This paper focuses on the question of income convergence among countries. While the methodology used to determine convergence differs from the common cross-sectional approach, it corroborates Baumol's finding of a convergence club among the world's wealthiest countries. It also shows that there is strong evidence in support of a second convergence club, however. This one is among the world's very poorest countries. These clubs exhibit different forms of convergence. The group of wealthy countries is characterized by what may be referred to as upward convergence, where the poorer group members catch up with the richer countries. The group of extremely poor countries exhibits downward convergence, or a reduction in income disparity brought about by nearly zero, or even negative, growth by the group's `wealthier' members. One of the attributes that sets these countries at the bottom apart is that they are very close to what Stigler once calculated as the least cost subsistence diet. Inserting this constraint into the neoclassical growth model produces two steady states, with divergence in between. An example of such a model is developed here.
