The Estimation of Dynamic Bivariate Mixture Models
纠正了Watanabe在估计动态二元混合模型时使用的多步抽样器中的错误，采用修正后的抽样器重新估计，结果与Liesenfeld和Richard的估计一致。
Watanabe estimated the dynamic bivariate mixture models introduced by Tauchen and Pitts and modified by Andersen using a Bayesian method via Markov chain Monte Carlo techniques. Based on a maximum likelihood method via efficient importance sampling, Liesenfeld and Richard obtained estimates that are significantly different from those of Watanabe. This note corrects the error in the multimove sampler used by Watanabe and reproduces all analyses in the work of Watanabe using a corrected multimove sampler. The estimates using the correct multimove sampler are found to be close to those obtained by Liesenfeld and Richard.