Heteroscedasticity in financial time series.
This thesis deals with two different topics, both related to
modelling time-varying variances in high frequency financial time
series. The first topic concerns the estimation of unobserved
component models with autoregressive conditional heteroscedastic
(ARCH) effects. The second topic concerns the quasi-maximum
likelihood estimation of stochastic variance processes. These are an
alternative to ARCH processes for modelling conditionally
heteroscedastic time series.
The motivation of the work is based on the increasing interest in the
financial area in modelling volatility. In financial markets, many
decisions are based on the volatility of a specific stock or index,
which is closely related to the variance. Therefore, it is important
to develop good statistical models able to describe time-varying