In summation, based on the results of the present analysis, five key conclusions may be drawn:
The productivity of the CGIAR investment has been shown in all five scenarios of aggregate benefits estimates to exceed normal standards of investment efficacy by a substantial margin.
Three main categories of innovations constitute the vast majority of plausibly estimated benefits: modern varieties of wheat, modern varieties of rice and the biological control of the cassava mealybug.
Selection of a single scenario as the most true minimum aggregate benefit-cost ratio of the CGIAR investment is arbitrary without client articulation of required evidence for impact claims.
More comprehensive data collection procedures, including a stakeholder-needs-based mechanism for priority-setting for impact-assessment, could help to render more reliable and meaningful results from IA studies.
Improving the average degree of methodological transparency would be an easy means to raise the confidence with which findings may be utilized and scaled-up in future analyses.
The effectiveness and efficiency of CGIAR research could be more substantially demonstrated through improved impact assessment procedures. In general, the realm of impact assessment within the CGIAR is in need of additional guidance at the System level, since methodological standards for impact assessment vary so greatly among Centres and programmes. Unfortunately, institutional dynamics have inhibited the development of this direction, as it appears that insufficient resources have been allocated to System level guidance.
With the range of potential benefit-cost ratios for the CGIAR investment ranging from less than two to over 17, it is logical to ask: What ratio should be considered as closest to the truth, and should be used for future reference? This study cannot provide a definitive answer to this question, as such is dependent upon the standards of causality demanded from IA for the acceptance of results. In turn, specification of such standards must come from the clients of IA studies, principally investors in the System, so that the kinds of proof required for the validity of impact claims can be identified.
It is crucial that the necessary data requirements for staking different kinds of impact claims be articulated by key stakeholders. Such would help to clarify the resources to be invested in IA activities, and would help to improve IA utility for priority-setting and allocative decisions. The foundation for such requirements should be interaction with the prospective clients of impact assessment studies, including donors, co-sponsors, NARS, NGOs, farmers organizations and even researchers within the CGIAR institutions, so as to achieve consensus on the minimum requirements for viable impact assertion, and the proper role of impact assessment. From this basis, a System level impact-assessment prioritization system should be derived, as was initially recommended by the Impact Assessment Task Force eight years ago. Once such a system has been developed, impact assessment for accountability should be planned at a System level in consultation and consensus with the IARCs.
To conclude, much credit should go to those IARCs that have successfully initiated endogenous economic impact assessment efforts, with little System-level guidance and limited resources. These efforts, even if somewhat uncoordinated, have helped to facilitate methodological progress, and have helped to highlight System successes. Now, to move forward, and help build upon these accomplishments, impact assessors and interested stakeholders should come together to define expectations and standards for the future.