arXiv subjects

Hector Tzavellas

Publications and source records attributed to Hector Tzavellas.

3 recordsLinked to original sources

arXiv2026

Do People Follow AI Advice? Evidence from a Pension Portfolio Choice Experiment

We study how differences in AI-generated financial recommendations are transmitted into individual portfolio choices. In an experiment with 400 employed adults enrolled in workplace defined contribution pension plans in South Korea, participants allocate a hypothetical pension balance across eleven products and may revise it after receiving one of two fixed AI-generated recommendations. A $2 \times 2$ design randomizes recommendation content and whether the recommendation includes a short rationale. Approximately 37$\%$ of the experimentally induced difference between the aggressive and conservative recommendations passes through to final portfolios. This causal contrast changes expected portfolio return, volatility, allocations across risk grades, and the number of products held, but produces no detectable difference in computed Sharpe ratios. 81$\%$ of participants revise. Among revisers, 95$\%$ move toward the assigned recommendation and implement about half of the suggested adjustment. Rationales do not detectably alter pass-through. These results show that users partially and selectively transmit recommendation content into economically meaningful differences in risk exposure while retaining substantial weight on their initial choices.

econ.GN↗

arXiv2024

Can an LLM Learn Preferences from Choice Data?

Can large language models (LLMs) learn a decision maker's preferences from observed choices and generate preference-consistent recommendations in new situations? We propose a portable Simulate-Recommend-Evaluate framework that tests preference learning from revealed-choice data by comparing LLM recommendations with optimal choices implied by known preference primitives. We apply the framework to choice under uncertainty using the disappointment aversion model. Recommendation accuracy improves as models observe more choices, but learning is heterogeneous across preference types and LLMs: GPT learns risk aversion better than disappointment aversion, Gemini performs best in high disappointment-aversion regions, and Claude shows the broadest effective learning across parameter regions.

econ.GN↗

arXiv2023

Network Beliefs and Behavior with Peer Effects

Individuals often act without knowing the full structure of the social network in which they are or will be embedded. We study how an individual's beliefs about their networks shapes their behavior when actions are peer interactive. Agents use what they know about the network to forecast their peers' actions. Those peers' actions depend on their beliefs, which then generate an iterative expression what we call "Iterative Belief Centrality." Agents' beliefs formed based on what they each see of the network are heterogeneous, depend on their network position, and can be correlated across connected agents. The resulting equilibrium behavior nests both complete-information and degree-based models as special cases, but more generally can differ systematically. If people's beliefs about the network satisfy a natural monotonicity condition in how connected they are, then belief iteration (fully rationally) amplifies behavioral differences across the network, increasing actions of more-connected and decreasing actions of less-connected agents relative to situations with homogeneous beliefs. We also show how positive correlation in people's positions in the network even further amplifies the variance of behavior. The framework provides a unified and tractable theory of network-based behavior with implications for many applications.

econ.TH↗