EconLit

A Simple Way States Can Reduce Healthcare Spending

James BaileyLeave a comment

Repeal Certificate of Need requirements, assuming you are one of the ~35 states that have them.

That’s what I find in a paper with Shishir Shakya just published in Health Economics.

Certificate of Need laws require healthcare providers to obtain government approval before building new facilities or expanding services. The original intent of the laws was that less building would mean less spending, but critics worry that it means less competition and higher prices leading to higher spending. Our paper finds that the critics are right, as states repealing their Certificate of Need laws see their healthcare spending fall relative to those that maintain them.

Longtime readers may recall that back in 2022 I posted about another paper of mine with a similar result, and even that wasn’t the first word on the subject. So what’s new in this paper? Besides newer data, we measure health spending as a percentage of the state economy rather than in raw dollars, to try to account for how CON changes the true burden of healthcare spending. We also use newer regression techniques that are probably more accurate but definitely help make prettier graphs.

Here’s what we found happening to health spending when states repeal Certificate of Need:

Full published paper is here, ungated working paper version here.

Paper Published on Persistent LLM Hallucinations

Joy Buchanan1 Comment

I am pleased to announce the publication of

“LLM Hallucination of Citations in Economics Persists With Web-Enabled Models”

By Joy Buchanan, Stephen Hill, and Olga Shapoval

ABSTRACT: Despite significant advances since 2023 in large language model (LLM) capabilities and the introduction of web-enabled search functionality, citation hallucination remains a persistent problem in AI-generated academic content. We test GPT-4o, DeepSeek V3, web-enabled GPT-4o, and web-enabled GPT-5.1 across prompts derived from Journal of Economic Literature categories. Using a hierarchical citation validation framework, we find that even an advanced web-enabled model produces non-existent citations. While web-enabled models show improvement over their predecessors, the error rates remain high for scholarly applications. Our findings suggest that the fundamental challenge of citation hallucination persists despite technological advances, with implications for academic integrity and the reliable use of AI in research contexts. We provide an online tool for researchers to run LLMs through the benchmark test using our prompts.

Suggested citation:

Buchanan, J., Hill, S. and Shapoval, O. (2026), LLM Hallucination of Citations in Economics Persists With Web-Enabled Models. Southern Economic J. https://doi.org/10.1002/soej.70069

After some time went by since we published our 2024 paper on this topic and GPT became web-enabled, we thought that this silly problem with hallucinating up fake citations might have gone away. So, we went back to do our count again. We found that a significant number of references are still made up.

I believe that this particular problem of fake citations will go away soon. However, I think the starkness of fake references illustrates something more general about the LLM technology and highlights the importance of verification. Feel free to cite our 2026 paper as evidence that hallucinations are still happening, despite improvements along many margins. Since we started down this track, OpenAI released a paper titled “Why language models hallucinate” at https://openai.com/index/why-language-models-hallucinate/ Excerpt:

Hallucinations persist partly because current evaluation methods set the wrong incentives. While evaluations themselves do not directly cause hallucinations, most evaluations measure model performance in a way that encourages guessing rather than honesty about uncertainty.

You can access our working paper of “LLM Hallucination of Citations in Economics Persists with Web-Enabled Models” on SSRN at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5318314

We are not in the slow-lane “LLMs are stochastic parrots” group. Stephen Hill does even more with AI in his personal workflow than I do. AI will probably soon qualify for all of the academic prizes we used to award to human writers and scientists. However, as we hand more responsibility over to AI, keep the issue of hallucinations in mind.

Is AI Taking Our Jobs? Signal Boosting Noah Smith

Joy Buchanan3 Comments

Artificial intelligence is supposed to transform the labor market. Depending on whom you ask, AI is either about to eliminate millions of jobs or usher in a new era of productivity and prosperity. I am reviewing the evidence before I teach an “unemployment” chapter…

I want to give a signal boost to economist Noah Smith, who has been following the emerging evidence about AI and employment at his blog, Noahpinion. His answer, so far, is that the evidence for widespread displacement is weak.

