EXECUTIVE SUMMARY Low effective demand is often cited as a major reason for the lack of private-sector involvement in the seed system for legume crops in developing countries. The viability of these seed systems depends on whether farmers perceive the seed product as a quality planting material, and whether they are willing to pay a premium for the seed relative to grain. To evaluate these issues, double blind field experiments and experimental auctions were conducted with more than 500 bean and cowpea farmers in northern Tanzania and northern Ghana. The experiments were designed to gauge the relative demand for three types of seed products: certified, quality declared (QDS), and recycled (i.e., grain saved from previous harvest). These three types of seeds differ in seed input (i.e., which generation of seed is used to produce them), the level of regulatory supervision they receive during production, and the technical conditions under which they are produced. Whether the production cost differential across these types of seeds makes them qualitatively different products as reflected in their perceived or actual performance of the plant, and whether that translates into sufficient price premiums paid by farmers for these better quality seeds are the research questions addressed by this study. Overall, the results of the field experiments indicate that, all else equal (i.e., the variety and management practices), plots planted with certified seed performed better on measures of objective indicators (i.e., yield, seed quality, and agronomic traits). However, the actual yield difference between the different bean plots in Tanzania was much smaller than the yield differential observed for the cowpea plots in Ghana. Irrespective of the magnitude of the yield differences (or yield deterioration observed over each generation of seed production), an important implication of this finding is that to increase productivity, it is not sufficient to promote only the adoption of improved varieties, but it is likely also necessary to promote the use of higher quality seed as an input. In both countries, plots planted with certified seeds were perceived, based on the observations of plots at the flowering and harvest stages, to be of the best quality by a majority of farmers. All else equal, farmers were willing to pay a premium for the higher quality seeds. The relative difference in farmers’ willingness-to-pay (WTP) for different seed types was highly correlated with the relative difference in their perceived quality. This willingness to pay a premium for quality legume seed by smallholder farmers is encouraging and indicative of the existence of an effective demand for self-pollinated crop seeds (as opposed to recycled grain) like beans and cowpeas. While a significant portion of farmers in the study (35-40%) were willing to pay a large enough premium to cover the higher production costs of certified seed, across both crop/country case studies, we found the rest of the legume growing farmers’ willingness to pay premium for quality seed was below the current local price of certified seed. Indeed, for a sub-set of these farmers in Tanzania (25%), the willingness to pay for quality seed was even lower than the local grain price. The implication of these findings is that there is no one-sizefits-all strategy to meet the seed needs of all the farmers. Current efforts to encourage the private sector to produce and supply certified seeds through agro-dealers can potentially meet the seed needs of at most 35-40% of farmers (if the quality of those seeds is substantially superior to recycled grain, and the seed is of a preferred variety). Clearly, more research and discussion is needed to assess the seed needs of the remaining (majority) farmers whose WTP for quality seed appears to be less than the current market price of certified seed. One avenue of research is to investigate mechanisms to lower the cost of producing quality seed, without reducing the incentives for seed producers, so that quality seed can be provided at prices closer to that of grain. While cost-reducing strategies through policy, programmatic and technological options should remain a high priority for governments and donor-supported programs such as the Alliance for a Green Revolution in Africa (AGRA), this study also indicates the need for continued support for innovative and smart subsidy-based approaches to meet the needs of the 15-20% of farmers whose WTP for seed is so low that for-profit seed production/marketing models will not work. In drawing this implication, we have not taken in to consideration the varietal preferences of farmers and how their absolute WTP for quality seed might be influenced by the type of variety, which was held constant in the experiments. More research is needed to understand farmers’ varietal preferences and their demand for seed that embodies both their preferred traits and quality. Across the two countries, results indicate that seed quality matters, and that on average certified seeds consistently outperformed both the QDS and recycled seeds. In Ghana, on average the QDS outperformed the recycled seed, but in Tanzania the QDS and recycled seed performed similarly. Community based QDS production and sale is often promoted as a lowcost option for increasing farmers’ access to quality seed within a community. However, if the lower cost also comes with lower quality, then the sustainability of QDS seed systems is questionable since farmers may not be willing to pay the price premium required. That said, this study also illustrates that when the advantages of planting better quality seed in place of recycled seed can be demonstrated, a significant portion of farmers appear to be willing to pay more, indicating that private sector seed systems can be viable.