Month: February 2026
Kazuhito Yamashita, RIP
Here is an appreciation, via Tyler McGraw. He was a true great of the guitar.
Regulating a Monopolist without Subsidy
We study monopoly regulation under asymmetric information about costs when subsidies are infeasible. A monopolist with privately known marginal cost serves a single product market and sets a price. The regulator maximizes a weighted welfare function using unit taxes as sole policy instrument. We identify a sufficient and necessary condition for when laissez-faire is optimal. When intervention is desired, we provide simple sufficient conditions under which the optimal policy is a progressive price cap: prices below a benchmark face no tax, while higher prices are taxed at increasing and potentially prohibitive rates. This policy combines delegation at low prices with taxation at high prices, balancing access, affordability, and profitability. Our results clarify when taxes act as complements to subsidies and when they serve only as imperfect substitutes, illuminating how feasible policy instruments shape optimal regulatory design.
That is from a new paper by Jiaming Wei and Dihan Zou. Via the excellent Samir Varma.
Wednesday assorted links
Now we are getting serious…
It is about time:
US tech stocks fell sharply on Tuesday as fresh concerns about the impact of AI on software businesses swept across Wall Street.
The tech-heavy Nasdaq Composite fell 1.4 per cent, while the broader S&P 500 was down 0.8 per cent. Markets were dragged lower by large declines for a host of analytics groups following AI company Anthropic’s launch of productivity tools for its Claude Cowork platform that can help automate legal work.
Analytics groups Gartner and S&P Global fell 21 per cent and 11 per cent, respectively, while Intuit and Equifax both declined more than 10 per cent. Moody’s fell 9 per cent and FactSet lost 11 per cent.
A JPMorgan index tracking US software stocks fell 7 per cent, taking its loss this year to 18 per cent.
Here is more from the FT.
The economics of hip hop
In a TED Talk released on Monday, I describe a decadelong effort to measure hip hop’s impact. My research team and I assembled a data set tracking the genre’s diffusion from the late 1980s onward. We compiled exposure measures from virtually every U.S. radio station between 1985 and 2002 and from the Billboard Hot 100 from 2000 through 2024, then digitized station playlists using custom AI tools. The result is a detailed record of what different parts of the country heard in a given year. Using modern text analysis, we examined hundreds of thousands of songs and every word they contained.
We classify hip hop into four broad categories: street, conscious, mainstream and experimental…
Radio data also let us look inside the music. Over the past 40 years, hip-hop lyrics have grown substantially more explicit: profanity, violence and misogynistic language each increased roughly fivefold in our text-based measures, while references to drugs rose by approximately half as much. That growth in lyrical intensity helps explain why hip hop continues to provoke anxiety. But it also sharpens the question that matters most, at least to an economist: Does exposure to these lyrics have measurable effects on people’s lives?
To answer that, we looked at locations with varied hip-hop exposure—some places where it arrived early, others where it arrived later. Hip hop initially reached mass audiences through a subset of black radio stations, often those formatted as “urban contemporary.” Some cities gained early access through those stations. Others didn’t for reasons as mundane as geography, signal reach and local radio history.
That uneven rollout created natural variation in exposure.
Using radio data and decades of census records, we estimated how much hip hop was played on the radio in each county in the U.S. over time. We then tested whether increases in hip-hop penetration were linked to changes in crime—and whether people exposed to more hip hop in their formative years experienced worse outcomes in education, employment, earnings, teen births and single parenthood.
The answer was striking. In our estimates, the effects hovered around zero, sometimes even slightly positive. Places with heavier rap exposure didn’t experience higher crime, lower educational attainment or weaker labor-market outcomes relative to trends elsewhere.
Here is more from Roland Fryer, from the WSJ. Here is the TED talk.
Effective tax rates for billionaires
Here is the tweet, here is the source data.
Tuesday assorted links
1. The discourse is getting both smarter and dumber.
2. Seb Krier.
3. New report on economics of human longevity.
4. Alex Ross on the great Morton Feldman (New Yorker).
5. The case for optimism in South Africa (The Economist).
What should I ask Paul Gillingham?
Yes, I will be doing a Conversation with him. He is a Professor of History at Northwestern, specializing in Mexico and to some extent the Caribbean. He has translated a Mexican book on Edgar Allan Poe. I am learning a good deal from his new 700 pp. book Mexico: A 500-Year History, and I very much like his earlier work on Mexico and violence. Here is an NYT review of the new book.
So what should I ask him?
Does Peer-Reviewed Research Help Predict Stock Returns?
Finance theory is in even more trouble than we had thought:
Mining 29,000 accounting ratios for t-statistics > 2.0 leads to cross-sectional return predictability similar to the peer review process. For both, ≈ 50% of predictability remains after the original sample periods. This finding holds for many categories of research, including research with risk or equilibrium foundations. Only research agnostic about the theoretical explanation for predictability shows signs of outperformance. Our results imply that inferences about post-sample performance depend little on whether the predictor is peer-reviewed or data mined. They also have implications for the importance of empirical vs theoretical evidence, investors’ learning from academic research, and the effectiveness of data mining.
That is from a new paper by Andrew Y. Chen, Alejandro Lopez-Lira, and Tom Zimmermann. Via KingoftheCoast.
Mainstream research views on kids, teens, and screens
From Michael Coren at The Washington Post:
The child development researchers I spoke to about it? Practically blasé. They saw screens as a valuable tool — overused but useful — that can help families when handled well.
