Canonical Cointegrating Regressions
提出一种新的协整回归统计推断方法，通过对数据进行简单变换得到典型协整回归，使普通最小二乘法能产生渐近有效的估计和卡方检验，适用于多种协整模型。
This paper develops a new procedure for statistical inference in cointegrating regressions. We introduce the concept of canonical cointegrating regressions, which are the regressions formulated with the transformed data. The required transformations involve simple adjustments of the integrated processes using stationary components in cointegrating models. Canonical cointegrating regressions therefore represent the same cointegrating relationships as the original models. They are, however, constructed in such a way that the usual least squares procedure yields asymptotically efficient estimators and chi-square tests. The methodology presented here is applicable to a very wide class of cointegrating models, including models with deterministic and singular, as well as stochastic and regular, cointegrations.