If the dependent variable is trended then the most logical model to begin is with is constant and trend. Nevertheless, It is not easy to answer this question because the correct answer will depend on the explicit form of the cointegrating equation that a researcher expects to find.Thie latter however is unkownis thus one can try various combinations of trend and intercept as well inclusion of structural breaks that may be present in the time series data.

I Think the most important criterion is whether the cointegratnig equation is meaningful i.e, according to economic theory, if this criterion is not met then the cointegrating equation cannot be accepted.

Dear Safwan Safi ,

First check with the option constant and trend, if you find the constant (c) staistically significant then you move forward.

Best wishes.

Thank you all for your precious comments

Hello Debasmita Dey and Assad Ullah ,

When i select Constant only, the constant c comes to be statistically significant and ardl bound test shows cointegration. But when i include "Constant & trend", the constant c becomes insignificant but trend is statistically significant and ardl bounds test shows no conintegration. So here lies my concern.

In that case you shall move forward with constant only, as it is statistically significant. Safwan Safi

The decision of whether to include “constant” only or “constant & trend” or “none” in ARDL bounds test for cointegration depends on the nature of the data and the research question being addressed.