Feasible Generalized Least Squares - Coefficient interpretations

#1
Hello,

I am doing a panel model analysis with 3 countries and 14 years. After a number of considerations, I came up with FGLS as the best method to use to obtain meaningful results. In past literature I found that the interpretation of coefficients is the same as for linear regression. I have 2 issues.

1. I would like to explain differences among the three countries with reference to the coefficients. That is to say, the coefficient results of FGLS explain the change in Y given a marginal change in X, taking into consideration the countries all together. However, I would also like to see how the effect of X on Y changes from one country to the other.

2. I tried to plot each explanatory variable vs Time to confirm my coefficient results. For instance, a get a negative coefficient for the independent variable "Bank Non-performing-loans/total gross loans". The dependent variable is "Share of Cashless payments at POS". This means that the higher "Bank Non-performing-loans/total gross loans", the lower "Share of Cashless payments at POS". However, the plot I obtain seems to say exactly the opposite. In fact, Germany has the lowest "Bank Non-performing-loans/total gross loans" over all the years, but also the lowest "Share of Cashless payments at POS". You can find attached the two plots for better understanding and visualization.

Could someone help with these 2 points? Thanks in advance

Filippo Zocco
 

Attachments