Edgeworth expansions in small noise asymptotics

June, 2005

Tech Report

Author(s)

Lan Zhang, Per A. Mykland, and Yacine Aït-Sahalia

Abstract

The paper considers Edgeworth expansions for estimators of volatility. Unlike the usual exanpsions, we have found that in order to obtain meaningful terms, one needs to let the size of the noise to go zero asymptotically. This is reflected in our expansions. The results have application to Cornish-Fisher inversion and bootstrapping.