The Sustainable Groundwater Management Act of 2014 (SGMA) often seems like THE biggest story in California agriculture. It must be affecting ag decisions already, right?
@ellenm_bruno
and I did a deep dive... and our best answer seems to be no
A long🧵on what we found and why
Here are the full materials for my data-science-in-economics course, for anyone who might benefit
The idea is to explicitly teach skills/tools that otherwise are only gained through RA experiences. Help level the playing field for research/PhDs/predocs
So tomorrow I was supposed to give a talk about whether California has been adapting to climate change
But now it's canceled -- precisely because PG&E has failed to adapt to climate change
...revealing my master plan all along!!! Performance art as economics
I gave a talk recently to the first-year PhD students, reflecting on my grad school experience. Basically what would I want to go back and tell myself?
Posting the slides here, in case they're at all helpful to others:
2 useful papers today on how you can get basically any answer you want from regressions using arcsinh(y) or log(y+1)
I made these graphs to show the problem: The mean of the transformed var (black lines) varies like crazy by what units you choose for y
🧵 on my own take-aways
Reminder: Using log(x+1) or arcsinh(x) is never coherent, and you probably want Poisson regression instead
This paper changed what I do and teach, and it's the rare econometrics paper that's actually clear and accessible. All applied social science researchers should read it!
To me, Esther and Abhijit's greatest impact has been to normalize intellectual humility -- the idea that even the "experts" often have no clue what the answer is, and it's OK for them to admit that.
(thread)
MIT News: Professors Esther Duflo and Abhijit Banerjee share the
#NobelPrize
in economics with Michael Kremer of Harvard, are cited for breakthrough antipoverty work.
A couple things from my econ data analytics course that might be helpful to someone:
* Exploratory Analysis slides
* Data Cleaning Checklist
I'll eventually post everything but am mostly using existing materials - these are the most original parts so far
A methods fallacy I've seen more than once recently:
IV does not "let you see" how much of the effect of one variable (Z) on an outcome (Y) operates through a particular channel (X)
Instead it ASSUMES that ALL of the effect of Z on Y operates through X and proceeds accordingly
Excited to develop an all-new course on data analytics for our econ master's students for the fall. I'd love to see tutorials/notes/other resources you've found useful - please share!
Aim is in-between PhD-level "data science for economists" and undergrad "statistical computing"
I'm excited to share that, barring societal collapse over the next few months, I'll be joining the awesome ag econ & econ faculty at Montana State
Looking forward to getting to know
@montanastate
students, environmental/ag issues in MT, and the trails around Bozeman. Come visit!
Excited to finally share a fully documented & updated version of my California Surface Water Data!
It gives volumes of deliveries, diversions, and allocations of surface water for all wholesale users in CA, by sector and year, for 1993-2021
We have the same great deal for an MS in Applied Economics here at Montana State (fully funded, with limited exceptions).
Are there other fully-funded master's degrees in econ/applied econ out there?
🚨🚨
Semi-annual psa:
Instead of a pre-doc, apply for a MS in ag/applied econ next year!!
At Illinois, our applied econ MS is fully funded - two years of classes and close mentoring with professors developing your own research.
Finish with a thesis, a grad degree, and no debt.
🚨Job Market Paper🚨
Can we adapt to environmental change?
How important is surface water to the economy?
I make progress on these questions in "The Scope for Climate Adaptation: Evidence from Water Scarcity in Irrigated Agriculture"
Paper:
[THREAD]
@Noahpinion
Noah, as someone working on climate economics and related topics, I think most of your critiques were spot on -- in the past.
Good thing we're already working hard on all 7 of your suggested steps, and have been for a few years now!
Good article summarizing one of the greatest achievements of modern empirical economics: a huge body of evidence on the horrific health effects of air pollution. Not obvious from casual observation; not possible to learn w/o big admin datasets + careful thinking about causality.
Air pollution kills an estimated ten million people each year. But it does much more than that, too. A long thread on what it means that more than 90 percent of the world's population is breathing dangerously polluted air. (1/x)
Did your student's grandmother really die? A back of the envelope calculation
Assumptions:
# of living GPs: 2-3
Age of GPs: 70-80
Annual Pr(death): 2-5% (CDC)
Term = 1/3 year
Each student's prob of at least 1 GP death during the term: 1-(1-.02/3)^2 to 1-(1-.05/3)^3 = 1-5%
Ok econ twitter: How do you find & choose coauthors? I'm curious how others approach it.
