When and why to choose unrestricted, restricted constant, or other options in that combobox? There is such a combobox option for selecting unrestricted and restricted constant in GRETL. From the menu of GRETL, Model -> Time Series -> VECM
It is always handy to have a look at the graph of timeseries data in the first place, which gives you a visual impression if the data exhibit upward or downward trend or not. The constant takes care of the unit of measurement used in data, hence we generally keep an unrestricted constant in the VECM. The question is whether we should keep a constant in the long run equilibrium part of a VECM equation or in the short run part. It is prudent to keep a constant in the long run part, otherwise the model will be too restricted. However, the 'Pantula Test' confirms which particular model a VECM equation belongs to and we can decide almost unambiguously whether you should keep a 'restricted' or 'unrestricted' constant either in the long run or short run part of the VECM equation in consideration.
It is always handy to have a look at the graph of timeseries data in the first place, which gives you a visual impression if the data exhibit upward or downward trend or not. The constant takes care of the unit of measurement used in data, hence we generally keep an unrestricted constant in the VECM. The question is whether we should keep a constant in the long run equilibrium part of a VECM equation or in the short run part. It is prudent to keep a constant in the long run part, otherwise the model will be too restricted. However, the 'Pantula Test' confirms which particular model a VECM equation belongs to and we can decide almost unambiguously whether you should keep a 'restricted' or 'unrestricted' constant either in the long run or short run part of the VECM equation in consideration.