Abstract
In this paper, we test the weak-form stock market efficiency for the Tunisian stock market (TSE). Our empirical approach is founded on the analysis of the behaviour over time of the Hurst's exponent. Thus, we computed the Hurst's exponent using a "rolling sample" with a time window of 4 years. The sample data covers in daily frequency the period (January, 1997-October 2007). Since the classical R/S analysis is strongly affected by short-range dependencies both in the mean and the conditional variance of TSE daily volatility, daily stock returns were filtered using the traditional AR-GARCH(1,1) model. Our results for Hurst's and filtered Hurst's exponents behaviour analysis show a strong evidence of long-range dependence with persistent behaviour of the TSE. However, during the last two years, the filtered Hurst's exponent seems to exhibit a switching regime behaviour with alternating persistent and antipersistent behaviour but where it was somewhat close to 0.5. The nonparametric statistic approach results reveal that some TSE reforms including the launching of the Electronic quotation system on April, 1998, the fiscal regime for holdings, the security reinforcement laws, the legal protection of minority shareholder may play a role in understanding the Hurst's exponent behaviour over time