Estadística

Permanent URI for this collectionhttp://98.81.228.127/handle/20.500.12404/757

Browse

Now showing 1 - 2 of 2

An empirical application of stochastic volatility models to Latin-American stock returns using GH skew student's t-distribution
(Pontificia Universidad Católica del Perú, 2015-07-17) Lengua Lafosse, Patricia; Bayes Rodríguez, Cristian Luis
This paper represents empirical studies of stochastic volatility (SV) models for daily stocks returns data of a set of Latin American countries (Argentina, Brazil, Chile, Mexico and Peru) for the sample period 1996:01-2013:12. We estimate SV models incorporating both leverage effects and skewed heavy-tailed disturbances taking into account the GH Skew Student’s t-distribution using the Bayesian estimation method proposed by Nakajima and Omori (2012). A model comparison between the competing SV models with symmetric Student´s t-disturbances is provided using the log marginal likelihoods in the empirical study. A prior sensitivity analysis is also provided. The results suggest that there are leverage effects in all indices considered but there is not enough evidence for Peru, and skewed heavy-tailed disturbances is confirmed only for Argentina, symmetric heavy-tailed disturbances for Mexico, Brazil and Chile, and symmetric Normal disturbances for Peru. Furthermore, we find that the GH Skew Student s t-disturbance distribution in the SV model is successful in describing the distribution of the daily stock return data for Peru, Argentina and Brazil over the traditional symmetric Student´s t-disturbance distribution.
Modelos Chain Ladder estocásticos y aplicaciones al cálculo de reservas en compañías de seguros
(Pontificia Universidad Católica del Perú, 2015-07-20) Mazuelos Vizcarra, Gisella Gabriela; Valdivieso Serrano, Luis Hilmar
This document is intented to deepen the study of univariate and multivariate Chain Ladder methods for estimating reserves in an insurance company. It presents from a theoretical and applicative perspective both the univariate deterministic and stochastic Chain Ladder methods. Although, the first is the most used method by insurance companies due to its simplicity and lack of probabilistic assumptions, the second, proposed by Mack (1993), allows the construction of confidence intervals for the estimated reserves, which is invaluable for researchers. We also develop the General Multivariate Chain Ladder model, which has the basic premise to analyze the possible relationship that may exist between different development triangles, thus providing another tool to improve inferences and predictions of reserves. These methods have been developed and applied to a database of 3 types of health insurance, thus showing the advantages and disadvantages of each of them in different scenarios and providing various tools for decision making in meeting the future obligations of insurance companies.