Intentional or serendipitous?
Start w/ the idea or the people?
How do you tell if you'll work well together?
Look for specific skillsets you're weaker at, or thought partners for everything?
(I think economists could benefit generally from thinking more about units! In physics EVERYTHING is about unit analysis, which gave me helpful intuition)
(I've been trying to tell people about this problem for years, though I've used the arcsinh too so I'm totally a hypocrite)
We're hiring a full-time RA/predoc in economics! Come work with
@carly_urban
, Justin Gallagher, and me at Montana State University
We'd like to fill this quickly so please apply soon / share with anyone who might be looking for a next step
@econ_ra
Looking again at my notes from NBER. I really liked this point in
@mikejrob
's discussion:
"Agriculture is a huge share of consumer surplus."
Sometimes hard to articulate why others should care about ag when it's only 1% of GDP, but it's the 1% we literally couldn't live without
Really gratified to see all the ag papers at NBER SI EEE, all from the angle of irrigation water. This classicly difficult topic is finally getting some modern data analysis. Water was so niche just a few years ago, and now all the cool kids are doing it.
👇👇👇
Often people say the econ job market is "noisy", as if its REAL purpose is external validation, to render the ultimate judgment of your abilities. But:
(a) that's wrong, the purpose is to get a job
1/5
𝐉𝐌𝐂𝐬:
1. Many factors affect your JM outcome beyond the quality of work → you cannot infer much re: its quality from outcomes
2. It says even less about your ability as a researcher
3. It says NOTHING about your value as a person
4. 1-3 more true this year than ever
@economeager
so much advice seems premised on the idea that people are complacent about external standards and need to be scared into working harder, when almost all of us have the exact opposite problem
Very cool paper by job market candidate
@yuanning_liang
showing that trucks crash *more* after getting inspected. One of the clearest event study figures I've seen
She suggests less likely re-inspection => less safe driving. Implies randomized inspections could reduce crashes
All 3 winners set good examples that more people in econ could learn from. Esther is generous with time and credit, Abhijit is constantly self-effacing, Michael Kremer ALWAYS offers a constructive suggestion when raising a concern to a seminar speaker (be like Michael Kremer!)
All the time, in reporting and public discussion, I see issues around climate adaptation framed in terms of, "How will we do X?"
How will we move to higher ground?
How will we reduce water demand?
How will we decarbonize?
How will we feed the world?
It's the wrong question.
Esther especially has invested like crazy in building institutions and in education, from highly dedicated PhD advising to civil service training in India and online courses that reach thousands. It's not about her; she's making sure she herself ultimately becomes unnecessary.
Essentially: The 1 in log(y+1) is arbitrary (why not 10? 0.1? 0.0001?), and the arcsinh formula contains a similar hidden parameter
So they're sensitive to scaling, which isn't great because our answers should be the same whether we work with $, hundreds of $, millions, etc.
Academics: Please stop announcing who your department has hired as faculty or accepted to a PhD. You are stealing their thunder! It's a far bigger deal to them than to you! Plus they might not be ready to publicly share their employment status. At least ask their permission first
@johnwhitehead81
Maybe we all could be less judgmental of minor stylistic choices and quit generalizing from own preferences to prescriptive mandates for everyone else
This is a situation we can help with at Montana State! Send us your students who could use a little more exposure to rigorous econ and math before applying to PhDs. Most students in our Applied Econ master's program are fully funded.
Consequently, and because I found economics late, by the time I figured out I wanted to go to grad school and professors told me how much math was needed, there was almost no time to take the math. I remember sitting in a profs office feeling so dejected when I learned this
5/13
Chen & Roth show that a "percentage" average treatment effect is *just not a well-defined estimand* when your outcome values can include 0.
I find this tremendously disappointing!! 😥 Though it seems obvious in retrospect, I think I'd been holding out hope for a way to get it
Going to read more carefully but this seems like the key. Most climate damages happen through channels other than local temperatures. Climate science has said this forever, yet most of the econ impacts lit ignores it. Global temps capture more of that broader climate system
𝗠𝗲𝗰𝗵𝗮𝗻𝗶𝘀𝗺: Global temperature shocks predict strong rise in damaging extreme events. From Deschênes-Greenstone 2011; Hsiang-Jina 2014 we know that such events are associated with substantial economic damages.
