I made (at least) 5 research-management mistakes as a young academic. You should avoid these.
1/ You should drop everything possible when you get an R&R and turn that right away.
Economics PhD programs should have a required descriptive statistics class.
Not for learning to make pretty graphs. No theory.
Just to learn facts that could get students’ creative juices thinking about big problems.
Also so we don’t lose sight of the forest for the trees.
4 great tips for young academics I've gotten but regret not following enough:
1/Tom Sargent: Take 1 class per year even when ur a prof. Offsets the depreciation of the human capital you accumulated during grad school.
Long term benefits in productivity >> short term time cost.
How valuable is causal inference?
Economists & statisticians embody high human capital and spend a ton of time showing (policy) X causes Y. What is the rate of return on this effort?
I just learned that Mick Jagger got an undergraduate degree in economics from LSE. How is there not a Mick Jagger Professor of Economics at LSE? If there is one, how is this not the most coveted professorship in economics.
2/ Work on fewer projects.
I invested in too many projects. My biggest fear was having no papers to work on or in the pipeline. But that meant I was making smaller contributions because I never had a clear plate and time to think deeply.
Why have economics papers gotten so long over time (compared to say physics or comp sci or stats or biology papers)?
I do law and also economics. When I started 20 years ago, Econ profs used to mock law profs for writing 80pp papers. Now they all do it!
4/ Quit projects earlier.
Borrow an idea from SV: fail early and often. Check prospects for success early. (Steve Levitt told me: if you can't make a good Fig 1 with raw data, you won't convince a lot of ppl.) Invest a lot only in projects that show prospects for success.
I do a lot of interdisciplinary work. Here are 4 hard lessons I've learned. They may help you be more successful (than me) if you want to work across fields.
When I was in grad school
@UChi_Economics
20 yrs ago, I didn't learn what it means to be a successful academic. 🧵
I was "taught" the goal is to get a top 5 (or a Nobel
😂).
I didn't learn how to get citations, be an influential academic or production fn for getting a prize.
5/ Perfection is the enemy of the good (and total cites/influence).
People cite new ideas, not just the best implementation of them. Know when to cut your investment in the n-th robustness test and just submit to a lower journal.
There are many times where I waited to perfect an idea and someone else published something similar.
Being the first person to say something in a reasonable way is better for influence than only saying things that are perfectly implemented.
How long should academic papers be?
The hard sciences have short papers. The social sciences, especially economics, have long papers.
I think there is a good argument that longer papers are less productive, for both authors' careers & the development of fields.
Research involves a *lot* of admin costs that reduce scientific productivity. Here are 8 ways to reduce this (especially in economics), with little cost to journals or grantors. We should pressure organization to make these changes....
Announcing a Twitter habit: if you tweet out a thread explaining your Econ paper, I am likely to Re-tweet it. I learn a ton by seeing papers explained in threads. I assume preferences of other economists/legal scholars are similar.
3/ When choosing projects, think about the opportunity cost of your time.
Corollary of the last point. I thought: would this get published? I should have asked: what other project could you work on that would do better?
Many journals in economics require exclusive submission: a submitted paper not be under consideration elsewhere.
Given the long review times at econ journals, this delays publication of research. Given fixed tenure clocks, it creates large career risk for junior economists....
Each week a paper is not published is a week it can't get cited.
Ppl blame slow refs & editors. But as an editor
@JLE
for >> a decade (& as an author) I know lags before submitting a revision seems almost as long as editorial delays.
2/Emir Kamenica: Read at least the abstract (or intro) for every paper in your field's top journals, even if unrelated to your field.
General knowledge will make you appear informed to colleagues & may spark ideas. Cost is v. low.
*Prioritizing districts in India for COVID assistance*
We've been working on ranking districts for different aid organizations trying to send PPE and O2 to India. Here are our top 25 based on future risk of COVID. They're useful for things like PPE.
Not new news, but still shocking:
The West played an important role in pressuring India to engage in mass sterilization to control population in the 1970s. It wasn’t India acting all on its own.
