Replacing Missing Values

Hi all,

I would like to run a Fixed Effects (FE) model but my sample is relatively small. On top, some values are missing which I would have to replace. Please find an overview of the data attached.
The cells with an "x" mean that I have data for this country in the respective year. Orange cells mean that I would have to calculate the missing value (e.g. by an average). I was thinking to constrain the analysis to Angola, South Africa, Egypt and Uganda. If I could replace the organge cells, this would give me 44 data points.

Does anyone know literature which states how much values I can replace and whether the data points would be sufficent to run a FE model?

Thanks & best,