT or tenured hires since 2016, 2 people on our hiring committee are so recent that they aren’t even on our website yet! (should update that...) Living in LA is pretty great, increasingly so as winter arrives... 🌴🏖️🧗🏄🌄🎞️
Why is this happening? Three main things. The largest part of the effect comes from murders, which are almost eliminated in prison. When not incarcerated, this population has 16x higher risk of being murdered than the general population, so this matters a lot.
Second, overdose deaths are reduced by more than 50%. We think this is due to some combo of lower availability of drugs and better monitoring/quicker treatment for overdose. This matters even more after ~2010, when the opioid epidemic took off
Third, deaths from natural causes (eg heart disease) decline during incarceration. Prison is the only place that Americans are constitutionally guaranteed healthcare. Even low-quality prison care can be better than nothing, which is often true for this population on the outside.
This reduction is a big deal because overdoses are the single largest cause of death for this population. But we don’t see any evidence this reduction is sustained after release, suggesting it’s not because of in-prison drug treatment.
We use court and death certificate data from Ohio to answer this question. We study how mortality evolves over the years after sentencing, when sentence length is determined
What’s going on? This fig compares mortality for the incarcerated relative to the similar non-incarcerated defendants before/after release. During incarceration, mortality risk is half that of the non-incarcerated (baseline 600 per 100k). But after release, mortality is the same.
Our takeaway: we do not see this as pro-prison. These results reflect the dangers of life outside prison for this pop and highlights possible non-carceral interventions. Better healthcare access, safer neighborhoods, and more access to Nalaxone could have big returns
And here is all the articles/authors in case anyone wants to put together the list of faculty affiliations from dept websites but doesn't want to rescrape it/is too shy to email me -- would love to see the correct absolute numbers without the repec issue!
… but I am biased since my department (USC) does well (
#17
). For that, it would be better to include non-top 5 journals using some weighting, but still, another metric to consider when deciding where to apply for PhD programs, etc.
Final big finding: not much evidence for effects of past incarceration on mortality risk after release. For the 15 years following release, the formerly-incarcerated have the same mortality rate as the not-incarcerated, leading to long-term increases in survival.
@tomer_stern
Someone who wants to spend more time on it definitely could but would need to pull in some extra data to get that (and would want to use dept rosters rather than repec).... this exercise was pretty much the limit of my interest in the topic!
Ex post realized this complements other measures of dept research production. It probably measures *recent* productivity better than citation-based measures (dominated by articles from 10+ years ago) and opinion-based rankings like USNews (subjective/lagging)
In this paper (), we use another, potentially better empirical approach (judge IV) and have very similar findings -- reductions in mortality from incarceration.
Data caveats: 1) I wanted current affiliation rather than at time of publication since people move a lot (esp in response to top pubs). Affiliations come from matching author names to department rosters on Repec, but Repec rosters are usually ~70% complete/not fully up to date
@EmilyNix100
I wonder if the editors may be constrained by the number of pages allocated to each journal, where each is supposed to be roughly the same number of pages, but there are more submissions to AEJ:A/Macro/Pol than Micro? That looks consistent with the reports for each journal.
Updated version here that uses department websites rather than Repec to construct rosters. Mostly the same (corr of 0.91 in counts, 0.95 in rank) but large increases for Princeton, Duke, and Wisconsin, who had uniquely bad Repec rosters:
We use village-level data on services like ration cards,
sanitation, workfare, education, many others. Across the board, smaller local gov'ts->greater service access in both short and long-run as well as better targeting of services. Lots of data here that others may find useful
@JlibDoesEcon
@ErzoLuttmer
To quash escalation of conflict (though not the discussion on the value/use cases for this data) -- I think the critique is reasonable and written in a totally respectful way so I take no offense at it!
2) Bigger departments do better on this since they have more authors. 3) Counts are at the author-article level, so a 4-author paper with 3 authors at the same uni will count as 3 rather than 1 or ¾. 4) I only looked at US econ departments. 5) Forthcoming articles aren’t included
This is also just one set of journals – I also got data from some other “general interest” econ journals, and the linked spreadsheet has that and more schools if you want to dig into it. Also feel free to take my code (below) to pick other journals as you wish
@JaminSpeer
@footeball45
haha, good memory Jamin. The main issue was that people no longer remember the Bush administration, so I got a lot of emails asking me if there was a typo in my title...
