therefore, to overcome any apparent effect regression (Arlt, 1997, 2007) which could skew the results statistically insignificant variables to statistically significant once I chose the generation of the first difference (annual)for all non-stationary time series