conclude that most events leading to volatility shifts tended to be local (e.g., the Mexican
peso crisis, periods of hyperinflation in Latin America), and that the only global event over
the sample that affected several emerging markets was the October 1987 crash.
However, recent literature has shown that the ICSS algorithm tends to overstate the
number of actual structural breaks in variance. Specifically, Bacmann and Dubois (2002)
point out that the behavior of the ICSS algorithm is questionable under the presence of
conditional heteroskedasticity. They show that one way to circumvent this problem is by
filtering the return series by a GARCH (1,1) model, and applying the ICSS algorithm to the
standardized residuals. Bacmann and Dubois conclude that structural breaks in
unconditional variance are less frequent than it was shown previously.
An alternative approach to testing for homogeneity of variance is wavelet analysis.
Wavelet analysis is a refinement of Fourier analysis that was developed in the late 1980’s,
and which offers a powerful methodology for processing signals, images, and other types of
data. In particular, the discrete wavelet transform allows for the decomposition of time
series data into orthogonal components with different frequencies. In finance, potential
applications of wavelet methods are quantification of spillovers between stock markets at
different time horizons, and testing for the presence of structural breaks in volatility in
detailed and smooth components of a time series.
Recent applications of wavelets in economics and finance are Ramsey and Lampart
(1998), Norsworthy, Li and Gorener (2000), Lee (2001a, 2001b), and Gencay, Whitcher,
and Selcuk (2003). Ramsey and Lampart (1998) study the permanent income hypothesis,
and conclude that the time-scale decomposition is very important for analyzing economic
relationships. Norsworthy, Li and Gorener (2000) and Gencay, Whitcher, and Selcuk