Hi! In Johansen cointegration, one has to ensure that residuals are normally distributed - at least for the ECM part. Is there a simialr requirement for ARDL cointegration? Is the Bounds test valid under non-normality? And are the t-values for the long-run coefficients reliable when the underlying ARDL has non-normal residuals?

More Lars Ahnland's questions See All
Similar questions and discussions