Food Policy 107 (2022) 102196 Available online 26 December 2021 0306-9192/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Viewpoint Viewpoint: Aligning vision and reality in publicly funded agricultural research for development: A case study of CGIAR Philip Thornton a,*, Jeroen Dijkman b,1, Mario Herrero c, Lili Szilagyi a, Laura Cramer a,d a CGIAR Research Programme on Climate Change, Agriculture and Food Security (CCAFS), International Livestock Research Institute (ILRI), P.O. Box 30709, Nairobi 00100, Kenya b Animal Sciences Group, Wageningen University & Research, P.O. Box 65, 8200AB Lelystad, The Netherlands c Department of Global Development, College of Agriculture and Life Sciences & Cornell Atkinson Centre for Sustainability, Cornell University, Warren Hall, Ithaca, NY 14850, USA d Alliance of Bioversity International and CIAT, Km 17 Recta Cali-Palmira, AA 6713, Cali, Colombia A R T I C L E I N F O Keywords: Food systems Agricultural research Research expenditure Poverty Vulnerability CGIAR A B S T R A C T Global food systems are currently facing unprecedented challenges with respect to production and nutritional targets, inclusivity and environmental footprint. Several recent reports highlight the need for major, rapid reconfiguration of our food systems as a result. International publicly funded agricultural research for devel opment will play an increasingly vital role in support of such goals as reducing poverty, improving food and nutrition security, and improving natural resources and ecosystem services. Here we take stock of the work over the last decade of CGIAR, one of the major players in the agricultural research for development arena, from the perspective of published, peer-reviewed science. We do this with respect to several elements of its vision as set out in 2011, elements that are shared by many other organisations that are also working towards achieving the Sustainable Development Goals. Overall, we found a strong association between number of CGIAR publications and countries with large numbers of rural poor and high child stunting prevalence. At the same time several countries were identified that are anomalous, being either relatively over- or under-represented in the peer- reviewed literature in relation to numbers of rural poor and stunting prevalence. On average, 30% of the cal ories consumed in national food baskets come from food sources that are not currently the commodity focus of CGIAR research, such as fruit and vegetables. We identify possible ways in which the alignment between the strategic objectives of an agricultural research for development organisation such as CGIAR and its publicly funded science outputs might be further strengthened, for maximum impact in the nine years that are left for the world to achieve the Sustainable Development Goals. 1. Introduction Over the last fifty years, the changes in the landscape of agricultural research for development have been profound. While the progress in food production to feed more than twice as many people has been impressive, there is growing recognition that there are serious failings associated with current food systems with respect to production and nutritional targets, inclusivity and environmental footprint (Lobo guerrero et al., 2020). Several recent reports highlight the fact that current trajectories are neither on track nor speedy enough to meet the Paris Agreement on climate change and the Sustainable Development Goals (SDGs), particularly SDG 2 (zero hunger), and that major reconfiguration of our food systems is needed (FAO et al., 2019; Willett et al., 2019; FOLU, 2019; De Cleene, 2019; Herrero et al., 2020a; Steiner et al., 2020, Meridian Institute, 2020; Barrett et al., 2020). With respect to the allocation of funds, the international research and development (R&D) literature has predominantly concentrated on the identification of relative expenditure by the private and public sector and its potential contribution to public goods such as improved nutrition and poverty reduction (e.g., Fuglie et al., 2011; Fuglie et al., 2012). Data from this literature show that the growth in private spending tripled between 1990 and 2014 (Fuglie, 2016) and that in high income coun tries this now constitutes the main investment in agricultural R&D. Evidence from other analyses also indicates that public R&D, unlike private R&D investments (Beintema and Stads, 2017), focusses more on * Corresponding author. E-mail address: p.thornton@cgiar.org (P. Thornton). 1 Current affiliation: Société des Produits Nestlé S.A., Route du Jorat 57, CH 1000 Lausanne 26, Switzerland. Contents lists available at ScienceDirect Food Policy journal homepage: www.elsevier.com/locate/foodpol https://doi.org/10.1016/j.foodpol.2021.102196 Received 17 March 2021; Received in revised form 13 November 2021; Accepted 20 November 2021 mailto:p.thornton@cgiar.org www.sciencedirect.com/science/journal/03069192 https://www.elsevier.com/locate/foodpol https://doi.org/10.1016/j.foodpol.2021.102196 https://doi.org/10.1016/j.foodpol.2021.102196 https://doi.org/10.1016/j.foodpol.2021.102196 http://crossmark.crossref.org/dialog/?doi=10.1016/j.foodpol.2021.102196&domain=pdf http://creativecommons.org/licenses/by-nc-nd/4.0/ Food Policy 107 (2022) 102196 2 issues and crops with smaller potential returns (Reynolds et al., 2017). In line with the private sector, however, most public and philanthropic agricultural R&D investment still targets commodity grains and cash crops (Anderson et al., 2017). Another related strand of investigation focusses more specifically on returns to food and agricultural R&D investments (e.g., Maredia and Raitzer, 2010; Hurley et al., 2016; Pardey et al., 2016). The methodol ogies used to estimate aggregate returns to research investments have been subject to significant debate for many years (e.g., Griliches, 1958; Rao et al., 2017). Although it is not the objective here to enter into these discussions, it should be noted that the internal rates of return quoted for research investments are sometimes implausible (Rao et al., 2017), are not recommended in comparing the relative profitability of investments (Daunfeldt and Hartwig 2014), and, particularly in sub-Saharan Africa, are not particularly congruent with the composition of agricultural production (Pardey et al., 2016). There is a large amount of information on agricultural research funding, although much of it is in aggregated form. The ASTI website (https://www.asti.cgiar.org/), for example, is a public resource with information on investments, human capacity, and the institutional structure of agricultural R&D in low- and middle-income countries (LMICs). In addition, many of the major philanthropic, multilateral and other investors in agricultural R&D liberally share their investment portfolios online; see, for example, https://www.gatesfoundation. org/What-We-Do/Global-Growth-and-Opportunity/Agricultural-Deve lopment; https://ec.europa.eu/eip/agriculture/en/about/pooling-fund ing-streams-boost-interactive; https://www.ers.usda.gov/data-produc ts/agricultural-research-funding-in-the-public-and-private-sectors/. This information on R&D investment flows is clearly essential to un derstanding the contribution of research to agricultural growth and development. In its current publicly available form, however, it does not provide the necessary detail to allow the different flows to be compared and contrasted, which effectively impedes their use in any analysis or recommendation related to the efficient allocation of funds. In a highly dynamic landscape in which research needs are shifting rapidly, this is a considerable impediment to the effective and efficient redirection or repurposing of funding flows. One input into analyses that could help to address this issue would be information on the congruence between shifting research for develop ment needs and actual research expenditure. Given the difficulties in obtaining disaggregated research expenditure data, here we investi gated the viability of using recent peer-reviewed literature as a proxy for research expenditure by CGIAR, a large, international agricultural research for development research system encompassing a wide research agenda in different countries. We then analysed the broad alignment of these proportions with global databases on population, poverty, food and nutrition security, and national food baskets (i.e., the aggregated make-up of the human diet at the national level). We iden tified some gaps in country and topic coverage that may warrant attention in the future from CGIAR and/or other partners working on agricultural research for development (AR4D). We discuss the limita tions of our analysis and make suggestions for some simple monitoring of research expenditures in the future that could provide information to help rebalance agricultural research for development portfolios rela tively quickly and effectively in highly dynamic environments. 