Modelos univariados e multivariados para cálculo do Valor-em-Risco de um portifólio / Multivariate and Univariate Models for Forecasting a Portfolios Value-at-Risk

AUTOR(ES)

Renato Fadel Fava

DATA DE PUBLICAÇÃO

2010

RESUMO

The present work consists of a comparative study of several portfolio Value-at-Risk models. Univariate models, which consider only the portfolio log-returns series, are compared to multivariate models, which consider the log-returns series of each asset individually and their conditional correlations. Additionally, recently proposed models such as PS-GARCH and VARMA-GARCH are tested. We also propose a new model that uses past cumulative returns as exogenous variables. All models are evaluated in terms of their compliance to Basel Accord and financial impact, in period that includes high volatility times. In general, univariate and multivariate models performed similarly. More complex models yielded more accurate results, with satisfactory performance including in crisis periods.

ASSUNTO(S)

volatilidade ps-garch egarch valor em risco varma-garch basel accord pgarch riskmetrics past cumulative returns. value-at-risk ps-garch varma-garch correlação condicional conditional correlation dvec pgarch garch riskmetrics dvec acordo de basileia retornos passados acumulados. volatility egarch garch

Documentos Relacionados

• Estudo comparativo dos modelos de value-at-risk para instrumentos pré-fixados.
• Evaluating Value-at-Risk models via Quantile regressions
• Comparing Value-at-Risk Methodologies
• Risk management in international financial market: a comparative analyze between volatility models to Value-at-Risk estimation
• Teoria de carteiras e value-at-risk â estudo de caso da Capef