Long-Term Memory and level shifts: Application of the modified GPH test in Tehran Stock Exchange

Subject Areas : Financial Knowledge of Securities Analysis

mansoor kashi ¹ , Mohammad Donyaei ² , Roholah Ahmadi ³

1 - M.A in Financial Management, University of Sistan and Bluchestan
2 - PhD Student of Financial Management, Allameh Tabataba’i University
3 - M.A in Business Management, University of Sistan and Bluchestan,

Received: 2014-04-04 Accepted : 2014-05-22 Published : 2014-12-22

Keywords: Long -Term Memory, modified GPH, Level Shifts,

Abstract :

Always one of the concerns of researchers in the study of time series, the long-termmemory and whether the observed long memory series is affected by level shifts or not?To avoid spurious long memory that may be caused by levels shifts, different methodshave been tested. In the present article, we will review it for return and volatility (Are twoapproaches: First approach: GARCH and ARMA-GARCH-filtered series. Secondapproach: GARCH-filtered-squared series and squared returns) based on modified GPHof smith (2005). The results indicate that long memory in return series TEPIX is acceptedby the GPH but modified GPH test (Bandwidth choice for classic GPH and modifiedGPH are based on Plug-in and1/ 2 J  T ) leads to the rejection of the existence of longmemory. Then, using the first approach, the presence of long memory is demonstrated bythe GPH. But modified GPH denies its existence, which reflects the impact of level shiftsto long-term memory. The second approach about how to volatility infer the existence oflong memory is demonstrated by both tests. It followed that the predictability of themarket as possible and Weak form efficiency would violate the Tehran Stock Exchange