2. CGIAR: A major public contributor to agricultural research for development CGIAR is the largest global agglomeration of international publicly funded AR4D institutes. It was established in 1971 as one response to mid-20th century concerns about widespread famine. In its five decades, CGIAR has spent about USD 60 billion in present value terms, an in vestment that is estimated to have returned tenfold benefits with respect to greater food abundance, cheaper food, reduced rates of hunger and poverty, and a smaller geographical footprint of agriculture (Alston et al., 2020). This is an outstanding return on investment. Funding to CGIAR represents <2 percent of total global agricultural research, and about 3 percent of public investment in LMICs (Beintema et al., 2020). Nevertheless, CGIAR has delivered considerable international public goods, as well as having a key role in building national research ca pacities to help deliver impacts at scale (Beintema and Echeverria, 2020). As GCA (2019) and others note, however, much more investment will be needed, possibly of the order of USD 1.8 trillion to 2030, if food and nutrition security is to be ensured for all in the face of economic and zoonotic shocks and a warming and increasingly variable climate. Total expenditure by CGIAR for 2016 amounted to just under USD 930 million (CGIAR, 2016), and this declined slightly in 2017 to just over USD 800 million (Alston et al., 2020). By region, 49% was spent in sub-Saharan Africa, 27% in Asia, 16% in the Americas, 5% in Central & West Asia and North Africa, and 3% in Europe (though the source of these numbers is unclear). The work program was implemented by some 10,720 staff, working on 19 mandated crops or crop groups (banana and plantain, cassava, potato, sweet potato, yam, maize, wheat, rice, groundnut, cowpea, soybean, chickpea, Phaseolus (dry) bean, pigeon pea, lentil, faba bean, barley, millet, sorghum); six livestock species (cattle, sheep, goats, pigs, fish and chickens); and livestock feed and trees. In what follows, we refer to these 26 crop, livestock and other Nomenclature Acronyms A4NH CGIAR Research Program on Agriculture for Nutrition & Health AfricaRica Africa Rice Center Bioversity2 Bioversity International CCAFS CGIAR Research Program on Climate Change, Agriculture & Food Security CIFOR Center for International Forestry Research CIAT2 Center for International Tropical Agriculture CIMMYT International Maize and Wheat Improvement Center CIP International Potato Center CRP CGIAR Research Program FISH CGIAR Research Program on Fish ICARDA International Center for Agricultural Research in the Dry Areas ICRAF World Agroforestry Center ICRISAT International Crops Research Institute for the Semi-Arid Tropics IFPRI International Food Policy Research Institute IITA International Institute of Tropical Agriculture ILRI International Livestock Research Institute IRRI International Rice Research Institute IWMI International Water Management Institute ODA Official Development Assistance PIM CGIAR Research Program on Policies, Institutions, and Markets RTB Roots, tubers and bananas WLE CGIAR Research Program on Water, Land & Ecosystems WorldFish World Fish 2 In 2020 Bioversity and CIAT combined to form The Alliance of Bioversity International and the International Center for Tropical Agriculture. P. Thornton et al. https://www.asti.cgiar.org/ https://www.gatesfoundation.org/What-We-Do/Global-Growth-and-Opportunity/Agricultural-Development https://www.gatesfoundation.org/What-We-Do/Global-Growth-and-Opportunity/Agricultural-Development https://www.gatesfoundation.org/What-We-Do/Global-Growth-and-Opportunity/Agricultural-Development https://ec.europa.eu/eip/agriculture/en/about/pooling-funding-streams-boost-interactive https://ec.europa.eu/eip/agriculture/en/about/pooling-funding-streams-boost-interactive https://www.ers.usda.gov/data-products/agricultural-research-funding-in-the-public-and-private-sectors/ https://www.ers.usda.gov/data-products/agricultural-research-funding-in-the-public-and-private-sectors/ Food Policy 107 (2022) 102196 3 resources as “commodities”. The share of annual CGIAR expenditure on each of these commodities in recent years is, as far as we know, not known with much accuracy, although we attempt to estimate it below. Fig. 1 shows the evolution in annual expenditure since the early 1990s for rice, wheat and maize, legumes and dryland cereals, roots and tu bers, livestock, fish and trees (ISPC, 2018). To estimate a breakdown in CGIAR research expenditure in more detail, we used the number of peer-reviewed publications as a proxy. As we show below, there is a strong, highly significant correlation between Centre expenditure and the number of peer-reviewed publications. There is some reasonably strong theoretical justification for this. Over the last 15 years, CGIAR research programs have made increasing use of theory of change to help bridge the gap between knowledge generation and development impact on the ground. R&D activities may result in a range of different outputs, and translating these into outcomes and impact requires broad engagement to ensure ownership and buy-in by partners; it also requires enhancing the capacity of next- and end-users to make best use of the outputs produced (Vermeulen and Campbell, 2015). Whether operating at global, national or local level, peer- reviewed science (for example, in relation to innovation testing and scaling or quantifying outcomes and impacts) is a cornerstone compo nent of the multiple pathways that may connect research with devel opment impact on the ground (Thornton et al., 2017). Peer-reviewed science outputs may contribute to development impacts through a wide range of different pathways involving many different actors (Gaunand et al., 2015; Temple et al., 2018; Tomich et al., 2019). Bib liometric analysis is widely used to assess research organisations’ quality of science, including that of CGIAR itself (Rünzel et al., 2021), though linking science quality with impact in a robust way requires additional elements such as impact narratives or case studies (Wilsdon et al., 2015; Pedersen et al., 2020). Using number of peer-reviewed publications as a proxy of both research expenditure and research quality has limitations. Nevertheless, as we argue in Section 5 below, the results still have utility: not to prioritize commodities or countries, or even to provide information that could be used to do so, but in a very broad way to reflect on the alignment between the goals of CGIAR, as expressed in 2011, and the research portfolio in the subsequent decade. We also highlight the necessity of monitoring research expenditures in some detail in future, to allow more nimble AR4D that responds to rapid shifts in circumstances. 3. Methods 3.1. Literature searching and allocation to Centre, country and commodity We searched the Scopus abstract and citation database (https ://www.scopus.com) for all peer-reviewed publications produced by CGIAR scientists over the period January 2010 to April 2020. We restricted the search to documents published after 2009; although shifts between commodities in resource allocations have occurred over the last 30 years (Fig. 1), the structure of CGIAR was relatively stable during this time, mostly reflecting the design and implementation period of the Collaborative Research Programs (CRPs). We chose Scopus as the reference database for two reasons. First, its coverage of the scholarly literature is broader than other options (Wilsdon et al., 2015; Tomich et al., 2019). Second, it allows the user to export author affiliations as well as title, keywords, authors, the abstract, and number of citations (up to the date of extraction). Records were extracted from Scopus using Centre name. All papers were included that had at least one author from one of the then-15 CGIAR Centres. The results from Scopus were exported in text format and scripts written to allocate publications and publication shares to the 26 commodities and to countries. Searching was done in three stages. The search terms used are shown in the Sup plementary Material, page 2. The first stage was to generate files that listed eligible papers that could be allocated to Centres, commodities and countries. For Centres, the scripts (in compiled FORTRAN) searched the author affiliation fields of the Scopus outputs, which contained one affiliation per line of the record. For example, if a paper had nine authors and six different affil iations, Scopus exported the affiliations in different lines. Searching was done on the full Centre name as well as on the usual acronym (e.g., “IITA” and “International Institute for Tropical Agriculture”). Scripts had to be adjusted as sometimes the Scopus record dropped a capital letter or used both “Centre” and “Center” for the same organisation. There are also examples of other non-CGIAR organisations with the same acronym: CIFOR is a national forestry organisation in Spain as well as a CGIAR Centre headquartered in Indonesia, and CIP is a biotech Fig. 1. Share of CGIAR average annual expenditure by commodity, 1992–2016. These data exclude several CRPs (PIM, A4NH, CCAFS, WLE, FISH), CGIAR’s genebanks, IFPRI, IWMI and Bioversity (ISPC, 2018). LDC: Legume and dryland cereals, RTB: Roots, tubes and bananas. P. Thornton et al. https://www.scopus.com https://www.scopus.com Food Policy 107 (2022) 102196 4 company in the UK as well as a CGIAR Centre headquartered in Peru, for example. These records were weeded out manually. For Centres with regional offices hosted at other Centres, the hosting Centres were ignored in allocating each paper to the appropriate Centre. After several iterations, 22,565 eligible papers were included in the analysis, “eligi bility” referring only to the fact that each could be allocated to at least one CGIAR Centre. Nearly 91% were single-Centre papers, and 9% were multi-Centre papers, though most were with two Centres. There was one paper with authors from 11 Centres. To allocate each of the papers to one or multiple commodities, the title and the paper’s keywords were searched. In Scopus, some papers have both author keywords and index keywords; both types were searched where they existed. At first, the abstract was also searched, but this proved difficult to do accurately in the absence of a more sophisti cated natural language processor. In almost all cases, searching the title and keywords provided the information needed. Several iterations were needed to do this reasonably accurately. For example, “bean” may refer to “soybean”, “soy-bean”, “soy bean”, “faba bean” or “Phaseolus bean”, and these had to be untangled. Of the 22,565 papers, 48% were allo cated to one commodity, 8% to two or more, and 44% to no commodity. The papers that had no mention of any of the commodities in the title or keywords were dealt with as described below for stage 2. To allocate each of the papers to one or more countries, a similar procedure was used as for commodities: titles and keywords were searched. As before, several iterations were required to do this accu rately; for instance, the text string “sudan” may refer to “Sudan”, “South Sudan”, or to “Sudanian Savanna”. Of the 22,565 papers, 56% did not mention a country in the title or keywords, 39% mentioned one, and 5% mentioned two or more. Papers