1 Introduction

It remains a puzzle why a rice Green Revolution has failed to take place in most areas of sub-Saharan Africa (SSA), even though it transpired in tropical Asia a half-century ago. It is still more puzzling if we recall the fact that Asian rice Green Revolution technology is highly transferable from tropical Asia to SSA. Historically, improved rice farming technology was transferred from Japan to Taiwan in the 1920s, to the Philippines in the 1960s, and further to other Southeast Asian countries and South Asia in the 1970 and 1980s (Barker and Herdt 1985; Hayami and Ruttan 1985; Otsuka and Zhang 2021). Otsuka and Larson (2013, 2016) argue that Asian rice Green Revolution technology is also highly transferable even to sub-Saharan Africa (SSA), given the high yield performance of Asian rice varieties in various parts of SSA. Therefore, the key question is: what are the major constraints that prevent the full-fledged success of the rice Green Revolution in SSA?

As pointed out in Chap. 1, policymakers, and even leading agricultural economists, have assumed that the Green Revolution requires only the use of modern inputs, such as modern varieties (MV) and chemical fertilizers (Gollin et al. 2021; Carter et al. 2021), ignoring the role of improved cultivation practices, such as transplanting in rows, land leveling, and bunding (Abe and Wakatsuki 2011). We argue strongly in this book that the widely held view that Green Revolution can be called a “seed-fertilizer revolution” is misleading. After reviewing the literature on the role of extension in Chap. 2, case studies of rice cultivation training programs in Cote d’Ivoire, Tanzania, Uganda, and Mozambique were examined in Chaps. 36. These case studies demonstrated that the key to disseminating such management-intensive technologies—consisting of not only the use of improved varieties and inorganic fertilizer but also improved cultivation practices—is the training of farmers by extension agents. It is truly remarkable to observe that rice yield, income, and profit increase significantly after rice production training programs are offered, even without changing the marketing system, input subsidies, irrigation conditions, or credit provision. We cannot over-emphasize this exceedingly important finding because, for the rice Green Revolution to be successful in SSA, the notion that the rice Green Revolution is management intensive must be widely shared among policymakers, researchers, and extension workers.

While the rice cultivation training program is a crucial entry point to increase the intensity and productivity of rice farming in SSA, there are important complementary strategies that are indispensable for realizing a full-fledged rice Green Revolution. After reviewing the literature on the role of mechanization and irrigation in the intensification of rice farming in Chap. 7, case studies from Cote d’Ivoire and Tanzania examining the impact of agricultural mechanization (particularly tractorization) on the intensity of rice farming are reported in Chaps. 8 and 9, respectively. The rate of return to a large-scale irrigation scheme in Kenya is discussed in Chap. 10, whereas the efficiency of large scale relative to small and medium-scale irrigation schemes in Senegal is assessed in Chap. 11. Chapter 12 reports on the critical role played by rice millers in improving the quality of milled rice in Kenya, while the effect of the information provision about quality-based pricing to farmers on paddy quality is discussed in Chap. 13.

In this concluding chapter, we summarize the significant findings of each case study to draw clear policy recommendations that will contribute toward a full-fledged rice Green Revolution in SSA.

2 Training as an Indispensable Entry Point

This section summarizes the results of five new case studies presented in this volume, three of which are concerned with longer-term impacts and the extent of information spillovers from participants to non-participants in training in irrigated areas (Chap. 3 on Cote d’Ivoire and Chap. 4 on Tanzania) and rainfed areas (Chap. 5 on Uganda). The remaining two case studies are concerned with the short-run impacts of rice cultivation training in rainfed areas (Chap. 4 on Tanzania and Chap. 6 on Mozambique).

