2019-2020

4月19日 (金): 16:30-18:00 富士見校舎3階303教室

発表者: 尾山大輔 (東京大学経済学研究科)

タイトル: Mean-Field Approximation of Forward-Looking Population Dynamics (joint with Ryota Iijima)

概要 :

We study forward-looking game dynamics in a large finite population. New agents constantly arrive and make irreversible choices to actions for stochastic length of durations after observing noisy signal about the aggregate state. We first show that, in the large population limit, (i) the aggregate dynamics under an equilibrium can be approximated by a continuous deterministic dynamics, and (ii) each agent ignores the effect of her action upon the population. The results provide a micro-foundation for non-atomic population dynamic models used in applications.

5月17日 (金): 16:30-18:00 富士見校舎3階303教室

発表者: 松島法明 (大阪大学社会経済研究所)

タイトル: Competitive Personalized Pricing (joint with Zhijun Chen and Chongwoo Choe)

概要:

We study a model where each competing firm has a target segment where it has full consumer information and can exercise personalized pricing, and consumers may engage in identity management to bypass the firm's attempt to price discriminate. In the absence of identity management, more consumer information intensifies competition since firms can effectively defend their turf through targeted personalized offers, thereby setting low public prices offered to non-targeted consumers. But the effect is mitigated when consumers are active in identity management since it raises the firm's cost of serving non-targeted consumers. When firms have sufficiently large and non-overlapping target segments, identity management can enable firms to extract full surplus from their targeted consumers through perfect price discrimination. Identity management can also induce firms not to serve consumers who are not targeted by either firm when the commonly non-targeted market segment is small. This results in a dead-weight loss. Thus identity management by consumers can benefit firms, and lead to lower consumer surplus and lower social welfare. Our main insight continues to be valid when a fraction of consumers are active in identity management or when there is a cost of identity management. We also discuss the regulatory implications for the use of consumer information by firms as well as the implications for management.

6月7日 (金): 16:30-18:00 富士見校舎3階303教室

発表者: 大和毅彦 (東京工業大学工学院)

タイトル: 実験経済学・実験ゲーム理論入門

概要 :

経済学とゲーム理論において実験の果たす役割について概説する。近年、経済学とゲーム理論に関する多くの学術雑誌には実験の論文が掲載され、ノーベル経済学賞も授与されている。セミナーでは市場実験と交渉ゲーム実験を実施し、実験結果と理論予測の比較と、他の実験結果の紹介も行う。

6月28日 (金): 16:30-18:00 富士見校舎3階303教室

発表者: 澤亮治 (筑波大学システム情報系)

タイトル: A Stochastic Stability Analysis with Observation Errors in Normal Form Games

概要 :

We perform a stochastic stability analysis with observation errors.Players recurrently play a symmetric two-player normal form game with one another and respond to the strategy distribution of other players. In each period, a revising player observes the strategy distribution and choose a best response to it. Her observation is perturbed with positive probability and she may respond to the misperceived strategy distribution. We examine the robustness of Nash equilibria to such observation errors. We find that the set of stochastically stable states under observation errors is robust to addition of strictly dominated strategies for a certain class of games given that observation errors are uniform, i.e., each misrepresented state is observed with uniform probability. We also examine the set of stochastically stable states under an alternate observation error model where the observation probability depends on the L1-norm between the true state and the observed state. For the local interaction model, we prove that the set of stochastically stable states under both observation error models is robust to addition of strategies that do not survive the iterative elimination of strictly dominated strategies provided that the population size is sufficiently large.

10月21日 (月): 16:30-18:00 富士見校舎 3階303教室

発表者: 岸下大樹 (東京大学大学院経済学研究科博士課程)

タイトル: Contagion of Populist Extremism (joint with Atsushi Yamagishi)

概要 :

To explore the propagation of populist extremism across countries, we construct a multi-country model wherein each country's politician sequentially implements a policy. Voters learn the incumbent politician's type and the desirable policy by observing foreign policies on top of the domestic policy. We first establish a preliminary result--- that populist extremism, wherein undesirably extreme policies are supported by voters, arises when the public opinion is sufficiently radical. We then show that populist extremism is contagious across countries through a novel mechanism: the dynamic interaction between the public opinion and implemented policies. The persistence of the domino effect depends on the correlation of the desirable policy across countries. Extremism eventually disappears under the perfect correlation, while either the convergence toward extremism hold or the cycles of extremism hold when the state of the world follows a Markov process without absorbing states. Our model is also applicable to various agency problems with observational learning.

11月15日 (金): 17:00-18:30 富士見校舎 3階303教室

発表者: 土橋俊寛 (大東文化大学経済学部)

タイトル: A Buyout Option Alleviates Implicit Collusion in Uniform-Price Auctions

概要 :

Theoretical research shows that in uniform-price auctions, a bidder will optimally submit bids below his valuations on all units except the first unit, which is referred to in the literature as demand reduction. Such demand reduction can be interpreted as implicit collusion among bidders when it causes extremely low revenues. In this paper, we investigate the potential of a buyout option for alleviating implicit collusion. We employ a two-stage model to study a uniform-price auction with a buyout option. We focus on the extreme case in which two bidders with two-unit demand submit a "single-unit bid," which yields a revenue of zero to the seller when there is no buyout option. Our main result is that the seller obtains a positive expected revenue unless the buyout price is high. We find that a bidder will exercise a buyout option even though he is risk neutral; that is, auction aversion is fully endogenous, in contradiction to the findings of previous work.

12月17日 (火): 16:30-18:00 富士見校舎 3階303教室

発表者: 中田里志 (東京理科大学経営学部)

タイトル: Stable Coalition Structures of Patent Licensing Games (joint with Ryo Shirakawa)

概要 :

We consider what coalition structures are stable in a Cournot oligopoly market with homogenous goods. Each firm's payoff is affected by a coalition to which it belongs as well as other firms' coalitions, so that we consider a coalition formation game with externality to deal with the problem. First, as a benchmark, we show that the pessimistic core is always non-empty, whereas the projective and optimistic core are always empty unless it is a duopoly market. Next, we consider an opportunity to make a contract with a patent holder of a new technology, which can be used to reduce each firm's production cost, and the effects on the stability of coalition structures. We show that, under certain conditions, the coalition structure where some firms contract with the patent holder to use the technology and other firms do not make any cooperation, which is a common assumption in the patent licensing games, can be a pessimistic core outcome. This result provides a theoretical foundation of the common assumption in the patent licensing games and clarifies when this assumption is plausible.

1月24日 (金): 16:30-18:00 富士見校舎 3階303教室

発表者: 福田恵美子 (東京工業大学工学院経営工学系)

タイトル: A discrete-time single-server Poisson queueing game: agent based simulation and laboratory experiments

概要 :

We consider a discrete-time single-server queue with a single acceptance period for a Poissonian population of homogeneous customers. Customers are served on a first-come first-served (FCFS) basis, and their service times are independent and identically distributed with a general distribution. We assume that each customer chooses her/his arrival-time slot with the goal of minimizing her/his expected waiting time in competition with other customers. For this queueing game, we derive a symmetric (mixed-strategy) Nash equilibrium. We also propose an agent-based model, which simulates the dynamics of customers who try to minimize their waiting times for service. Through numerical experiments, we confirm that this agent-based model achieves, in steady state, an arrival-time distribution similar to the equilibrium arrival-time distribution analytically obtained. Moreover, we conducted a series of laboratory experiments to examine the symmetric mixed-strategy equilibrium of two games with Poissonian population and with a fixed number of players.