DocumentCode :
3625978
Title :
Long-Term Dependence Analysis in Index Return and Absolute Return Series of Turkish Stock Market
Author :
Cengiz Bektas;Suleyman Baykut;Tayfun Akgul
Author_Institution :
Elektronik ve Haberle?me M?hendisli?i B?l?m?, ?stanbul Teknik ?niversitesi, ?stanbul. bektasc@itu.edu.tr
fYear :
2007
fDate :
6/1/2007 12:00:00 AM
Firstpage :
1
Lastpage :
4
Abstract :
This study analyzes the long-term dependence in daily index "Return" and "Absolute Return" series of ISE National-All, National-100, National-30 and additionally 17 sectoral indices of Turkish stock market. Long-term dependence can be measured by a single Hurst (W) or relatedly the fractional differencing parameter (d). The data are analyzed by using four different parameter estimation methods namely wavelet based estimation method, periodogram based estimation method, Kettani-Gubner Methods for SOSS processes and FAR1MA (0, d, 0) processes. For the return series; long-term dependence is not observed in ISE National-All, National-100 and National-30 data. Long-term dependence is also not found in 13 of 17 sectoral index return series. Remaining 4 sectoral indices namely XFINK, XTRZM, XTEKS and XYORT are found long-term dependent. For the absolute return series; long-term dependence is found in all of the index data.
Keywords :
"Stock markets","Internet","Data analysis","Parameter estimation","Wavelet analysis"
Publisher :
ieee
Conference_Titel :
Signal Processing and Communications Applications, 2007. SIU 2007. IEEE 15th
ISSN :
2165-0608
Print_ISBN :
1-4244-0719-2
Type :
conf
DOI :
10.1109/SIU.2007.4298853
Filename :
4298853
Link To Document :
بازگشت