were allocated to 145 countries in all. In the second stage, to deal with papers that explicitly addressed no commodity and no country, we proceeded as follows. The number of papers for each Centre was counted, for which it was possible to allocate to one or more commodities and to one or more countries (see Supple mentary Material Figure SM1). There were 8955 of these single-Centre papers. It was assumed that this would give a reasonable representa tion of the breadth of commodities and countries that each Centre was working with. Over the last decade, many Centres have increasingly been pursuing a broader systems orientation in their work, and their scientific publications may not be confined strictly to their mandate crops. From this subset of papers, we defined the countries and com modities that are “reached” by each Centre. For the remaining papers that did not mention either commodity or country, the paper’s country share was split pro rata between the countries of the sub-sample and its commodity share between the commodities in the sub-sample. As an example, more than 90% of AfricaRice’s papers were allocated to rice (not surprisingly), but 3% of them concerned maize, either alone or in addition to rice. For a paper from AfricaRice that did not mention rice explicitly, 90% of the paper was allocated to rice, and 3% to maize (and so on, for the other commodities). The same process was used to allocate papers to countries for those that did not explicitly mention any country. One of the main reasons for stage two of the allocation of papers was to address the several Centres that do not have an explicit food com modity focus, namely Bioversity International (agricultural biodiversity research), CIFOR (forestry research), ICRAF (agroforestry research), IFPRI (policy research) and IWMI (water research). The Scopus data set included papers published by Bioversity, IFPRI and IWMI on each of the 26 CGIAR mandate products (thus giving a complete commodity dis tribution that could be used to allocate papers to these Centres that did not explicitly mention any of the mandate commodities). Some 60% of the Scopus papers published by IFPRI, for example, mentioned at least one commodity or at least one country or both. As well as some Centres focussing on cross-cutting issues such as policy and water resources, all Centres carry out some social science and policy research. We acknowledge that our focus on commodities and countries may intro duce some biases in allocating papers. However, as most papers from the non-commodity Centres can still be allocated to commodities and countries using our methods, this suggests that these biases may not in fact be large. Stage 3 of the process involved allocating all papers, single- and multi-centre ones, whether they mentioned any countries and com modities or not, using the allocations derived from stage 2. Results for all 22,565 papers found in Scopus are shown by Centre and by commodity in Table SM2 in the Supplementary Material), by Centre and by country (Table SM3), and by commodity and by country (Table SM4). Country codes used are shown in Table SM1. Paper citations were calculated by Centre, country and product in the same way. The probability distribution of citations was found to be very long-tailed (Table SM5), with ten of the top 12 cited papers being in the field of nutrition. To address the imbalance that would result in using the actual number of citations, we allocated them using a “cite factor” based on the deciles of the citation distribution (Table SM6). At the time of data extraction, some 17% of all CGIAR papers had not been cited at all, and 23% had only one citation. At the end of the third stage, and as a final check, we carried out a series of random selections of 30 papers each and the allocation to Centre, commodity and country was checked manually. Where we found errors, we repeated the entire process (as shown in Figure SM1) until the random selection of articles achieved 100% accuracy in allocation, which occurred in the fifth iteration. Given the impracticality of manually checking nearly 22,600 papers, some allocation errors may still exist, but these are likely to be few. 3.2. Indicators of alignment with CGIAR strategic objectives We used the strategy and results framework of the CGIAR as formulated in 2011 (CGIAR, 2011) as the basis for selecting indicators with which to compare the publication record from 2010 to 2020. This framework was revisited in CGIAR (2015) and has recently been rede veloped considerably with respect to the One CGIAR reorganisation (CGIAR, 2021). For the period of the analysis, the elements of the strategy and results framework remained similar, with four goals or system-level outcomes being sought: reducing rural poverty, improving food security, improving health and nutrition, and fostering sustainable management of natural resources (CGIAR, 2011). Table 1 shows these outcomes along with their rationale, as given in CGIAR (2011). Also shown are examples from the possible impact pathways from science outputs to outcomes as set out by Tomich et al. (2019). For simplicity, we identified just one indicator for each of these outcomes to use in subsequent analysis, for which global coverage at the national level was complete or nearly complete. We acknowledge that these are proxy in dicators only, and they do not comprehensively cover all the elements of the outcomes. These were the following (Table 1). Number of rural poor people. We used data of the Oxford Poverty and Human Development Initiative (Alkire and Kanagaratnam, 2018), which gives rural and urban population and poverty headcount break downs for more than 100 countries. These data, for the most recent year reported in the dataset (which ranged from 2006 to 2016), were sup plemented with information from World Bank and OECD country pro files, Crespo Cuaresma et al. (2020), and The Borgen Project (https://bo rgenproject.org/). There were seven countries for which we could find no disaggregated rural poverty data for any year; for these, we used the national poverty rate and the percentage of rural population (i.e., we assumed that rural and urban poverty rates were the same). Prevalence of child stunting. We used data from the WHO’s Global Health Observatory data repository (https://apps.who.int/gho/data/ node.main.CHILDSTUNTED?lang = en) on prevalence of stunted chil dren aged under 5 years (percentage height-for-age more than two standard deviations below the mean) as a proxy for food security status at the national level. The data were for the year 2019. There are a few countries for which this indicator does not exist. Different food groups in national diets. We used national food baskets for 2019 from FAO (www.fao.org/faostat) and assembled data P. Thornton et al. https://borgenproject.org/ https://borgenproject.org/ https://apps.who.int/gho/data/node.main.CHILDSTUNTED?lang https://apps.who.int/gho/data/node.main.CHILDSTUNTED?lang Food Policy 107 (2022) 102196 5 on calorie and protein consumption per person per day for 123 coun tries. So that we could compare these food baskets with CGIAR com modities, we amalgamated data into four food groups: cereals (maize, wheat, rice, sorghum, millet, barley), root, tuber and banana crops (bananas and plantains, cassava, potato, sweet potato, yams), pulses (groundnut, chickpea, pigeon pea, cowpea, dry bean, lentil, faba bean, soybean) and animal products (cattle, sheep and goats, chicken, fish, pigs, poultry, feeds). Vulnerability to climate change. We used the ND-GAIN country index for 2018 (Chen et al., 2015), which summarizes a country’s vulnerability to climate change and other global challenges in combi nation with its readiness to improve its resilience. The index combines a country’s exposure, sensitivity and capacity to adapt to the negative effects of climate change, with a measure of a country’s ability to leverage investments and convert them to adaptation actions. The ND- GAIN country index has been calculated and published each year from 1995 onwards. The index has been criticised as containing somewhat arbitrary elements, although it performs comparably with other global, national-level indicators related to climate change and development (Miola and Simonet, 2014). It is widely used, with caveats, in the literature (Chen et al., 2018) and we deemed it appropriate for this analysis. Annual expenditures on national agricultural research. We included a fifth indicator, annual expenditure on agricultural research by national agricultural research systems (NARS). The close relation ships that CGIAR has developed with many NARS over time are critically important to its effectiveness. The CGIAR’s strategy and results frame work highlights the complementary nature of CGIAR and NARS in vestments all along the impact pathway, from strategic and applied research, through piloting and innovation platforms, to scaling up and out (CGIAR, 2011). This indicator can be seen as cross-cutting as it contributes to the four outcomes in Table 1. We used the annual national expenditure on agricultural research per 100,000 farmers (million constant 2011 PPP dollars) as a proxy, for the most recent year available (in the range 2012–2017) in the ASTI database (https://www.asti.cgiar. org/). These data cover 91 countries. 