Non-Linear Moving-Average Conditional Heteroskedasticity

Heteroskedastic Conditional Variance Modeling is a §particularly rich branch of Econometrics. Despite its §young age, the literature embracing variance models §is indeed impressive. Statistical treatment of §financial time series has been profoundly enlarged by §this class of models. Many characteristics of §financial variables (leptokurticity, asymmetry,...) §can now be modeled. Of course, in the search for §better adjustment, the complexity has been§significantly increased at a non-trivial cost. In §this book we introduce three simple specifications §based upon Robinson s (1977) NLMA model: QMACH, §NLMACH and PLMVES. The first proposal belongs to the §NLMA class whilst the second one is closely related §to it. Both possess similar properties as those of §the ARCH-class. Our last proposal, PLMVES, also §considers long-memory by means of a new concept: §Pseudo Long-Memory, an artifact allowing for long-§memory modeling. These variance models have §advantages none of the inspiring sources (ARCH and §NLMA) possess. This book may help economists in their §quest for a better comprehension of the variance of §economic variables, particularly in the modeling of §financial volatility.