Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/238468 
Year of Publication: 
2020
Series/Report no.: 
ADBI Working Paper Series No. 1111
Publisher: 
Asian Development Bank Institute (ADBI), Tokyo
Abstract: 
We introduce the Credit Risk Database (CRD) and its contribution to financial inclusion efforts in Japan. By collecting financial data about small and medium-sized enterprises (SMEs), the CRD contributes to the overall understanding of the SME sector, to the adaptation of risk-based lending and to a fairer loan guarantee system. In addition to financial data, the CRD also includes alternative data, bank account transaction data, when assessing SME credit. A machine learning model is adopted to process the extremely large body of transaction data. The best performing predictors of default include cash balance and cash outflow related to repayments. The machine learning model outperforms the logistic model and is highly accurate in predicting the probability of short-term default. The alternative data and model can serve as both an enabling short-term monitoring instrument and a credit assessment tool for SMEs without financial statements.
Subjects: 
SME credit assessment
machine learning
transaction data
JEL: 
G21
G28
G32
Creative Commons License: 
cc-by-nc-nd Logo
Document Type: 
Working Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.