Is involuntary hospitalization working?
From Natalia Emanuel, Valentin Bolotnyy, and Pim Welle:
The involuntary hospitalization of people experiencing a mental health crisis is a widespread practice, as common in the US as incarceration in state and federal prisons and 2.4 times as common as death from cancer. The intent of involuntary hospitalization is to prevent individuals from harming themselves or others through incapacitation, stabilization and medical treatment over a short period of time. Does involuntary hospitalization achieve its goals? We leverage quasi-random assignment of the evaluating physician and administrative data from Allegheny County, Pennsylvania to estimate the causal effects of involuntary hospitalization on harm to self (proxied by death by suicide or overdose) and harm to others (proxied by violent crime charges). For individuals whom some physicians would hospitalize but others would not, we find that hospitalization nearly doubles the probability of being charged with a violent crime and more than doubles the probability of dying by suicide or overdose in the three months after evaluation. We provide evidence of housing and earnings disruptions as potential mechanisms. Our results suggest that on the margin, the system we study is not achieving the intended effects of the policy.
Here is the abstract online at the AEA site. I am looking forward to seeing more of this work.
“AI is everywhere but in the productivity statistics…”
These people are saying it is there too. Though I am not quite sure what they (or anyone, for that matter) mean by AI:
First, we argue that AI can already be seen in productivity statistics for the United States. The production and use effects of software and software R&D (alone) contributed (a) 50 percent of the 2 percent average rate of growth in US nonfarm business labor productivity from 2017 to 2024 and (a) 50 percent of its 1.2 percentage point acceleration relative to the pace from 2012 to 2017. Second, taking additional intangibles and data assets into account, we calculate a long-run contribution of AI to labor productivity growth based on assumptions that follow from the recent trajectories of investments in software, software R&D, other intangibles, and productivity growth in both US and Europe. Our central estimates are that AI will boost annual labor productivity growth by as much as 1 percentage point in the United States and about .3 percentage point in Europe.
That is from Bontadini, Corrado, Haskel, and Jona-Lasinio, here is the complete abstract online.
Sunday assorted links
1. What Stanley Kubrick learned from chess.
3. Rob Wiblin and Dean Ball podcast on AI. And ChinaTalk surveys Chinese AI in 2025.
4. Stagnant construction productivity is a worldwide problem.
5. “Announcing Progress in Medicine, a high school summer career exploration program.”
How bad was British “austerity” anyway?
Chris Giles writes in the FT:
The main periods of measurement error came in the austerity years of 2012 to 2014, in 2017 during the early period after the Brexit referendum and in recent post-pandemic years. The truth is that a huge pessimistic bias in our national accounts has led us to be fed with contemporary reports of doom and gloom, which subsequently turn out to be nonsense.
But it is the first version of economic events that enters the national debate — and the national consciousness — for the entirely understandable reason that initial releases of economic data make news. You cannot expect people to care deeply about a revision to data that is three years old. Psychologically, they have made up their mind by then.
We are still told that 2010s austerity destroyed growth, but the data no longer supports that story: growth between David Cameron’s election victories of 2010 and 2015 now registers an annualised average of 2 per cent.
Somehow I am not seeing people jumping all over this story? Is it even correct? I have not seen anyone refute or counter it. Here is the analysis from 5.2 Pro, largely confirming, though it suggests 1.8% to 1.9% is a better estimate than 2%. I am very open to alternative points of view here, but at the moment it appears the correct stance was a) the British economic problems were largely structural and would not just be fixed by an aggregate demand boost, and b) fiscal consolidation was necessary, and while done imperfectly, not a disaster relative to the alternatives available.
The dust has not yet settled, but perhaps most of you were basically just wrong on this one?
