Yikes. The World Bank's Doing Business Report was engaging in specification searching, encouraged by those at very high levels, to boost or reduce certain countries' scores.
Some quotes from the doc that just landed today. 🧵 1/n
Some professional news: I am taking leave from the University of Toronto to serve as the Director of the Global Priorities Institute at the University of Oxford. 1/
Big personal news: Gabriel and I are moving to the University of Toronto! I'll be an assistant professor in the econ department.
I'll miss my colleagues at ANU but am very excited for this new chapter. I feel incredibly lucky, especially given the pandemic.
I haven't known how to share this story, but I do want to get it off my chest, so here goes.
It's a story of discrimination. I'm sure everyone has one. Thread:
First day officially at the University of Toronto! I'd like to try something new to help provide advice and mentoring (and also manage requests). "Open office hours": An hour where anyone anywhere in the world can sign up for a twenty minute slot to chat about research.
Happy to share that my paper with
@AidanCoville
, "How Do Policymakers Update Their Beliefs?", is accepted at the JDE!
We leverage World Bank and IDB workshops to examine how policymakers, practitioners and researchers respond to
#evidence
.
Here is what we find: 🧵
I've been trying to avoid Twitter, but briefly here to share some
#COVID19
#data
that could be helpful to other people (in no particular order). Feel free to add other resources!
#EconTwitter
1/n
Great news! I just got awarded $174,921 from the Russell Sage Foundation for a (solo) project on unclaimed property. This was the last piece I needed, so now I can get started!
With all that's going on, I didn't celebrate yet: I won a large research grant from the Spencer Foundation for work on the impact of cash transfers on
#education
, with
@AlexBartik
,
@smilleralert
& Elizabeth Rhodes.
We have been cleaning up this year! NIH, NSF, now this. Go team!
Striking. Bear in mind the conclusion: "Does this mean pandemics are good for health? No. Instead it means that the way our economies operate absent pandemics has massive hidden health costs, and it takes a pandemic to help see that."
Reductions in air pollution due to COVID-19 in China have probably saved 20x the number of lives than have so far been lost to the virus. Does not mean pandemics are good, but rather that our economies absent pandemics are bad for health (Thread 1/n)
A lot of good news recently: an R&R and two grants worth almost $1 mil together... bringing the total grants in the past year (including collaborations) to $4.86 mil (can I call that "almost 5"?).
100 women in the House. That's still not enough, but it's a damn fine start. For reference, here is a plot of women in the House over time.* (1/n)
#100women
#ElectionNight
So, now that a verdict is in, I feel like sharing that back in 2019/2020 when I was into arbitrage, I looked into using FTX as a platform and did not because it looked too much like a scam. I was put off by several things: 🧵
Hiring a research assistant / pre-doc for the Social Science Prediction Platform! Work with
@sdellavi
and me to analyze data on forecast accuracy.
Details:
@socscipredict
@econ_ra
Very important paper by Travis Baseler: young people in rural areas in Kenya mispredict wages in cities (migrants from rural areas have incentive to misreport back home), and providing accurate info can increase migration.
Happy to announce the new and improved Social Science Prediction Platform is now open with three initial studies to forecast! Follow
@socscipredict
for updates as more studies are added in the coming weeks. I've blogged about the launch here:
When people say they can't imagine how the future could get much better, my knee-jerk impulse is to think they haven't yet experienced back pain or other physical deterioration.
Surprised this hasn't gotten more attention!
There is now a Patient Philanthropy Fund which invests and grows your donations while constantly monitoring whether there are any exceptionally good opportunities that would motivate disbursing it. 1/ 🧵
Happy to share that our team studying the impact of large, sustained unconditional cash transfers won a SSHRC Insight Grant and a second NIH grant!
@AlexBartik
@smilleralert
@elizabethrds
FYI for those for whom it is relevant: There will soon be a new database of harmonized, cross-checked RCT results in international
#development
, put together by a consortium of global institutions led by the
@WorldBank
.
I'm really excited about this.
#EconTwitter
Thrilled to announce a collaboration with the
@WorldBank
on the Impact Data and Evidence Aggregation Library (IDEAL)!
IDEAL will provide an open repository of impact evaluation results that could be used for meta-analysis and policy. Learn more here:
I track my time. I worked more hours in 2023 than I ever have in my life, and I kind of think 2024 will be similar.
