| 大会名称 |
|---|
| 2022年 情報科学技術フォーラム(FIT) |
| 大会コ－ド |
| F |
| 開催年 |
| 2022 |
| 発行日 |
| 2022-08-30 |
| セッション番号 |
| 1e |
| セッション名 |
| 人工知能・ゲーム |
| 講演日 |
| 2022/09/13 |
| 講演場所(会議室等) |
| 12棟-105教室 |
| 講演番号 |
| CF-003 |
| タイトル |
| 制約付きマッチングのためのデータ駆動型課税規則に関する研究 |
| 著者名 |
| 松下 旦, 池上 慧, 奥村恭平, 冨田燿志, 岩崎 敦, |
| キーワード |
| 制約付きマッチング, 計量経済学, 反実仮想分析, マーケットデザイン |
| 抄録 |
| Real-world matching markets often regulate the number of matches for specific groups or types of agents. In Japanese medical residency matching, for example, the policymaker (PM) restricts the number of doctors matched in urban areas in order to maintain the minimum standards for health care services in rural areas. This paper proposes a tax scheme that regulates matching outcomes to satisfy the upper and lower bound constraints in the transferable utility setup, extending. Our framework enables the PM to (1) estimate agents' preferences merely from observed matching patterns and (2) compute the welfare-maximizing levels of taxes and subsidies that the PM must impose to satisfy the constraints. We prove that the unique optimal taxation can be easily obtained by convex programming. Moreover, we show how it performs better than the naive cap adjustment via simulation. |
| 本文pdf |
| PDF download (334.1KB) |