SSRN

The Decision Economy: How Organizations Rely on Decisions They Cannot Check by Alexander Barrett.

Organizations routinely act on decisions made by others. Insurers rely on assessments they did not perform, banks accept diligence conducted elsewhere, and healthcare systems act on clinical determinations they cannot independently verify.

This paper argues that modern economies increasingly function through the exchange of accepted decisions rather than directly verifiable evidence. When decisions move across organizational boundaries, their credibility often depends on trusted institutions, accreditation systems, regulatory frameworks, or established channels, rather than on the receiving party's ability to verify the underlying facts itself.

The paper distinguishes different forms of decision acceptance and examines how institutions cope with non-verifiability, drawing on examples from healthcare, conformity assessment, public-health credentials, and financial services. Its central insight is that existing systems succeed not because they make everything verifiable, but because they create structured ways to manage trust when verification is costly or impossible.

Read: http://spkl.io/61817Me4P

The Microstructure of Wealth Transfer in Prediction Markets by Jonathan Becker.

Prediction markets are often praised for producing accurate forecasts. This paper looks beneath that success and asks a different question: who actually makes money within these markets?

Using 72.1 million transactions on Kalshi between 2021 and 2025, the study finds that while market prices are generally well-calibrated, a persistent wealth transfer occurs between different types of participants.

A striking finding is that YES contracts systematically underperform equivalent NO contracts, particularly for longshot events. The author labels the resulting premium paid by traders attracted to affirmative outcomes the "optimism tax."

The paper also shows that the main beneficiaries are not necessarily the best forecasters, but rather liquidity providers. Liquidity takers lose, on average, what liquidity makers gain, suggesting that much of the transfer arises from market microstructure and the accommodation of behaviorally biased trading rather than informational advantages.

The broader implication is that prediction markets may generate accurate aggregate forecasts even when many individual participants trade on biased beliefs. Market efficiency, in this view, emerges from a structure that enables sophisticated liquidity providers to absorb and monetize those biases.

Read: http://spkl.io/61847MeeI

The Help Me Search Era: Do We Choose "Better" with AI? by Raluca Ursu, Anita Rao, and Jake Embrey.

As AI-powered search tools increasingly shape how consumers discover products, an important question emerges: does AI actually help people make better choices?

This paper examines how tools such as ChatGPT and Google's AI search features present shopping information. Analyzing more than 27,000 AI-generated responses, the authors find that AI often uses a "best-for" format, highlighting the strengths of each product while leaving other attributes unstated.

The authors argue that this format can be highly effective but also potentially misleading. Consumers may receive only a partial picture unless they ask follow-up questions, creating situations where important drawbacks remain hidden. The paper develops a theoretical model and tests its predictions through a pre-registered experiment comparing AI-assisted search with traditional e-commerce search.

A key insight is that AI does not simply provide information differently; it may fundamentally reshape decision-making by determining which product attributes receive attention and which remain implicit.

Read: http://spkl.io/61807Menq

Uniform Rate-Setting and the Provision of Quality: Evidence from Financial Advisors by Ellen Longman.

Why do some financial advisory firms employ advisors with records of misconduct?

This paper points to an unexpected factor: uniform pricing. Using new data on advisory fee schedules, the author finds that many Registered Investment Advisor (RIA) firms charge nearly the same rates across clients with different wealth levels. As a result, some clients generate substantially more revenue than others.

The study shows that branches serving high-wealth areas, where revenue per client is higher, tend to employ advisors with cleaner records. In contrast, branches in less profitable markets are more likely to hire advisors with histories of misconduct. To explain this pattern, the paper develops a model in which firms balance quality against costs, leading them to rely more heavily on lower-cost, lower-quality advisors in markets with weaker economics.

The broader implication is that advisor quality may be shaped not only by regulation and oversight, but also by the economics of service provision and competitive pressures within the advisory industry.

Read: http://spkl.io/61827Me30

News Narratives and Omitted Credit-Cycle States in Corporate Bond Markets by Yiyuan Wang.

What if important parts of the credit cycle are not fully captured by traditional financial indicators?
Using more than 40 years of Wall Street Journal business news, this paper investigates whether news narratives contain information about credit-market conditions that standard credit-risk variables miss. The author constructs topic-based measures of narrative attention and examines their relationship with corporate-bond spreads across investment-grade and high-yield markets.

A key finding is that a substantial common component of credit-spread movements remains unexplained after controlling for conventional factors such as interest rates, yield-curve slope, equity markets, and option-implied measures. News narratives help capture part of this omitted variation, particularly in high-yield (HY) bonds, where credit risk is greatest.

The paper also finds a clear rating gradient: narrative measures add little explanatory power for higher-rated A bonds, somewhat more for BBB bonds, and the strongest explanatory power for high-yield credit. This suggests that public news narratives may be especially informative when assessing riskier segments of the corporate-bond market.

The broader implication is that credit cycles may be driven not only by measurable economic fundamentals, but also by evolving narratives that shape perceptions of risk and financing conditions.

Read: http://spkl.io/61887zhyM

Governing Agentic AI: Why Legal Personhood is Neither Necessary nor Sufficient by Shruti Rajagopalan.

