Regresar

FRvarPSO as an Alternative to Measure Credit Risk in Financial Institutions

Abstract:

When assessing potential clients and their credit status, financial institutions must use methodologies and/or internal procedures as a prerequisite for credit approval, instrumentation and disbursement. In this regard, good management in the credit analysis and knowing the risk of the potential client should increase guarantees or collaterals, this would lead to a reduction in risk by the financial institution. The objective of this work is to present an alternative to reduce credit risk, when expert assessment is not available. To this end, FRvarPSO is used as a methodology that allows obtaining fuzzy classification rules through a technique called Fuzzy C-means aimed at assigning the membership degree of the variables to each of the fuzzy sets. It also uses a neural network and an optimization technique for a variable population particle cluster, which uses a voting criterion reflected in particle speed. Each of the fuzzy rules obtained is assigned a value that is associated with the risk that the client represents, which will allow the financial institution to make appropriate decisions. This method was assessed with three different databases from Ecuadorean financial institutions – a savings and credit cooperative and two banks that grant productive credit, non-productive consumer credit and microcredit. The results obtained allow stating that the value associated to the fuzzy rule directly affects credit risk assessment. Finally, we conclude suggesting possible research lines.