Saturday morning links
Today I’m teaching the first day of a two day workshop on synthetic control. If you’re still interested, feel free to come. We accept registrations until the workshop ends because many people like having the recordings and so that’s the only way you can get them as we don’t sell the recordings as a stand alone. In the meantime, I thought I’d post some links from the week.
Yesterday I interviewed Daniel Chen from Toulouse (along with coauthors Elliott Ash and Suresh Naidu) for the Mixtape podcast series. Among other things, we discussed his paper evaluating the impact that this law and economics program that had been run for decades had had. The name of the group was the Manne Economics Institute for Federal Judges and the trained almost half of all federal judges from 1976 to 1999. Here’s the finding:
“We find that after attending economics training, participating judges use more economics language in their opinions, issue more conservative decisions in economics-related cases, rule against regulatory agencies more often, favor more lax enforcement in antitrust cases, and impose more/longer criminal sentences. The law-and- economics movement had policy consequences via its influence on U.S. federal judges.”
A review of a new memoir by Anna Jacobson entitled “How to Knit a Human” tells the story of the author’s awakening from a series of electroconvulsive therapy treatment after an extreme psychotic episode. It’s a true story and she retells the reintegration of her mind which had splintered into amnesia and a complete loss of her self. The metaphor is one of knitting as she became in her healing a knitter within a community of knitters. It looks interesting and reminds me of a genre books about madness, like one the classic An Unquiet Mind: A Memoir of Moods and Madness by Kay Redfield Jamison.
Matthew Kraft, who I have not met but see post on LinkedIn, is a professor in the education and the economics departments at Brown University, and a new paper of his caught my eye. It’s entitled “Teacher Shortages: A Framework for Understanding and Predicting Vacancies” and it’s coauthored with Danielle Sanderson Edwards, Alvin Christian and Christopher A. Candeleria. I’d like to study this and apply it to prison and jail officers, which in some states also suffer from severe shortages. So I’m going to leave this one open for now.
Governor Greg Abbott announced that Texas homeowners can use lethal force under castle doctrine reforms to get squatters out of their homes, but legal experts warn against that. Every day, google news sends mew another story or two about squatting and while I don’t plan to use lethal force on anyone, I would be remiss if I didn’t say that the thought of coming back from travels to find someone in the house doesn’t make me worried. But thankfully I have four cats to fight off the intruders by meowing all night and taking dumps in their beds.
Ethan Mollick of the excellent substack, One Useful Thing, has a new book out entitled Co-Intelligence: Living and Working with AI. You’ll be seeing his name from time to time now, I think, as it’s getting some press. Ezra Klein interviewed him for his podcast which I heard was a great interview. Here’s an opinion piece in The NY Times connected with that interview in which Mollick recommended spending around 10 hours experimenting with it, and other practical tips. And here’s his wonderful Substack.
We are seeing AI moving into sports too. The Hawk-Eye review system in professional tennis appears to be making umpires more accurate, but not always. These “not always” results have shown up before in studies. A Harvard study on radiologists found heterogenous responses too, though I think there it was when the radiologists rejected the AI finding. The Hawk-Eye study was coauthored by the great Lionel Page who has a new book out called Optimally Irrational, a provocative sounding name, which you can download from Amazon and other places.
Colin Cameron at UC-Davis and author of a great textbook on micro econometrics with Trivedi, has what might end up being a new book on causal inference. But here’s his slides in the meantime. Let a thousand flowers bloom. I studied for the econometrics field exam using Cameron and Trivedi, as well as Bill Greene’s book, as my main text and will always have a fondness for it. Bill’s book went into its 8th edition and in our interview he said that’s the last one.
Will AI boost firm productivity? I bet we see fairly large carving out of the workforce personally as the capital/labor ratios shift and we move along the isoquant. Unless AI increases demand for goods and services, then it’s unclear what will happen other than that firms will become more efficient, profits will rise and savings increase. Perhaps we will see new products in the longer run, but I think that depends on what these technologies do.
This CEO claims “Brain Boosters” turned his ADHD around. Color me skeptical. I think what turns adult ADHD around is cognitive behavioral therapy focused on executive functioning, but popping pills and promises seem to the more attractive draw. Oh, he’s also the CEO of the company that sells these brain boosters. In the meantime here’s a small meta-analysis on CBT for treating adult ADHD finding promising results.
For those interested in development, NYT discusses the future, not so much of the academic field of development so much as the challenges of develop in lower income countries going forward. Dani Rodrik makes a short appearance. Basically, large changes in trade, supply chains and technology are having disruptive effects and maybe the old playbook is needing a second look.
A new serial killer starring Denzel Washington was, at least as of April 2nd this week, #1 on Netflix, but I haven’t seen it. Though I did watch the Gentlemen by Guy Ritchie and loved it. So I’m going to recommend that one too while I’m at it. But here’s the trailer for the Denzel movie from 2021 named The Little Things.
And all the people said Amen. Tim Robinson from Netflix “I Think You Should Leave”, one of the funniest skit comedy I’ve seen in years which I’ve rewatched a dozen times, is getting his own HBO show. Robinson along with his frequent collaborator Zach Kanin have the Midas touch. Watch ITYUSL first a hundred times and then leave a comment. My first one was “Brian’s Hat”.
Scientists say something weird is happening with death.
Oregon governor is calling for a do-over on decriminalizing drugs. I’d love to see the synthetic control or diff-in-diff on this experiment, though. Maybe it’s been done and I didn’t see it, but it needs to be done pronto.
I submitted two sessions to the Southerns (November 2024, located in DC) this year in the Presidential session on “Artificial Intelligence and Algorithms”. I think it’s a solid group of presenters, and if it’s accepted I’ll post it. Fingers crossed. This will be my first live workshop to attend in person in what feels like five years. But I’ve decided to get back out there.
