Bayesian Analysis of Asymmetric Stochastic Conditional Duration Model

Abstract

This paper proposes Markov chain Monte Carlo methods to estimate the parameters and log durations of the correlated or asymmetric stochastic conditional duration models. Following the literature, instead of fitting the models directly, the observation equation of the models is first subjected to a logarithmic transformation. A correlation is then introduced between the transformed innovation and the latent process in an attempt to improve the statistical fits of the models. In order to perform one‐step‐ahead in‐sample and out‐of‐sample duration forecasts, an auxiliary particle filter is used to approximate the filter distributions of the latent states. Simulation studies and application to the IBM transaction dataset illustrate that our proposed estimation methods work well in terms of parameter and log duration estimation. Copyright © 2014 John Wiley & Sons, Ltd.

Conditional Duration Models for High‐frequency Data: A Review on Recent Developments

“…The studies by Men et al . (, b, ) put forward interesting questions. Selection of the correlation specification in ASCD and MSCD models and developing a formal statistical test to select the optimal number of components in the MSCD model are interesting problems.…”

“…Further, in an attempt to improve the statistical fit of the standard SCD model, Men et al . (, b) allow a correlation similar to Feng et al . () between the innovations driving the observed durations and the innovations driving the stochastic latent variable that generates the durations.…”

Threshold Stochastic Conditional Duration Model for Financial Transaction Data

“…Indeed, in our study, we find that even with only five iterations of our slice algorithm, we can feasibly and efficiently estimate the TSCD models by the MCMC. 2 The single-move simulation method is popular in the literature and used in Jacquier et al 2004; Yu et al (2006); Zhang and King (2008), Men et al (2015Men et al ( , 2016aMen et al ( , 2016b among others. The advantage of the slice sampler is that five iterations can give us a point from the underlying distribution unlike the MH algorithm where many generated points must be discarded.…”

Forecasting the duration of short‐term deflation episodes

“…In economics, the most popular approaches are grounded within extreme value theory (e.g., Gilli & Këllezi, 2006) and proportional hazard duration modeling. For a discussion on other approaches see Men, Kolankiewicz, and Wirjanto (2015). For a discussion on other approaches see Men, Kolankiewicz, and Wirjanto (2015).…”

“…The latter is a development from survival modeling, and is often used for analyzing the duration of unemployment (e.g., Bover, Arellano, & Bentolila, 2002;Lancaster, 1979). For a discussion on other approaches see Men, Kolankiewicz, and Wirjanto (2015).…”