Bootstrap Prediction Intervals for Conditional Heteroskedastic Models with Cyclically Varying Unconditional Variance

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Malaka Thilakaratne*+ and V A Samaranayake
• Companies: Missouri University of Science & Technology and Missouri University of Science and Technology
• Keywords: Volatility Forecasting ; Financial Time Series ; Bootstrap ; Cyclical Volatility ; Daily Volatility

Abstract

A computationally fast method for obtaining bootstrap-based prediction intervals for returns and volatilities in ARCH and GARCH processes was introduced in a recent publication. This method utilizes a constrained least squares approach to estimating the GARCH parameters. We extend this technique to ARCH type processes having unconditional variance that changes cyclically. Results of a Monte Carlo study that looks at the finite sample properties of the proposed intervals are presented. Simulation results indicate that the intervals for returns provide reasonable coverage probabilities in most circumstances, but that the intervals for volatilities are liberal.