Mitsubishi Research Institute and Mebuki Financial Group Agree to Officially Introduce Screening AI Service

Mitsubishi Research Institute, Inc.
Mitsubishi Research Institute and Mebuki Financial Group Agree to Officially Introduce Screening AI Service
Promoting the DX of examination work for housing loans and unsecured loans through automatic examination by AI

Mitsubishi Research Institute, Inc. (Headquarters: Chiyoda-ku, Tokyo, President: Kenji Yabuta, hereinafter referred to as MRI) and Mebuki Financial Group, Inc. (Headquarters: Chuo-ku, Tokyo, President: Tetsuya Akino, hereinafter referred to as Mebuki Financial Group) has agreed to introduce the “Screening AI Service” provided by MRI for the home loan and unsecured loan screening business of Mebuki Financial Group.
1. Background
As the environment surrounding financial institutions such as the business environment and the competitive situation becomes more severe, there is a need to promote DX with the aim of further improving “response speed”, “efficiency” and “objectivity” in loan screening operations for retail loan products. increase. Also, in order to adapt to the new post-corona normal, it is essential to rebuild the BCP (Business Continuity Plan).
From 2021, MRI and Mebuki Financial Group will use MRI for the screening of projects handled by Mebuki Credit Guarantee Co., Ltd. for housing loans and unsecured loans of Joyo Bank, Ltd. and Ashikaga Bank, Ltd., both members of Mebuki Financial Group. We have been conducting demonstrations toward the introduction of the “Screening AI Service” provided by (*1). As a result, it was possible to build an AI model with high discrimination performance, and we were able to evaluate that AI screening could replace conventional human screening, so we agreed to introduce the “Screening AI Service”.
*1: Mitsubishi Research Institute, Mebuki Financial Group and Mebuki Financial Group Commence Demonstration of Screening AI Service – Aiming to Automate Home Loan and Unsecured Loan Screening Operations Using AI – (News Release, December 15, 2021)
https://www.mri.co.jp/news/press/20211215.html
2. Overview of screening AI service
The “Screening AI Service” provided by MRI replaces the “Screening AI System” equipped with an AI model that learns the approval or rejection judgment (judgment of loan availability) that was
conventionally performed by humans (examiners). By linking with the examination business system, it promotes the automation of examination and supports the DX of examination work.
The AI ​​model calculates the approval probability for each project that receives an application, and it is possible to control the business flow according to this value. As a result, about 50% to 80% of the total cases can be automatically approved in real time without human review.
The AI ​​model of the “Screening AI Service” reproduces the credit policy of each financial institution using all the information necessary for acceptance/rejection judgment, so compared to the conventional method of credit scoring + rule checking on the system, etc. , enables a significant increase in automation rate.
Since the examination AI system and the examination business system are linked by API (*2), the examination business system needs only minor modifications such as interfaces, making it possible to introduce an AI model at a realistic time and cost. is. In addition, since an independent examination AI system is built for each financial institution, it is easy to monitor and maintain the AI ​​model, realizing high maintainability.
*2: API (Application Programming Interface) is a specification that defines an interface for linking data, etc. between different systems.                                    Vision
[Image 1

Source: Mitsubishi Research Institute
Inter-system cooperation using API
[Image 2

Source: Mitsubishi Research Institute
3. Future plans
From November 2022, we plan to conduct system development and acceptance tests for the use of this service, and to start practical application from December 2023. Through this initiative, we will work to further enhance the “Screening AI Service”, such as improving the performance and usability of AI models.
[Reference] Mebuki Financial Group, Inc. Corporate Overview
(1) Name Mebuki Financial Group Co., Ltd.
(2) Head office location: 2-1-1 Nihonbashi Muromachi, Chuo-ku, Tokyo (3) Title and name of the representative President and CEO Tetsuya Akino (4) Business content: Business management of banks and other companies that can be made subsidiaries under the Banking Act, and all incidental and related operations
(5) Capital 117.4 billion yen
(6) Date of establishment: April 1, 2008 (Trade name changed on October 1, 2016) [Reference] The Joyo Bank, Ltd. Corporate Profile
(1) Name The Joyo Bank, Ltd.
(2) Head office location: 2-5-5 Minamimachi, Mito City, Ibaraki Prefecture (3) Title and name of representative President (Representative Director) Tetsuya Akino
(4) Capital 85.1 billion yen
(5) Date of establishment July 30, 1935
[Reference] Ashikaga Bank, Ltd. Corporate Profile
(1) Name: The Ashikaga Bank, Ltd.
(2) Head office location: 4-1-25 Sakura, Utsunomiya City, Tochigi Prefecture (3) Title and name of representative Kazuyuki Shimizu, President (Representative Director)
(4) Capital 135 billion yen
(5) Date of establishment: October 1, 1895
[Reference] Mebuki Credit Guarantee Co., Ltd. Corporate Profile (1) Name: Mebuki Credit Guarantee Co., Ltd.
(2) Head office location: 4-1-25 Sakura, Utsunomiya City, Tochigi Prefecture (3) Title and name of representative President (Representative Director) Hiroo Shimada
(4) Capital 50 million yen
(5) Date of establishment: December 21, 1978 (Trade name changed on October 1, 2020)
【Related information】
・Mitsubishi Research Institute Launches Screening AI Service
–Promoting Automation of Retail Loan Screening Operations and Supporting Screening Operations DX- (News Release, December 3, 2020)  https://www.mri.co.jp/news/press/20201203.html
・Introduction of retail loan screening AI service (screening AI service leaflet)  https://www.mri.co.jp/news/press/dia6ou000002jbtj-att/nr20201203_leaflet.pdf

rehow

Leave a Reply

Your email address will not be published.

%d bloggers like this: