基于经验似然的自回归条件久期模型的统计推断

Statistical Inference for Autoregressive Conditional Duration Models Based on Empirical Likelihood

利用经验似然方法对自回归条件久期(ACD)模型参数进行统计检验,给出了自回归条件久期模型参数的经验似然比统计量,并证明了该统计量渐近服从χ2-分布.数值模拟结果表明,经验似然方法优于拟似然方法.

This paper solves the statistical test problem of an autoregressive conditional duration（ACD） models based on an empirical likelihood method.We construct the log empirical likelihood ratio statistics for the parameters of ACD model.it is showed that the proposed statistics asymptotically follows an χ2-distribution.A numerical simulation demonstrates that the performance of the empirical likelihood method are better than that of the quasi-likelihood method.