Concrete evidence on the sustainability of training impacts and the spillover of technological knowledge is provided by the experimental study of irrigated areas in Cote d’Ivoire in Chap. 3. Eligibility to participate in training was randomly allocated to avoid any imbalance in pre-training characteristics between potential participants and non-participants. Thus, there were no significant differences between treated (eligible) and non-treated (ineligible) households in paddy yield, fertilizer use, and the adoption rates of leveling and transplanting in the pre-training year of 2014. When training was offered in 2015, treated and non-treated households were initially requested not to communicate with one another in order to identify the pure impact of the training. It was revealed that the yield of treated households, their fertilizer application, and the adoption of leveling and transplanting in rows significantly increased from 2014 to 2015 compared with non-treated households. After the 2015 season, treated and non-treated households were advised to communicate and spread the new practices. Consequently, the adoption rates of improved management practices of non-treated households increased from 2015 to 2016, indicating spillover effects. As a result, the yield gap between the two groups of farmers disappeared.Footnote 1

Longer-term and spillover effects of rice cultivation training can also be confirmed by a study of the diffusion of improved rice production practices in irrigated areas in Tanzania, as reported in Chap. 4. In the training program, competent and motivated farmers, called key farmers, were initially selected and directly trained by extension agents at nearby training institutes for 12 days before the main crop season in 2009. Each key farmer was requested to choose five intermediary farmers and train them in the improved rice production methods. Intermediary farmers were then expected to train other ordinary farmers. As expected, key farmers’ performance was better than the intermediary and other farmers in 2008 before the training program commenced (Table 4.3). The performance of key farmers substantially improved immediately after they took the training program in 2009, including an increased adoption rate of MVs. A critically important observation is that this high performance of key farmers was sustained for the next three years, indicating that the impact of the rice production training program is sustainable. Also noteworthy is that the performance of intermediary farmers improved gradually, followed by improved performance of other farmers in later years. As shown in Table 4.3, paddy yields of four to five tons per hectare were achieved, which are very high by any standard (see Fig. 1.5 for yield trends of SSA and India). Thus, there is no question that a rice Green Revolution occurred in irrigated areas in Tanzania due to the rice cultivation training program.

A spectacular example of high paddy yield triggered by training in rainfed conditions is found in the Kilombero Valley in Tanzania (see Chap. 4). Rice cultivation management training was offered by a large private rice plantation to nearby farmers in 2012 and 2013. The production management approach was called a “system of rice intensification (SRI).” However, unlike its original definition,Footnote 2 the use of MVs and chemical fertilizer was recommended, the use of irrigation was not assumed, and straight-row dibbling was promoted.Footnote 3 This is why Chap. 4 referred to this approach as a “modified SRI” or MSRI. As can be seen from Table 4.1, paddy yield was as high as 4.7 tons per hectare on plots where the trainees adopted MSRI technologies. To the best of our knowledge, this is higher than the highest paddy yield under rainfed conditions in tropical Asia. There were no changes in technology adoption on other plots. Thus, the significant difference in paddy yield between MSRI plots and others can be attributed not only to the difference in the adoption of MVs and the use of chemical fertilizer but also to the difference in cultivation practices. While substantial yield gains were observed, trained farmers did not adopt MSRI practices on all plots they had access to. The authors also did not find systematic evidence of information spillover from the trained to non-trained farmers, even though their ongoing study anecdotally shows some signals of spillover.Footnote 4 Thus, in this study site, it remains unclear how sustainable and widespread the impact of MSRI training in rainfed conditions will be.

According to Fig. 5.1 in Chap. 5, the average yield per hectare of participants in the rice cultivation training in rainfed areas in Uganda increased by roughly 50% from 2008/09 to 2011/12. This high yield was maintained for five to six years, suggesting that the impact of rice cultivation training could be substantial and sustainable. In contrast, the average yield of non-participants did not increase as much as that of participants in 2011/12 but caught up in 2015/16. In this study site, however, participants in the training program were not randomly selected: participants were those who expressed interest in the training program, whereas non-participants did not show interest or were not informed about the training by village leaders. Thus, rigorous comparisons were made based on the propensity score matching method between participants in training villages and farmers with similar characteristics in non-training villages, as well as between non-participants in training villages and farmers with similar characteristics in non-training villages. According to this analysis, while participants are found to improve yields and adopt transplanting in 2015/16 relative to their counterparts in non-training villages, there is no evidence that non-participants in the training villages improved their performance relative to farmers in non-training villages. The results thus do not provide supporting evidence on information spillovers from participants to non-participants in the training villages, even though non-participants improved the planting method from broadcasting to transplanting, which indicates that non-participants partially learned technology from participants.