The impact of local shocks on extreme events is much weaker
New paper describing Raj Chetty's exciting alt intro econ course:
Basic ideas could be adopted widely:
* Motivate theory with real-world questions, not vice versa
* Treat econ as an ongoing empirical science
* Labs/projects engage & show what we really do
My coauthor
@_AnshumanTiwari
unearthed this (non-exhaustive) list I wrote up early in grad school of "big" questions at the intersection of environmental and development economics
We've def made progress but there's clearly much more to learn. Interested in others' reflections
Really gratified to see all the ag papers at NBER SI EEE, all from the angle of irrigation water. This classicly difficult topic is finally getting some modern data analysis. Water was so niche just a few years ago, and now all the cool kids are doing it.
Last week
@Interior
announced a new program to pay for water conservation in the Southwest. But the plans don't look aggressive enough.
To save Lake Mead and avoid painful water cuts,
@bryleonard
and I call for Interior to conduct a reverse auction
I'm worried the algorithm is leading people to miss the point (since the first tweet has 14x the views of the others)
It's truly very common for college students to lose elderly relatives!!! Most students who claim this are likely telling the truth
Did your student's grandmother really die? A back of the envelope calculation
Assumptions:
# of living GPs: 2-3
Age of GPs: 70-80
Annual Pr(death): 2-5% (CDC)
Term = 1/3 year
Each student's prob of at least 1 GP death during the term: 1-(1-.02/3)^2 to 1-(1-.05/3)^3 = 1-5%
It has come to my attention that this would not be a good joke to share with econ classes because it is based on an event that happened when college students were...
4
In earlier econ culture, certain great men held special wisdom that was inaccessible to others without their help.
Esther and Abhijit (+ plenty of others) not only pulled back this curtain but also showed everyone ways to start figuring out what we CAN know (however narrowly).
@Noahpinion
The main things I think you have wrong:
1. The point about bad research is outdated -- we haven't had stuff like that published in the last decade
2. Climate economists exist, publish a lot, and have non-oversimplified policy ideas -- they just have less prominence & influence
All of these chapters look great, but I want to draw attention to "Chapter 83 - Machine learning in agricultural economics" by Baylis, Heckelai, and Storm. Exhaustive, up-to-date list & discussion of ML applications in enviro, ag, devo, related fields
The Handbook of Agricultural Economics, vol. 5, is now published! An absolute all-star roster of authors span 10 chapters and >800 pages defining current key research frontiers in the field. And look for volume 6 in summer 2022 (chapters now in review).
Ok I am nearly 12 months late to this game but I just saw
@MeeraMahadevan
present this paper at
@emLabUCSB
and it is so cool. It's a whole detective story disguised as an econ paper. Gripping, with a super clever and tenacious protagonist/author. Read it if you haven't!
Check out this blog post on my job market paper about political corruption in the Indian electricity sector, and its welfare implications! Thanks
@DaveEvansPhD
,
@dmckenzie001
,
@BerkOzler12
, & others at
@wb_research
for this awesome way for more people to access job market papers!
Twitter is fun because we can promote the work of friends and people we admire!
Today I want to call your attention to a brand new paper by
@jondr44
and
@jiafengkevinc
, available at:
What is the gist?
Check the abstract!
Going to AERE? Don't miss the first results from my RCT with
@ariel_zucker
We offered smallholder farmers in Gujarat, India payments to voluntarily reduce their groundwater pumping. Did it work? Come see!
Flagging esp. for development folks since it's a general water session
With 1 instrument and 1 endogenous variable, IV is analogous to the Chain Rule
You think Z causally affects both X and Y, but also X is a necessary stopover on the way to Y
So dividing by the first stage lets you express the
effect on Y in "units" of X
I was lucky to take Marty Weitzman's grad course, with only 3 students enrolled. One day both
@ToddGerarden
and
@JesseJenkins
were absent, so Marty says, "It doesn't seem right to lecture; do you have any questions?"
I still wish I'd been better prepared for that opportunity!
The
@nytimes
has a big article today on groundwater depletion across the US, it's a nice summary with some cool data collection
Thread on 3 important points of context that I thought didn't quite come through in the article
Love to see a measure of Vulture Suitability in an econ paper. Of course only
@Eyal_Frank
could be responsible
But seriously loss of wildlife & biodiversity is a big deal and can have massive human consequences, as this article explains
Today we released our first column of 2023, featuring research by
@Eyal_Frank
&
@anantsudarshan
:
One particularly striking fact is that approximately 100,000 extra deaths per year can be attributed to collapsing
#vulture
populations in
#India
. 🧵(1/6)
Leah and Matto's case against carbon pricing is much more sophisticated than usually seen
Not sure where I come down, but I do think economists should take their arguments seriously
Here's a 🧵 on a couple key points, and an analogy about why hidden costs might be desirable.