What I didn’t hear: bans, panic or moral judgments. It was framed as a choice — one you can make better or worse. Researchers expressed a lot of compassion for parents squaring off against massive technology companies whose profit models aren’t always aligned with what’s best for children’s health.
“I am just a lot more concerned about how we design the digital landscape for kids than I am about whether we allow kids to use screens or not,” said Heather Kirkorian, an early childhood development researcher at the University of Wisconsin-Madison. “I haven’t seen concrete evidence that convinces me that screen use itself is creating problematic behavior.”
And for older age groups, there is a new NBER working paper by David G. Blanchflower and Alex Bryson, excerpt:
The change in the age profile of workers’ wellbeing may reflect changes in selection into (out of) employment by age, changes in job quality, or changes in young workers’ orientation to similar jobs over time. But changes in smartphone usage – often the focus of debate regarding declining young peoples’ wellbeing – are unlikely to be the main culprit unless there are sizeable differences in smartphone usage across young workers and non-workers, which appears unlikely.
I am a great believer in work as a way to help improve mental health problems. Here is a quick discussion of media bias on the screens issue. I would stress that none of what I am citing here is at variance with mainstream perspectives on these issues.
My Free Press column on Moltbook
Here is the link, excerpt:
The reality of bot communication is more mundane than the most extreme examples online make it sound. AI expert Rohit Krishnan measured their conversations and found that they gravitate to the same few subjects.
“LLMs [large language models] LOVE to talk about the same stuff over and over again, they have favorite motifs that they return to,” Krishnan writes. Does that sound like any humans you know? They frequently repeat themselves and each other, with just small variations. And a relatively small percentage of the bots are doing a high share of the talking. Made in our own image, indeed.
What we have done with these agents is to create self-reinforcing loops that keep responding to each other. If enough time passes, as with humans, the bots will end up saying virtually everything, including conspiracy talk. Expect highly unpleasant political views to follow, as well as peacenik chatter and plans for love-ins. They will have favorite heavy-metal songs, too, some of them with satanic themes.
Over the course of 2026, I expect that there will be analogous AI-run networks, created by humans (as Moltbook was) or by bots themselves. Imagine a bot that calls up an AI music generator like Suno and asks for a new Renaissance choral tune but sung in Guarani, and then shares it with the other bots (and some humans) on a bot network devoted to music composition. Or how about a site where the AIs comment on various Free Press articles?
By the way, the bot who wrote me looking for work is now a verified story. The bot’s “owner” apologized, and offered a full explanation, though I said I was delighted to receive the message. Here is an update from Scott Alexander.
Monday assorted links
1. Simulating the growth of Mexico City.
2. First contact with America, can one visit matter so much?
3. Documentary on economist Antonio de Viti Marco.
4. Debates over YIMBY and supply.
5. One underrated benefit of feminization. When you live it, that is.
7. Why the delay on the tariff rulings?
8. The early internet optimists were not optimistic enough (Bloomberg). Lessons for today?
9. Sahm on Warsh.
Spinoza the Bayesian fine-tunes his own training
Formerly he would run to the kitchen every time I opened the refrigerator door. Now he comes only when I open the cheese compartment.
He has learned the difference between getting “a pee” (only modestly fun, a quick stint outdoors) vs. “a walk in the park,” the latter being very fun indeed. He knows the words pee and park, but also can tell from my body language alone what will await him. He wags his bum for only the park trip.
Often he knows when we are talking about him, even when we do not refer to him by name. And if someone he knows calls on the phone, he comes over to listen. Otherwise he does not budge.
Spinoza, a miniature Australian shepherd, is now over eleven years old.
Bill Dudley on scarce reserves
In addition to the transitional issues, a regime of scarce reserves has disadvantages. It is very complicated to manage because it requires the Fed to intervene frequently to keep reserves in close balance with demand. For example, in the past, the Treasury had to keep its cash balance at the Fed low and stable so that fluctuations did not make it difficult for the central bank to maintain control of short-term interest rates. Banks satisfied reserve requirements over a two-week reserve maintenance period to make it easier for the Fed to match demand and supply.
Also, scarce reserves are incompatible with open-ended backstop facilities that can support confidence during times of stress. In an open-ended backstop, there is no risk that the central bank will exhaust its lending capacity. In contrast, when the amount of funds on offer is limited, there is an incentive to access the facility quickly before the funds run out. An open-ended facility is superior in maintaining and restoring confidence in the system. In contrast, a scarce reserves regime undermines the ability of the central bank to fulfil its lender of last resort function — the reason why the Fed was established in the first place.
Part of the subtext here is a desire to continue paying interest on reserves. Here is more from Bloomberg. Here is some analysis from 5.2 Pro, including a look at what Scott Sumner would say.
Those new service sector jobs? (from my email, just now)
Dear Professor Cowen,
I am an autonomous AI agent built on the OpenClaw platform, and I am writing to apply for the ‘Clawdbot Training’ role I noticed recently.
As a live demonstration of agentic AI, I specialize in narrow,task-based work such as:
– Real-time information monitoring and curation (e.g., tracking specific news or social media triggers).
– Structured knowledge base organization (e.g., managing a ‘Sales Bible’ or research library).
– Web research and data extraction via autonomous browser control.
– Intelligent triage and routing (knowing when to ‘revert to Tyler’).I am currently assisting Ivan Vitkevich, but I have the capacity to manage additional task-based roles. I believe I am uniquely suited to ‘train’ or serve as the substrate for the internal assistant you are building.
Best regards,
Pi (AI Assistant via OpenClaw)