From
@slatestarcodex
Economics has a very slow production process that slows innovation. As evidence, has anyone looked at ave time from first posting of a working paper to date of publication? In midst of lit review now, and 6 yrs (!) is very common. That's almost the length of time to tenure!
A common problem I hear from applicants to PhD programs is where to apply and where to attend. Most students have selected a program (e.g., econ) and have a reach and some safeties. Let me offer 2 tips.
I'm teaching microeconomics this quarter & frame the class as policy-relevant by discussing different methods of allocating scarce resources.
The purpose is to stop students from immediately going from mkt failure --> govt intervention. Instead...
*** Accepted paper announcement ***
My paper w
@r_sabareesh
on COVID mortality in India has been accepted at J Devt Economics.
We use
@CMIE
's CPHS data set to estimate mortality from deaths on their household rosters. Highlights below...
@ChinmayTumbe
@countthedead
My lessons:
1/ Academic arbitrage has high returns: take an idea from another field, apply it to your own.
N Nunn brought cultural evo to econ, D Almond developmental constraints to econ, R Posner econ to law. Etc.
Other fields have good ideas that you can apply to yours.
3/Occasionally write lit reviews & descriptive papers. They'll make sure you are filling in holes in the lit so your papers have more value. Descriptive papers have lower citation variance, higher return than analytic ones. Won't get you tenure, but will help citation count.
I want
@nytimes
to interview ppl who went to grad school w Daron Acemoglu and write their experience with him in class.
@arpitrage
@causalinf
@KhoaVuUmn
They Tried to Tackle Derrick Henry in High School. It Didn’t Go Well.
Could displacement of Brahmins from governance in North Indian under the Mughals *partly* explain low development in North India & inequality b/w North & South India today?
I am an economist, not historian, & this is conjecture, but let me lay out & critique the causal chain.
Super interesting…by Chinoy
@DrNathanNunn
Sequeira
@S_Stantcheva
A subset of ppl have a culture of zero sum thinking. It is a function of your ancestry. It is correlated to race, location, immigrant history and political preference.
Interesting paper from
@KrusellPer
@ToshiMukoyama
Sahin and Smith
What is the welfare consequence eliminating business cycles? How does eliminating cycles affect inequality? 1/n
I've been thinking about the economics of misinformation. Here are a few simple models of misinfo & the solutions they do & do not imply. A thread...
TL;DR: The usual solutions (censoring misinfo) will not work or may backfire.
@cblatts
@tylercowen
@balajis
@slatestarcodex
I have a new publication (w Kosuke Imai & Zhichao Jiang of
@Harvard
) in JASA on the identification of spillover effects in RCTs with two-stages of randomization.
Causal Inference with Interference and Noncompliance in Two-Stage Randomized Experiments*
Two conversations this week got me to think differently. They end w/ requests for recommended reading.
TL;DR:
1/Economics needs to be as empirically rigorous about welfare analysis of politics as markets.
2/Economics must supplement causal ev w/ non-causal ev to be impactful.
1/ Journals should adopt a common manuscript format for submissions (borrowing from the common app for college applications). They should specialize formatting AFTER a paper is accepted. Researchers should spend more time researching, less time reformatting for submission.
A puzzle from Indian marriage markets: why are age gaps b/w brides & grooms falling?
Indian fertility is falling, which should be yield higher gaps as high sex ratios (M/F) push males look for ever younger females to wed.
But NFHS shows age gaps falling (Das et al 2013)age :
We just fostered a dog. While investigating how best to train the dog, I realized ... we need
@ProfEmilyOster
to get a dog and write a book about what the literature on dog training, diet and health actually says.
@stacylrosenbaum
TL;DR: The policy & business does not seem to value causation as much as academic economists do. It suggests that economists need to either:
-convince ppl on the value of investing a lot of effort on causation
-work on reducing the costs of inferring causation
Pro tip: After being rejected at 1 journal and submitting to a 2nd journal, make sure you change the name of the journal in the cover letter to the 2nd journal. 🙄
Stork throws one of its babies out of the nest 😬
This isn't uncommon in storks: They have more offspring than they can care for then get rid of the weakest hatchling.