In his September 7 post, “AI keeps stubbornly refusing to take our jobs,” Smith points out a contrast. Many technology executives expect AI to make human workers obsolete, and public opinion surveys from Pew Research Center indicate that Americans share these concerns. Yet employment has held up remarkably well, especially among prime-age workers. AI might still take our jobs soon, but we have not seen widespread displacement yet. Economists have been saying for a long time that automating tasks is not the same thing as eliminating jobs.

In “Automation and New Tasks: How Technology Displaces and Reinstates Labor,” Daron Acemoglu and Pascual Restrepo explain that automation can reduce demand for labor performing particular tasks while also increasing productivity and creating new tasks for workers.

Consider software developers. AI can now write substantial amounts of code. One might conclude that fewer programmers will be needed. But programmers also design systems and decide what software should do. Making code less expensive could expand the amount of software that firms want to produce. Greater productivity may increase rather than decrease the demand for skilled programmers. (my old paper on who wants these jobs is “Willingness to be Paid: Who Trains for Tech Jobs?“)

Smith points readers to Guy Berger’s analysis of software developer employment and reporting on firms that have reconsidered replacing programmers with AI. These examples challenge the assumption that better coding tools necessarily mean fewer coding jobs.

One piece of evidence comes from the U.S. Census Bureau’s Business Trends and Outlook Survey, highlighted by Alex Tabarrok at Marginal Revolution and discussed by Smith. Among firms using AI, 44 percent reported that it supplemented or enhanced existing employee tasks. Ten percent reported that AI performed tasks previously done by employees, while 11 percent reported that AI introduced tasks nobody had previously performed. Most firms reported no employment change attributable to AI.

Smith also highlights research by Ara Kharazian, Jonathon Simon, and David Stevens who find that firms adopting AI tend to expand their employment, including entry-level employment. These findings contrast with the popular narrative that businesses are replacing junior employees with chatbots.

What about young workers?

Continue reading →

“Whinging” or Merely Whining?

Scott BuchananLeave a comment

I have heard the term “whinging” at least twice so far in 2026, both times from fairly sophisticated speakers. I don’t recall ever hearing it before this year. Is this a new trend?

“Whinging” is a British/Commonwealth term that means much the same as “whining,” although (surprisingly) they derive from completely different Old English roots. However, it feels a bit sharper and more dismissive than mere complaining. Thus, whinging is a more intellectual-sounding put-down than whining. And goodness knows how much we need more effective put-downs around here.

ChatGPT informs me that in the realm of finance, the word fills a useful niche. Calling someone a whinger often implies not just that they’re wrong, but that they’re engaging in repetitive, unproductive complaining:

In investing discussions, you’ll often see phrases like: “perma-bear whinging”, “shareholder whinging”, “whinging about management”, “stop whinging and buy the index”

where an American 30 years ago would more likely have written: whining, griping, bellyaching, complaining.

In economics, practitioners often want to distinguish between a legitimate complaint or critique, versus repetitive, emotionally-driven complaining. Thus:

Calling an argument “whinging” implies that the speaker is expressing dissatisfaction without offering analysis or solutions. That rhetorical function is useful in debates over inflation, trade, housing, inequality, central banking, or academic economics itself.

I suppose the first one who unleashes the term “whinging” wins; it is tough to dig out from being buried by a term like that. Knowing all this hopefully leaves you better armed for your next testy exchange over trends or policies.

Journal Crisis at Games and Economic Behavior

Joy Buchanan3 Comments

I used Grok for this query because economists were discussing the issue on X(Twitter) this week and Grok would have quick access to that discourse. Note that Grok provided links to legitimate web sources, not just Twitter speculation. I am providing those same links in the place where Grok put them. I do not know how much the chatter of the week affected the response below.

Prompt from J. B. on 8/22/26 to Grok: What happened to the economics journal Games and Economic Behavior?

