Improvement in Bank Financing Access for Peruvian MSE using prediction and classification models

Abstract

Peruvian economic development has grown 149% in the last decade-15% yearly-mainly because of micro and small enterprises (MSE) which stands for about 42% GdP thus MSE becomes a key driver for national development and growth. However, as MSE can acquire various business know-how and transform into influential partners for big enterprises, they still face obstacles that limits their mid and long-term survival rate (avg. life cycle of 8.3 years, 6% of annual mortality rate), mainly because of the lack of access to ways to gain funds in order to invest in new machinery or upgrades in business infrastructure to be more productive and, in consequence, be more profitable and contribute to national growth by more tax payments. The current research assesses this problem by understanding which are the main difficulties that MSE face when trying to apply to a loan for a financial institution. A dedicated study was performed to identify and prioritize which variables are statistically relevant to the approval or rejection of a loan application and then use data mining techniques such as logistic regression and neural networks to model the probability of approval of a loan application and in consequence give highlights of the most important parameters that MSE should improve in their business model.