This week in , we have a discussion between and Michael Kremer about Kremer’s approach to economics, and Tim’s reflections on how economics has changed in the last decade.1

Brazilian families spend nearly 30% of their incomes servicing debt.

No, really, the immigrants are not causing an increase in the crime rate.

Economics, you’re broken: “The analysis reveals a trend of declining concentration in the institutional affiliations of award-winning researchers [from the 1820s to the 2020s], shifting from a few science-strong universities in high-income countries to a more diverse set of institutions across the world. The decline in concentration is observed in all fields except one: economics.”

Man, it’s weird to see your tiny subculture on the front of NYMag.

Physician performance improves with LLM access.

Subject acceleration is good and leads to more people continuing math education.2

The Kremlin spends less than $2 billion per year on disinformation. Honestly, this is… shockingly little?

I appreciate P. Aronow’s commitment to being a menace.3 He has just put out a paper with the following abstract:

“In contemporary mathematical and quantitative research, the absence of a disclosure regarding LLM use is cause for suspicion. If a large language model was used, the absence of a disclosure misrepresents how the research was done. If no large language model was used, the authors declined an inexpensive and effective check of their results without acknowledgment or justification. Thus a failure to disclose is evidence of either dishonesty or carelessness.”

Countries with more cats are also happier. OK, there is no identification strategy to be found here, but I am choosing to believe this is causal.

is back! This week, we have letters to the editor - and a debate about the quality of the South Sudanese census.

The Economist thinks effective altruism is the century’s biggest idea. I also laughed a lot at this line in a related article: “the movement is tiny, with an informal membership probably in the tens of thousands, many of whom claim to be only ‘EA adjacent’.”

• Speaking of EA, I made it onto a weird conspiracy website about the shadowy network of EAs.1

White South Africans make as much as Danes. Black South Africans make as much as Bangladeshis.

Americans no longer throw dinner parties (or, really, even have people over).

Apollo predicts agentic bank runs.

Economics, by and large, contains calculation errors.

I quite liked this piece on America A and America B. It seems to understand America B in a way that most pieces2 do not.

2-6% of aid money gets shunted into crypto.

Neural similarity predicts if strangers become friends.

Children of immigrants to Britain do better on standardized tests than the native-born. Truly, I need to sit down and write my Anglosphere Immigration Really Is Different1 post.

The US has more data centre2 capacity than the next 14 countries combined.

Coefficient Giving has started a $165M fund to help countries transition as international health aid amounts decrease.

- ’s in-house illustrator - argues that the third wave of philanthropy must consider political economy.

My friend Freddy Soames is starting a UK progress & politics news roundup.

“The number of countries and territories in the low- or no-growth club has risen from 68 in 2014 to 80 in 2024.” The average low-income country actually had GDP/capita decrease from 2014 to 2024. Seems not ideal.

British productivity growth has been less bad than previously thought (though still not amazing).

Why does everyone freak out about inheritance tax in the UK3 when relatively few estates pay it? Because one in five households with members over 65 would trigger it if the member(s) died now.

A new paper finds that 1) Europe increasingly takes in migrants from female-seclusion societies, and 2) this makes migration increasingly costly because women are less likely to work.4

is taking pitches for their “secrets” issue.

The provost of Dartmouth wrote an op-ed about how to fight AI cheating. It was entirely written with AI.

>50% of German long-distance trains are delayed.

Scientists have discovered a new kind of cat. I approve of this science.

The Houthis use Claude.

Economists disagree on the returns to education.

It’s all coming up Michael Kremer - he’s the new chief economist of the World Bank and also he and coauthors have raised $60M for AI for weather forecasting in LMICs.

has a new post about the non-material consequences of growth (or lack thereof). I… very nearly sent this post to the degrowther in my apartment building’s group chat.1

writes on how to build an AI team in government.

and don’t think AI will lead to explosive growth.

Female students are given information about the work/life balance of career options far more than male students. Guys. Guys. I know you’re trying to support women in their career ambitions. If you assume only one gender is going to balance home production and paid labor, that’s still sexist. Some women do not care about home production!2 Some men really care about the ability to balance work and family! Assuming women are The Childcare Gender is still a problem.