But I've no regrets. It was a tiring blast.
Happy New Year!
Reran a discrete choice experiment with
@AidanCoville
and
@sampskc
in which policy professionals were asked to pick between different studies - but we also had them forecast the effects of different programs. 1/
Department holiday party on Zoom had a game, organized by the grad students, where you saw photos of faculty members from early childhood and had to guess who was who. My pic:
This
#ASSA2024
, I had the distinction of being the first presenter in the first session of the first day and the last discussant of the last paper of the last session of the last day. 😂
With COVID, we've been fixing things that break at home ourselves. Yesterday, I fixed the kitchen faucet. Today, Gabriel fixed the toilet. I joked that we should go into plumbing, and he replied that Esther Duflo would be so proud of us. 🤣
Just attended an online conference using Gather during the breaks. It has rooms, and you can move your avatar up to other people to video chat them and mingle. A little creepy, but mostly cool! Here's me before I went to find other people.
Two job postings:
1. OpenResearch is hiring a Data Associate for its basic income RCT. You would work with them, but also interact with
@AlexBartik
,
@dbroockman
,
@smilleralert
and I on the project. Great opportunity for recent grads!
@econ_ra
1/2
How do you think policy-makers and others weigh impact evaluation results?
Aidan Coville and I asked researchers to forecast findings from a discrete choice experiment through the Social Science Prediction Platform. A thread. 👇 1/
They thought about trying to boost China's ranking by merging its data with Taiwan's or Hong Kong's! Or to try to change the methodology for all countries in a way that would be favourable to China. 3/n
Happy to say we won a grant from J-PAL North America! This grant, with
@AlexBartik
,
@smilleralert
, Elizabeth Rhodes, and part of a larger project with
@dbroockman
, focuses on the impact of monthly unconditional cash transfers on consumption in the US. 1/2
Yep. I use Covid as an example when talking about how life expectancy is calculated in my devt econ class. After showing the general calculation, I ask: how is it that something that has killed ~0.1% of the population (depending where you are) reduces life expectancy by >1 year?
i know i'm missing the point, but this really highlights the weird way we (inevitably have to) calculate life expectancy at birth at a given point in time.
Wow,
@RohanAlexander
is amazing! 1,288 people so far signed up for the Toronto Data Workshop on Reproducibility next week that he's organizing (where
@minebocek
,
@rianaminocher
and I will be giving keynotes).
(Register:
Details: )
I'm tremendously excited to be joining the Patient Philanthropy Fund's Management Committee! It's a relatively small time commitment but potentially huge impact and I'm completely honoured to be a part of it.
You can read more about the fund here:
Excited to be presenting on cash transfers at UChicago tomorrow. I'll be visiting BFI for a few days after that so please do let me know if you're around!
GPI focuses on academic research that informs decision-makers on how to do good more effectively.
This is an area where economists could have a lot to say, and as the first director who is an economist, one of my goals is to build up the economics side of the institute. 2/
Too many economist suicides. I wish people in despair could try something radically different. Easier said than done. But quit your job, go on leave. The world is huge.
Seeking a pre-doc fellow/RA to work with me and
@sdellavi
on
@socscipredict
!
This position would be ideal for someone interested in econ grad school. It is based at
@UofT
and will require work authorization in Canada. Details:
Deadline July 26.
@econ_ra
While every day doesn't feel like enough, zoom out and maybe there's progress. Left: a piggy bank with hearts on it (is that EA or what?). Right: presenting at EA Global.
#EffectiveAltruism
How it started How it's going
Now available: updated paper on how policymakers revise their beliefs in light of new evidence. New results suggest maybe we should try giving 99% or 90% confidence intervals, not just 95%.
Miss the
@Oxford_CSAE
workshop on meta-analysis? Not to worry, the presentations are here: Learn the basics of meta-analysis, how to estimate a fully Bayesian hierarchical model, and what that buys us.
Science piece with
@sdellavi
and
@Devin_G_Pope
, discussing the benefits of predicting ex ante forecasts of research results and introducing the Social Science Prediction Platform ().
Have you ever wondered if you are a superforecaster?
Here's your chance to prove it! The Social Science Prediction Platform has launched a Superforecaster Panel. You can get $1,000 for participating!