Art as Data in Political History
From Valentine Figuroa of MIT:
Ongoing advances in machine learning are expanding opportunities to analyze large-scale visual data. In historical political economy, paintings from museums and private collections represent an untapped source of information. Before computational methods can be applied, however, it is essential to establish a framework for assessing what information paintings encode and under what assumptions it can be interpreted. This article develops such a framework, drawing on the enduring concerns of the traditional humanities. I describe three applications using a database of 25,000 European paintings from 1000CE to the First World War. Each application targets a distinct type of information conveyed in paintings (depicted content, communicative intent, and incidental information) and a cultural transformation of the early-modern period. The first revisits the notion of a European “civilizing process”—the internalization of stricter norms of behavior that occurred in tandem with the growth of state power—by examining whether paintings of meals show increasingly complex etiquette. The second analyzes portraits to study how political elites shaped their public image, highlighting a long-term shift from chivalric to more rational-bureaucratic representations of men. The third documents a long-term process of secularization, measured by the share of religious paintings, which began prior to the Reformation and accelerated afterward.
Here is the link, via the excellent Kevin Lewis.
China fertility facts of the day
A Chinese billionaire was seeking parental rights to at least four unborn children, and the court’s additional research showed that he had already fathered or was in the process of fathering at least eight more—all through surrogates.
When Pellman called Xu Bo in for a confidential hearing in the summer of 2023, he never entered the courtroom, according to people who attended the hearing. The maker of fantasy videogames lived in China and appeared via video, speaking through an interpreter. He said he hoped to have 20 or so U.S.-born children through surrogacy—boys, because they’re superior to girls—to one day take over his business.
Several of his kids were being raised by nannies in nearby Irvine as they awaited paperwork to travel to China. He hadn’t yet met them, he told the judge, because work had been busy…
Some Chinese parents, inspired by Elon Musk’s 14 known children, pay millions in surrogacy fees to hire women in the U.S. to help them build families of jaw-dropping size. Xu calls himself “China’s first father” and is known in China as a vocal critic of feminism. On social media, his company said he has more than 100 children born through surrogacy in the U.S.
Another wealthy Chinese executive, Wang Huiwu, hired U.S. models and others as egg donors to have 10 girls, with the aim of one day marrying them off to powerful men, according to people close to the executive’s education company.
…“Elon Musk is becoming a role model now,” said Zhang. An increasing number of “crazy rich” clients are commissioning dozens, or even hundreds, of U.S.-born babies with the goal of “forging an unstoppable family dynasty,” he said.
Here is the full WSJ article.
Addendum: In the comments Gilligan writes: “On the positive side, we will be able to tax the heirs’ world wide income for the rest of their natural lives.”
Building a cohesive Indonesia
Building a cohesive nation-state amid deep ethnic, linguistic, and religious diversity is a central challenge for many governments. This paper examines the process of nation building, drawing lessons from the remarkable experience of Indonesia over the past century. I discuss conceptual perspectives on nation building and review Indonesia’s historical nation-building trajectory. I then synthesize insights from four studies exploring distinct policy interventions in Indonesia—population resettlement, administrative unit proliferation, land reform, and mass schooling—to understand their effects on social cohesion and national integration. Together, these cases underscore the promise and pitfalls of nation-building efforts in diverse societies, offering guidance for future research and policymaking to support these endeavors in Indonesia and beyond.
That is from a new NBER working paper by Samuel Bazzi. As I have noted in the past, Indonesia remains a remarkably understudied and also undervisited country (Bali aside), so efforts in this direction should be appreciated.
Saturday assorted links
1. Polygenic overlap across psychiatric disorders?
2. Did humans make fire earlier than we had thought?
3. “In Development is a new magazine dedicated to exploring how progress actually happens in the developing world. We publish narrative-driven essays on ideas, policies, and technologies that have the possibility to, or are already, improving global well-being.” A call for pitches.
