Testing for Serial Correlation in Regression Models with Lagged Dependent Variables
研究了用Bootstrap方法近似临界点来检验含滞后因变量的回归模型中的序列相关性，模拟显示Bootstrap比传统方法更准确估计检验的零分布，并报告了检验的调整后功效。
Bootstrap methods are investigated for approximating critical points to several widely used tests of serial correlation in regression models with lagged dependent variables. Simulation results suggest that the bootstrap accurately estimates the null distributions of the tests, in contrast to conventional approximations. Results of some studies on the size-adjusted power of the tests are also reported. Copyright 1993 by MIT Press.