In panel data estimation if t=20 and i=20, do we necessarily need to test for stationarity, co-integration and causality. What is best method to estimate the model with this data set, if one is interested to find a long-term relationship? The key variables are social infrastructure and economic growth.