An examination of volatility of some selected countries exchange rates using ARCH models

In financial series, the nonlinear conditional heteroscedastic models are more commonly used than the linear time series models since the properties they have. Time series analysis requires the models that take into account the heteroscedasticity since the prediction errors have unconstant variances. Robert F. Engle (1982) generalizes the assumption about the heteroscedasticity and proposes a new stochastic models class ?Autoregressive Conditional Heteroscedasticity (ARCH)?. In this study, some ARCH models; GARCH, GARCH-M, EGARCH, and TGARCH are examined in statistical properties and estimation methods and applied to ten countries exchange rate series which are selected randomly by development level. The resulting models are compared with each other and the best model for the problem in question is defined.