Haus has PhD Economists and former Amazon and former Googlers. Among other things they are making causal inference technologies that can help businesses, thus reducing dependence on consultants. The future is near and they’re coming for us (me). I believe they’re hiring so check them out.
“An analysis by the National Institute on Alcohol Abuse and Alcoholism shows that people over 65 accounted for 38 percent of that total. From 1999 to 2020, the 237 percent increase in alcohol-related deaths among those over age 55 was higher than for any age group except 25- to 34-year-olds.”
That’s from the NYT entitled “Why Are Older Americans Drinking So Much?” I on the other hand have cut alcohol almost entirely from my drinking. Not to brag. Just don’t have the metabolism I once had.
As I continue to think about transitioning to a library based subscription model entitled Mixtape Academy where people can sign up and receive regular video and writing updates about causal inference, I am looking at others who have done something sort of different (though not causal inference). One is this Digital Marketing firm. The style and the design, but also the architecture, are what’s catching my mind. If I can somehow make the substack the portal into it, and I can figure out the concept, I want to try it. But it’s still just something I’m thinking about at this moment.
NYT reports on a new Scottish Hate Crime Law which some are concerned about.
Here’s 6 books for adults living with ADHD. If I ever recommend any book for adults living with ADHD, it’s things by Russell Barkley, and then I typically stop there. He’s made a career out of these “pop ADHD” books. When things get added to the DSM-5,
And yet, in the last few years, a debate has arisen as to the validity and utility of bipolar II as a diagnostic category. Some authors have suggested its elimination (hence, the existential crisis) (Malhi et al. 2019a, b) while others support its continued inclusion in our diagnostic systems (Ha et al. 2019; Nierenberg 2019; Ostacher 2017; Post 2018, 2019; Schaffer 2018; Vieta 2019). In a field as imprecise as ours, it is not surprising that these debates occur. As always, there is at least some merit to both sides of the debate. This paper will summarize both sides of this discussion, highlighting the not always congruent needs of clinicians, patients, researchers and others and the role that this distinction plays in the debate.
I would love to know what happens to the appearance of new books educating the public on it. Bipolar II, for instance, was once considered “not a real disorder” but was added to the DSM-4 back in the 1990s. I wish I could find it now, but one of the authors of that part of the DSM-4 has since walked it back claiming it has led to over diagnosis among men. You can read a little about that here. Here’s a summary of some studies addressing the overdiagnosis of bipolar two.
And here’s a study entitled, “The Existential Crisis of Bipolar II Disorder.
One of the most influential social scientists ever died recently — Daniel Kahneman. RIP. For those who don’t of him, he was a winner of the Nobel prize in economics, despite being a psychologist. His long term collaborator, Amos Tversky, had died before the award was given, and since the award is not awarded posthumously, Kahneman won it alone of that group (though shared it with Vernon Smith).
I may be repeating myself, but I can’t remember if last week I posted this or not. But MIT Technology Review has an article by Will Douglas Heaven entitled “Large Language Models Can Do Jaw-Dropping Things. But Nobody Knows Exactly Why.” Ethan Mollick says 10 hours playing around and experimenting with it so you can figure out how weird and powerful this technology is, or do what I do and spend 10,000 hours.
Large language models are very persuasive. An RCT was done by Francesco Salvi and three coauthors with a title similar to what I just wrote. Here’s the abstract.
The development and popularization of large language models (LLMs) have raised concerns that they will be used to create tailor-made, convincing arguments to push false or misleading narratives online. Early work has found that language models can generate content perceived as at least on par and often more persuasive than human-written messages. However, there is still limited knowledge about LLMs' persuasive capabilities in direct conversations with human counterparts and how personalization can improve their performance. In this pre-registered study, we analyze the effect of AI-driven persuasion in a controlled, harmless setting. We create a web-based platform where participants engage in short, multiple-round debates with a live opponent. Each participant is randomly assigned to one of four treatment conditions, corresponding to a two-by-two factorial design: (1) Games are either played between two humans or between a human and an LLM; (2) Personalization might or might not be enabled, granting one of the two players access to basic sociodemographic information about their opponent. We found that participants who debated GPT-4 with access to their personal information had 81.7% (p < 0.01; N=820 unique participants) higher odds of increased agreement with their opponents compared to participants who debated humans. Without personalization, GPT-4 still outperforms humans, but the effect is lower and statistically non-significant (p=0.31). Overall, our results suggest that concerns around personalization are meaningful and have important implications for the governance of social media and the design of new online environments.
On that note, another provocative paper in a similar vein found something similar.
And then I’ll conclude with a paper I printed out and hope to read very soon. It’s by an Allstar cast of excellent economists like Marcella Alsan, Ariel Barnett, Peter Hull and Crystal Yang. They find a positive effect of a program in jails called the IGNITE Program reducing misconduct and recidivism. As I mentioned on Twitter, our paper recently published at the Journal of Human Resources, found that randomized clinicians at booking marking the marginal inmate as having elevated mental health problems had a negative effect: higher suicidality and suicide attempts, longer stays in jail, and increased recidivism. Definitely more puzzles to unpack.
And that my friends closes out the week of open tabs. I need to now go run over and start looking over my notes for today’s working on synthetic control. In the meantime, let me leave you with some photographs of new artwork in the house.
That last one is a picture by Ariel Rubinstein, the great game theorist. It’s a picture of great coffeeshops all over the world. You can get your own copy if you want one for free by writing him. I may send him some photographs of some great local Waco ones like Dichotomy and Pinewood.
Have a blessed day and weekend. Don’t look at the eclipse. Sadly, Waco may have some cloud cover and rain. But I’m still looking forward to it.