Carbon pricing has dominated climate policy debates for decades. But the solution is ineffective and politically toxic.
We need standards, investments and justice to meet this crisis at its scale.
My latest in
@BostonReview
with
@mmildenberger
. 🧵...
Also: Mullahy & Norton show that when you place the 0's far from the rest of the distribution (when units are small or c in log(y+c) is large), you're basically just estimating a linear probability model!
Intuition: you're putting more weight on the extensive margin
@thetahat
@shoshievass
Yes (just tested), reghdfe prioritizes the absorbed vars which might have fixed this case. Or a more general option to absorb controls in general (would cut down table length…)
But this case would still have loopholes for a user to slip through
New paper by
@JohnMullahy
and
@healtheconnort1
:
"Why Transform Y? A Critical Assessment of Dependent-Variable Transformations in Regression Models for Skewed and Sometimes-Zero Outcomes"
So about 1 student out of every 20 to 100 will see a grandparent die during a usual semester
Which means that in most college courses there will typically be about 1 real GP death
Or, in a 300-student intro course, even 15 GP deaths are perfectly plausible, actuarially speaking
We're excited to announce our 2023 Applied Economics Summer Conference! Join us in the beautiful mountains of Bozeman, Montana from June 20-21. Application portal coming soon.
If there are ANY other channels through which Z affects Y, it's a textbook violation of the exclusion restriction. Your estimate is not necessarily any closer to a causal estimate than OLS, and it's arguably now harder to interpret
2. (M&N) Use Poisson regression.
Poisson helps concern (b) but not (a). It directly estimates the log difference of the means in the treatment & control groups: log(E[Y(1)]) - log(E[Y(0)])
Not the average log difference: E[log(Y(1)) - log(Y(0))], which is what I tend to prefer
4. (C&R) Take a stand on how much you value the extensive margin. Estimate ATE for m(y) =
=log(y) for y>0 and
=-x for y=0,
where you choose x based on theory.
This is exactly what we're doing already, but without hiding the parameter choice!
It's a total pain to constantly adjust to new methods and norms and I feel that. But I'm also glad/excited/relieved when the econometricians find problems with common methods and point us in better directions. It means our hard work is more likely to actually reveal some truth
5. (Both papers) Use a two-part model to separate the extensive and intensive margins.
This requires some further structural assumptions (though they can be weak if you're OK with bounds instead of a point estimate)
Think I need to review/learn more how to implement these!
It is really really important for people to understand that the sky-high rates of depression and anxiety in economics departments are not at all the same thing as the inclination to write cruel and bigoted things on an anonymous internet forum
I'm just really sorry to all my international friends/colleagues/students/followers. Even if you aren't directly hit by this policy I know the uncertainty and fear is awful. The US has never quite lived up to its welcoming image but I hope we can at least start trying again soon.
All you're doing is scaling one causal effect by another. There's no magic bullet
I find this way of thinking about IV a lot more intuitive than the usual "find the variation in X due to Z and use that predicted value"
And a lot clearer for evaluating the assumptions required
Another note for any econometricians who might be listening... a "proof sketch" that involves supermodularity and Frechet-Hoeffding bounds is not exactly, uh, accessible to us applied researchers 😂
(I understand your incentives of course -- this is a structural complaint lol)
Here's a preliminary topic list:
R basics
coding practices
version control
data cleaning
describing data
data visualization
spatial analysis
webscraping
prediction vs. causal inference
some basic ML
(clearly, I'm going to be learning a lot myself)
@AnthonyLeeZhang
I spent 3 weeks driving across the country and 6 weeks wandering South America, used up my meager life savings, came back just in time for math camp, and have never had a single regret
This debate ceases to exist when you realize that "numbers that ought to be replicable" and "numbers that ought to be communicated in the first place" are the exact same numbers
1. (Mullahy & Norton) Forget the proportional treatment effects, just estimate OLS on the untransformed outcome.
