Based on these impressive median or average rates of return, many have argued that there is widespread underinvestment in agricultural research and development, including in Africa. Others, however, have been skeptical about the rate-of-return evidence and have pointed to two weaknesses: (a) the sample of rate-of-return results is biased towards the success cases and ignores failures; and (b) rate-of-return methodologies are seriously flawed (Roseboom, 2002). Although the defenders of the underinvestment hypothesis tend to share these concerns, they argue that the median or average rate of return is so high that even after adjusting for the various uncertainties the evidence will still hold (Oehmke and Crawford, 1996; Evenson, 2001: 573-628).
An argument frequently put forward in discussions of why the Green Revolution seems to have by-passed Africa is that the diversity in production systems in Africa is many times greater than in irrigated South Asia where the Green Revolution occurred. This gives Asia important economies of scale advantages in the generation and diffusion of new technology. This would explain why Asia has performed better in terms of agricultural (land and labour) productivity than Africa with an on-average lower research intensity ratio. Evenson and Gollin (2001), however, argue that the Green Revolution has not by-passed Africa but that its impact has started later as can be shown by the area cropped under modern varieties across regions and over time (Figure 5.11).
Hence, African farmers have only recently started to benefit from crop genetic improvement. This is consistent with relatively stagnant average yields for most crops across the continent for the past 40 years. Moreover, the increase in agricultural production in Africa has been, until recently, mainly due to the shortening of fallows and the conversion of forests and rangelands into cropland. By bringing poorer land into cultivation and by not allowing cropland to recuperate its fertility (either naturally or artificially), average yields will remain low and the impact of modern varieties will not be evident.
Because in Africa we are dealing with highly diversified, primarily rainfed agricultural systems - and this will be the case for decades to come - it is likely we will have many smaller 'green evolutions' than pervasive Green Revolutions of the South Asian type. In fact the numerous examples now documented in Table 5.4 show there are very high returns to agricultural R&D investments already evident in many individual African research programs and projects. These indicate that there is substantial underinvestment in African agricultural research and development and many more opportunities for profitable investments may be made. Because of its heterogeneity, primary reliance on smallholder rainfed systems, immense size and poor infrastructure, Africa is likely to require greater R&D investments per unit of agricultural productivity improvement than was the case in mostly irrigated South Asian Green Revolution rice-wheat systems. Governments and the international community will have to recognize this and modify their expectations and priorities accordingly.