Title: Default probability models applied to a Mexican peasant institution
Authors: Maria Rosa Nieto Delfin - Investigaciones y Estudios Superiores SC (Mexico) [presenting]
Jose Morfin Tarasco - Investigaciones y Estudios Superiores SC (Mexico)
Abstract: Credit risk models failure to forecast crises has become especially important since the global economic crisis of 2008. The mortgage backed assets that started the crisis had excellent credit ratings. Since then, global credit risk regulators, are converging to have more efficient regulations which aim to develop accurate credit risk models. It is postulated that the Value at Risk forecast of a peasant financial company is more accurate if an endogenous stochastic model is used to calculate the default probability. A variety of models were tested on the historic data of the peasant financial company to find the best fit one. The results show that a Zero-Adjusted-Inverse-Gaussian model is the best fit for this type of credit institution. Hence, the Value at Risk forecast of the peasant financial company is improved. It was also found that Mexican credit risk regulations are undesirable, as they prohibit institutions from calculating their credit risk parameters through internal models. Mexican regulators give a generic value for the default probability vector to each institution. If the current credit risk regulations in Mexico changed to allow institutions to calculate their risk parameter through internal models, they would improve the calculation of its Value at Risk.