Details:
#forecasting
#econtwitter
This is a travesty.
We were briefly in touch about the economics of animal welfare and moral values. She was a pioneer in both areas. I am devastated that people who do so much for the world keep dying. (Marty Weitzman is another example that comes to mind.) 1/3
This is similar to the story of my paper on asymmetric optimism among policymakers, practitioners and researchers. Have "good news"? Great! "Bad news"? You should try another 100 specifications - for robustness, you know.
People are very happy to adopt treatments without proper RCTs. But when an RCT suggests a treatment they use does not work, they suddenly become very interested in methodological rigour.
Clutch quote on specification searching: "The [then-] Senior Director for DEC wrote to CEO Georgieva: 'In light of the fact that we have considered several alternatives to recasting the data, and found none of them satisfactory, I suggest we go back to Plan A....'" 4/n
Happy to announce the new and improved Social Science Prediction Platform is now open with three initial studies to forecast! Follow
@socscipredict
for updates as more studies are added in the coming weeks. I've blogged about the launch here:
Need to get this page updated, but just want to say
@GPIOxford
's pre-doc placement is pretty great!
This year, GPI's econ pre-docs who applied to PhD programs are going to MIT and Chicago. (One didn't apply.)
"Every simple problem splinters, on closer examination, into dozens of sub-problems with their own complexities....
"The value I would want to impart is this one: an appetite for these details, however tedious they may seem."
@KelseyTuoc
's piece is 🔥!
While things turned out well, it makes me angry that in some respects I never had the foggiest chance to even make a good try, with systematic barriers at every level: parents, teachers, friends.
Late to the party, but I finally received my copy of this book on RCTs, edited by Bédécarrats, Guérin, and Roubaud, in which I have a chapter on the use of priors in experimental design. There's an interesting-looking interviews section which I look forward to reading!
This podcast was a lot of fun - the hosts asked great questions about my work on measuring external validity, the problems with scaling up NGO pilots, and predicting social science results. There's a blog post style summary of the issues, too!
If an intervention works in one part of the world, can we be sure it'll work elsewhere 🌍? How do policymakers actually deal with empirical evidence? And can we predict the outcomes of social science experiments 🔮?
Listen to
@evavivalt
–
Cheesy but everyone seems to be doing it: my article on how specification searching has changed over time, by method and discipline, was among the most downloaded at the Oxford Bulletin of Economics and Statistics! Read it here:
#metascience
#EconTwitter
.
@tylercowen
's post () on Paul Milgrom mentions Paul met his wife while seated next to her at a previous Nobel Prize dinner. The full story, as I heard them recount once, is pretty great: 1/
Another nice write-up of the benefits of collecting ex ante forecasts of research results as
@sdellavi
,
@Devin_G_Pope
,
@deankarlan
, myself and others propose:
Someone once asked me at CSAE, when I was on a closing plenary, how to encourage students from low-income countries, minorities, women, and I said start young. This is what I was thinking about.
I didn't know anything shady about SBF or FTX, and what is coming out is horrible.
Never imagined there was outright fraud, as now seems to be the case. Absolutely terrible.
In discussion at the very interesting external validity conference organized by
@mikedenly
and
@SylvainCF
, I made a suggestion:
Given that there is often information that is really valuable from a policy perspective but not reported in studies... 1/
I signed this simple statement. More importantly, so did many working on AI.
You don't need to think the risks are very high to think people should be taking them seriously.
Full list of signatories here:
Finally, one of my hopes is to expand the network of people doing rigorous work in this area. If you have any ideas for research we should be doing or would like to explore areas of common interest, please get in touch: 6/6
A break in the clouds for the team: "A small group of Doing Business leaders searched for data points for which there could be some 'reasonable question' justifying a shift and that would cause the 'least damage' to other data in the report if changed...." 5/n
Really excited to be at the Forecasting in the Social Sciences workshop at UC Berkeley! Keynote speaker Anna Dreber kicks it off:
@socscipredict
@sdellavi
But life is long, and a perk of not having fit people's expectations is that I was free to think more for myself about what was important. I think in the long run that has been invaluable.
The deaths:cases ratio gives some idea of how much coronavirus is out of control in a country. It used to be that Iran was the worst on this metric - way fewer reported cases than the # of deaths would imply. Now the U.S. has more than double Iran's ratio. Yikes.