Um Procedimento Para Análise De Persistência Na Volatilidade

Abstract

Diebold (1986) suggested that conditional heteroskedastic models tend to overestimate persistence of volatility when there are instabilities at the unconditional second momento We propose a procedure for analysing this phenomenon based on recent techniques developed by the sudden changes of variance literature. Applying the ICSS algorithm (Inclán & Tiao, 1994) to the standardised residuals of the conditional heteroskedastic model, it is possible to detect break points at the unconditional variance. By incorporating dummies to capture these instabilities, a more accurate and reliable estimate of the volatility persistence can be computed. The procedure is successfully applied to Brazilian financial data.