It is known that this test shows whether the estimated values ​​of the positive and negative shocks are equal. The problem is that when one of the estimated values ​​is not significant, this test treats it as significant and therefore the test result may show that the differences between the effect of positive and negative shocks are not significant (meaning that the test captures the symmetry in the estimated relationship), despite one of the estimators Morale and the second is no different from zero.

Wald's NARDL International test is misleading on this point, especially if the significance level of the parameter is large, so the confidence interval range will be large enough to include the range of the two confidence intervals for the ability with a low significance level (eg, less than 0.05).

Here, I ask colleagues what is the possible solution to confront this contradiction. With my sincere greetings