I am trying to measure the impact of Geographical Indications in international trade with a gravity model. I am focused in two equations. 1) The value of imports of GI products from the EU as dependent variable and GDP, Population and transport costs of the importing countries as independent variables. 2) The quantity of the imports as dependent vraible and the same independent variables. I first did a random effects model but my both dependent variables showed signs of non-stationarity. I have been looking for models for models that could work over the stationarity. First I tried by using the first differences with the same specification model but all of my parameters were not significant and it was very hard to interpretate the results with the differences. Now I am thinking on using an autoregressive model with an AR(1) disturbance or an Arellano-Bond Dynamic model. I undertsand the theory of the models but I have never ran them in Stata, how can I know that this options are fitting my data by the results in Stata? And is there another spevification that could be use for working on non-stationary panel data?