4. Results Table 2 shows the total number of papers by Centre, allocated as outlined in section 3 above. The total number shown exceeds the actual number of papers because of allocation to multiple Centres. The number of multi-Centre papers associated with each Centre is also shown in Table 2. Between 10% and 28% of a Centre’s scientific output, defined in terms of peer-reviewed papers in Scopus, was in collaboration with at least one other Centre. The number of papers by commodity is shown in Table 3 along with citation numbers. Between them, rice, wheat and maize account for nearly one third of all papers and citations. Detailed breakdown of number of papers by Centre, commodity and country can be found in the Supplementary Material (Tables SM2-4). Fig. 2 shows the relationship between the average number of Scopus papers published by each Centre per year over the period 2010 through the first third of 2020 and centre expenditure as reported for 2017. Centre expenditure is a reasonably good predictor of number of Scopus papers, by and large. On average, during this period the 15 Centres published 2422 papers each year. If Scopus papers were the only sci entific output of the centres (which of course is not the case), the average cost per paper published would be nearly USD 384,000, although the range among the Centres is wide: from USD 210,000 to USD 580,000 per paper published. Table 1 CGIAR system level outcomes and their rationale for the period 2010–2020 (from CGIAR, 2011), examples of possible impact pathways from science outputs to outcomes (from Tomich et al., 2019), and the proxy indicators of each outcome used in this analysis. Outcome Rationale Possible impact pathways Indicator of alignment1 Reducing rural poverty Agricultural growth through improved productivity, markets and incomes can be an effective contributor to reducing poverty • Breeding and practice innovations that increase productivity • Innovations that help address market imperfections & failures Number of rural poor people by country Improving food security Access to affordable food is a challenge for millions of poor people and it requires increasing supplies of key staples and containing potential price increases and price volatility • Innovations to minimize production risks • Food safety nets Percentage of under- fives stunted Improving nutrition and health Poor populations suffer particularly from diets which are insufficient in micronutrients affecting health and development, particularly in women and children • Innovations that increase farm enterprise diversification • Breeding innovations that increase the nutrient value of staples Make-up of national diets Sustainable management of natural resources Improved management of natural resources is needed to ensure both sustainable food production and provision of ecosystem services to the poor, particularly in light of climate change • Natural resource governance, property rights, livelihoods • National food and agricultural policy Vulnerability to climate change 1. A fifth indicator was included in the analysis, annual national expenditure on agricultural research per 100,000 farmers (https://www.asti.cgiar.org), in view of the importance of strong NARS-CGIAR partnerships for impact at scale, one of the key elements of the CGIAR’s Strategy and Results Framework throughout the impact pathway from strategic and applied research, through piloting and innovation platforms, to scaling up and out (CGIAR, 2011). Table 2 Scopus papers published by Centres, January 2010 through April 2020: total and multi-centre, and total number of Centre citations. Centre Number % of the total Number with other Centres % of Centre total Total Centre citations AfricaRice 465 1.9 130 28.0 4,175 Bioversity 1074 4.3 275 25.6 11,593 CIAT 1710 6.8 460 26.9 25,856 CIFOR 1745 7.0 220 12.6 34,811 CIMMYT 2770 11.1 529 19.1 48,875 CIP 995 4.0 152 15.3 11,829 ICARDA 1311 5.2 278 21.2 11,943 ICRAF 1804 7.2 358 19.8 28,900 ICRISAT 2379 9.5 393 16.5 32,340 IFPRI 2347 9.4 224 9.5 34,295 IITA 1931 7.7 435 22.5 15,957 ILRI 2257 9.0 475 21.0 32,892 IRRI 2197 8.8 300 13.6 40,682 IWMI 1460 5.8 196 13.4 18,933 WorldFish 579 2.3 69 11.9 25,914 Total 25,024 100 378,928 Note: numbers shown are counts for Centre authorship, including multiple- Centre papers. P. Thornton et al. https://www.asti.cgiar.org/ https://www.asti.cgiar.org/ https://www.asti.cgiar.org Food Policy 107 (2022) 102196 6 4.1. CGIAR publications by country and the number of rural poor We examined the relationship between the number of rural poor per country and the number of CGIAR publications (2010–2020) for the 132 countries for which we were able to estimate rural population size. The thirteen countries excluded in the analysis accounted for only 30 pub lications. There is a strong positive relationship between publication number and number of rural poor by country, which is statistically highly significant (p < 0.001) (see Figure SM2 in the Supplementary Material). There is substantial regional spread, however, which is shown in Figure SM3 for ten regional aggregations and by country for each region in Figure SM4 in the Supplementary Material. The strong correlation between number of publications and the number of rural poor per country suggests that CGIAR research effort has been focused in appropriate countries, by and large, with respect to the goal of fostering agricultural growth through improved productivity, markets and incomes to contribute to reducing poverty along appro priate impact pathways (Table 1). There appear to be some country outliers – countries with fewer or more research publications than in other countries with similar numbers of rural poor – but there may be several reasons for this, such as a country’s instability or the legacy of previous research capacity and infrastructure, for example. Neverthe less, the analysis here is one step in identifying possible candidate countries for ramping up AR4D activities to help prevent their falling behind in efforts to achieve SDG 1 (no poverty). We acknowledge that future AR4D portfolio planning could benefit from more sophisticated analysis to investigate the links between research effort and poverty reduction in specific contexts. 4.2. Publications and the intersection of rural poverty, malnutrition and under-funded NARS Fig. 3A and 3B show publication number by country against child stunting prevalence (as a proxy of food and nutrition insecurity) and the number of rural poor (panel 3A, 132 countries) and national research expenditure USD per 100,000 farmers (panel 3B). The relationship be tween the number of publications and child stunting prevalence is sta tistically significant but considerably weaker than that between number of publications and the number of rural poor per country (these are reported in the caption of Fig. 3). CGIAR publications tend to be concentrated in countries with both large numbers of rural poor and high stunting prevalence (Fig. 3A, 85 countries), although there are several countries in the top right-hand part of the figure that might be said to be under-represented in the publication record. By the same token, some countries in the lower right-hand part of the figure are perhaps over-represented. The absence of countries in the top left-hand of Fig. 3A (relatively few rural poor and high rates of stunting) is noteworthy. Access to affordable food remains a challenge for many millions of poor people, and the impact pathways to which AR4D can contribute, such as minimizing production risks and provision of food safety nets (Table 1), may need many other elements too, if beneficial food access outcomes and impacts are to be achieved (Blesh et al., 2019). With respect to the relationship between stunting prevalence and national research system expenditure, Fig. 3B suggests a strong negative correlation between the two, with increased NARS research expenditure being associated with reduced levels of malnutrition. From our analysis here, not much can be said about causality, but the result is highly suggestive, nonetheless, in view of the large literature on the relation ship between agricultural R&D and development metrics such as poverty reduction (De Janvry and Sadoulet, 2010) and nutrition (Gil lespie and van den Bold, 2017), for example. The relationship between number of CGIAR Scopus papers and levels of malnutrition is not as strong, with the great majority of Scopus papers focussing on countries that fall within the midranges for national research expenditure and stunting prevalence. With respect to the number of papers, Fig. 3B highlights some outliers: Nigeria for one, which also bucks the trend in having a high level of research expenditure but still substantial levels of malnutrition. Fig. 3B reflects some uncomfortable questions, though: for example, despite decades of high ODA (Official Development Assis tance) investment in countries such as Ethiopia, Tanzania and Malawi, malnutrition remains high and national research expenditure low. Other countries in the same situation, such as Mozambique, Madagascar, Burundi and Chad, reflect what may be considerable under-investment in agricultural research. There are many reasons for such under- Table 3 Scopus papers published by Centres, January 2010 through April 2020, by commodity: number and number of citations. Commodity Code Number of papers % of total papers Number of citations % of total citations Banana & plantain BP 776.4 3.4 9237.0 2.4 Barley BA 361.1 1.6 4651.5 1.2 Cassava CS 733.2 3.2 9342.3 2.5 Cattle CT 1192.4 5.3 18,351.8 4.8 Chicken CK 335.9 1.5 8535.4 2.3 Chickpea CH 509.9 2.3 8784.2 2.3 Cowpea CP 300.4 1.3 3204.2 0.8 Dry bean BN 718.0 3.2 12,424.6 3.3 Faba bean FB 147.1 0.7 1773.4 0.5 Feed FE 896.5 4.0 13,153.9 3.5 Fish FI 756.1 3.4 26,305.8 6.9 Groundnut GN 449.2 2.0 6930.4 1.8 Lentil LT 198.2 0.9 2274.9 0.6 Maize MZ 2161.5 9.6 37,673.0 9.9 Millet ML 526.6 2.3 6837.7 1.8 Pigeon pea PP 262.4 1.2 3865.1 1.0 Pigs PI 451.1 2.0 7340.5 1.9 Potato PO 724.1 3.2 10,085.4 2.7 Rice RI 2972.7 13.2 50,808.6 13.4 Sheep & goats SH 656.5 2.9 8339.2 2.2 Sorghum SG 607.5 2.7 8343.8 2.2 Soybean SB 317.5 1.4 5164.1 1.4 Sweet potato SP 423.6 1.9 5785.5 1.5 Trees TR 3790.3 16.8 70,086.1 18.5 Wheat WH 2034.2 9.0 36,685.0 9.7 Yam YA 265.6 1.2 3007.1 0.8 Total 22,579.8 100.0 378,928.2 100.0 Note: numbers shown are split across multiple commodities where appropriate (see text for details) so are not necessarily integers. Fig. 2. Average number of Scopus publications per Centre per year (2010–2020) as a function of 2017 Centre expenditure. Linear regression shown in yellow. P. Thornton et al. Food Policy 107 (2022) 102196 7 Fig. 3. Number of CGIAR Scopus publications by country (2010–2020) in relation to child stunting prevalence and the numbers of rural poor (Panel A, top) and in relation to child stunting prevalence and national research expenditure per 100,000 farmers (Panel B, bottom). Country codes are show in the Supplementary Material, Table SM1. Note: Linear regression of A (number of Scopus publications) on S (stunting prevalence, %): Ln (A) = 0.017 * S + 1.23, r2 = 0.073, p < 0.01 Linear regression of A on R (number of rural poor): Ln (A) = 0.603 * Ln(R) – 2.26, r2 = 0.480, p < 0.01. P. Thornton et al. Food Policy 107 (2022) 102196 8 investment, including the effects of civil unrest, war and social dislo cation, for example. Whatever the reasons for this under-investment and under-study, these are noteworthy gaps in the country portfolio. As for SDG 1 (no poverty) in section 4.1 above, the analysis here can constitute a preliminary step in identifying possible candidate countries for ramping up or rebalancing AR4D activities with respect to their efforts to achieve SDG 2 (no hunger). 