since the Nikkei 225 index constitutes the underlying asset of financial derivatives traded on
two other major Asia Pacific derivatives markets, namely the Chicago Mercantile Exchange,
and Singapore Exchange Derivatives Trading Division. The regime-switching models are
estimated over a sample period that covers important events such as the Asian financial
crisis, the Russian debt crisis, the Long Term Credit Management crisis, the burst of the
information technology bubble, and the Japanese economic recession, among others.
The remainder of the paper is structured as follows. The next section describes the
Markov regime-switching models used to examine the nonlinearities in implied volatility
dynamics. Section 3 presents the sample data and distributional properties of implied
volatility indices. Section 4 discusses the empirical results for the Japanese and US markets.
The regime-switching models include various conditioning variables and are estimated
using both the levels and first differences in implied volatility. Section 5 concludes the
paper.
2. Regime-Switching Modelling of Implied Volatility
In the absence of perfect knowledge of when structural breaks in implied volatility
can take place, regime shifts are incorporated in the volatility-generating process following
the Markov regime-switching model by Hamilton (1989). The two-regime Markov process
used in the present study accounts for latent states of the relationship between implied
volatility and the set of past information. The sign and significance of model parameters
which describe the dynamics of implied volatility are driven by discrete switches in the
indicator variable. This unobservable variable takes the value ofzt = i, for i = 1, 2 , and
determines which regime governs the volatility dynamics at timet . With each observation
being drawn from a distribution conditional on the prevailing regime, the model parameters
are likely to differ in sign and/or magnitude across regimes. These regime-dependent