ASYMPTOTIC BEHAVIOR OF THE CUSUM OF SQUARES TEST UNDER STOCHASTIC AND DETERMINISTIC TIME TRENDS
将平方累积和检验推广到含确定性时间趋势的非平稳自回归分布滞后模型，可用普通最小二乘残差或标准化预测误差实施，模拟显示检验差异很小且受干扰参数影响微弱。
We generalize the cumulative sum of squares (CUSQ) test to the case of nonstationary autoregressive distributed lag models with deterministic time trends. The test may be implemented with either ordinary least squares residuals or standardized forecast errors. In explosive cases the asymptotic theory applies more generally for the least squares residuals-based test. Preliminary simulations of the tests suggest a very modest difference between the tests and a very modest variation with nuisance parameters. This supports the use of the tests in explorative analysis.