4.3. Publications and the ND-GAIN country index With respect to the number of Scopus paper and the ND-GAIN Country Index for 2018 for 136 countries, three things stand out (Figure SM5 in the Supplementary Material). First, there is no statisti cally significant correlation between the two variables at the 5% level. This is perhaps to be expected, given that the ND-GAIN Country Index covers a broad range of readiness variables that describe a country’s business, governance and social environments. Second, there is a group of countries with a large output of scientific publications coupled with a low readiness index for addressing the future challenges posed by climate change. For some countries such as Ethiopia and India, the country index has remained the same or increased slightly since 1995. For others such as Kenya, Madagascar, Malawi, Mozambique, Tanzania and Uganda, their ND-GAIN country index has actually declined since 1995. This suggests that there may be challenges in translating research outputs to effective action on the ground. Understanding what these challenges have been, and perhaps developing modified, nationally- appropriate theories of change to address them in future, would seem to be a priority action. Third, there is a group of countries with low readiness and limited research outputs with which to inform national and local investment priorities. The country index for many of these countries has also declined since 1995 – Chad, Eritrea, Somalia, Lesotho, Eswatini and DR Congo, for example – but it appears that the local science base on which to draw is quite limited, and this may hamper efforts to effectively pri oritize investments and action to address agricultural development challenges. 4.4. Publications and calories & protein in national diets A comparison of the proportion of the five food groups (cereals; pulses; roots, tubers and bananas; animal products; and others) in na tional food baskets with the proportion of Scopus papers covering ostensibly the same food groups is shown in Fig. 4, in terms of both calories (Panel A) and protein (Panel B). All dietary components that are not CGIAR mandate crops such as fruit and vegetables are shown as “other”, and the Scopus paper proportions were scaled accordingly. Taking Central Africa as an example, there are four countries that have national food basket data in FAOSTAT. The average proportions, weighted by population, of the different food groups in these food bas kets are shown in the top of Fig. 4A; 30% of calories come from the “other” category. On average, 12% of the Scopus papers that were published by CGIAR centres on six countries of the Central Africa region addressed cereals, while FAOSTAT indicates that on average 36% of calories are from cereals. For the same region on a protein basis, the patterns are relatively similar: 35% of protein comes from cereals, and 13% of CGIAR papers addressed cereals, when scaled to protein intake. It is worth noting that for 2013, the most recent year in the dataset of Ritchie and Roser (2017), we estimated that CGIAR commodities made up 81% of the calories in the average “global diet”; though if alcohol and sugar and other sweeteners are omitted (which make up 13% of the global diet), then CGIAR commodities account for 93% of calories. We recognise that even if CGIAR focused 100% on the foods that people consume, this would not necessarily imply full alignment with the goal of improving nutrition and health (Table 1). Nevertheless, country-level comparisons on a protein basis may be useful for several purposes, depending on the outcomes sought. These include identifying potential protein “coldspots” (areas with a likely deficit in future) in countries where national diets are heavily dependent on protein sources that are projected to suffer yield and nutrient density decreases as a result of climate change. An example is Uganda: currently, beans and maize make up 20% and 18%, respectively, of national protein supply; both these crops may see yield reductions of 20% or more to mid-century because of climate change (Ramirez-Villegas and Thornton, 2015). Commodities that can replace or supplement protein supply (and farm income) will almost certainly be needed as a result. Country-level comparisons may also be useful from the perspective of AR4D that seeks to promote healthy and nutritionally-balanced diets, for instance through increasing dietary diversity in both rural areas and in rapidly- urbanising societies. 5. Discussion & conclusions Commitment to the SDGs represents a turning point with profound implications. Achieving the goals not only implies reforming existing production and consumption to create step changes in economic per formance, but it also demands step changes in social and sustainability performance as central pillars of the restructuring process. Enabling the necessary shifts along these different axes requires wide-ranging tech nical, institutional and policy change for the emergence of fundamen tally different consumption and production systems (Steiner et al., 2020; Herrero et al., 2020b; Barrett et al., 2020) and research priorities. This evolution and broadening of the framing demanded by the sustainable development agenda suggest a form of innovation that embeds social and environmental concerns, involves new patterns of governance and coalitions of interest, draws from a range of existing and diverse analytical and policy frameworks, and demands proactive public sector leadership and investment (Hall and Dijkman, 2019). The magnitude and rapidity of the performance changes needed to supply growing amounts of safe and healthy food and to support rural livelihoods, whilst at the same time reducing resource use and environmental footprint, are daunting, but not impossible. Our analysis indicates that at the regional level, there is a reasonable fit between the numbers of rural poor people and CGIAR research effort, as measured by number of Scopus publications. Within each region, countries are widely spread, and regional differences in the relationship are substantial. We found that CGIAR publications tended to be concentrated in countries with large numbers of rural poor and high stunting prevalence (Fig. 3A), albeit with some outliers of both under- and over-representation. We demonstrated strong negative correlation between stunting prevalence and national research system expenditure (Fig. 3B), indicating that increased research expenditure is associated with reduced levels of malnutrition, on average. In contrast, we found no statistically significant correlation between the number of Scopus papers and the ND-GAIN Country Index (a combination of a country’s vulner ability to global challenges and its readiness to improve its resilience). We compared the proportion of publications on cereals, RTB crops, pulses and animal products in Centres’ national publication records with the proportion of these food groups in national food baskets, at the regional level (Fig. 4). Across regions, some 30% of all calories and about 15% of all protein in national food baskets come from sources that are not part of CGIAR’s current mandate, such as fruit and vegetables. From the results, we identified three specific issues that warrant attention. These issues may be relevant to the reorganisation and reor ientation of CGIAR into One-CGIAR that is currently underway and the associated opportunity to rebalance the country and commodity port folio. First, there are several countries with both relatively high rates of expenditure on agricultural research and relatively large publication records, coupled with a low (and in some cases declining) readiness index for addressing the future challenges posed by climate change and other global issues. As indicated above, part of the reason for this may be the historical and continuing difficulties in translating the published outputs of research into effective action on the ground at scale. The P. Thornton et al. Food Policy 107 (2022) 102196 9 Fig. 4. Proportion of CGIAR Scopus publica tions by commodity group (2010–2020) and the proportion of those food groups in FAO national food baskets, compared on a regional basis. CAF Central Africa; CAM Central Amer ica; EAF East Africa; EAP East Asia & the Pa cific; SA South Asia; SAF Southern Africa; SAM South America; SEA South-east Asia; WAF West Africa; WANA West Asia & North Africa. Panel A: on the basis of calories in the national food basket, Panel B: on the basis of protein in the national food basket. P. Thornton et al. Food Policy 107 (2022) 102196 10 reasons for bottlenecks between research and uptake may be many, but revisiting and possibly modifying the theories of change that address local conditions and contexts in these countries so that the bottlenecks can be addressed in future would appear to be a priority action. To that end, Hall and Dijkman (2019) suggested that the CGIAR consider four new narratives to frame critical areas of its activities and role: i) a new scaling and impact narrative that adopts an agri-food system innovation perspective; ii) a new partnership and value network narrative that emphasises commitment to advancing the sustainable development agenda; iii) a new social licence narrative that proactively addresses issues of social acceptability and the need to create a platform to host these discussions; and