Time-varying long-term Memory in the Tehran Stock Exchange: the Generalized Hurst Exponents and the Rolling Window Approach

This study is the first to examine the issue of time-varying long-term memory in the Tehran Stock Exchange, using a new efficiency index through a rolling window technique. To test the robustness of the results, this estimation technique is repeated with time windows with 5-day shifts. Furthermore, the wild bootstrap versions of the Automatic Portmanteau test (AQ) and the Automatic Variance Ratio test (AVR) have been performed with 14- and 5-day shifts.

The sample employed in this paper consists of daily observations on the Tehran Stock Exchange Index (TEPIX), covering the period from December 2008 to October 2019, making up a total of 2621 observations. The TEPIX series are collected from the Rahavard Novin software. The hypotheses are analyzed by Excel, EViews, R, and MATLAB software.

The findings show that in the Tehran Stock Exchange the estimated values of the generalized Hurst exponents (GHE) for all windows with 14-day shifts are over the efficiency indicator 0/5. Therefore, a kind of long-term memory exists on the Tehran Stock Exchange. Moreover, a high degree of inefficiency ratio is observed in the market. Furthermore, the Tehran Stock Exchange does not become more efficient over time. Finally, the results from the time windows with 5-day shifts as well as wild bootstrap versions of the AQ and AVR tests with 14- and 5-day shifts indicate that the previously-mentioned empirical results are robust. The findings from the study provide important implications for investors, portfolio managers, and policy-makers.