Wavelet analysis of Asian FX markets using high frequency data
Date of Issue2000
College of Business (Nanyang Business School)
Speculative pricing process are nonstationary and do not conform to geometric Brownian motion since they exhibit scaling law. FX series reveal fractal dimensions lower than 2 which explains their features of self-similarity. This paper uses Mallat's (1989) time-scale multiresoultion analysis with Haar (1900) orthonormal filters to analyze non-stationarity (time-dependence) and self-similarity (scale-dependence) of min-by-min indicative quotes of Asian spot currency prices during the period May 1997- July 1997. The fractal dimension of each FX market is identified by its Hurst coefficient, which shed light on the persistence and the stability of the fractal nature of the various market pricing mechanisms. The Hurst coefficient is estimated from the Wavelet coefficients following Kaplan-Kuo algorithm. In comparison with the efficient markets of the YEN and D-Mark, Asian FX markets show evidence of non-linear complex structure. The results show that the increments are different from white noise; i.e they are not random. Most FX noise is pink or anti-persistence. This research indicates features of multi-fractality which opens more avenue for further research.
Nanyang Technological University