I don't like this because
(a) I often care more about the avg proportional change than the avg level change across individuals
(b) estimates can be noisy for skewed y
My two favorite
@nberpubs
workshops, Environmental and Development, are livestreaming today & tomorrow
Tune in to Environmental today at 3p eastern to see me present "The Scope for Climate Adaptation: Evidence from Water Scarcity in Irrigated Agriculture"
Excited for tomorrow's
@nberpubs
Environment and Energy Summer Institute. Thank you to the authors for writing such great papers, my co-organizer
#ReedWalker
(UCB) for carrying me, and most of all to the discussants for the selfless work!!!! Looking forward to learning a ton.
RFF's new estimate of the Social Cost of Carbon ($185/ton) reflects the incredible amount of new research on climate impacts that's been done since the Obama Admin first set the SCC at $51. Excited to dig in, and hope the federal gov takes this seriously.
One benefit of teaching on Zoom: remote guest talks are no less natural than normal lectures
This fall I was super lucky to snag
@danaehernandezc
,
@MeeraMahadevan
, and
@gabeenglander
to speak to my natural resource econ undergrads about their awesome research for 30-40 min each
📢
@JPAL
co-founder & Nobel laureate
#EstherDuflo
(
@MIT
) has been appointed to the newly constituted Economic Advisory Council to the
@CMOTamilnadu
. The Council will provide guidance in promoting inclusive economic growth in
#TamilNadu
[1/4].
Time for a fun game! Which is closer to the freezing point of water?
1. the temperature of the human body, a sweltering 98.6 degrees Fahrenheit
2. the air outside right now in Montana
Hint: it's not
#2
The impact of field experiments in development (which E+A have promoted) isn't just about what we learn from these few hundred studies. It's about starting to budge the way major organizations all around the world think about evidence in their decision-making.
Come for the political effects of ag subsidies
Stay for a great practical explanation of how to analyze a shift-share natural experiment using the new "design-based" paradigm (this is probably the future)
New WP: "The Political Benefits of the Monoculture"
@BobbyGulotty
and I took a look at whether Trump's ag subsidies helped him in 2020. tl;dr - they did.
Today in Montana problems: Found a marmot in our car
It hitched a ride home from our hike in the front of the engine compartment for a 2 hour drive. It then refused to come out and kept emitting ear-piercing chirps
@Jabaluck
This is a great thread for what to do practically, but it's distinct from how to handle the emotions. It's easy to get overly discouraged when you don't yet feel secure/established in your abilities/career. Need tools to deal with the feelings before you can make rational choices
Focus is on practical tools (they take econometrics separately). Some students go on in research, others go into industry.
Obviously I am going to steal heavily from
@grant_mcdermott
(thanks Grant!)
Too bad SCOTUS missed
@ctaylor463
and
@HannahDruck
's 2022 paper in the AER
"We evaluate wetland location relative to the surface water network and find that the most valuable wetlands for flood mitigation are those located 500 to 750 meters from the nearest stream or river."
Insightful & useful piece by
@RichaelYoung
on water markets
A couple key points:
* Most transfers are informal - more reporting would help markets work better
* Regulatory complexity isn't the enemy - the problem is when it isn't clear or predictable
In fact it's more plausible that the toxic culture in econ causes mental health problems than the other way around
Some habitually cruel people might be mentally ill, but most mentally ill people are not habitually cruel
When weather fluctuations affect people's later decisions, what do we learn?
* Many papers interpret as adaptation
* This article shows why that's often wrong. Evidence of irrationality! Temporary shocks shouldn't affect beliefs
Except in general that may be incomplete too (1/2)
📢 Out now 📢 in the July 2024 issue of
@JaereAere
:
"Positive Rainfall Shocks, Overoptimism, and Agricultural Inefficiency in China" by Kaixing Huang, Jingyuan Guo, and Da Zhao
Read it here:
Maybe a better way of asking this question:
What have you learned through experience about choosing and working with coauthors, that led you to make (or intend to make) changes in your process for the next time?
Ok econ twitter: How do you find & choose coauthors? I'm curious how others approach it.
Intentional or serendipitous?
Start w/ the idea or the people?
How do you tell if you'll work well together?
Look for specific skillsets you're weaker at, or thought partners for everything?
Huge thanks to these authors, who have generously published prior materials under open licenses and/or given me permission to adapt their stuff
*
@grant_mcdermott
*
@edrubin
*
@rafalab
* UC Berkeley's D-Lab