Conditional Variance Parameters in Symmetric Models

摘要: The parameters and are restricted to be non-negative in GARCH model, which have some consequences for the stationarity condition, although disturbances mean 0, they clearly not white noise because of their time-varying asymmetric probability density functions. Thus, this stationary process is capable capturing well known phenomena present financial markets such as volatility clustering, marginal distributions having heavy tails thin centres (Leptokurtosis); return series appearing almost uncorrelated over time but dependent through higher moments. possibility dependence between conditional moments, most notably variances, involves examining nonlinear stochastic processes from a more realistic perspective data. This motivates consideration models. results obtained Monte Carlo simulations established practicability small sample performance symmetric models under Gaussian distributions. also showed that corresponding standard errors very indicating estimators asymptotically unbiased, efficient consistent at least within sample.