Inference for Duration Models Using Estimating Functions

• Activity Number: 416
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Julieta Frank*+ and Melody Ghahramani and Aerambamoorthy Thavaneswaran
• Companies: University of Manitoba and University of Winnipeg and University of Manitoba
• Keywords: Estimating Functions ; Random Coefficient Autoregressive Conditional Duration (RCACD) model ; Nonlinear Time Series ; Information

Abstract

A class of martingale estimating functions provides a more convenient framework for studying inference for nonlinear time series models relative to other widely used methods such as maximum likelihood estimation. For example, the estimating function approach does not assume any particular distribution for the innovation. Liang et al. (2011) have recently shown that quadratic estimating functions are more informative than linear estimating functions for Random Coefficient Autoregresive (RCA) models. Duration models are commonly used to model the behaviour of irregularly time-spaced financial data. The method is used to study the inference for the parameters of a new class of multiplicative Random Coefficient Autoregressive Conditional Duration (RCACD) models.