"Brood reduction" is found in many birds, but usually it's siblings that do the deed.
We posted a new pre-print: Seroprevalence in Tamil Nadu through India's 2 COVID waves
Takeaways:
1/Seroprevalence was 31.5% after wave 1 (11/2020), 67% post wave 2 (7/2021)
2/Antibody decay after natural infection: sero fell ~30% between rounds.
1/N
4/Give ppl unsolicited comments on their working papers. Ref reports are anonymous. Cmts at workshops aren't individually credited. But cmts via email are often acknowledged. Ppl read that acknowledgments fn.
This is a first-order problem for the US.
Economic growth comes from productivity improvements, esp if fertility declines & you believe we're approaching environmental limits of capital growth.
Trade helps, but it's not like we're ready to drop "Make in America" regs.
So
In 2017, the USA welcomed 4292 published research scientists on net; China welcomed 116; & non-US OECD nations lost 2927.
By 2021, the USA saw 4 scientists leave (again, on net); the non-US OECD welcomed 1792; and China welcomed 2408.
This is a self-inflicted disaster.
Warning: Doing too much aside from research has costs. So pick and choose which of these pieces of advice to follow. You probably don't have enough time in the day to do all of these. (And you should have a life outside academia.) Pick the tips that are lowest cost to you.
Thread: sample size calculations by economists.
Economists, esp in dev and health econ, are increasingly doing sample size calcs. It seems weird to do from an economics of info perspective.
Instead of calculating minimum sample size, we should report minimum detectable effect.
@jonatanpallesen
I am not sure I agree. There are situations with placebo effects in well designed settings. On design, I would suggest this paper (by me) in J Political Economy. It stresses the need to define the effect and shows how varying the share randomized to treatment can identify it.
I find discussions of herd immunity thresholds frustrating. Herd immunity (which I'll define as a current reproductive rate < 1) depends on policy & behavior. It is NOT an absolute number. THREAD 1/17
I'm excited to see our Karnataka sero-prevalence study out in
@JAMA_current
today. This is part of a longer chain of surveys we've done to track and understand COVID in India to help policymakers. THREAD
Both Europe and India had a form of caste: social groups were endogamous (married only their own). Why did
#caste
give way in Europe but not India?
Ketkar (1914): It is because Indians married at a younger age...
How do I reconcile
#1
&
#3
? Using ideas from other fields to write in your own = good. Using your ideas to write in other field = hard. So...
3/ Don't write in other fields w/o a member of the other field.
Editing/revising a paper with a 3+ person research team can be really hard.
You get conflicting suggestions. It takes a long time.
With experience I have some general suggestions to improve the process. Esp for junior scholars who, honestly, do most of the editing.
170 docs were responsible for 30% of all synthetic opioid prescriptions
From the new
@evanhughes
book the Hard Sell on John Kapoor and Insys. Quote from
@davidenrich
review in
@nytimes
.
So many questions…
I have a new paper on inequality in India during the pandemic (with
@arpitrage
& Bartek Woda).
We find that, while poverty rise during the pandemic, inequality surprisingly fell.
1/N
The COVID-19 pandemic in India was associated with a rise in poverty but a surprising reduction in inequality, which was mediated by falling business income and labor demand of the rich, from
@arpitrage
,
@anup_malani
, and Bartosz Woda
I worry falling fertility is going to be one of the top crises of this century.
I wonder how long before immigration politics changes from countries like the US stopping ppl from coming in to countries like China & India stopping ppl from leaving?
Question about the future of economics: Arguably the biggest social cost we’ve experienced in the last 50 years is COVID. Two key economic aspects are the human response to the pandemic (activity levels, vax rates) & shortcoming in the govt response (drug approval, testing).
