재무연구 제 권 호 (2008년 5월)
Asian Review of Financial Research, Vol., No..
pp.1520~1546
pp.1520~1546
Intraday Periodicity and Long Memory Property in High Frequency Data
Sang Hoon Kang University of South Australia
Seong-Min Yoon Pusan National University
This paper examines the nature of long memory or self-similarity in temporally aggregated data of KOSPI and KRW-US $, such as 10-min, 30-min, 1 hour and 1.5 hours. Apart from the commonly observed U-shaped pattern, inverse J-shaped patterns appear, due to market opening effects. The autocorrelations of absolute and squared normalized returns decay very slowly, and are associated with the long memory property. From empirical results from the FIGARCH(1, d,1) model, the 10- min and aggregated intraday returns exhibit long memory in volatility. Finally, the long memory property is invariant to temporal aggregation data, supporting the theory of self-similarity in Korean financial data.
Sang Hoon Kang
Seong-Min Yoon
This paper examines the nature of long memory or self-similarity in temporally aggregated data of KOSPI and KRW-US $, such as 10-min, 30-min, 1 hour and 1.5 hours. Apart from the commonly observed U-shaped pattern, inverse J-shaped patterns appear, due to market opening effects. The autocorrelations of absolute and squared normalized returns decay very slowly, and are associated with the long memory property. From empirical results from the FIGARCH(1, d,1) model, the 10- min and aggregated intraday returns exhibit long memory in volatility. Finally, the long memory property is invariant to temporal aggregation data, supporting the theory of self-similarity in Korean financial data.
Intraday periodicity,Volatility,Temporal aggregation,High frequency data,Long memory