The leading game theory journal Games and Economic Behavior (GEB), published by Elsevier since its founding in 1989, remains active and is still publishing issues (including volumes into 2026), but it is in a major editorial crisis due to a mass resignation of its leadership. (https://www.sciencedirect.com/journal/games-and-economic-behavior)

In late July 2026, Editor-in-Chief Hervé Moulin (who had held the role since January 2021) was informed by Elsevier that his 3-year contract would not be renewed, despite his willingness to continue. Elsevier cited a desire to move “towards a closer alignment with Elsevier strategic priorities and publishing policies.” Requests for further explanation (from Moulin and then collectively from the other six editors) were dismissed; Elsevier simply asserted its contractual right to make the decision and began searching for new editors without input from the existing board. (https://gametheorysociety.org/message-from-herve-moulin-geb-editor/)

In response, by August 14, 2026, all seven editors (Moulin plus Sushil Bikhchandani, Aislinn Bohren, Nicole Immorlica, Roberto Serrano, Emanuel Vespa, and Nicolas Vieille) resigned with the required three-month notice. Nearly all of the advisory editors (the equivalent of associate editors) also resigned in solidarity. The outgoing team is handling papers already under review and new submissions until November 14, 2026 (with some flexibility for papers already assigned by mid-August); after that, they will only finish evaluations already in progress. Elsevier has publicly thanked Moulin for his service. (https://gametheorysociety.org/message-from-herve-moulin-geb-editor/)

The journal continues under Elsevier ownership (it is one of the two official journals of the Game Theory Society, alongside the International Journal of Game Theory). Community discussion has focused on the implications for the field, the role of commercial publishers, and possibilities such as launching a successor journal. Similar recent mass resignations have occurred at other Elsevier titles over governance and editorial control issues. (https://casrai.org/news/2026-editorial-board-mass-resignations)

In short, GEB has not shut down, but its longstanding editorial team has largely departed over Elsevier’s non-renewal of the Editor-in-Chief and related decisions.

Prompt from J. B. on 8/22/26 to Grok: What will be the implications for the reputation of the journal?

The reputation of Games and Economic Behavior (GEB) is likely to face significant short- to medium-term damage, with longer-term outcomes depending heavily on how Elsevier handles the transition and whether a strong successor journal emerges.

GEB has long been viewed as the leading specialized journal in game theory (and one of the stronger outlets in theoretical economics more broadly). Its prestige rests on attracting high-quality, creative work; the expertise and independence of its editorial team; rigorous standards; and community trust. The mass resignation of all seven editors (including Editor-in-Chief Hervé Moulin) plus nearly the entire advisory board in August 2026, following Elsevier’s non-renewal of Moulin’s contract, directly undermines those foundations. (https://gametheorysociety.org/message-from-herve-moulin-geb-editor/)

Short-term effects

  • Loss of trust and submissions: Top researchers often choose journals partly based on the known editors and the expectation of fair, expert handling. With the board depleted (current editors only fully process papers until mid-November 2026), many authors are likely to pause or divert submissions. Community discussion already includes calls for boycotts of the Elsevier GEB and support for alternatives. @Afinetheorem
  • Perception of publisher interference: Elsevier’s stated rationale (“closer alignment with Elsevier strategic priorities and publishing policies”) and its decision to search for new editors without board input have fueled views that commercial priorities overrode academic ones. This echoes other recent mass resignations (e.g., Journal of Approximation Theory at Elsevier, where resigning editors declared the journal “as we have known it has effectively ceased to exist”). casrai.org
  • Operational disruption: Delays in reviews, potential quality variation under interim or new leadership, and uncertainty can further erode confidence among authors, referees, and readers.

Metrics such as impact factor or AJG ranking (currently 3) change slowly and will not immediately reflect the crisis, but subjective reputation in the game theory and economics communities can shift quickly.

Medium- to long-term scenarios

Historical patterns from similar mass resignations offer useful analogies:

  • “Zombie journal” risk: The title, ISSN, indexing, and historical impact factor stay with Elsevier. If the new editorial team is perceived as less distinguished or more aligned with volume/publisher goals, submissions from leading researchers can decline, citation patterns can shift (sometimes toward more geographically concentrated or lower-prestige work), and the journal’s standing within the field can permanently weaken. Cases like Lingua (Elsevier, 2015) and others show that the brand can persist while the scholarly niche contracts. sciencedirect.com
  • Recovery path: If Elsevier rapidly appoints highly respected new editors who maintain rigorous standards and rebuild community confidence, the damage can be limited. Some journals have stabilized after board changes, though full restoration of prior prestige is uncommon when the exit was acrimonious.
  • Successor journal effect: Strong community interest already exists in launching a new outlet (possible names floated include variations on “Theory of Games and Economic Behavior” or society-backed alternatives; comparisons are drawn to how JEEA or other breakaways have functioned). If a credible successor is established—especially one endorsed by the departing editors and the Game Theory Society—and is treated as the “true” continuation by tenure/promotion committees, the original GEB’s relative prestige would likely decline further. Successful breakaways (e.g., Glossa after Lingua, or Imaging Neuroscience after NeuroImage) have often captured much of the field’s attention and high-quality papers. journalmetrics.org

