The Asian Bankers Association (ABA), the Association of Credit Rating Agencies in Asia (ACRAA) and CRD Association Japan wish to invite fellow bank members, investors and associates to join the 90 minutes webinar on “Enhancement of Banks’ Lending and Credit Risk Assessment: Scoring Model Using Machine Learning & Transaction data” to be held on 13 April, 2022 at 2PM Taipei time.
The webinar will feature Jiro Tsunoda, Senior Advisor CRD Association Japan (Former Principal Portfolio Management Specialist, Asian Development Bank), and Dr Lan Nguyen, Deputy Director, International Department CRD Association Japan (Former Research Associate, Asian Development Bank Institute).
Megumi Sagara, Director, International Department, CRD Association Japan and Dr. Khaliun Dovchinsuren, Senior Analyst International Department, CRD Association Japan will join the Q&A session.
I. Welcome Remarks by Moderator Grace Lee, ACRAA Board Director (3 minutes)
II. Review of approaches for a scoring model, and the significance of CRD system by Jiro Tsumoda (15 minutes)
This introductory presentation will elaborate on the following issues:
· Methodology applied for the scoring models and the User’s goal of the model.
· Data availability (public or in-house) and a scoring model.
· Significance of the CRD system in Japan, members and the financial statements/default data approach – the traditional CRD scoring model.
· Snapshot of using the bank account transaction – the machine learning scoring model.
· Summary of two types approaches.
II. Scoring Model Using Machine Learning (ML) & Transaction data by Dr. Lan Nguyen (45 minutes)
This part of the presentation aims to assist banks’ initial effort to utilize the bank’s own data, specifically transaction data, to build a ML scoring models. This is the data which is often left unexplored due to limited resources to gather a team of data scientists, financial and credit risk experts.
This part of presentation will cover the following issues:
· Benefits of ML in lending & credit risk management.
· Applications of ML in lending & credit risk management.
· Comparison of different machine learning algorithms.
· Development of a ML model (1): Data exploration.
· Development of a ML model (2): Model selection and training.
· Evaluation of a ML model: Performance assessment criteria and interpretability.
· Proof of Concept: Process & Results.
III. Q&A session (17 minutes)
Audience will be able to issue questions to the speakers.
We therefore encourage you and your colleagues to register for this webinar by clicking HERE.
ACRAA was organized on 14 September 2001 at the Asian Development Bank headquarters, by 15 Asian credit rating agencies from 10 countries. As of September 2022, membership has increased to 28 members from 15 countries. An initiative of the ABA, ACRAA unites domestic credit rating agencies in a regional cooperative effort to develop and promote interaction and exchange of ideas, information, and skills among credit rating agencies in Asia to enhance their capabilities and their role of providing reliable market information. ACRAA also aims at promoting the adoption of best practices and common standards that ensure high quality and comparability of credit ratings throughout the region, following the highest norms of ethics and professional conduct.
CRD (Credit Risk Database) was founded in March 2001 as a non-profit and a membership organization (credit guarantee corporations and financial institutions) in response to the policy measures for promoting the financing to the small and medium enterprises (SMEs) initiated by the Japanese government agencies (Ministry of Economy, Trade and Industry and Bank of Japan). As of 1 April 2022, Membership (168 entities in total) is comprised of 96 private financial institutions, 51 credit guarantee corporations, 4 government affiliated financial institutions and 17 credit rating agencies and others. Several government agencies such as Small and Medium Enterprise Agency, Bank of Japan and Financial Service Agency are also utilizing the services of CRD. Our services to the members are comprised of (i) the scoring, (ii) the statistical information, (iii) the sample data, (iv) the management consulting support system as well as (v) the consulting services such as the validation and reconstruction of internal rating system, the credit risk measurement, the housing loan database, the apartment loan database, and the education. CRD’s database is composed of the financial statements of 25,210,000 incorporated SMEs with 3,612,000 default information, and those of 6,428,000 sole-proprietor SMEs with 964,000 default information.