To run ARDL, you firstly need to provide the results indicating  that ARDL is an appropriate method for what you want to confirm. Firstly, unit root test need to be performed and to  apply ARDL  integration order of exports series and GDP series as well needs to be less than or equal to one. Neither exports nor GDB must be integrated of order two or higher. To decide on potential causality direction(s) you can run Granger causality test.

Dear Mile

Thanks for your response But i am interested in long- and short-run causality which is not obtain directly from causality test, therefore, i want to derive from ARDL model. However,  i seen in many paper, they have derived the long run causality from ARDL  and reported the results in the case of causality run from gdp to exports, like  ECT(-1)/ d(GDP) = 4.5 for long - run but i am little confuse how 4.5 is come from. Either it is impose restriction in ECM model or it is obtain from some other way. please guide me in this regard if you know this.

thanks and best regards

irshad