**end of Grok responses **
If you want some chatter, you can put “GEB” into the search bar at x.com, LinkedIn, or BlueSky this week and see some more speculation.

There is a lot happening with publishing, including a huge shock of strangeness from the ability to both write and review papers with LLMs. Tumult will be normal. Read Mike on Will AI kill the research paper? or Mike on The actual AI problem in academic economics or Mike on Academic economists are overcommitted . The real perennial problem is that nothing is free but no one wants to pay.

Lastly, dear reader, you might wonder why so much of my post was AI-generated this week. I see this as burning my tokens so you don’t have to. You can find this with a free web search. Some people predict that “cheap AI” is coming to an end, so we might find it worthwhile to publish responses as well as prompts for each other. Regards, fellow humans and AI readers.

AI Innate Preferences Paper on Arxiv

Joy BuchananLeave a comment

Please check out my new paper, with Joshua Foster

The Innate Economic Preferences of Language Models (arXiv link)

Abstract: Language models increasingly settle real resource tradeoffs on behalf of principals yet their economic preferences remain unobserved. We demonstrate their generation rule is isomorphic to the random utility model of discrete choice. This allows internal logit scores to structurally identify preferences. Estimating risk attitudes across twelve models in a portfolio task reveals universal but heterogeneous risk aversion. Although models reject strictly dominated options, their elicited preferences fail invariance tests and violate the independence of irrelevant alternatives across varying experimental prompts. Finally, fine tuning establishes that a principal can explicitly engineer a target risk attitude.

I hope you will refer to the manuscript for details, but I will share one picture here. This is panel (a) of Figure 3: Empirical indifference curves for open-weight models mapped over the portfolio space.

In simple language, what the red/blue picture shows is that the Qwen language model is picking the portfolios that offer more money (in expectation, with a distaste for excessive risk). That’s basically what a rational actor should do. We find that the language models make fairly consistent choices and rarely violate the monotonicity requirement for a well-behaved utility function.

How we describe this figure in the paper: “Starting from a base bundle with expected return µ = 10 and risk σ = 30, we sweep over the dense grid of alternative portfolios from our experimental protocol and record the position-corrected logit gap between each grid portfolio and the base. The yellow dashed line overlays the indifference curve implied by the mean-variance structural estimates, and the heatmap colors encode the sign and magnitude of the logit difference, with blue regions preferred to the base and red regions dispreferred. Several patterns emerge from these plots. All six models produce upward-sloping indifference curves, confirming that higher risk must be compensated by higher expected return.”

We think this basic research on behavior is important, for alignment research and for business applications with delegating work to AI agents. The first question to ask, before testing whether we can impose our preferences on AI agents, is whether those agents have preferences at all in a consistent sense.

Suggested citation: Buchanan, J., & Foster, J. (2026). The innate economic preferences of language models [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2607.26288

Uncertainty Increases Profits?

Zachary BartschLeave a comment

Most people have an intuition that uncertainty can harm economic outcomes. Baker, Bloom, & Davis (2016) and Bloom (2009) demonstrated that industrial production and manufacturing decline in the face of policy uncertainty. The typical mechanism that people suggest is that uncertainty about the future causes people to engage in precautionary saving, resulting in fewer sales.

The theory continues that firms consequently decrease production as demand for their output declines. Firms aren’t interested in causing the quantities supplied and demanded to be equal. Rather, they don’t want to produce too many goods that don’t get sold or don’t get sold at an adequate markup. Production is costly.  A related theory is that more persistent or longer-run uncertainty can also depress investment, since the riskier future increases the tail risk of losses.

