Information Aggregation in Markets with Analysts, Experts, and Chatbots

summary

The gist

The present paper shows that it can be advantageous for traders to publish their information on the true value of an asset even if they (i) cannot build a position in the asset prior to the…

In short

The study examines how traders should publish their true asset values even without payment or prior position building. It shows that while analysts and experts can hinder information aggregation, a single informed trader gains an advantage. The informational content of prices is U-shaped regarding the number of published reports, suggesting a small group of experts can be an obstacle.

Key concepts

Benchmark Model Foundation
This baseline model uses the Grossman and Stiglitz (1976) framework to show how markets should work when information is perfectly aggregated. In this ideal scenario, equilibrium prices fully reveal the asset's true fundamental value to all traders.
Extension and Information Degradation
The model adds a second private signal that contains common errors. This error term causes demand functions to become correlated, preventing prices from efficiently aggregating all private information and leading to adverse selection in the market.
Incentive for Publication
An individual trader publishing their true value gains an informational edge. If only one person publishes, they can infer the true fundamental, while others relying on research notes cannot. Traders are incentivized to publish even without payment.
Informational Content of Prices
This measures how much true information prices convey. The content decreases when agents receive an analyst report, but it increases with the precision of both private signals and how well agents understand the analyst's findings.

Terminology used across episodes

This episode discusses

The paper

Information Aggregation in Markets with Analysts, Experts, and Chatbots · Read on arXiv

Wolfgang Kuhle, Corvinus University of Budapest, Max Planck Institute for Social Law and Social Policy

Corvinus University of Budapest · Max Planck Institute for Social Law and Social Policy

We study information aggregation when traders interpret public news, such as analyst reports, differently. This differential interpretation injects noise into equilibrium prices, compounding the informational bottleneck caused by noise traders. Next to analysts and experts, we consider the role of ``AI" chatbots, which (i) provide free financial advice and (ii) analyze traders' queries to learn about the asset's fundamental. Our model predicts that (i) chatbots aggregate information more efficiently than markets, (ii) increase trade volume, and (iii) enhance the informational content of prices. Ultimately, the AI platform extracts an absolute informational advantage over other market participants, highlighting a structural shift from information aggregation through market prices to ``AI" platforms.

Transcript

Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.

Ruth: We're looking at the paper: "Information Aggregation in Markets with Analysts, Experts, and Chatbots".

Warren: Here's the gist of the paper. The present paper shows that it can be advantageous for traders to publish their information on the true value of an asset even if they (i) cannot build a position in the asset prior to the publication of their information and (ii) cannot charge for the provision of information.

Ruth: The model also shows that the informational content of prices is U-shaped in the number of traders who publish their information.

Warren: Benchmark Model Foundation The paper establishes a benchmark model using the Grossman and Stiglitz (1976) model to show efficient information aggregation where equilibrium prices fully reveal the asset’s true fundamental.

Ruth: This baseline model assumes that the error terms in traders’ private signals are uncorrelated, such that equilibrium prices fully reveal the asset’s true fundamental.

Title and authors: Ruth: Let's talk about the paper Information Aggregation in Markets with Analysts, Experts, and Chatbots. It’s interesting because it tests how different types of private advice—analysts versus chatbots—impact market efficiency when you have all those players involved >

Warren: And it specifically looks at the results showing that if only one trader publishes their information about the true value, they can fully figure out the asset's fundamental value, but everyone else who got that report can't >

Iris: That’s a strong finding because it challenges the idea that more private information always leads to better market prices, suggesting a small group of distinguished experts can actually become an obstacle to aggregation >

Ruth: Exactly. It seems like the paper is arguing that the benefit of having analysts or chatbots isn't always positive for everyone in the market if you don't structure things right >

Warren: So what we saw was that when traders receive an analyst report, the informational content of those prices actually decreases because they no longer reveal the true fundamental value fully >

Iris: The paper also points out that this information content goes up if the private signal gets more precise, like when you have a better underlying analyst report, but it drops if agents don't understand that report very well >

The paper's summary: Ruth: So to summarize what the paper is showing us about this dynamic of information aggregation in markets with analysts and chatbots, it’s that the informational content of prices actually goes down when agents get these additional private signals >

Warren: It also shows that the informational content is U-shaped when you look at how many traders publish their information about their true asset value >

Iris: And the key takeaway there is about structure: information aggregation works best if either no trader publishes anything, or if every single trader publishes their private information >

Ruth: That sounds counterintuitive because we usually think having more opinions helps you get closer to the truth, but here it seems like an extreme case works better than a medium one >

Warren: And that brings us to the idea of expert groups: small groups of distinguished experts act as an obstacle rather than a help for aggregating information >

Iris: That is what Section four demonstrates about traders' incentives: they have an incentive to publish their private information even if they aren't getting paid for it > <ref:2411.01938#pg1>

The paper's improvements: Ruth: The paper suggests a few ways we can improve this situation, like ensuring that either no trader publishes their information at all, or making sure every single trader publishes their information to keep things efficient >

Warren: And another idea is that the AI system could actually help us identify when a small group of distinguished experts is acting as an obstacle instead of being a benefit for automated decision-making >

Iris: That's practical. It suggests we might strategically deploy expert advice, not just blindly trust it, by using the AI to detect when that group is causing problems in the information flow >

Ruth: It also seems like this framework could help us model the impact of publishing private information on price precision by showing that when just one trader publishes their information, the informational content of prices falls >

Warren: So we can use this as a way to assess the risk associated with spreading proprietary data because it clearly shows a trade-off between having that one piece of info and how much transparency you lose >

Conclusion: Ruth: To wrap up, the paper Information Aggregation in Markets with Analysts, Experts, and Chatbots suggests that a small group of distinguished experts is an obstacle to information aggregation >

Warren: It also shows that the informational content of prices is U-shaped based on how many traders publish their information >

Iris: And it confirms that in the "expert-free limit," where every trader publishes their opinion on the asset's value, prices once again reflect fundamentals accurately >

Ruth: It's a bit of a surprise because we see so much literature suggesting that getting better private information should improve how well prices reveal what's actually fundamental >

Warren: This paper provides a counterexample to that idea, showing that the expert-free limit works as expected when everyone participates in publishing their info >

Iris: It’s also worth noting the finding about how agents place a greater weight on an analyst report once its precision increases, which is interesting for how people use new information >

Ruth: And this whole study, Information Aggregation in Markets with Analysts, Experts, and Chatbots, shows that a trader who publishes their private information gains an information advantage over all other traders >

More episodes

← Home