Abstract:
|
Since realized measures of volatility are affected by measurement errors, the study considers a new class of discrete-time stochastic volatility (SV) models, which can relate many realized volatility measures to the latent conditional variance. We propose a hybrid estimator for this class of models that combines a generalized least square (GLS) type transformation and instrumental variable (IV) approach. A simulation study reveals that the hybrid estimation method has excellent finite-sample properties. We illustrate the proposed method's empirical relevance using mixed frequency IBM stock returns and options prices.
|