Relations Between Serial Correlation and Volatility: Is There a LeBaron Effect in Brazil?

Abstract

This paper examines the relation between serial correlation and volatility of the Ibovespa index returns and extends the empirical evidence of the LeBaron effect for higher orders of serial correlation. We employ an exponential general autoregressive conditional heteroskedastic model to estimate volatility and an automatic variance ratio statistic to calculate serial correlation. The results support some stylized facts from behavioral finance and help us to explain evidences from empirical studies. We show that (i) serial correlation in weekly returns are negative related with volatility, (ii) this negative relation is found in daily returns only if we use first order serial correlation, and (iii) the effect for weekly returns was not intensified by the 2008 crisis, but a positive relation between volatility and serial correlation for daily returns was identified during that time.