iv) a new science narrative that accommodates transdisciplinarity and the role of social and systems sciences in the innovation process. Second, there are several countries with a low readiness index; in several cases, as noted in section 4.3 above, the readiness index has been declining over time. Some of these latter countries, such as Mozambique and Madagascar, also have a relatively small publication record and relatively low levels of research expenditure (Fig. 3B). It may be difficult for such countries to effectively prioritize investments and action to address agricultural development challenges, and thus could be candi dates for more targeted AR4D investment in future. Under-investment in agricultural research and worsening readiness to address the challenges suggest serious problems that both international and national agricul tural research for development organisations could be in a strong posi tion to help rectify. The new One CGIAR investment strategy (CGIAR, 2021) could actively contribute to redirecting funding flows towards the new partnerships, innovation systems and research modalities that will be needed if food systems are to be transformed (Baranski and Ollen burger, 2020; Klerkx and Begemann, 2020; Stringer et al., 2020). As noted above, public-domain information such as ASTI does exist but it is currently insufficiently granular. We return to this below. Third, at an aggregated level, there are substantial proportions of countries’ diets, in terms of calories and protein, that CGIAR research has not addressed in the past. The absence of fruit and vegetable research, for example, is particularly noteworthy, reflecting a long- standing imbalance in agricultural research investment in general (Keatinge et al., 2011; Haddad et al., 2016). With an increasing focus on food systems in R&D, there is a compelling case to be made for much greater attention to both the production and the consumption of more balanced, nutritious diets (Willett et al., 2019). Whether research or ganisations such as One CGIAR should in future expand or modify their mandate in response, or whether they should seek new alliances and research partnerships with other organisations to address such issues, are questions beyond the scope of this viewpoint paper. In any case, the critical importance of partnership and engagement is clear: in the absence of silver bullets and one-size-fits-all solutions, setting actions and interventions in local contexts can only be done effectively and efficiently via concerted effort involving broad coalitions of develop ment actors. These include the international and national agricultural research and extension systems, the private sector, and the NGO and disaster risk management communities. We undertook the work here with the aim of estimating where and on which commodities agricultural research resources have concen trated in the recent past, using CGIAR as a case study. We acknowledge weaknesses of the methods used. First, peer-reviewed publications in the literature are but one output of science. Although there is a strong correlation between numbers of Scopus papers by CGIAR Centre and Centre expenditure, a considerable amount of agricultural research and development activity may not be fully captured with such a metric. Examples would include capacity development and engagement efforts (Vermeulen and Campbell, 2015), indirect spill-over effects from in formation and technology development and implementation in one place to other places, and regional and global policy research, for instance, as well as the publication record of other research organ siations. Second, there is a considerable legacy in the fifty years since CGIAR was founded in hard infrastructure and country focus. This is not surprising, in view of the breeding and agronomic focus of CGIAR’s original vision. This legacy persists in the country profile of the recent publications record: just six countries account for nearly 40% of all CGIAR Scopus papers since 2010 (Bangladesh, China, Ethiopia, India, Indonesia and Kenya). Despite these weaknesses, we argue that such analysis is still useful. Segmented national and organizational research investment data allow the identification of possible gaps in commodity coverage and help to target potential mismatches between future challenges and current na tional and international research foci. The “hotspots” of poverty, dietary inadequacy and climate-related insecurity may change with increasing speed. Investment data could be used to rapidly identify protein and micronutrient hotspots and coldspots at higher spatial resolution than presented here, based on local diets under different climate change scenarios, linked to current knowledge about likely future changes in staple crop suitability (Manners and Van Etten, 2018), for example. We know where some of these hotspots are likely to be located. The chal lenge is then to ensure that international and national agricultural research for development efforts are covering the bases adequately, so that livelihoods and diets can be sustained and improved, even as global population swells to (and possibly peaks at) just under 10 billion people by mid-century (Vollset et al., 2020). Although we have not done this in the analysis reported here, similar analysis could highlight situations in which food systems might need to provide improved nutritional di versity to local populations in ways that meet environmental sustain ability objectives as well as adaptation and mitigation targets. Such analyses would be greatly facilitated, and could be made considerably more nuanced than those discussed here, if there were publicly-available, disaggregated and regular data on financial alloca tions of public agricultural R&D and development indicators. This would seem to be a prerequisite for justifying fund allocation and its alignment to current and future sustainable dietary requirements in regions and countries. This could also facilitate an analysis of research publications from other national and international organisations, which would allow a more comprehensive view of the alignment between AR4D visions and reality in specific countries. Given the likely pace of future technical innovation in food systems, if we are to have any chance of meeting the SDGs on time (Herrero et al., 2020a; 2020b; Barrett et al., 2020), actionable data with which to evaluate innovation in agri-food systems would seem to be crucial for making agricultural research in vestment portfolios better fitted to an extremely dynamic environment. An international initiative targeting the development of a balanced global public, private and tertiary sector agricultural research invest ment agenda that is both responsive to a range of current needs and that anticipates the future seems a logical next step. The collation of annual expenditure data would greatly facilitate this. One CGIAR already has a platform set up (see https://www.cgiar.org/dashboards) with some aggregated financial information, and this could be greatly expanded, perhaps via links with ASTI, as the mission of One CGIAR continues to evolve in response to the calls for food system transformation (Klerkx and Begemann, 2020). There is also the possibility of focussing “big data” initiatives, including activities under One CGIAR’s Big Data plat form (see https://bigdata.cgiar.org/) on in-depth analyses of how the research portfolio addresses specific agricultural R&D challenges in LMICs, such as the location and characterisation of sustainable diet and food security hotspots. Nimble agricultural research resource allocation will ultimately depend on appropriate metrics that link research expenditures, outputs and development impacts, and that can be measured relatively easily and quickly. This is proving to be a tough ask. In the context of the UK University Research Excellence Framework, for example, the relation ship between research excellence and societal impact overall is fairly weak and heavily dependent on the research area (Woolley and Robinson-Garcia, 2017). Research outputs may contribute to develop ment impacts through a wide range of different pathways involving P. Thornton et al. https://www.cgiar.org/dashboards https://bigdata.cgiar.org/ Food Policy 107 (2022) 102196 11 many different actors (Gaunand et al., 2015; Temple et al., 2018). If bibliometrics alone is unable to make robust, critical connections be tween research outputs and impact (Wilsdon et al., 2015), there may be opportunities to combine bibliometric approaches with social media data that can map communication and interaction patterns (Noyons and Ràfols, 2018). Such work is greatly needed. That our food systems are facing far- reaching and radical change is beyond doubt: one need look no further than burgeoning private-sector investment in alternative protein sources in livestock feed and human diets (one estimate is that alter native meats will make up 60% of the global market by 20403). Food system transformation beckons (Herrero et al., 2020b; Barrett et al., 2020), but we must ensure that progress is equitable. What such trans formation will mean for international and national agricultural R&D and both rural and urban poor in LMICs, is still very much to be determined. There is an enormous opportunity for highly-responsive and well- targeted agricultural research for development to help guide this transformation for the global good, and for One CGIAR to maximise its input into an agricultural science agenda that serves contemporary global development ambitions. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments Without implicating them, we are grateful to Jessica Fanzo, Andy Jarvis, Anne Miki, Vinay Nangia, Bernard Vanlauwe and Anthony Whitbread for assistance in the early stages of this work, and to anon ymous reviewers who made helpful comments on previous versions of the paper. PT, LS and LC acknowledge support provided to the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) from CGIAR Fund Donors, and through bilateral funding agreements (please see ccafs.cgiar.org/donors). Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.foodpol.2021.102196. References Alkire, S., Kanagaratnam, U., 2018. Multidimensional Poverty Index Winter 2017–18: Brief methodological note and results. University of Oxford, OPHI Methodological Notes, Oxford Poverty and Human Development Initiative, p. 45. Anderson CL, Reynolds T, Biscaye P, Callaway K, 2017. Funding for Agricultural Research and Development Public Goods. EPAR Technical Report #339. Evans School Policy Analysis & Research Group (EPAR). Alston JM, Pardey PG, Rao X, 2020. The Payoff to Investing in CGIAR Research. SoAR Foundation, https://supportagresearch.org/assets/pdf/Payoff_to_Investing_in_ CGIAR_Research_final_October_2020.pdf. Baranski, M., Ollenburger, M., 2020. How to Improve the Social Benefits of Agricultural Research. Issues Sci. Technol. XXXVI, Spring 2020, 47–53. Barrett, C.B., Benton, T.G., Cooper, K.A., Fanzo, J., Gandhi, R., Herrero, M., James, S., Kahn, M., Mason-D’Croz, D., Mathys, A., Nelson, R.J., Shen, J., Thornton, P., Bageant, E., Fan, S., Mude, A.G., Sibanda, L.M., Wood, S., 2020. Bundling innovations to transform agri-food systems. Nat. Sustainability 3 (12), 974–976. https://doi.org/10.1038/s41893-020-00661-8. Beintema, N., Stads, G.J., 2017. A comprehensive overview of investments and human resource capacity in African agricultural research. IFPRI, Washington D.C, ASTI Synthesis Report. Beintema, N., Echeverria, R.G., 2020. Evolution of CGIAR funding. ASTI Program Note. International Food Policy Research Institute (IFPRI), Washington, DC. Beintema, N., Nin Pratt, A., Stads, G.J., 2020. Key trends in global agricultural research investment. IFPRI, Washington D.C, ASTI Program Note. Blesh, J., Hoey, L., Jones, A.D., Friedmann, H., Perfecto, I., 2019. Development pathways toward “zero hunger”. World Dev. 118, 1–14. https://doi.org/10.1016/j. worlddev.2019.02.004. CGIAR, 2011. CGIAR strategy and results framework 2016-2025. Online at www.cgiar. org/wp/wp-content/uploads/2011/08/CGIAR-SRF-Feb_20_2011.pdf. CGIAR, 2015. CGIAR strategy and results framework 2016-2025. Online at https:// cgspace.cgiar.org/handle/10947/3746. CGIAR, 2016. Financial Report 2016. Online at https://cgspace.cgiar.org/bitstream/ handle/10947/ 4666/2016-CGIAR-Financial-Report.pdf. CGIAR, 2021. CGIAR 2030. Research and innovation strategy: transforming food, land, and water systems in a climate crisis. Online at https://www.cgiar.org/how-we- work/strategy/. Chen C, Noble I, Hellmann J, Coffee J, Murillo M, Chawla N, 2015. University of Notre Dame Global Adaptation Index Country Index Technical Report. Online, https:// gain.nd.edu. Chen, C., Hellmann, J., Berrang-Ford, L., Noble, I., Regan, P., 2018. A global assessment of adaptation investment from the perspectives of equity and efficiency. Mitig. Adapt. Strat. Glob. Change 23 (1), 101–122. https://doi.org/10.1007/s11027-016- 9731-y. Crespo Cuaresma, J., Danylo, O., Fritz, S., Hofer, M., Kharas, H., Laso Bayas, J.C., 2020. What do we know about poverty in North Korea? Palgrave Commun 6, 40. https:// doi.org/10.1057/s41599-020-0417-4. Daunfeldt, S.O., Hartwig, F., 2014. What determines the use of capital budgeting methods? Evidence from Swedish listed companies. J. Finance Economics 2 (4), 101–112. https://doi.org/10.12691/jfe-2-4-1. De Cleene, S., 2019. We can completely transform our food systems to ensure everyone has a seat at the table - here’s how. World Economy Forum, Available at: https:// www.weforum.org/agenda/2019/11/food-systems-agriculture-sustainable-sdgs/. de Janvry, A., Sadoulet, E., 2010. Agricultural growth and poverty reduction: Additional evidence. The World Bank Research Observer 25 (1), 1–20. FAO, IFAD, UNICEF, WFP, WHO, 2019. The State of Food Security and Nutrition in the World 2019. Safeguarding against economic slowdowns and downturns. Rome, FAO. Available at: http://www.fao.org/3/ca5162en/ca5162en.pdf. FOLU, 2019. Growing Better: Ten Critical Transitions to Transform Food and Land Use. The Global Consultation Report of the Food and Land Use Coalition, Available at: https://www.foodandlandusecoalition.org/wp-content/uploads/2019/09/FOLU- GrowingBetter-GlobalReport.pdf. Fuglie K, Heisey P, King J, Pray C, Day-Rubenstein K, Schimmelpfennig D, Wang S, Karmarkar-Deshmukh R, 2011. Research Investments and Market Structure in the Food Processing, Agriculture Input and Biofuel Industries Worldwide. Economic Research Report 130, Economic Research Service, U.S. Department of Agriculture, Washington, DC. Fuglie K, Heisey P, King J, Pray C, Schimmelpfennig D, 2012. The contribution of private industry to agricultural innovation. Science 338 (23 November), 1031-1032. DOI: 10.1126/science.1226294. Fuglie, K., 2016. The growing role of the private sector in agricultural research and development world-wide. Global Food Security 10, 29–38. https://doi.org/10.1016/ j.gfs.2016.07.005. Gaunand, A., Hocde, A., Lemarié, S., Matt, M., de Turckheim, E., 2015. How does public agricultural research impact society? A characterization of various patterns. Res. Policy 44 (4), 849–861. https://doi.org/10.1016/j.respol.2015.01.009. Gillespie, Stuart, van den Bold, Mara, 2017. Agriculture, food systems, and nutrition: meeting the challenge. Global. Challenges 1 (3), 1600002. https://doi.org/10.1002/ gch2.v1.3. Global Commission on Adaptation (GCA), 2019. Adapt now: a global call for leadership on climate resilience. GCA, https://gca.org/global-commission-on-adaptation/ report. Griliches, Z., 1958. Research costs and social returns: Hybrid corn and related innovations. J. Political Economy 66 (5), 419–431. Haddad, Lawrence, Hawkes, Corinna, Webb, Patrick, Thomas, Sandy, Beddington, John, Waage, Jeff, Flynn, Derek, 2016. A new global research agenda for food. Nature News 540 (7631), 30–32. https://doi.org/10.1038/540030a. Hall, A., Dijkman, J., 2019. Public Agricultural Research and Development in an Era of Transformation Independent Science Council of the CGIAR and Commonwealth Scientific and Industrial Research Organisation. Herrero, M., Thornton, P.K., Mason-D’Croz, C., Palmer, J., Bodirsky, B.L., Pradhan, P., Barrett, C.B., Benton, T.G., Hall, A., Pikaar, I., Bogard, J.R., Bonnett, G.D., Campbell, B.M., Bryan, B.A., Christensen, S., Clark, M., Fanzo, J., Godde, C.M., Jarvis, A., Loboguerrero, A.M., Mathys, A., McIntyre, L., Naylor, R.L., Nelson, R., Obersteiner, M., Parodi, A., Popp, A., Ricketts, K., Smith, P., Valin, H., Vermeulen, S. J., Vervoort, J., van Wijk, M., van Zanten, H.H.E., West, P.C., Wood, S.A., Rockstrom, J., 2020a. Articulating the impact of food systems innovation on the Sustainable Development Goals. Lancet Planetary Health. https://doi.org/10.1016/ S2542-5196(20)30277-1. Herrero, Mario, Thornton, Philip K., Mason-D’Croz, Daniel, Palmer, Jeda, Benton, Tim G., Bodirsky, Benjamin L., Bogard, Jessica R., Hall, Andrew, Lee, Bernice, Nyborg, Karine, Pradhan, Prajal, Bonnett, Graham D., Bryan, Brett A., Campbell, Bruce M., Christensen, Svend, Clark, Michael, Cook, Mathew T., de Boer, Imke J.M., Downs, Chris, Dizyee, Kanar, Folberth, Christian, Godde, Cecile M., Gerber, James S., Grundy, Michael, Havlik, Petr, Jarvis, Andrew, King, Richard, Loboguerrero, Ana Maria, Lopes, Mauricio A., McIntyre, C. Lynne, Naylor, Rosamond, Navarro, Javier, Obersteiner, Michael, Parodi, Alejandro, Peoples, Mark B., Pikaar, Ilje, Popp, Alexander, Rockström, Johan, Robertson, Michael J., Smith, Pete, Stehfest, Elke, Swain, Steve M., Valin, Hugo, van 3 https://www.thefuturescentre.org/signal/at-kearney-expects-alternative- meats-to-make-up-60-market-in-2040/. P. Thornton et al. https://doi.org/10.1016/j.foodpol.2021.102196 https://doi.org/10.1016/j.foodpol.2021.102196 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0010 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0010 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0010 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0020 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0020 https://doi.org/10.1038/s41893-020-00661-8 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0030 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0030 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0030 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0035 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0035 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0040 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0040 https://doi.org/10.1016/j.worlddev.2019.02.004 https://doi.org/10.1016/j.worlddev.2019.02.004 https://doi.org/10.1007/s11027-016-9731-y https://doi.org/10.1007/s11027-016-9731-y https://doi.org/10.1057/s41599-020-0417-4 https://doi.org/10.1057/s41599-020-0417-4 https://doi.org/10.12691/jfe-2-4-1 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0090 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0090 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0090 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0095 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0095 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0105 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0105 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0105 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0105 https://doi.org/10.1016/j.gfs.2016.07.005 https://doi.org/10.1016/j.gfs.2016.07.005 https://doi.org/10.1016/j.respol.2015.01.009 https://doi.org/10.1002/gch2.v1.3 https://doi.org/10.1002/gch2.v1.3 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0140 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0140 https://doi.org/10.1038/540030a https://doi.org/10.1016/S2542-5196(20)30277-1 https://doi.org/10.1016/S2542-5196(20)30277-1 https://www.thefuturescentre.org/signal/at-kearney-expects-alternative-meats-to-make-up-60-market-in-2040/ https://www.thefuturescentre.org/signal/at-kearney-expects-alternative-meats-to-make-up-60-market-in-2040/ Food Policy 107 (2022) 102196 12 Wijk, Mark, van Zanten, Hannah H.E., Vermeulen, Sonja, Vervoort, Joost, West, Paul C., 2020b. Innovation can accelerate the transition towards a sustainable food system. Nature Food 1 (5), 266–272. https://doi.org/10.1038/s43016-020-0074-1. Hurley, T.M., Rao, X., Pardey, P.G., 