Jump to main content
Back25/07/2025

Game Theory Miniworkshop 

The Corvinus Institute for Advanced Studies is organizing a game theory mini workshop on the 29 July from 14:00-17:00 at Corvinus. 

Date: 29 July 2025, 14:00-19:00 

Location: Covinus University of Budapest, building C, C.714 

Language of the event: English 

Preliminary Program 

Attila Ambrus (Duke): Learning from Past Negotiations: Theory and Evidence from Somalian Piracy  

Abstract: We analyze the time pattern of bargaining outcomes in ransom negotiations with Somali pirates, using a unique data set with a comprehensive coverage of kidnapped ships in 2002-2012. We find that, even when controlling with ship characteristics, negotiated ransoms initially increased by a large magnitude, followed by negotiation durations sharply increasing as well. In the last years of the time period, both average ransom levels and negotiation durations seemed to stabilize. We argue that the main force behind these changes was learning by the pirates about the distribution of valuations of the buyers (ship owners). To investigate this issue theoretically, we analyze a model involving a sequence of negotiations with different buyers and sellers, in which buyers’ valuations are drawn independently from the same distribution, initially unknown to the sellers. Sellers observe past negotiations and update their beliefs on the distribution accordingly. We provide conditions under which over time sellers learn the true distribution of valuations. We use our model framework for structural estimations and find that pirates’ beliefs over time did move closer to the true distribution of valuations, although not all the way. We use the estimated parameters of the model for welfare analysis and investigation of counterfactual scenarios such as bargaining outcomes with pirates starting out with correct beliefs. 

Eszter Kabos (Oxford): Unstoppable 

Abstract: Whether, when, and how Artificial Intelligence (AI) will substitute for human labor are all active debates. This paper evaluates the ability of several leading Large Language Models (LLMs) to solve strategic decision-making problems from the social sciences. On average, we find that LLMs perform similarly to highly competent humans, but the type of problem affects solution rates. LLMs perform better when faced with problems that require numerical calculations of expected values. They perform worse when given tasks that are: a) not based on well-known ‘textbook’ problems, b) involve complex equilibrium reasoning (such commitment problems) or c) require novel economic insights. 

Mariann Ollar (NYU Sanghai): Incentive Compatibility and Belief Restrictions 

Abstract: We study a framework for robust mechanism design that can accommodate various degrees of robustness with respect to agents’ beliefs with the belief-free and Bayesian settings as special cases. For general belief restrictions, we characterize incentive compatible direct transfer mechanisms in general environments with interdependent values. Our main result provides a first order approach to incentive design under belief restrictions. It informs the design of transfers via belief-based terms to attain incentive compatibility. Using the resulting design principle, we provide possibility results in environments that violate standard single-crossing and monotonicity conditions. We discuss a robust version of the revenue equivalence theorem that holds under a notion of independence generalized to non-Bayesian settings. Further, to contrast implications with the well-known anything goes results from Bayesian mechanism design, within our framework, we study a fairly general model of comovement of payoff types and beliefs. We show that while under comovement implementation possibilities are rich, full rent extraction typically can not be attained. 

Kristof Madarasz (LSE): TBA 

For further information, please contact Péter Vida (peter.vida@uni-corvinus.hu). 

Copied to clipboard
×