An EM Algorithm for Conditionally Heteroscedastic Factor Models
展示EM算法如何大幅降低条件异方差因子模型的计算负担，使研究者能处理大量序列，并用11和266只股票收益的实证验证了速度优势，但指出在最优解附近需切换为导数方法。
This article discusses the application of the EM algorithm to factor models with dynamic heteroscedasticity in the common factors. It demonstrates that the EM algorithm reduces the computational burden so much that researchers can estimate such models with many series. Two empirical applications with 11 and 266 stock returns are presented, confirming that the EM algorithm yields significant speed gains and that it makes unnecessary the computation of good initial values. Near the optimum, however, it slows down significantly. Then, the best practical strategy is to switch to a first-derivative-based method.