Rice farming in Mozambique is underdeveloped and rainfed, with direct seeding of local varieties without any fertilizer. Chapter 6 evaluated a randomized controlled trial (RCT) of rice cultivation training in the Central Region implemented by the Japan International Cooperation Agency (JICA). After the baseline survey was conducted in the 2016–17 season, rice cultivation training programs were offered to six farmer’s associations in 2017 and another six in 2018, referred to as Demo groups 1 and 2, respectively. According to Table 6.2, paddy yield was lowest among the control farmers (i.e., no assignment of training) in the post-training year of 2018–19, even though it was highest in the pre-training year of 2016–17. Since paddy yield depends on rainfall in the study sites as well as household and plot characteristics, it is challenging to identify the impact of rice cultivation training from the descriptive data. According to the regression analysis, which controls for relevant characteristics, yield increased significantly by 450–550 kg per hectare among farmers in the treated groups compared with the control group. This result is consistent with higher adoption of improved practices, such as plot leveling and straight-row transplanting, among the treated groups more than in the control groups. Yield gain of 450–550 kg may appear modest, but this accounts for an approximately 30% improvement compared to the control group. This was achieved without applying any additional modern inputs, such as improved varieties or chemical fertilizer. These results are consistent with earlier studies of the impact of rice cultivation training on the rice production performance in rainfed areas in northern Ghana by de Graft-Johnson et al. (2014) and eastern Uganda by Kijima et al. (2012). Note also that rice farmers learn new cultivation practices from participating in the training program at the demonstration plot or from extension workers. In contrast, there is no clear evidence of “social learning” or information spillover from participants to non-participants, at least in the short run in the Mozambique sites.

To sum up, the evidence reviewed in this section indicates that the impacts of cultivation training are significant and sustainable in both rainfed and irrigated areas, as well as being transmissible from participants to non-participants, especially in irrigated areas. It must be emphasized that such impacts were realized without any improvement in irrigation, marketing, or credit programs, among others. Thus, our findings can be taken to imply that rice cultivation training is a crucial entry point to the rice Green Revolution in SSA.

3 Complementary Development Strategies

3.1 Impacts of Tractorization

Power tillers were introduced in Asia intensively in the 1980s to reduce the use of draught animals and labor (David and Otsuka 1994). However, SSA differs in this respect because manual labor has primarily been used for land preparation due to the unavailability of draught animals in most areas (Chap. 7). In other words, power tillers are substitutes for draught animals as well as labor in Asia, whereas they are substitutes for human labor in SSA.

According to a study on the use of power tillers in Cote d’Ivoire reported in Chap. 8, the average paddy yield is significantly higher for power tiller users (4.7 tons/ha) than non-users (3.6 tons/ha). Furthermore, both family and hired labor were more intensively applied on plots plowed and harrowed by power tillers. The use of power tillers also increased fertilizer application and enhanced the implementation of improved cultivation practices. Cultivation size is also significantly larger for power tiller users (0.9 ha) than non-users (0.7 ha), indicating that the use of power tillers contributes to both intensification and extensification, which is consistent with the conceptual framework shown in Fig. 1.8.

The case of Tanzania is unique because not only power tillers and hand hoes but also four-wheel tractors and draught animals are used in land preparation. While four-wheel tractors were most common in 2018, the use of power tillers has been increasing more sharply. Several important observations can be made from the analysis of Chap. 9. First, the adoption of power tillers is associated with the higher yield and the higher adoption of modern inputs and straight-row transplanting compared with draft animals. Second, neither the adoption of power tillers nor four-wheel tractors is associated with a larger rice cultivation area per household. Third, there is no evidence that using four-wheel tractors contributes to intensification compared to using draught animals. It may well be that it is difficult to maneuver heavy four-wheel tractors, which bog down in muddy small paddy fields. Judging from this analysis, the introduction of power tillers seems to be highly conducive to intensifying rice farming and improving rice yield in Tanzania.

