Optimal expenditures for agricultural research and extension: implications of underfunding