Being in public in India as a woman is grim. “Past-year prevalence of groping, stalking, or sexual assault is 41% in New Delhi and 14% in Bengaluru.”

In 1976, Brazil allowed people to describe their own skin color in a survey. Answers included “sunburnt”, “regular” and “yellowish”.

The last piece of issue 2 of is out: and on how AI is affecting IT outsourcing in India - and what conclusions we can (and can’t) draw from that.

is launching a new program for thesis-driven philanthropic funds in global health.

is back!

argues most think tank events are a complete waste of time. I largely agree, but good events are so good it’s worth attending the terrible ones.

Anglosphere immigration continues to be different1 than immigration elsewhere - second-generation immigrants in the UK score better than the native-born on tests (while this is not true in much of the rest of the OECD).

A paper shows that people watch nearly twice the amount of short-form video that they intend to.

Meetings: yes, we all hate them, but they might be important for on-the-job learning and contribute to wage growth.

New coinage: nitpacking — when an AI can’t find a problem in your work and therefore brings up irrelevant things.

I have a new personal website. I suggest clicking over to the bookshelf for a surprise.

This week in , and explain what on earth is happening in US foreign aid these days.

, and argue that the standard RCT toolkit needs to be tweaked for AI — mostly because RCTs are slowwwww.

Narrative violation: cousin marriage may decrease violence against women.

Another Dan Ariely paper is probably fraudulent. Quelle surprise.

Being diagnosed with cancer makes you more likely to commit crimes.

More than a third of Britons buy a meal deal every week.

Dan Sullivan has endorsed Dan Sullivan over Dan Sullivan in the Alaska Senate race.

This week in , we decided to make economists angry by pointing out that they should do more qualitative work.

There’s a new $100k fee on H-1Bs.1 I’m so tired.

GovAI is looking for entrepreneurs in residence. The “AI Bellingcat” idea seems particularly useful.

BREAD and IGC have a free PhD-level course in political economy. I’m v tempted to show up to the crime session.

Yale is hiring an econ comms person, and you get to work with the awesome Jeff Mosenkis.

Via , this is a lovely essay about love in a country where a quarter of the population is a carrier of a genetic disease.

• Also: recessive genes often do affect phenotype a bit.

Pocketbook voting is alive and well in southern Africa.

When people are looking for a job, they edit their LinkedIn profiles to emphasize AI and de-emphasize DEI.

analyzes why economists and technologists disagree on the growth boost from AI in .

Free idea: instead of Inkhaven, where everyone has to produce an essay a day, a progress-y writing workshop where you spend two weeks writing one really good essay.2

’s story this week focuses on disaster data - and how it is probably worse than you think.

gave a TED talk! I was privileged to be on the trial audiences for this talk, and it is indeed excellent.

Some of USAID’s data was saved because a high school student was doing a school project and had saved 4000 PDFs.

There’s a new David Roodman (and Maxim Massenkoff) meta-analysis (on job retraining).

AI might read our science fiction and use that to determine its behavior.

Some advice on how to maintain long-distance friendships.

finds we are all breaking French law.

Someone did poll childless cat ladies on how they feel about JD Vance. Spoiler alert: they’re not fans.

Our essay of the week is by Dan Honig, on the management practices that make governments work better.

The blockbuster essay of the week is clearly ’s piece on catch-up growth in developing countries and how AI could increase the gap between rich and poor countries.

Though is giving Rafa competition for essay of the week with her piece on how every disease is a policy failure.

Talking to an AI chatbot does not relieve loneliness; it actually increases it.

Akhil Bansal writes on fractional dosing for the mpox vaccine.1

has announced their 2026 fellows. Excited to see some development writers in there!

The county my family is from is the one county in western Kansas that voted against women’s suffrage. No comment.

The more you procrastinate, the less money you make and the more medical problems you have. Yikes.

Supposably is apparently a word - it means “as may be conceived or imagined”.

There is a neologism in China to describe “women perceived to have no money, employment, education, prospects, looks, friends, or sophistication”. It is 沒女.

This week’s story is by , on the mystery of short Indian children.

