Resumo

GIRALDO GOMEZ, Norman. Beta and VaR prediction for stock portfolios using Kalman's filter and Garch models. Cuad. Adm. [online]. 2005, vol.18, n.29, pp.103-120. ISSN 0120-3592.

This paper studies the performance of an estimator used to calculate VaR for a stock portfolio using stock Beta coefficients. Such coefficients are recursively projected using Kalman's filter and modelling the market-volatility restrained term according to conditionally heteroscedastic GARCH, EGARCH and GJR models. A low Beta oscillation model is used to carry out diverse simulations and the results are compared with RiskMetrics mean-variance. Two hypothesis tests were introduced. We conclude that the new estimator is as effective as the one used by RiskMetrics to foresee those cases in which extremely negative yields overflow the VaR margin. However, our estimator is a better predictor of the excess amount on VaR margin than the one used by RiskMetrics.

Palavras-chave : CAPM; value at risk; Kalman's filter; betas; heterocedasticity; GARCH.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )