On the power of the durbin-watson test under high autocorrelation
Helmut Zeisel · Communication in Statistics- Theory and Methods · 1989
In the linear regression model without an intercept, it is known that the limiting power of the Durbin-Watson test (as correlation among errors increases) equals either one or zero, depending on the underlying regressor matrix. This paper considers the limiting power in the model with an intercept, and proves that it will never equal one or zero.