ASYMMETRY EFFECT IN DUAL LONG MEMORY: BIST BANK CASE

The aim of this paper is to test the efficient market hypothesis by examining the dual long memory feature of the Turkish banking sector index in return and volatility with the ARFIMA-FIGARCH and ARFIMA-FIEGARCH models. For this purpose, closing prices of 2008-2017 period Stock Exchange Istanbul Bank Index (XUBANK) were used as data set in the model. According to the ARFIMA-FIGARCH model estimates established according to different error distribution assumptions to test the dual long memory, while no findings can be obtained about the long memory feature in the return; the volatility has long been supported by findings that support long memory. Moreover, it has been determined that structural break has no statistical effect on the long memory related to the volatility in the mentioned period. In order to measure the asymmetric effect of the information shocks, the ARFIMA-FIEGARCH model was estimated according to the Student-t distribution and it was found that there was no long memory in the return. However, it was observed that the rate of long memory in the volatility of return was 0.74 and the negative information shocks caused more volatility than the positive information shocks