3.2 Impacts of Irrigation

There is no question that irrigation has significant impacts on rice farming performance because rice plants rely on a steady water supply. According to Balasubramanian et al. (2007), the average paddy yield in 16 countries with less than a 10% irrigated ratio in SSA was 1.6 tons per hectare in 2004. By contrast, the average of four countries with more than a 90% irrigation ratio (i.e., Cameroon, Kenya, Mauritania, and Swaziland) was 3.9 tons per hectare in the same year. A similar tendency was found in tropical Asia in the late 1980s (David and Otsuka 1994).

One of the critical questions is whether the rate of return to investment in large-scale irrigation schemes is high enough to justify the investment. Chapter 10 estimated the rate of return to irrigation investment in the Mwea Irrigation Scheme in Kenya by asking the hypothetical rate of return if the Mwea Irrigation Scheme was constructed as a new scheme now. The estimated internal rates of return are reasonably high (10.7–14.9%) if the value-added ratio of 0.8 is assumed but lower than 7% if the value-added ratio of 0.5 and low rice prices are assumed.Footnote 5 Note that the Mwea Irrigation Scheme is considered one of the most successful irrigation schemes in SSA because of its extremely high yield. The apparent conclusion is that the rate of return to investment for such a successful large-scale irrigation scheme as Mwea is not necessarily very high, mainly because world rice prices have remained low after the success of the Asian Green Revolution.Footnote 6

We want to make a couple of additional comments on the rate of return to investment in irrigation in SSA. First, because of the possible complementarity between improved cultivation practices and the availability of irrigation water, well-designed training on appropriate rice cultivation may significantly enhance the rate of return to large-scale irrigation investments. Second, rates of return would be higher if we assume that the benefits of irrigation schemes are accrued not only to producers, rice millers, and traders but also to various economic sectors, including input supplies and other related businesses, through economic linkages and transactions. If such multiplier or market-wide “general-equilibrium” effects are taken into account, the net benefits for the entire economy could be enlarged to justify investment in irrigation.

While rates of return to investment in large-scale irrigation schemes in SSA are considered to be generally low, as discussed in Chap. 10, there is no evidence that larger-scale irrigation schemes are less efficient than smaller ones in Senegal River Valley, as reported in Chap. 11. It is also identified that government-invested irrigation schemes perform worse than private ones. Thus, it seems that not only the scale of irrigation schemes but also the type of investors and management capacities of users critically affect the efficiency of irrigated rice farming.

3.3 Role of Upgrading Milled Rice and Grading Paddy Quality

Aside from the proper timing of paddy harvesting and its impeccable drying, the quality of milled rice depends on the quality of milling machines, particularly the use of destoners and color sorters. Rice millers are in a good position to provide information about appropriate harvesting and drying to farmers and local traders and provide information about milled rice quality to urban traders and consumers through branding and marketing for supermarkets in both Asia (Reardon et al. 2014) and SSA (Ogura et al. 2020).

Chapter 12 observes significant improvements in rice milling machines in the Mwea Irrigation Scheme in Kenya. As in many other places in SSA, a major factor impairing the quality of milled rice is the inclusion of small stones and other impurities. However, these can be removed by installing destoners. Three rice millers adopted destoners in 2011 and 34 millers had done so by 2019.Footnote 7 The estimated market share of non-adopters was 80% in 2011 but decreased to less than 10% in 2019. Rice millers adopting destoners have a greater milling capacity, can charge higher milling fees, and fetch 10% higher prices for milled rice sold than non-adopters. Furthermore, 50% of the early adopters and 25% of the late adopters had brand names as of 2019, and only these millers sold milled rice to supermarkets in large cities.

