Title: On the unbiasedness of financial volatility proxies
Authors: David Kreiberg - BI, Norwegian Business School (Norway) [presenting]
Steffen Groenneberg - BI Norwegian Business school (Norway)
Genaro Sucarrat - BI Norwegian Business School (Norway)
Abstract: Models for conditional volatility play an important role in asset pricing and in risk analysis. Since the conditional volatility is an unobserved quantity, a problem is how to undertake forecast comparisons. A common way to handle this problem is to use some form of volatility proxy such as realized volatility or a range-based volatility measure. However, such proxies may not represent unbiased measures of the true volatility. We device simple tests to evaluate hypotheses concerning the unbiasedness of these proxies. The proposed tests hold under mild assumptions and accommodate various hypotheses, depending on the form and nature of the observed volatility proxy. Simulations confirm that the empirical size corresponds well to the theoretical size, and that it has power under the alternative. The results are further illustrated in an empirical application.