The Trump administration is considering adding a $100,000 fee to OPT. OPT is the program that allows non-citizen graduates of US universities to stay and work in the US for one year after graduation.1 This would be catastrophic for both US universities and US competitiveness. No one is going to pay $100,000 to employ a new grad,2 so this makes it much harder for international students to stay in the US after their degree. I’ve written before on how much of US innovation has come from immigrants; this fee would be an excellent way to harm the US economy.

• Relatedly, we have a new paper this week showing that H-1B exposure raised native wages.

LMICs spend roughly twice as much on energy subsidies as they do on social protection.

A new paper argues ~50% of lead exposure globally comes through diet.

argues that AI philanthropists should endow institutes at universities in the Global South.

More evidence that the UK is London with some irrelevant things attached:3 the borough of Bromley4 collects more income tax than the city of Birmingham.5

It is taking longer and longer to become wealthy. That is, we are transitioning from an income society to a wealth society - where high income may not be enough to make you wealthy.

says you should meet your mutuals. I endorse this message, largely because Larissa is one of my mutuals who I met through the internet, and now she is one of my best friends.

The M1 motorway has higher net favorability in the UK than AI does.

The Home Office has suggested that asylum seekers should pay a £10k fee to reimburse the UK state for part of their costs. Or… asylum seekers could just be allowed to work and not be kept in forced inactivity for a year?

outlines what Ed Miliband can do to make British aid more effective without spending more money.

The US has changed regulation around duration of status for international students. In practice, this means that international students will have to get a new visa after four years in the country. Given that PhD programs are six years, this is incredibly disruptive to international graduate students and will likely reduce the number that choose to study in the US. Since immigrants make up a large percentage of the innovation workforce, this will ultimately be bad for the US.

The US is moving towards requiring colleges to show that their graduates make more than they would if they had not gone to college in order for students to be eligible for financial aid. This seems largely sensible; I’ve never bought the argument that one goes to college simply to expand one’s mind. That’s lovely, I’m sure, but I’m not paying 200 grand for it.1

In the US, 65% of homes are owner-occupied. But due to the rise of young adults living with their parents, this does not mean 65% of people own their homes; actually, only 53% of people live in a home they own.

begs us to stop saying “correlation is not causation”. I second this; I’d get this comment all the time on QI facts. Any time I’d report on a causal paper and use causal language, someone would comment that correlation is not causation. Yes, I know, that’s why I went through the identification strategy to see if I thought it was good…

This is old, but it is good to remember that you are a crazy person and should choose your career based on your particular kind of crazy.

Li Keqiang’s wife translated a book on Yes, Minister into Chinese…

For America’s 250th birthday, listed 25 times the United States has been a force for good in the world. I agree with most of this - particularly the US’ place as a creedal nation.

To quote Ronald Reagan, “you can go to live in France, but you cannot become a Frenchman. You can go to live in Germany or Turkey or Japan, but you cannot become a German, a Turk, or a Japanese. But anyone, from any corner of the Earth, can come to live in America and become an American.”

There is a paper that finds the median married Indian woman leaves the house for only 30 minutes a day. There have been some methodological criticisms of this paper, but another paper finds that only 16% of the people on the streets of Mumbai are women. I don’t really have witty commentary on this; life is more than household production.

writes about the migration hump - and how people who would most benefit from migrating for work can’t afford to do so.

has a reading list to understand Chinese biotech. I know very little about Chinese biotech, but can endorse many of the books listed.

It is likely that 20,000 Europeans died because of the recent heat wave. Seems bad.1

I love Kazakhstan so much. According to the constitution, you can only be president for one seven-year term. The current president would like to run again. So the constitutional court ruled that this rule applies once per constitution, and Kazakhstan just passed a new constitution. So Tokayev is all good to run again.2

Pitches for issue 3 (or 4) for close tomorrow! Information here.

is also accepting pitches on the theme of flourishing. If I was not busy running a magazine of my own, I would be pitching them.

has announced the awardees for their Research Revival Fund - including one translating and uploading Soviet science.

wrote on how we got COVID vaccines so far - and included some very excellent footnotes.

has a post about AI writing - and why it should be shunned.