Chapter 13 is concerned with the improvement of the quality of paddy produced in northern Ghana. It was hypothesized that the lack of knowledge of paddy quality and its relationship with price causes farmers to continue producing low-quality paddy. The authors conducted a field experiment in northern Ghana to verify this hypothesis. They randomly selected 108 villages and ten rice producers from each sample village. From this group, they randomly chose 54 treatment villages and provided the farmers with information about paddy quality-enhancing technologies and quality parameters appreciated by the market. It was found that the intervention significantly influenced farmers to adopt some quality-enhancing practices. Moreover, the intervention induced significant behavioral changes among the treated farmers: they sold more aromatic varieties of paddy outside the village than the control farmers and received a higher sales price. Thus, Chap. 13 concludes that providing information about paddy quality and quality-based pricing improved farmers’ paddy quality management and market sales.

4 Concluding Remarks

To conclude, our major findings and policy implications toward realizing a full-fledged Rice Green Revolution in SSA are summarized as follows. First, we found that the impact of rice cultivation training programs is significant not only in the short run but the results indicate that it is likely to be long-lasting. It is also being disseminated through information spillovers from training participants, particularly in irrigated areas. Thus, we advocate the rice cultivation training program as a critical entry point. This strategy is different from tropical Asia because Asian farmers largely adopted basic rice cultivation practices at the dawn of the Green Revolution, most likely due to the long tradition of rice farming. However, whether and to what extent new technology is disseminated from participants to non-participants in the training program in rainfed areas must be further analyzed. Such participant-to-non-participant information dissemination may be more difficult in rainfed areas than in irrigated areas, partly because of the greater heterogeneity of agro-ecological farming conditions and partly because of the weaker social interaction among farmers in rainfed villages without a water user association.

Second, we found that the use of power tillers promotes the intensification of rice cultivation by thorough plowing, harrowing, and leveling and by inducing increased labor use for subsequent care-intensive activities in both Cote d’Ivoire and Tanzania. To overcome the handicap arising from the unavailability of draft animals in many parts of SSA, the use of power tillers should play a significant role in the intensification of rice farming. However, when soil is very hard, powerful four-wheel tractors may be more suited for plowing than power tillers. The defect of four-wheel tractors is their heavy weight: they bog down in muddy paddy fields when harrowing is performed. To what extent power tillers can be successfully disseminated in SSA must be further examined.

Third, we found that the rice farming intensification, including the application of modern inputs and adoption of recommended rice management practices induced by rice cultivation training—coupled with intensive land preparation with oxen and four-wheel tractors—improved the return to large-scale irrigation investment in Mwea. However, large-scale irrigation investment is often not economically viable because of the low rice price resulting from the Asian rice Green Revolution. Yet, relatively large irrigation schemes perform better than smaller ones in Senegal River Valley, which does not indicate that large-scale irrigation schemes are inherently inefficient. Although evidence is not decisive, it seems to us that large- and medium-scale irrigation schemes in SSA significantly contribute to the improvement of productivity of rice farming in SSA, judging from the extremely high paddy yield in our study sites in Kenya and Senegal. An under-explored issue is whether small-scale irrigation schemes are more cost-effective than large- and medium-scale schemes.

Fourth, we also found evidence that the introduction of improved milling machines has a significant impact on the quality of milled rice. Rice millers must be trained or provided information about improved milling machines to produce high-quality milled rice that can compete with imported rice from Asia. Similarly, to produce high-quality paddy, rice farmers must be trained to learn paddy quality-enhancing technologies and informed about the relationship between paddy quality and market prices.

In conclusion, we argue firstly that strengthening the public extension system for improved rice cultivation must be a central strategy to realize the rice Green Revolution in SSA. Secondly, we argue that the promotion of power tillers must play a complementary role in supporting the rice Green Revolution. Thirdly, since the availability of irrigation water is a decisive factor affecting the performance of rice cultivation, the benefits and costs of large-scale irrigation projects in SSA should be carefully reconsidered. Finally, we recommend the training of rice millers in the use of improved rice milling machines to enhance the quality of African-milled rice. There is little doubt that a full-fledged rice Green Revolution can take place in SSA if the rice extension system is adequately strengthened, power tillers are widely diffused, irrigated areas expand significantly, and the quality of milled rice is improved.