@DanielYiXu
@bassi_vittorio
@DanielYiXu
I redid it using faculty rosters from dept websites rather than Repec. Princeton, Duke, and Wisconsin had atypically bad Repec rosters so jump considerably but the rest are mostly the same (new/old versions have a corr of 0.91):
Could prison be good for your health? Incarceration has been linked to infectious diseases, mental illness, cancer, and violence. But new research suggests it can extend some people’s lifespans. Bapu investigates the paradox of prison time.
@RWacziarg
Yep, just econ. Totally agree that it would be nice to expand to b-schools (and outside the US), though it may be that the repec rosters are worse for those, so could be even noisier
Lots more in the paper–an alternative RD approach, tests of many mechanisms, robustness checks – hope you check it out! Ungated version here: , short write-up here: . Big thanks to the editors/refs, who really improved the paper
@cubic_logic
Thanks for taking a look! There are actually around 10,000 data points on either side that have been put into 20 bins so the graph is readable, not n=40. We then conduct statistical tests of the likelihood of seeing a jump like this by chance and find that it is extremely low.
@LiglyCnsrvatari
The reason is that due to random assignment, the “treated” individuals will have the same characteristics on average as “control” individuals prior to the intervention. If we see differences emerge, we can attribute this to the treatment.
@lore_magno
@lore_magno
I updated it using faculty rosters from dept websites. Mostly didn't change (corr of 0.91 in counts, 0.95 in ranks) aside from Wisconsin, Duke, and P'ton change a lot b/c their Repec rosters were atypically incomplete. updated here:
@PriscilliaHunt
@jenniferdoleac
@ProbCausation
In general so many factors are at work with parental incar - almost any mechanism you can think of operates in at least some cases, our estimates aggregate over all of the mechs. That's why i think the qual work is so valuable -- it goes into the factors we aggregate over.
@LiglyCnsrvatari
In general, it isn't necessary to include baseline measures like IQ with studies involving random assignment such as ours (assuming the sample size is large enough).
If missingness is similar across unis, this should be roughly right in relative terms but not absolute levels. Hopefully not a big deal, but not sure. Using dept websites would be better but take WAY more time – I didn’t want to spend more than an hour on this.
@O_Goss_
the original version depended on the affiliations listed on repec so it depends on if kellog profs are listed there w econ. I have redone it here with the faculty listed on the dept's website, which maybe does that less. It is pretty similar (corr of 0.91)
👇👇 Highly recommend this post-doc opportunity for job market candidates interested in development. Great folks to work with and San Diego/the devo community at UCSD is wonderful!
More opportunities to work with
@Prof_Karthik_M
@PaulFNiehaus
and me, and enjoy the camaraderie and weather @ UCSDEcon: we're looking for a post-doc to join the team. The irreplaceable
@Jeff_Weaver_
can hopefully attest to this being a great experience...
@LiglyCnsrvatari
In our study, the children whose parents are randomly assigned stricter judges (and thus more likely to have incarcerated) will on average have the same baseline IQ at the time of assignment. So it is unnecessary to add it as a control.
@DanielYiXu
@bassi_vittorio
A PhD student with 3-4 hours could make more accurate ranking using faculty lists on dept websites matched to the names to the data in my previous reply... but I didn't want to spend that much time on this, I was more curious about the rate of decline (which is prob approx right)
@Ingar30
Others are welcome to do that! I just did this out of personal curiosity, I am not planning to do any research on this topic or keep updating this as a rankings system.
@lore_magno
Yeah, it is def a problem; from quick looks at other depts, I think that Wisconsin's measurement error is particularly bad, but hoping a grad student/someone with more time will scrape dept websites to fix this. Happy to send anyone all the publications/author names to match to!
@DBaqaee
I broke down and ended up getting the faculty rosters from dept websites myself. See link for updated data using that instead of Repec. Turned out to not matter much -- correlation of 0.91 with the old counts though.
@ryanbedwards
@USC_Econ
Oh no! Send it to my usc email address, and I'll see if i can get it added in -- we really should have had a midnight deadline.
@LiglyCnsrvatari
One certainly could study the role of IQ at age 7 on later criminal activity (though IQ is a product of environment during early childhood so when talking about the role of IQ, you’re really also talking about early childhood environment). But that isn’t what our study is about.
@glenwhitman
Yep, totally agree -- my interest in doing this was actually not about editorial bias (though it seems like that is what most readers were interested in) but on the distribution after any sorting. But I see how both are interesting!
@DanielYiXu
@bassi_vittorio
The absolute #'s are definitely wrong since the repec dept rosters are incomplete (but the relative levels are more right). but if anyone wants to calculate the correct # for their dept, the list of papers/authors are here: