There are many options:

1. First evaluate the consistency of the within country (the 4 time points), and if this is reasonably consistent, you can take the mean and standard deviation (or some other appropriate enumerative statistic) of those 4 points. You will be able to model the 5 independent variable effects on the mean and on variation.

2. As you suggest, you can model each time separately and compare the 4 models.

3. Do you predict any of the 5 independent variables will have a latent or lagged effect? If so you might want to look at the slope of the 4 points in time series and use that as a response variable.

"All models are wrong, some are useful" G.E.P. Box