2023 Fiscal Year Final Research Report
Construction of the global inference theory in high-dimensional macroeconometrics
Project/Area Number |
19K13665
|
Research Category |
Grant-in-Aid for Early-Career Scientists
|
Allocation Type | Multi-year Fund |
Review Section |
Basic Section 07030:Economic statistics-related
|
Research Institution | Hitotsubashi University (2022-2023) Tohoku University (2019-2021) |
Principal Investigator |
UEMATSU Yoshimasa 一橋大学, 大学院ソーシャル・データサイエンス研究科, 准教授 (40835279)
|
Project Period (FY) |
2019-04-01 – 2024-03-31
|
Keywords | ファクターモデル / ベクトル自己回帰モデル / 高次元統計学 / 偽発見率 / スパース性 |
Outline of Final Research Achievements |
During the period of this research project, we primarily conducted studies related to large-scale factor models and vector autoregressive models. The research outcomes obtained are as follows: (1) We proposed a weak factor model induced by sparsity, which had not been considered before, and its efficient estimation method. (2) We proposed a statistical inference method to verify the sparsity. (3) We proposed a statistical inference method to detect Granger causality networks in large-scale vector autoregressive models.
|
Free Research Field |
統計学
|
Academic Significance and Societal Importance of the Research Achievements |
上記(1),(2)の学術的意義は,より実データに沿った弱いファクターモデルの理論を発展させた点にある.これにより,例えばより正確な経済予測が可能になる.こうした成果は2つの論文にまとめられ,共にJournal of Business & Economic Statisticsに掲載された.上記3の学術的意義は,大規模時系列に潜む新たな知見の発見につながるネットワーク関係を安定的に検出できる点にある.この成果は海外専門誌からの改訂要求を受けて改訂し再投稿中である.
|