Rather than make a risky investment, one could instead just hold off and wait for some of that uncertainty to get resolved. There’s tradeoffs to this, of course. As future costs and benefits become clearer, they also get priced-in to asset values. So, there is an optimization problem. The possible downside outcome is big and uncertain. If the risk of the investment gets resolved and the downside outcome is still too likely or harmful, then a project manager did the ex-post ‘right thing’ by waiting.

But, if the downside risk disappears or is found to be very small, then waiting to invest in the project incurs an economic cost. Either 1) the profitable project and its associated profits will occur later and less valuably, or 2) other firms also resolve their uncertainty and bid up the price of the project’s inputs. Invest too early, and the downside is large and uncertain. Invest too late, and you may lose the potential upside partially or entirely.

But can uncertainty systematically increase profits?

Walter Oi said yes.

Continue reading →

The Journal of Healthcare Finance Is Back

James Bailey1 Comment

Most academic journals are run by big for-profit publishing companies, and most of the rest are run by universities or big academic societies. The Journal of Healthcare Finance was an extreme outlier from this norm, run single-handedly by Editor-In-Chief James Unland since 1994. It was the rare journal that was free both for readers and authors.

I loved the idea of having a single person truly in charge and accountable without being slowed by a complex bureaucracy. But eventually a single person will want to, or have to, move on. Having an institution run a journal can ease this process, though an individual can still try to find their own successor.

In this case, The Journal of Healthcare Finance had been on hiatus since its Editor-In-Chief stepped back, with its last issue published in 2023. Their old website domain had expired- not a great look for anyone who published there and was going up for a job or tenure.

But now it is officially back at a new domain, with the single Editor-In-Chief replaced by a full editorial team, and accepting submissions again with the hope of releasing a new issue this year.

Selfishly, I’m happy to see this both because it means they will continue hosting my past publication, and to have a potential outlet for my future work. I recommend that other health economists and health services researchers give it a try, though as of now I have no personal experience with the new editorial team.

Do NBA Teams Play Worse In Back-To-Back Games?

James BaileyLeave a comment

The conventional wisdom is that the NBA regular season has too many games. Teams play worse because they are tired, or injured, or resting their stars so they can be ready to actually play hard in the playoffs.

New research shows that the conventional wisdom is…. probably right. In particular, teams play worse by many measures when they have to play two days in a row. That’s what Max Aicardi and I found in a paper published today, “Running on Empty: How Back-to-Backs Impact Pace and the Four Factors of Basketball Success“:

Teams on the second night of a back-to-back shoot less efficiently (lower eFG%), grab fewer offensive rebounds, and play at a slower pace. On defense, they allow opponents to shoot more efficiently, force fewer turnovers, and give up more free throw attempts and second-chance opportunities. Turnover percentage and offensive free throw rate did not change significantly, consistent with our conceptual framework’s distinction between effort-dependent and execution-dependent metrics. While not every metric changed significantly, the overall pattern is clear: second-night back-to-back scheduling is associated with a measurable decline in team performance

The effect sizes here tend to be small, around 0.5-2%, but they are statistically significant given that we studied over 20,000 games, and practically significant given how close NBA games are.

Max had the idea for this paper and wrote the first draft as a student in my Economics Senior Capstone class in 2025. After he graduated, I joined the paper as a coauthor to get it ready for journals. We share the data and code for the paper here.

Video on You Wouldn’t Steal a Car

Joy Buchanan2 Comments

The brilliant content creator economist Matt Hill has posted a video “How Piracy Accidentally Created AI” to the @EconNerds channel on YouTube.

The video is so funny (and smart!) that I encourage you to sit back and watch it all the way through. Around minute 2, he gets to the topic of online piracy.

The 4 minute mark is where I am featured to explain my paper with Bart Wilson: You Wouldn’t Steal a Car: Moral Intuition for Intellectual Property (SSRN link)

The @EconNerds channel on YouTube has over 100 engaging videos like this to help you learn economics. You can also find Econ Nerds updates on X/Twitter

I summarized our findings in a previous blog “Summary of You Wouldn’t Steal a Car,” but Matt’s video is more fun and quite technically accurate, so now everyone can just watch it.