This paper looks at past trends of agricultural R&D allocation in developing countries, projects future performance of agriculture in these regions based in past investment and determines the optimal allocation of R&D investment across regions to maximize global welfare using a dynamic linear programming model of global agriculture. Results suggest that present allocation of agricultural R&D in SSA is highly inefficient and substantial gains could be obtained by increasing investment in East Africa in the next twenty years. At the global level, differences between efficient and present investment allocation are smaller than those observed within SSA due to the importance of China as an innovator in agriculture.