Correspondence: David G. McMillan, School of Management, University of St Andrews, The Gateway, North Haugh, St Andrews, KY16 9SS, UK. Tel: +44 0 1334 462800, Fax: +44 0 1334 462812, E-mail: dgm6@st-andrews.ac.uk

¹David G. McMillan is Reader in Finance in the School of Management at the University of St Andrews, UK. His research interests are in the time-series modelling of intra-day financial data, volatility forecasting, interest rate dynamics and the relationship between financial and macroeconomic variables. He has published widely on these topics in national and international journals

²Alan E.H. Speight is Professor of Finance in the School of Economics and Business at the University of Wales, Swansea, UK. His research interests are in empirical finance, volatility modelling and forecasting, high frequency financial data, non-linear time series analysis and business cycle dynamics. He has published widely on these topics in national and international journals

Received 16 October 2006; Revised 16 October 2006.

Abstract

Daily 'realised' volatility for two UK bond futures is obtained from high frequency, five-minute, returns data. In contrast to previous results using daily data, which have reported that daily returns volatility and daily standardised returns exhibit conditional non-normality, we find that the 'realised' logarithmic standard deviation and returns standardised by 'realised' volatility are approximately normal. These results support the mixture-of-distributions hypothesis that daily returns are characterised by a normal–lognormal mixture. Finally, realised volatility exhibits highly persistent temporal dependence that can be characterised by a long memory process.