• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Development of a Practically Applicable Estimation Method for Expected Loss Ratio of SMEs: Using Integrated Big Data of Regional Banks

Research Project

Project/Area Number 20K13581
Research Category

Grant-in-Aid for Early-Career Scientists

Allocation TypeMulti-year Fund
Review Section Basic Section 07080:Business administration-related
Research InstitutionThe Institute of Statistical Mathematics

Principal Investigator

Nagahata Hideaki  統計数理研究所, リスク解析戦略研究センター, 外来研究員 (00815128)

Project Period (FY) 2020-04-01 – 2023-03-31
Project Status Completed (Fiscal Year 2022)
Budget Amount *help
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2022: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2021: ¥2,470,000 (Direct Cost: ¥1,900,000、Indirect Cost: ¥570,000)
Fiscal Year 2020: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Keywords信用リスク / 機械学習 / データ構造化 / データ結合 / デフォルト時損失率 / ビックデータ / 転移学習
Outline of Research at the Start

本研究は地銀複数行の統合ビッグデータを用いた期待損失率推計を扱い、全銀行に対応できる汎用性のある推計手法を開発・提案する。
銀行の抱える貸出リスクは、貸出先企業の期待損失率の推計によって補足される。期待損失率の要因はデフォルト確率とデフォルト時の貸出残高損失率(Loss Given Default; LGD)に分解される。しかし、統合データベースの欠如が根本原因となり、LGD推計はごく一部の手法しか提案されていない。本研究によって期待損失率推計に対し統計・機械学習を用いた推計手法が確立され、その成果は信用リスク研究の発展、銀行の融資審査の高度化、金融行政の合理化、中小企業金融の円滑化に貢献する。

Outline of Final Research Achievements

Using the database for estimating loan loss ratio (LGD) at default, we aimed to realize data structuring that would contribute to improving the accuracy of LGD estimation. Specifically, we structured the data to include the time point t before and t after the default point. The data structure was confirmed to be effective for estimation by learning statistical models and machine learning, since it can incorporate default-related information more deeply and extensively in order to improve estimation accuracy. We believe that we have made clear progress not only in improving the accuracy of LGD estimation through the sophistication of statistical models and machine learning, but also as a result of our review from the database used for estimation, which pivoted to focus on improving estimation accuracy and gaining new knowledge.

Academic Significance and Societal Importance of the Research Achievements

本研究は一般では公開されていない担保・保証・債務回収の情報を含む、地銀5行統合データベースを用いた期待損失率推計を扱い、汎用的で高精度な推計手法を開発・提案します。
これによって、期待損失率推計に対し統計的・機械学習的接近法を用いた推計手法が確立・実務利用され、その成果は信用リスク研究の発展、銀行の融資審査の高度化、金融行政の合理化、中小企業金融の円滑化に貢献することを目指します。

Report

(4 results)
  • 2022 Annual Research Report   Final Research Report ( PDF )
  • 2021 Research-status Report
  • 2020 Research-status Report
  • Research Products

    (2 results)

All 2023 2020

All Journal Article (1 results) (of which Peer Reviewed: 1 results) Book (1 results)

  • [Journal Article] Comparison study of two-step LGD estimation model with probability machines2020

    • Author(s)
      Tanoue Yuta、Yamashita Satoshi、Nagahata Hideaki
    • Journal Title

      Risk Management

      Volume: 22 Issue: 3 Pages: 155-177

    • DOI

      10.1057/s41283-020-00059-y

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Book] ANOVA with Dependent Errors2023

    • Author(s)
      Yuichi Goto, Hideaki Nagahata, Masanobu Taniguchi, Anna Clara Monti, Xiaofei Xu
    • Total Pages
      100
    • Publisher
      Springer Nature
    • Related Report
      2022 Annual Research Report

URL: 

Published: 2020-04-28   Modified: 2024-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi