Identificación de conglomerados espaciales de acuerdo a niveles de morosidad de empresas en el Perú
Date
2021-11-07
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú
Acceso al texto completo solo para la Comunidad PUCP
Abstract
El cumplimiento de las obligaciones financieras que tienen las empresas es respaldado por
una correcta gestión de riesgo de crédito, esto evita problemas de liquidez y solvencia. Por
ello es importante detectar los niveles de riesgo de morosidad en las empresas. La presente
tesis tiene como objetivo identifi car conglomerados de provincias del Perú, en funciona de la
tasa de incumplimiento de pagos, conocida también como la tasa de morosidad. Para ello se
propone un modelamiento en dos niveles. En el primer nivel se usan modelos aglomerativos
jerárquicos para seleccionar n conglomerados candidatos a priori, donde el número fi nal de
conglomerados se escoge mediante criterios de selección de modelos. Posteriormente, en un
segundo nivel, modelaremos el nivel de riesgo haciendo uso del modelo de Poisson y prioris
condicionales autoregresivas en base a los conglomerados de nidos en el primer nivel e incluyendo
covariables. Los modelos pueden ser reescritos como modelos Gaussianos latentes, y se
puede usar inferencia bayesiana para estimar sus parámetros, específicamente a través de la
aproximación de Laplace anidada integrada. Finalmente, como resultado de la aproximación
se obtienen conglomerados de provincias de acuerdo a sus niveles de morosidad, permitiendo
clasi ficar las provincias en conglomerado de alto, medio y bajo nivel de riesgo de morosidad.
Compliance with the nancial obligations of companies is ensured by proper credit risk management, this avoids liquidity and solvency problems. For this reason, it is important to identify the risk level of default in peruvian companies. The goal of this thesis is to identify clusters of provinces of Per u with regard to the default rate of payments, also known as probability of default. Thus it is proposed a model in two stages. In the rst stage hierarchical agglomerative models select prior candidate clusters, and the nal number of clusters is selected through selection criteria of models. In the second stage it is proposed the Poisson model considering autoregressive conditional prioris, the clusters de ned in the rst stage, and also including covariates. This model ll in the class of Gaussian latent models, therfore its paremeters were estimated using bayesian inference, speci cally through integrated nested Laplace approximation. Finally, as a result, we found clusters in accordance with the default level, allowing to classify provinces into clusters of high, medium and low risk level.
Compliance with the nancial obligations of companies is ensured by proper credit risk management, this avoids liquidity and solvency problems. For this reason, it is important to identify the risk level of default in peruvian companies. The goal of this thesis is to identify clusters of provinces of Per u with regard to the default rate of payments, also known as probability of default. Thus it is proposed a model in two stages. In the rst stage hierarchical agglomerative models select prior candidate clusters, and the nal number of clusters is selected through selection criteria of models. In the second stage it is proposed the Poisson model considering autoregressive conditional prioris, the clusters de ned in the rst stage, and also including covariates. This model ll in the class of Gaussian latent models, therfore its paremeters were estimated using bayesian inference, speci cally through integrated nested Laplace approximation. Finally, as a result, we found clusters in accordance with the default level, allowing to classify provinces into clusters of high, medium and low risk level.
Description
Keywords
Estadística bayesiana, Variables latentes
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/openAccess