TOP > 研究活動 > 研究会

研究会

信州大学経法学部において月1回のペースで開催される「研究会」は、経済学、経営学、法学、政治学など、社会科学諸分野の研究成果 について報告・議論する場を提供し、さまざまな研究トピックスに関して相互理解を深めるとともに、研究者間でのコミュニケーションの促進を図ることを目的としています。
構想段階の研究や調査進行段階の研究も発表可能であり、研究者間の意見交換を通 じて研究内容の発展を図るなど、建設的な議論が展開されています。また、報告者は信州大学の教員にとどまらず、他機関の研究者も積極的に招き入れ、より広範なトピックスを取り扱うことを目指しています。
開催スケジュールと内容については、本ホームページに随時掲載する予定です。

  • 講演者 明石 郁哉氏
    所属 東京大学経済学部
    日 時 2023年2月17日(金) 13:30~15:00
    場 所 信州大学経法学部 研究棟4階 研究会室
    題目 Spatial-median based self-weighted GEL and robust low-rank regression for VAR models
    概 要 本講演の前半では、ベクトル自己回帰モデル(VARモデル)に対して中央値回帰の手法を多次元化して適用し、これまでスカラー値確率過程に対して用いられてきた手法を拡張する。また自己加重法を用いて統計量を頑健化し、漸近正規性をもつことを示す。一方で係数行列に対する非線形仮説の検定は、モデル診断や因果性解析の際に基礎的になるが、非母数的な尤度を構成する手法である一般化経験尤度法を用いて統計量を構成し、実行可能な検定方式を構成する。本講演の後半では、VARモデルの係数行列にある種の低ランク性を仮定し、次元削減と推定を同時に行う手法を構成する。特に有限分散過程で用いられてきたアルゴリズムを最小距離回帰の手法を用いて頑健化し、シミュレーションによりそのパフォーマンスを既存手法と比較する。
  • 講演者 栗野 盛光氏
    所 属 慶應義塾大学経済学部
    日 時 2022年9月13日(火)13:30~15:00
    場 所 信州大学経法学部 研究棟4階 研究会室
    題 目 Inter-attribute equity in assignment problems: Leveling the playing field by priority design (co-authored with Tetsutaro Hatakeyama)
    概 要 Priorities over agents are crucial primitives in assignment problems of indivisible objects without monetary transfers. In this paper, we introduce a prioritization problem: there are several exogenous attributes and each agent is equipped with one such attribute, and the priority is initially determined only for agents with the same attribute. This leads to a partial priority order. Our problem is to construct a complete priority, which is needed to implement a known mechanism such as the serial dictatorship. We propose a simple prioritization rule called the relative position rule. We formulate three equity axioms and an invariance property; the priority preservation law, the equal treatment of equal positions, the equal split, and the attribute-wise consistency. We show that the relative position rule is characterized by these equity axioms. The result is applicable to general assignment problems with partial priorities. In the context of college students' exchange programs, the rule levels the playing field in the sense that inequality across attributes is partially reduced.
  • 講演者 島田 夏美
    所 属 信州大学情報・DX推進機構
    日 時 2022年9月13日(火)10:00~11:30
    場 所 信州大学経法学部 研究棟4階 研究会室
    題 目 Factors Causing Non-sincere Bidding in the Vickrey auction (co-authored with Shigehiro Serizawa and Tiffany Tsz Kwan TSE)
  • 講演者 長島 稜剣氏
    所 属 横浜国立大学経済学部(学生)
    日 時 2022年2月2日(水)17:30~18:00
    場 所 オンラインミーティング
    題 目 Nash implementation of weakly stable correspondence with distributional constraints
    概 要 In this study, we consider many to one matching problems with distributional constraints. For the theory of two-sided matching, stability is a key solution concept and the main objective of studies. In contrast to the conventional model, however, the notion of stability varies in a matching model with distributional constraints since some deviation sometimes violates feasibility. Kamada and Kojima (2017) propose two notions of stability, strong stability and weak stability. They show that the former does not necessarily exist and the latter always exists. However, it is unknown how to achieve a weakly stable matching in some strategic solution concepts. This paper focuses on Nash equilibria and asks whether a weakly stable correspondence is Nash implementable or not. Surprisingly, Nash implementability of the weakly stable correspondence depends on the total number of doctors. We show that, on one hand, the weakly stable correspondence is Nash implementable, when the number of doctors is large enough to overflow the seats, or when it is small enough not to violate feasibility constraints, on the other hand, when the number of doctors does not fall in the above, the weakly stable correspondence is generally not Nash implementable.
  • 講演者 平野 飛鳥氏
    所 属 横浜国立大学経済学部(学生)
    日 時 2022年2月2日(水)16:30~17:30
    場 所 オンラインミーティング
    題 目 Compatibility between incentives and efficiency in the probabilistic assignment problem on the full preference domain
    概 要 We consider the probabilistic assignment problems of n agents and n objects in which every agent receives probabilities that he is allocated each object on. For those problems, there are two main desiderata: strategyproof-ness and ordinal efficiency. When all agents have strict preferences, the random priority (RP) mech-anism satisfies strategyproofness, ex post efficiency and equal treatment of equals. The probabilistic serial (PS) mechanism which Bogomolnaia and Moulin (2001) propose is weakly strategyproof, ordinally efficient and envy-free. Thus, efficiency and incentives are compatible. Katta and Sethuraman (2006) extend PS mechanism to the full preference domain, which indifferences are allowed. Their extended PS mechanism satisfies ordinal efficiency but not even weak strategyproofness. They further show that there is no mechanism that is weakly strategyproof, ordinal efficient and envy-free. Consequently, it is still unknown if there exists a mechanism which possesses both efficiency and incentives on the full preference domain. Finally, we propose a novel extension of the RP mechanism to the full preference domain, which satisfies strategyproofness, ex post efficiency and equal treatment of equals. Therefore ex post efficiency and strategyproofness are compatible even on the full preference domain.

pagetop