Risk Perception, 2025, JEEA 23(4), 1309–1349, DOI, with Nick Netzer, Arthur Robson and Jakub Steiner. (paper)
In a model inspired by neuroscience, we study choice between lotteries as a process of encoding and decoding noisy perceptual signals. The implications of this process for behavior depend on the decision-maker’s understanding of risk. When the aggregation of perceptual signals is coarse, encoding and decoding generate behavioral risk attitudes even for vanishing perceptual noise. We show that the optimal encoding of lottery rewards is S-shaped and that low-probability events are optimally oversampled. Taken together, the model can explain adaptive-risk attitudes and probability weighting, as in prospect theory. Furthermore, it predicts that risk attitudes are influenced by the anticipation of risk, time pressure, experience, salience, and availability heuristics.
Boundedly Rational Demand, 2024, Theoretical Economics 19(4), 1415–1442, DOI, with Jakub Steiner and Colin Stewart. (paper)
Evidence suggests that consumers do not perfectly optimize, contrary to a critical assumption of classical consumer theory. We propose a model in which consumer types can vary in both their preferences and their choice behavior. Given data on demand and the distribution of prices, we identify the set of possible values of the consumer surplus based on minimal rationality conditions: every type of consumer must be no worse off than if they either always bought the good or never did. We develop a procedure to narrow the set of surplus values using richer data sets and provide bounds on counterfactual demands.
Fixed point theorems and ergodic theorems for nonlinear mappings in Banach spaces, 2011, Advances in Mathematical Economics 14, 67–87, DOI, with Wataru Takahashi and Jen-Chih Yao. (paper)
Weak and strong convergence theorems for generalized hybrid nonself-mappings in Hilbert spaces, 2011, J. Nonlinear Convex Analysis 12(3), 453–470, with Wataru Takahashi and Jen-Chih Yao. (paper)
Fixed point theorems and weak convergence theorems for generalized hybrid mappings in Hilbert spaces, 2010, Taiwanese Journal of Mathematics 14(6), 2497–2511, DOI, with Wataru Takahashi and Jen-Chih Yao. (paper)
An elementary new proof of the determination of a convex function by its subdifferential, 2010, Optimization 60(1–2), 3–12, DOI (paper)
Revealing Private Information in a Patent Race, with Eugen Kovac (slides, paper)
In this paper we investigate the role of private information in a patent race. Since firms often do their research in secrecy, the standard assumption in the patent race literature that firms know each other's position in the race is questionable. We analyze how the dynamics of the game change when a firm's progress is its private information. Further, we address the question whether revealing it might be to a firm's advantage. We find that a firm has an incentive to reveal its breakthrough only if its rival has not done so, and only if research is inefficient.
Multi-Player Discrete All-pay Auctions (paper)
In this paper I study all-pay common-value auctions in which bids are restricted to non-negative integers. I prove that the game has a unique symmetric Nash equilibrium in mixed strategies given that there are three or more players. Although players bid on average lower as more players are present, they always randomize on the whole set of bids smaller than the value of the prize, so long as there is sufficiently many players. Players always receive a positive expected payoff which is bounded by a constant regardless of how large the value of the prize is. Finally, I prove that in the limit the equilibrium converges to the equilibrium of a continuous all-pay auction.
Secrecy vs. Patenting in Innovation Races, with Eugen Kovac
This study examines the trade-off between patenting and secrecy in innovation races, considering a model where two firms simultaneously compete in developing two products that can be substitutes or complements. Patenting ensures a claim on the product but discloses information to rivals, while secrecy may delay immediate profits for future technology leadership. We find that firms have more incentives to patent if they are impatient and if there are no significant technological spillovers. In a scenario where firms are patient and moderate technological spillovers exist, they exhibit a greater tendency to patent products acting as perfect complements rather than perfect substitutes. These findings are in line with the empirical evidence by Cohen, Nelson, and Walsh (2000), who argue that firms are more likely to keep the innovation secret in "simple" industries, where goods have many potential substitutes, as opposed to "complex" industries, where a new product involves many complementary components.
Frontier AI Lab Competition and Model Release Decisions
This project studies when a firm that develops a state-of-the-art AI model might choose to use it only internally rather than releasing it to the public (e.g., via an API). AI inference has a distinctive feature as a product: it serves as an input into further R&D — for instance, it can augment human labor in programming and other tasks — and the firm selling access benefits from observing how users employ the model. I study under what competitive conditions a frontier AI lab would keep its most powerful model behind closed doors. This question has important policy implications: if leading labs withhold their most capable models, policymakers and the public may underestimate the true state of AI capabilities and the associated risks, leaving critical infrastructure unprepared for threats such as an accidental leak or theft of a powerful model's parameters.
Convergence or Divergence? The Future of Frontier AI Capabilities and Implications for Catastrophic Risk
Will the capabilities of frontier AI models continue to converge, or will they diverge? I study this question using game-theoretic tools from the innovation race literature, with a focus on the distinct catastrophic risks each trajectory entails — proliferation under convergence, concentration of power under divergence. The analysis examines the strategic role of training data, algorithmic diffusion, and compute costs in shaping frontier AI market structure, and draws implications for AI governance.