Extension and verification of the asymmetric autoregressive conditional duration models

Abstract

ABSTRACTWe extend the asymmetric autoregressive conditional duration (AACD) model of Bauwens and Giot [Asymmetric ACD models: Introducing price information in ACD models, Empir. Econ. 28 (2003), pp. 709–731] which allows describing the joint dynamics of financial durations, i.e. times between selected micro-events (trades, price, or volume changes) and the corresponding event categories. The generalized beta distribution of the second kind (GB2) is proposed as the distribution for the error terms in the model. The GB2 distribution allows for flexible non-monotonic shapes of baseline hazard functions, which facilitates a plethora of temporal dependence patterns for event arrival times. We propose the procedures for checking the adequacy of the parametric specification with the help of simulation methods that allow computing the model residuals. Moreover, we show how to assess the goodness of fit of the AACD model using probability integral transforms. The model is used to investigate striking regularities in the manner currency dealers execute their market orders in the interbank foreign exchange spot market.