Credit Risk Assessment for Small and Microsized Enterprises Using Kernel Feature Selection-Based Multiple Criteria Linear Optimization Classifier: Evidence from China

Abstract

Credit risk assessment has gained increasing marked attention in the recent years by researchers, financial institutions, and banks, especially for small and microsized enterprises. Evidence shows that the core of small and microsized enterprises’ credit risk assessment is to construct a scientific credit risk indicator system, and the key is to establish an effective credit risk prediction model. Therefore, we analyze the factors that influence the credit risk of Chinese small and microsized enterprises and then construct a comprehensive credit risk indicator system by adding behaviour information, supervision information, and policy information. Furthermore, we improve the multiple criteria linear optimization classifier (MCLOC) by introducing the one-norm kernel feature selection and thereby establish the kernel feature selection-based multiple criteria linear optimization classifier (KFS-MCLOC). As for experiments, we use real business data from a Chinese commercial bank to test the performance of these models. The results show that (1) the proposed KFS-MCLOC has greater advantages in predictive accuracy, interpretability, and stability than other models; (2) the KFS-MCLOC selects 10 features from 53 original features and gives selected features their weight automatically; (3) the features selected by the KFS-MCLOC are further verified and compared by the features selected by the logistic regression model with stepwise parameter, and the indicators of “quick ratio; net operating cash flow; enterprises’ abnormal times of water, electricity, and tax fee; overdue days of enterprises’ loans; and mortgage and pledge status” are proved to be the most influencing credit risk factors.

How Far Has Our MSMEs Credit Underwriting Assessment in Indonesian Commercial Banks Progressed?

“…However, the statistical modeling must be at a representative number of samples to get a model near the population, which sometimes takes time and cost. The statistical modeling relies on assumptions, cannot deal effectively with high-dimensional data, and has costly computation time (Wang & Zhang, 2020). This has been a classical statistical method challenge since credit scoring was first used for assessing MSMEs credit applications as the debtor's condition, credit facilities, and loan purpose are more heterogeneous than consumer credit or mortgage (Mester, 1997).…”

Credit risk assessment of small and micro enterprise based on machine learning

“…For instance Ref. [ 15 ], proposes a “two-dimensional” credit evaluation paradigm, blending traditional financial metrics with micro-behavioral insights. Pushing the envelope further [ 16 ], introduces DeepRisk, amalgamating demographic and financial behavioral data [ 17 ].…”

“…One pressing need is the refined usage of available informational components to portray these enterprises' more intricate credit landscape. Existing literature has drawn from corporate financial information [ [4] , [5] , [6] , [7] , [8] , [9] ], micro-level enterprise behavior information [ [15] , [16] , [17] , [18] ], public credit information [ [19] , [20] , [21] ], and information obtained from third-party sources [ 22 , 23 ] for crafting assessment frameworks. However, there is a lack of comprehensive consideration of these information elements.…”