Log scales are good for when data has a wide range of magnitudes. To fit Zimbabwe on the scale, about 90% of the countries would be bunched up in the first 10% on a linear scale. Additionally it's a log-log plot, so the relationship is linear