Evaluación de proyectos de inversión de la Banca de Desarrollo, mediante modelos credit scoring destinados a pymes.El caso de la provincia de Pichincha (Ecuador)

  1. Cerdá Prado, Nelson Alberto
Supervised by:
  1. Marcelo Sánchez-Oro Director
  2. Antonio Jurado Málaga Co-director

Defence university: Universidad de Extremadura

Fecha de defensa: 17 June 2020

  1. Juan Carlos Díaz Casero Chair
  2. Germán Jaraíz Arroyo Secretary
  3. María de la Cruz del Río Rama Committee member

Type: Thesis

Teseo: 621173 DIALNET lock_openTESEO editor


The present work, intends to contribute with a credit scoring model that uses complementary criteria of evaluation of investment projects, in the field of market, engineering and financial, for the process of credit granting of development banking to the sector pymes of the province of Pichincha Ecuador and in this way mitigate the risk for both the financial institution and small and medium enterprises. Credit scoring as a definition according to some authors, is classified as good or bad payers to potential customers of the bank, whether these individuals or companies, on the other hand is also defined as algorithms that allow customers to qualify, using statistical methods with techniques parametric and nonparametric to measure the risk. Parametric techniques support their analysis in a known distribution function, estimating criteria that explain the dependent variable and nonparametric techniques do not require information about the distribution function (neural networks, decision trees, search algorithms and artificial intelligence). The Ecuadorian financial system is regulated by the Monetary Financial Code, the resolutions of the Monetary Board and is under the supervision of the Superintendency of Banks, all the regulatory framework includes activities related to credit risk and the application of methodologies to your determination In the research process, 33 variables grouped in 5 dimensions are analyzed statistically of 315 companies, which through the use of two techniques, one non-parametric (decision trees) and another parametric (logit), a credit scoring model was built. The credit scoring model was validated with favorable results, through the following tests: omnibus using the maximum likelihood procedure, selecting the parameter estimates that allow the observed results to be adequate, Hosmer Lemeshow, global adjustment and Negelkerke index.