点击头像下方的“+加关注”或者订阅【Shibor文库】可以第一时间了解我的资料更新，您的鼓励是我前进的动力！！！

Simple Estimators for ARCH Models

Todd Prono

Abstract: Covariances between contemporaneous squared values and lagged levels form the basis for closed-form instrumental variables estimators of ARCH processes. These simple estimators rely on asymmetry for identi…cation (either in the model’s rescaled errors or the conditional variance function) and apply to threshold ARCH(1) and ARCH(p) with p < ∞ processes. Limit theory for these estimators is established in the case where the ARCH processes are regularly varying with a well-de…ned third and sixth moment of the raw returns and rescaled errors, respectively. The resulting limits are highly non-normal in empirically relevant cases, with slow rates of convergence relative to the thin-tailed √n-case. Nevertheless, Monte Carlo studies of a heavy-tailed ARCH(1) process show the simple IV estimator to outperform standard QMLE in (relatively) small samples when the data are (heavily) skewed. Methods for determining con…dence intervals for the ARCH estimates are also discussed.

附：技术性附录——