29 Pages Posted: 12 Apr 2005
Date Written: December 2004
This paper proposes a simple method to decompose the variance of returns into noise and information components, while allowing the two components to be correlated. Economically, noise is the real friction or transaction costs discussed in the literature. To apply the method, this paper examines noise in the Taiwan Stock Exchange, which is a call auction market. It also studies the cross-sectional and time-series variation of noise. The results find that noise has a distinct diurnal effect: the transaction price is less noisy at the open, but is noisier near the close. Trading mechanisms can also affect noise: a larger relative tick size and a longer time interval will increase noise. We also find that individual investors help to reduce noise.
Keywords: Noise, trading interval, volatility, tick
JEL Classification: G12, G14, G15
Suggested Citation: Suggested Citation