Long Memory or Structural break: Evidence from the Tehran stock market

Abstract

The process of short memory that meet structural break in a mean shows the slower hyperbolic rate of decay in autocorrelation function and fractional integrated processes. In this paper, we investigated the potential of structural break and long memory properties with semi-parametric and parametric estimators of long memory and diagnosed the break points by the OLS-based tests. Our results showed there are simultaneously long memory and break points in TEPIX series. Therefore, to correct deduction in market memory, statistical properties and long memory tests was examined after and before of break dates. The results exhibit the same statistical pattern and long memory in all subseries divided by structural breaks. Finally, it supposed the TEPIX memory is long and true and suggested Tehran stock market cannot be considered as efficient market in transformation of information speed. Hence, there is the possibility of obtaining unusual profits in such a market and consequently the assumption of a weak form of market efficiency will also be violated.