As AI agents increasingly act autonomously by transacting, publishing, and interacting with external systems, some scholars have proposed granting them legal personhood. This paper argues that this debate is asking the wrong question.

The central claim is that the real challenge is not whether AI systems should have legal status, but whether legal systems can identify a responsibility-bearer when something goes wrong. Historically, nonhuman legal entities, from corporations and trusts to other legal constructs, have always operated through identifiable humans who can be monitored, sanctioned, replaced, or held accountable. Agentic AI can break that link.

The paper distinguishes between situations where developers and deployers are identifiable and those where no clear human actor exists. It then evaluates existing liability frameworks and argues that many struggle when responsibility becomes difficult to trace.

Rather than creating AI legal personhood, the author proposes a governance framework built around registration, identification, verification, financial responsibility, traceability, and suspension mechanisms. The goal is to ensure that a human or organization remains accountable throughout the lifecycle of an AI system.

Read: http://spkl.io/61837zhP3

Confirmation Bias in LLM Pricing Recommendations by Maxime C. Cohen and Eddy Hage-Youssef.

As businesses increasingly use LLMs for decision support, an important question arises: Are AI recommendations genuinely independent, or do they simply reinforce user suggestions?

In a large-scale experiment involving 350,000 pricing recommendations, the authors test how susceptible leading LLMs are to confirmation bias. The results are nuanced. Both models demonstrate meaningful independent judgment, particularly by discounting suggested prices when they become economically implausible.

However, that independence proves surprisingly fragile. The study finds that recommendations can be significantly influenced by conversational cues, source credibility, and, most notably, timing. A simple follow-up question such as "Are you sure?" can materially alter recommendations, even when the original answer was delivered confidently and with detailed justification.

The paper's key insight is that persuasive AI outputs may appear highly reasoned while remaining sensitive to subtle prompt framing and conversational pressure. For organizations deploying LLMs in pricing and other decision-making contexts, robustness may depend as much on interaction design as on model capability.

Read http://spkl.io/61887zhb2

Ownership, Reputation, and Trade in the Art Market by Gurgen Aslanyan and Sergey V. Popov.

Why do some artworks command extraordinary prices beyond their intrinsic artistic value?
This paper argues that reputation in the art market is shaped not only by artists, but also by who owns their works. The authors develop a network-based model in which the reputations of artists and collectors are jointly determined through ownership relationships, creating powerful reputation spillovers throughout the market.

A key insight is that collectors do not simply acquire art for its consumption value. The value of ownership can also depend on the status and reputation of other collectors linked to the same artist. These network effects influence trading patterns, prices, market entry, and specialization, helping explain several well-known features of art markets that are difficult to reconcile with standard economic models.

The paper's broader contribution is showing how ownership networks themselves become economic assets, shaping value creation and market outcomes. In this view, reputation is not merely an individual characteristic but a property of the network connecting artists, collectors, and artworks.

Read: http://spkl.io/61817zho5

Shadow Work and Protocols: A Contingent Case for Negligence in Agentic AI by Veronica Paternolli and Ryan Calo.

As autonomous AI agents begin acting on behalf of users in digital and physical environments, a key legal question emerges: should firms be held strictly liable for harms caused by these systems, or should liability depend on negligence?

Rather than arguing for a single rule, this paper identifies two overlooked factors that should shape the debate.

First, the authors focus on "shadow work" and "sludge": the administrative burdens, paperwork, and self-service tasks that organizations have gradually shifted onto consumers and citizens. Agentic AI could potentially reduce these burdens by handling routine tasks on people's behalf, creating significant social value.

Second, the paper highlights the rapid emergence of protocols and operational standards governing how AI agents interact with users and external systems. These protocols could provide practical benchmarks for determining whether firms acted reasonably, making negligence-based liability more workable.

The paper's central argument is that the choice between strict liability and negligence should depend on empirical realities: whether agentic AI genuinely reduces burdens on individuals and whether industry protocols evolve into credible standards of care. In this view, liability doctrine should not only respond to risks but also consider the social benefits that agentic AI may generate.

Read: http://spkl.io/61847zh9a

From Talk to Walk: Fiscal Communication and Asset Prices in China by Mohan Xu, Yixin Zhang, Runheng Li, and Yao Tang.

How much can government communication move markets before any policy is actually implemented?

This paper uses a RAG-enhanced large language model to analyze official Chinese government communications and construct a high-frequency measure of fiscal policy stance. The resulting indicator is validated against subsequent fiscal actions, enabling the authors to study how markets respond to fiscal signals in real time.

The findings suggest that China's fiscal transmission mechanism differs from that typically observed in advanced economies. Expansionary fiscal communication is associated with lower equity prices, weaker household consumption, stronger investment, a flatter yield curve, and an appreciating RMB. At the same time, it raises sovereign and corporate bond yields and supports economic output.

The authors argue that these seemingly unusual asset-price responses reflect the nature of China's investment-led fiscal model. In their framework, fiscal expansion boosts productive public capital and output, but also increases real interest rates and consumption risk premia, reducing asset valuations even as economic activity strengthens.

Read: http://spkl.io/61827zAaj