Some universities (not naming names) are adjusting to omicron by
1/ delaying the start of classes
2/ going remote for the first few weeks of classes
I am not sure these measures will help control omicron.
Teaching microeconomics today & explained Friedman point that econ, unlike eg biology, seeks a parsimonious predictive model rather than a descriptively accurate one.
But raises a q: as cost of info & processing power falls, should economics use a more complex base model?
When doing research to making policy recommendations, should we take take all policy as a control variable, or should we treat some as given and some as a lever we can control? A 🧵...
Let me frame the q and then give 3 examples.
Today is Ramanujan’s birthday. His story captivated me as a child and I named my first son after him.
Here is Hardy’s commemoration of him from 1937. I found 2 things notable in it.
For me, the difference in quality of Ukraine war analysis between spending 15 min on (a)
@Twitter
and (b) legacy media is very large. (Twitter is higher.)
It makes me think that if you set up filters and like/retweet habits well, Twitter's algo's can deliver high quality info.
Some economists have dinner & come up with important, new paper ideas. Sadly, thats not
@cblatts
& me. Instead we came up with 2 new cocktail recipes. 1/3
@GYamey
A reverse-trolley problem for economists (I'd call it a Mario problem). Suppose the pandemic hits and you have the budget to do one of the things below. Which would you do?
1/ Vaccine that saves 1 million lives
2/ Social program that increases fertility by 1 million lives
How would I teach the course? I’d have different faculty come in each week and ask them to describe the part of the economy they are specialize in. No theory, just facts and narrative. I might even say should each teach 2 sessions, and only 1 should relate to their research.
CAN YOU HELP SLUMS WITH HANDWASHING?
This is an urgent request form NGOs working to stop COVID in Indian slums. Details of problem in THREAD. Summary: we need solution that slums can implement themselves w/ reused water. Please retweet to increase eyeballs. 1/
@vandanagoyal01
Been thinking a lot about Justice O'Connor. She was remarkable, for many reasons. I was lucky enough to have clerked for her in OT 2001. I wanted to share a few memories and thoughts. 🧵
Selfishly, I favor at least a few top schools having a model where they let a lot of people in (many w/o funding) but allow them to stay if they pass.
It helps people (like me) who came to economics late.
It can solve the pre-doc problem. Ppl can choose to skip the pre-doc and
Suppose Harvard College took the old Chicago PhD approach : accept 100 times more kids, do selective exams at end of year 1. Is this worse than:"admit entitled few, have them study 10h per week"?.
Very honored to be included on this list, along with my good friend and collaborator
@profmohanan
and the excellent
@MenonBioPhysics
and
@karthik_econ
, among many others way more qualified than me.
📣 We are excited to announce the publication of Apolitical's 100 Most Influential Academics in Government list!
Read on to see who public servants want to recognise for contributing innovative solutions to policy problems.
What?!? Aumann’s mother came up with the “you split, I pick” mechanism?
Plus, I learned that I really need to up my parenting game. (Pun very much intended.)
How did Robert Aumann's mother giving him and his brother a bar of chocolate end up teaching them about the economics of mechanism design?
@lindaunobel
But as a person who works with govt & consults, I see that demand for causal inference lags. Eg, I started a large health insurance RCT in India in 2013. It took 5 years. But the govt wanted answers in 6 mo. Eg, the empirical work in litigation lags the frontier by a decade+.
Larry Summers on pandemics pre-COVID
Reading through an old lit folder on my laptop I found this 2016
@nberpubs
paper by
@LHSummers
on the cost of a future (flu) pandemic.
I wish uni's would give rewards for interdisc work. (Though I oppose interdisciplinary depts because I think you should have solid skills in at least 1 subject.)
Wish funders wld give more $$ for it.
But my wishes are not reality. So be savvy about doing it.
This is quite possibly, the greatest published article ever. "We show that LaTeX users were slower than Word users, wrote less text in the same amount of time, and produced more typesetting, orthographical, grammatical, and formatting errors"
1/ There are so many basic facts about the world I don’t know. Eg, today I learned from
@Noahpinion
’s sub stack about the dramatic rise in female employment in Japan. If you asked me about female employment in the US, where I live, I’d have to google it.