5. How to revive biopharma productivity.
6. In the NBA, the average 2-point shot attempt now has higher yield than a three-pointer.
Origins and persistence of the Mafia in the United States
This paper provides evidence of the institutional continuity between the “old world” Sicilian mafia and the mafia in America. We examine the migration to the United States of mafiosi expelled from Sicily in the 1920s following Fascist repression lead by Cesare Mori, the so-called “Iron Prefect”. Using historical US administrative records and FBI reports from decades later, we provide evidence that expelled mafiosi settled in pre-existing Sicilian immigrant enclaves, contributing to the rise of the American La Cosa Nostra (LCN). Our analysis reveals that a significant share of future mafia leaders in the US originated from neighborhoods that had hosted immigrant communities originating in the 32 Sicilian municipalities targeted by anti-mafia Fascist raids decades earlier. Future mafia activity is also disproportionately concentrated in these same neighborhoods. We then explore the socio-economic impact of organized crime on these communities. In the short term, we observe increased violence in adjacent neighborhoods, heightened incarceration rates, and redlining practices that restricted access to the formal financial sector. However, in the long run, these same neighborhoods exhibited higher levels of education, employment, and social mobility, challenging prevailing narratives about the purely detrimental effects of organized crime. Our findings contribute to debates on the persistence of criminal organizations and their broader economic and social consequences.
That is a new paper in the works by Zachary Porreca, Paolo Pinotti, and Masismo Anelli, here is the abstract online.
The Tech Labs initiative
…the National Science Foundation’s Technology, Innovation and Partnerships directorate at long last announced its Tech Labs initiative, which is intended to provide $10-$50 million a year to independent research teams (and yes, that is a per team dollar amount, not the initiative’s entire budget).
The intent is to provide “entrepreneurial teams of proven scientists the freedom and flexibility to pursue breakthrough science at breakneck speed, without needing to frequently stop and apply for additional grant funding with each new idea or development.”
The idea has many precursors, including all of the independent research labs and organizations going back several decades, the recent burst of philanthropy for new institutes and organizations, the idea of focused research organizations (here’s a good piece from today), Caleb Watney’s excellent piece proposing X-Labs, and Jeffrey Tsao’s proposal for Bell Labs X.
But this is the first time the federal government has gotten into the business of actively pushing for institutional diversity and for scientific funding at the team level.
Huge, if it works.
Here is more from Stuart Buck. Here is Caleb Watney in the WSJ.
Uh-oh
Among the cohort who began college in 2004, 38% ever took economics. Among the 2012 cohort, only 26% took economics.
That is from Wendy Stock.
Friday assorted links
1. An AI class created from a talk I gave.
2. A quite good essay on how to find books to read. And “if I quit social media, will I read more books?” (New Yorker)
3. Google Maps as market maker.
4. NYT advice on how to find a good restaurant when traveling abroad, the piece feels decades out of date?
5. Casey Handmer energy predictions.
8. Huge undersea wall dating from 5000 BC found in France.
9. How much did Oliver Sacks distort his stories? (New Yorker)
Gans and Doctorow on AI Copyright
Josh Gans had written what I think is the first textbook of AI. Instead of the “big issues” like will AI result in the singularity or the end of the human race, Gans treats AI as a tool for improving predictions. What will better predictions do in legal markets, economic markets, political markets? He generally avoids conclusions and instead explores models of thinking.
I especially enjoyed the chapter on intellectual property rights which maps out a model for thinking about copyright in training and in production, how they interact and the net costs and benefits.
Gans’s chapter usefully pairs with Cory Doctorow’s screed on AI. It’s a great screed despite being mostly wrong. I did like this bit, however:
Creative workers who cheer on lawsuits by the big studios and labels need to remember the first rule of class warfare: things that are good for your boss are rarely what’s good for you.
…When Getty Images sues AI companies, it’s not representing the interests of photographers. Getty hates paying photographers! Getty just wants to get paid for the training run, and they want the resulting AI model to have guardrails, so it will refuse to create images that compete with Getty’s images for anyone except Getty. But Getty will absolutely use its models to bankrupt as many photographers as it possibly can.
…Demanding a new copyright just makes you a useful idiot for your boss, a human shield they can brandish in policy fights, a tissue-thin pretense of “won’t someone think of the hungry artists?…
We need to protect artists from AI predation, not just create a new way for artists to be mad about their impoverishment.