2016. Re-examining the reported rates of return to food and agricultural research and development: Reply. Am. J. Agric. Econ. 99 (3), 827–836. https://doi.org/10.1093/ajae/aaw079. ISPC, 2018. Estimating Historical CGIAR Research Investments. SPIA Technical Note 5 SPIA, S Elven & L Krishnan. ISPC Secretariat, Rome, Italy. Keatinge, J.D.H., Yang, R.Y., Hughes, J.D.A., Easdown, W.J., Holmer, R., 2011. The importance of vegetables in ensuring both food and nutritional security in attainment of the Millennium Development Goals. Food Security 3 (4), 491–501. https://doi.org/10.1007/s12571-011-0150-3. Klerkx, Laurens, Begemann, Stephanie, 2020. Supporting food systems transformation: The what, why, who, where and how of mission-oriented agricultural innovation systems. Agric. Syst. 184, 102901. https://doi.org/10.1016/j.agsy.2020.102901. Loboguerrero, A.M., Thornton, P.K., Campbell, B., Wollenberg, L., Zebiak, S., Millan, A., Dinesh, D., Huyer, S., Jarvis, A., Wadsworth, J., Mason-D’Croz, D., Herrero, M., 2020. Actions to reconfigure food systems. Global Food Security 26. https://doi.org/ 10.1016/j.gfs.2020.100432. Manners, R., van Etten, J., 2018. Are agricultural researchers working on the right crops to enable food and nutrition security under future climates? Global Environ. Change 53, 182–194. https://doi.org/10.1016/j.gloenvcha.2018.09.010. Maredia, M.K., Raitzer, D.A., 2010. Estimating overall returns to international agricultural research in Africa through benefit-cost analysis: a “best-evidence” approach. Agricultural Economics 41 (1), 81–100. https://doi.org/10.1111/j.1574- 0862.2009.00427.x. Meridian Institute, 2020. Food systems of the future: a synthesis of reports on food systems transformation. Global Alliance for the Future of Food. Available at: https:// futureoffood.org/wp-content/uploads/2020/03/GA-Meridian-Synthesis-Reports- FINAL-2020-03-16.pdf?fbclid=IwAR1QWVm77pNWD 02MDZ6G2Jq3ayNw2LQyH0GuJLe46ZfXuCkirtCm37wJLaQ. Miola, A. and Simonet, C., 2014. Concepts and metrics for climate change risk and development: towards an index for climate resilient development. Luxembourg: Publications Office of the European Union. doi: 10.2788/44142. Noyons, E., Ràfols, I., 2018. Can bibliometrics help in assessing societal contributions of agricultural research? Exploring societal interactions across research areas. Proceedings of the science and technology indicators conference. Pardey, P.G., Andrade, R.S., Hurley, T.M., Rao, X., Liebenberg, F.G., 2016. Returns to food and agricultural R&D investments in Sub-Saharan Africa, 1975–2014. Food Policy 65, 1–8. https://doi.org/10.1016/j.foodpol.2016.09.009. Pedersen, D.B., Grønvad, J.F., Hvidtfeldt, R., 2020. Methods for mapping the impact of social sciences and humanities—a literature review. Research Evaluation 29 (1), 4–21. https://doi.org/10.1093/reseval/rvz033. Ramirez-Villegas J, Thornton PK, 2015. Climate change impacts on African crop production. CCAFS Working Paper no. 119. CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS). Copenhagen, Denmark. Available online at: www.ccafs.cgiar.org. Rao, X., Hurley, T.M., Pardey, P.G., 2017. Recalibrating the Reported Returns to Agricultural R&D: What if We All Heeded Griliches? Am. J. Agricultural Resource Economics 64 (3), 977–1001. Reynolds T, Anderson CL, Biscaye P, Fowle M, 2017. Funding for Agricultural Research and Development Public Goods. EPAR Technical Report #329. Evans School Policy Analysis & Research Group (EPAR). Ritchie, H., Roser, M., 2017. Diet Compositions. Published online at OurWorldInData. org. Retrieved from. [Online Resource], ’https://ourworldindata.org/diet- compositions’. Rünzel, M., Sarfatti, P., Negroustoueva, S., 2021. Evaluating quality of science in CGIAR research programs: Use of bibliometrics. Outlook on Agriculture. https://doi.org/ 10.1177/0030727021102427. Steiner A, Aguilar G, Bomba K, Bonilla JP, Campbell A, Echeverria R, Gandhi R, Hedegaard C, Holdorf D, Ishii N, Quinn K, Ruter B, Sunga I, Sukhdev P, Verghese S, Voegele J, Winters P, Campbell B, Dinesh D, Huyer S, Jarvis A, Loboguerrero Rodriguez AM, Millan A, Thornton P, Wollenberg L, Zebiak S. 2020. Actions to transform food systems under climate change. Wageningen, The Netherlands: CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS). https://cgspace.cgiar.org/bitstream/handle/10568/108489/Actions%20to% 20Transform%20Food%20Systems%20Under%20Climate%20Change.pdf. Stringer, L.C., Fraser, E.D.G., Harris, D., Lyon, C., Pereira, L., Ward, C.F.M., Simelton, E., 2020. Adaptation and development pathways for different types of farmers. Environ. Sci. Policy 104, 174–189. https://doi.org/10.1016/j.envsci.2019.10.007. Temple, L., Barret, D., Blundo Canto, G., Dabat, M.H., Devaux-Spatarakis, A., Faure, G., Hainzelin, E., Mathé, S., Toillier, A., Triomphe, B., 2018. Assessing impacts of agricultural research for development: A systemic model focusing on outcomes. Research Evaluation 27 (2), 157–170. https://doi.org/10.1093/reseval/rvy005. Thornton, P.K., Schuetz, T., Förch, W., Cramer, L., Abreu, D., Vermeulen, S.J., Campbell, B., 2017. Responding to global change: a theory of change approach to making agricultural research for development outcome-based. Agric. Syst. 152, 145–153. https://doi.org/10.1016/j.agsy.2017.01.005. Tomich, T.P., Lidder, P., Coley, M., Gollin, D., Meinzen-Dick, R., Webb, P., Carberry, P., 2019. Food and agricultural innovation pathways for prosperity. Agric. Syst. 172, 1–15. https://doi.org/10.1016/j.agsy.2018.01.002. Vermeulen S, Campbell B, 2015. Ten principles for effective AR4D programs. CCAFS Info Note. CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Copenhagen, Denmark. https://hdl.handle.net/10568/67897. Vollset, Stein Emil, Goren, Emily, Yuan, Chun-Wei, Cao, Jackie, Smith, Amanda E, Hsiao, Thomas, Bisignano, Catherine, Azhar, Gulrez S, Castro, Emma, Chalek, Julian, Dolgert, Andrew J, Frank, Tahvi, Fukutaki, Kai, Hay, Simon I, Lozano, Rafael, Mokdad, Ali H, Nandakumar, Vishnu, Pierce, Maxwell, Pletcher, Martin, Robalik, Toshana, Steuben, Krista M, Wunrow, Han Yong, Zlavog, Bianca S, Murray, Christopher J L, 2020. Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100: a forecasting analysis for the Global Burden of Disease Study. The Lancet 396 (10258), 1285–1306. https://doi.org/10.1016/S0140-6736(20)30677-2. Woolley, R., Robinson-Garcia, N., 2017. The 2014 REF results show only a very weak relationship between excellence in research and achieving societal impact. LSE Impact Blog. https://blogs.lse.ac.uk/impactofsocialsciences/2017/07/19/what-do- the-2014-ref-results-tell-us-about-the-relationship-between-excellent-research-and-s ocietal-impact/. Willett, W., Rockström, J., Loken, B., Springmann, M., Lang, T., Vermeulen, S., Garnett, T., Tilman, D., Wood, A., DeClerck, F., Jonell, M., Clark, M., Gordon, L., Fanzo, J., Hawkes, C., Zurayk, R., Rivera, J.A., De Vries, W., Sibanda, L., Afshin, A., Chaudhary, A., Herrero, M., Agustina, R., Branca, F., Lartey, A., Fan, S., Crona, B., Fox, E., Bignet, V., Troell, M., Lindahl, T., Singh, S., Cornell, S., Reddy, S., Narain, S., Nishtar, S., Murray, C., 2019. Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from sustainable food systems. The Lancet 393 (10170), 447–492. https://doi.org/10.1016/S0140-6736(18)31788-4. Wilsdon, J., Allen, L., Belfiore, E., Campbell, P., Curry, S., Hill, S., Jones, R., Kain, R., Kerridge, S., Wouters, P., Hill, J., 2015. The metric tide: report of the independent review of the role of metrics in research assessment and management. https://doi. org/10.13140/RG.2.1.4929.1363. P. Thornton et al. https://doi.org/10.1038/s43016-020-0074-1 https://doi.org/10.1093/ajae/aaw079 https://doi.org/10.1007/s12571-011-0150-3 https://doi.org/10.1016/j.agsy.2020.102901 https://doi.org/10.1016/j.gfs.2020.100432 https://doi.org/10.1016/j.gfs.2020.100432 https://doi.org/10.1016/j.gloenvcha.2018.09.010 https://doi.org/10.1111/j.1574-0862.2009.00427.x https://doi.org/10.1111/j.1574-0862.2009.00427.x http://refhub.elsevier.com/S0306-9192(21)00175-5/h0210 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0210 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0210 https://doi.org/10.1016/j.foodpol.2016.09.009 https://doi.org/10.1093/reseval/rvz033 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0230 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0230 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0230 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0240 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0240 http://refhub.elsevier.com/S0306-9192(21)00175-5/h0240 https://doi.org/10.1177/0030727021102427 https://doi.org/10.1177/0030727021102427 https://doi.org/10.1016/j.envsci.2019.10.007 https://doi.org/10.1093/reseval/rvy005 https://doi.org/10.1016/j.agsy.2017.01.005 https://doi.org/10.1016/j.agsy.2018.01.002 https://doi.org/10.1016/S0140-6736(20)30677-2 https://blogs.lse.ac.uk/impactofsocialsciences/2017/07/19/what-do-the-2014-ref-results-tell-us-about-the-relationship-between-excellent-research-and-societal-impact/ https://blogs.lse.ac.uk/impactofsocialsciences/2017/07/19/what-do-the-2014-ref-results-tell-us-about-the-relationship-between-excellent-research-and-societal-impact/ https://blogs.lse.ac.uk/impactofsocialsciences/2017/07/19/what-do-the-2014-ref-results-tell-us-about-the-relationship-between-excellent-research-and-societal-impact/ https://doi.org/10.1016/S0140-6736(18)31788-4 Viewpoint: Aligning vision and reality in publicly funded agricultural research for development: A case study of CGIAR 1 Introduction 2 CGIAR: A major public contributor to agricultural research for development 3 Methods 3.1 Literature searching and allocation to Centre, country and commodity 3.2 Indicators of alignment with CGIAR strategic objectives 4 Results 4.1 CGIAR publications by country and the number of rural poor 4.2 Publications and the intersection of rural poverty, malnutrition and under-funded NARS 4.3 Publications and the ND-GAIN country index 4.4 Publications and calories & protein in national diets 5 Discussion & conclusions Declaration of Competing Interest Acknowledgments Appendix A Supplementary data References