• He also has a fun chart for the types of AI use he thinks are acceptable vs. the type of writing it is used for.

finds that Arnold Ventures spends about $35k per new collaboration created via convenings.

argues that more evidence couldn’t have saved USAID. He’s right, but is also correct that more evidence can save some lives.

Stripe has announced a new fund trying to end respiratory diseases. As an immunocompromised person living in a major city, this day cannot come fast enough.

You may have noticed that Europe had a heat wave last week. In London, temperatures were in the low 90s,1 with the heat index peaking around 100.2

• 11% of Britons planned to use air conditioning to combat the heat.3 Only 3% of households containing a person over 75 have AC. This is… too few.

• This is at least partially because most British homes have no ductwork - making retrofitting AC much more expensive than it is where there are existing ducts for heating.

• I have previously written about how air conditioning is a survival technology, and how it is actually useful for Britain’s Net Zero goals. If I was God of All UK Policy, I would remove all planning restrictions on AC installation and increase the air-to-air heat pump Boiler Upgrade Scheme subsidy.

The Economist encounters Indian government websites: “Before you read beyond this paragraph, grab a glass of water and 1,000mg of paracetamol. Walk over to your laptop—the supercomputer in your hand is not up to the task—and make yourself comfortable. Now navigate to indianvisaonline.gov.in and see if you can figure out how to apply for a visa.”

is still taking pitches.

Coefficient Giving has a new RFP for projects to improve global health and well-being in an era of transformative AI.

Current AI systems appear to make it much easier to break into (US) classified systems. Given that the US almost certainly has more robust cyber defenses than almost any other country,1 I would feel very nervous if I were India or Pakistan…

Young biomedical scientists are much less likely to want to stay in academia than they were pre-Trump, and they are much more likely to want to leave the US. Seems bad.

The speed of progress is determined by bottlenecks, not the most-accelerated fields.

is branching out into reported stories.2

is doing microgrants in hyperspectral biology.

Legalizing sports betting resulted in 284,000 more households becoming food-insecure. Seems bad.

In Argentina, if you’re behind on child support, you can’t buy soccer tickets. This seems good.

Almost half of G- or PG-rated movies contain torture????

The application for Managing Editor of In Development closes TODAY (at 11:59 PM Anywhere on Earth). Apply! If I may say so myself, we’ve assembled a rockstar editorial team and a hell of a contributor list; it’ll be a fun time.

I forgot to link this last week, but Jonathan Portes and I have a new paper about the disconnect between the UK’s labor force need for immigrants and the UK population’s concern about too much immigration. It was cited in this FT piece.

Global migration has tripled since 2000.

Speaking of migration, has a post about where migrants1 in the UK work and what they make. Her dataset is tech-biased but in general: they make a lot and live in London. Americans apparently are particularly likely to live in Kensington and Chelsea.2

I largely agree with this essay by Jason Crawford on using AI as a public intellectual.

• My essays are 100% human produced not because I hate AI, but because I think AI can’t produce what I want. AI can’t make fun jokes in the footnotes,3 and the fun jokes in the footnotes are a load-bearing part of my work.4

There’s a new AI microsite - Europe 2031 is about what happens if Europe gets left behind in AI.5 Given recent developments with Fable, I wish this was less prescient…

• There is a less serious version by Tim Hwang - he projects such things as “The High-Level Expert Group on Expert Groups delivers its finding: there are too many expert groups.”

argues AI probably won’t be a leapfrog technology. I largely agree with this; the lack of training data seems particularly likely to bite in LMICs.

Standard deviations of learning are not comparable across studies.

The Social Market Foundation has a new proposal to allow Brits to shift state spending earlier in life. It would allow people to take out one year of their state pension in their 30s - for instance, for a down payment - in exchange for delaying the start of their state pension.

This seems broadly sensible;1 in general, state spending in the UK is very skewed to pensioners, while people often struggle to buy property and afford children.

The Economist argues we should spice up soccer by allowing the goalie “to use less and less of his body to make saves” as the match progresses, specifying that “if a player is judged to have feigned injury, the next foul on them will be unpunishable”, and making referees able to “award a green card to any player who dribbles past an opponent in games played on American soil”. I would watch this game.

Labour voters think they’re richer, smarter, kinder and funnier than the average Brit.2