And incredibly enough, there’s a really simple way to do that. After 20+ years of being consistently wrong and terrible for artists’ rights, the US Copyright Office has finally done something gloriously, wonderfully right. All through this AI bubble, the Copyright Office has maintained – correctly – that AI-generated works cannot be copyrighted, because copyright is exclusively for humans. That’s why the “monkey selfie” is in the public domain. Copyright is only awarded to works of human creative expression that are fixed in a tangible medium.
And not only has the Copyright Office taken this position, they’ve defended it vigorously in court, repeatedly winning judgments to uphold this principle.
The fact that every AI created work is in the public domain means that if Getty or Disney or Universal or Hearst newspapers use AI to generate works – then anyone else can take those works, copy them, sell them, or give them away for free. And the only thing those companies hate more than paying creative workers, is having other people take their stuff without permission.
The US Copyright Office’s position means that the only way these companies can get a copyright is to pay humans to do creative work. This is a recipe for centaurhood. If you’re a visual artist or writer who uses prompts to come up with ideas or variations, that’s no problem, because the ultimate work comes from you. And if you’re a video editor who uses deepfakes to change the eyelines of 200 extras in a crowd-scene, then sure, those eyeballs are in the public domain, but the movie stays copyrighted.
AI should not have to pay to read books any more than a human. At the same time, making AI created works non-copyrightable is I think the right strategy at the present moment. Moreover, it’s the most practical suggestion I have heard for channeling AI in a more socially beneficial direction, something Acemoglu has discussed without much specificity.
GDPR is worse than you had thought
We examine how data privacy regulation affects healthcare innovation and research collaboration. The European Union’s General Data Protection Regulation (GDPR) aims to enhance data security and individual privacy, but may also impose costs to data collection and sharing critical to clinical research. Focusing on the pharmaceutical sector, where timely access and the ability to share patient-level data plays an important role drug development, we use a difference-in-differences design exploiting variation in firms’ pre-GDPR reliance on EU trial sites. We find that GDPR led to a significant decline in clinical trial activity: affected firms initiated fewer trials, enrolled fewer patients, and operated at fewer trial sites. Overall collaborative clinical trials also declined, driven by a reduction in new partnerships, while collaborations with existing partners modestly increased. The decline in collaborations was driven among younger firms, with little variation by firm size. Our findings highlight a trade- off between stronger privacy protections and the efficiency of healthcare innovation, with implications for how regulation shapes the rate and composition of subsequent R&D.
That is from Jennifer Kao and Sukhun Kang, here is the online abstract for the AEA meetings.
Agentic interactions
Do human differences persist and scale when decisions are delegated to AI agents? We study an experimental marketplace in which individuals author instructions for buyer-and seller-side agents that negotiate on their behalf. We compare these AI agentic interactions to standard human-to-human negotiations in the same setting. First, contrary to predictions of more homogenous outcomes, agentic interactions lead to, if anything, greater dispersion in outcomes compared to human-mediated interactions. Second, crossing agents across counterparties reveals systematic dispersion in outcomes that tracks the identity and characteristics of the human creators; who designs the agent matters as much as, and often more than, shared information or code. Canonical behavioral frictions reappear in agentic form: personality traits shape agent behavior and selection on principal characteristics yields sorting. Despite AI agents not having access to the human principal’s characteristics, demographics such as gender and personality variables have substantial explanatory power for outcomes, in ways that are sometimes reversed from human-to-human interactions. Moreover, we uncover significant variation in “machine fluency”-the ability to instruct an AI agent to effectively align with one’s objective function-that is predicted by principals’ individual types, suggesting a new source of heterogeneity and inequality in economic outcomes. These results indicate that the agentic economy inherits, transforms, and may even amplify, human heterogeneity. Finally, we highlight a new type of information asymmetry in principal-agent relationships and the potential for specification hazard, and discuss broader implications for welfare, inequality, and market power in economies increasingly transacted through machines shaped by human intent.
Here is the full paper by Alex Imas, Kevin Lee, and Sanjog Misra. Here is a thread on the paper.