volatility is characterized by faster mean reversion, symmetric impact of news and negative
relationship with past changes in realized volatility.
Judging from the inferred probabilities exhibited by Figure 4 for the US market,
regime 1 prevails with insignificant drift, significant mean reversion, weaker asymmetric
impact of news and weaker adjustment to the dynamics of realized volatility. Similar to
patterns revealed by Figure 2, the frequency of regime shifts seems to increase since the mid
1990s, in response to more turbulent periods including the Russian debt default and LTCM
debt crises. The onset of financial crises increases the likelihood of the alternative regime of
expectations for significant decreases in market volatility, with slower mean reversion,
stronger asymmetric reaction to news, and significant positive adjustment to the dynamics
of realized volatility.
As illustrated by Figure 5, the inferred probabilities for the Japanese market suggest
also that regime 2 is more likely to prevail with expectations of decreasing volatility, slower
mean reversion, asymmetric impact of news and significant positive adjustment to changes
in realized volatility. Compared to the US market results, there are less frequent switches
toward the alternative regime featuring expectations of increasing volatility, with stronger
mean reversion, significant though not asymmetric impact of news, and inverse adjustment
to changes in realized volatility. These regime changes are seemingly associated with
significant events such as the Asian financial crisis and the burst of the Japanese bubble,
which heralded a decade-long period of mounting bad debts, deflationary pressures, and
economic recession. Arguably, the results suggest that the worsening economic prospects
may have been conducive, particularly in the early 1990s, to expectations of increasing
volatility even in the presence of positive returns and marginal decreases in market
volatility.
15