Finite Sample Properties of Classcial Testings on Long Memory HYGARCH Models

• Activity Number: 515
• Type: Contributed
• Date/Time: Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Muyi Li*+
• Companies: The University of Hong Kong
• Keywords: Hyperbolic decay ; long memory GARCH process ; Likelihood ratio statistics ; Wald statistics ; Lagrange multiplier statistics

Abstract

Davidson(2004) proposed Hyperbolic GARCH (HYGARCH) model to unify the previously defined ARCH infinite models including GARCH, IGARCH, FIGARCH. He used the decay rate of the lag coefficients as another definition of memory and partitioned GARCH and IGARCH models as geometric-decayed models while FIGARCH as hyperbolic-decayed ones. Obviously it is a paradox with the former recognition that FIGARCH is the intermediate between IGARCH and GARCH, which analogical to I(d) compared with I(0) and I(1). In this paper, we apply there classical tests (non-robust and robust) to do some basic model testing on HYGARCH models. Derive the sizes and powers under finite sample. Compare the disadvantages and advantages of each test statistics in this case.