Working Papers 2003 – Abstracts

Bayesian Analysis of the Stochastic Conditional Duration Model

Chris M. Strickland, Catherine S. Forbes and Gael M. Martin

A Bayesian Markov Chain Monte Carlo methodology is developed for estimating the stochastic conditional duration model. The conditional mean of durations between trades is modelled as a latent stochastic process, with the conditional distribution of durations having positive support. The sampling scheme employed is a hybrid of the Gibbs and Metropolis Hastings algorithms, with the latent vector sampled in blocks. The suggested approach is shown to be preferable to the quasi-maximum likelihood approach, and its mixing speed faster than that of an alternative single-move algorithm. The methodology is illustrated with an application to Australian intraday stock market data.

Keywords: Transaction data, Latent factor model, Non-Gaussian state space model, Kalman filter and simulation smoother.

Next Abstract