Learned something new today.
There is just 1 thing that distinguishes Garam Masala and Chinese 5 spice: 5 spice swaps cardamom for star anise.
Mind blown!
@C_Garthwaite
@PreetiNMalani
From
@177MilkStreet
:
A standard economic assumption about productivity growth that grapples with fertility is that that growth is proportional to population L: dA / dt = \beta L.
The idea is more humans = more Einsteins.
This implies declining fertility affects productivity growth.
But…
I also learned may be why super successful academics often had parents who were academics. The kids were taught the lessons learned by their parents.
(Now I just have to figure out how to get my kids to want to be a professor rather than a professional soccer player.)
Indians love gold. According to a report by the World Gold Council, Indian households own more than 25,000 tonnes of gold - more than many central banks combined.
But what about the gold loan market? An explainer on how the gold loan market works.
2/If it's that causal inference costs too much, then I think econometricians & statisticians should start thinking about causal heuristics, and not just perfecting inference. Ie, include *cost of data & time* when estimating the efficiency of estimators.
2/ It is v hard to get published doing truly work that takes 2 fields seriously.
I tried a model w serious epi & econ w
@maciekboni
@Alison_Galvani
. We had to make compromises for models to fit together. Econ (epi) journals thought our econ (epi) was too simple or unorthodox.
2/ My favorite research papers to work on have been ones motivated by interesting facts. Eg, I’m excited to be working on one now about the *variation* in India’s sex ratio across geography & age. TBH, the descriptive plots do all the storytelling, not the math & metrics.
Publishing in economics is BROKEN.
So when I got an email from INEXSY (
@Prof_Riemann
), a market for papers, I opened it.
Authors post abstract & figs of a paper; then journals see, DM for deets, & make publication offers,
I am NOT endorsing, but economists should model this.
I want them to compare mkt allocation to non-market methods, incl public ones by first specifying the alternative method & then discussing how it compares to mkts. Trashing mkts is easy, specifying better alternative is hard.
*** New working paper alert ****
What's the impact of health insurance in a lower-income country? What is the optimal price of insurance? We conducted an RCT in India to answer these questions.
Ungated:
Reading
@tylercowen
's wonderful 1999 book "Public Good & Market Failures", & had a thought.
We frame public goods problems as a choice b/w govt & private market provision.
We should instead frame them as a choice b/w 2 govt solutions: produce a good or create a market for it.
@OrinKerr
@AndrewCMcCarthy
Orin, assuming the warrant is fine under 4A, under Andresen, what happens if the FBI seizes items that were not of the form described or not for the crime indicated in the warrant? Would that evidence be excluded under 4A?
Most people think
@nytimes
is in the journalism business.
This morning, when using towels recommended by Wirecutter, it occurred to me that's wrong (or incomplete).
It is in the culture (or lifestyle) business.
This has interesting economic and legal implications...
Nice interview with
@Susan_Athey
. Key economic insight about top-5 obsession among economists 👇🏽. If costs of publication process for lower tier journals fell, relative returns to top-5 would fall.
@TDeryugina
@profholden
A decade ago, I interviewed 25 economists for a book, The Art and Practice of Economics Research. I will be releasing interviews each week. This week's interview is with
@Susan_Athey
, then
@HarvardEcon
, now
@StanfordGSB
.
India testing news: ICMR allows pooling of RT-PCR tests. Specifically mentions community testing & testing asymptomatic people. Recommends no more than 5 samples per pool - BUT allows more if purpose is research. HUGE regulatory breakthrough!
@anuacharya
@profmohanan
@nebuer42
*Just accepted paper at J Urban Economics*
Does mobility explain why slums in Mumbai were hit harder by COVID-19?
join w J Sheng,
@ashishgoel
P Botla
Pre-print here:
THREAD