Assessing technological change in agri-food systems of the Global South: A review of adoption-impact studies in wheat

Abstract

Agricultural research-and-development (R&D) systems generate and disseminate new technologies, aiming to enhance crop and livestock production sustainably. Adoption-impact studies analyze farmer acceptance of these innovations and their impacts to identify what was taken up, how it works in farmers’ fields, and the implications thereof. In this paper, we focus on adoption-impact studies in wheat agri-food systems in the Global South and review 47 empirical studies published during 2008–2017. The review focuses on the assessment of technological changes rather than the nature of these changes per se. We document the breadth and adequacy of empirical approaches and methodologies, as well as the diversity and coverage of thematic areas for adoption-impact analyses. The coverage and scope do not fully capture the complex and dynamic technological changes taking place in wheat systems. The reviewed studies typically conceive the adoption process as a dichotomous choice representing the replacement of old inferior practices with new superior ones. Most reviewed impact studies address only the short-term changes associated with innovation adoption on yield and profitability, often assuming that improvements in the living standards of farm households follow de facto. Labor productivity and the ecological and social impacts of technological innovations were largely ignored. We propose several directions to enhance both data availability and analytical rigor, and to reorient future research toward the less-explored but socially, economically, and ecologically relevant topics.

Keywords

Technological change impact assessment emperical microeconomics wheat system developing countries systematic review rural development

Introduction

Agricultural research and development (R&D) systems aim to generate and disseminate various technologies (or innovations) to enhance crop and livestock production sustainably. A body of empirical social science research aims to understand the process of farmers’ adoption decisions and measuring the impacts of innovations for sustainable intensification in various agro-ecological and institutional contexts. The associated literature provides the basis for performance evaluation and estimated returns to investments in agricultural R&D (Glover et al., 2016). Furthermore, social science research helps draw lessons from past failures and building on achievements and positive experiences (Douthwaite et al., 2003). Innovations that are promising in researcher-managed plots may not be necessarily successful in farmers’ fields (Orr and Ritchie, 2004). In this regard, farmers can actively contribute to the development of appropriate technologies and the creation of practical knowledge by providing feedback on their perception and preferences for different technology traits and adapting innovations to their needs. By potentially facilitating this, social science research can strengthen the feedback loops between farmers, researchers, and other stakeholders (Jones and Kondylis, 2018).

In this paper, we focus on the case of adoption-impact studies in wheat agri-food systems in the Global South. Wheat is a particularly important staple food crop for the world and a significant source of dietary energy, protein, and essential micronutrients (Shiferaw et al., 2013). In 2018, global wheat production was 734 million t, with the Global South contributing about half (FAOSTAT, 2020). Wheat is widely cultivated by marginal and resource-poor smallholder farmers of the Global South, and a sizeable population depends directly or indirectly on the crop for their livelihoods. Wheat productivity growth has been relatively low in the Global South in the recent past, linked inter alia to limited investments in wheat breeding and agronomic research, climatic vagaries, and biotic and abiotic stresses (George, 2014). Some of the critical challenges that the wheat agri-food systems of the Global South face in the recent past are:

Stem rust (caused by the fungus Puccinia graminis tritici), which has reemerged as a renewed threat to global wheat production after the emergence of the race Ug99 in Uganda (Bhavani et al., 2019)

Wheat blast (caused by the fungus Magnaporthe oryzae pathotype Triticum), which affected more than 3 million hectares of wheat in Latin America in the first decade of this century, and recently became a significant threat to South Asia’s wheat production (Islam et al., 2016; Mottaleb et al., 2019)

Heat and drought stresses, which appear to be increasing in intensity over recent years, with climate change. For each degree Celsius increase in atmospheric temperature, global production of wheat is estimated to decline by about 6% (Asseng et al., 2015)

An increasing disconnect between wheat consumption and production and increasing import dependence. Wheat consumption is increasing in some regions where it has not been traditionally considered as a major food item (e.g., Africa). Low-income food-deficit countries produce only 16% of the global wheat production, and these countries import 134 times more wheat than they export (FAOSTAT, 2020).¹ Due to the high dependency on imports, changes in international wheat supply and prices can have significant adverse impacts on the livelihoods of the poor consumers in these countries (Cudjoe et al., 2010). The sheer importance of wheat as the staple food for the poor also offers prospects of enhancing the nutritional quality through biofortification to address hidden hunger from an inadequate supply of micronutrients in diets (Müller et al., 2014; Velu et al., 2014).

Numerous wheat innovations have been generated by collaborative research of the CGIAR Centers and various National Agricultural Research Systems (NARS). Some innovations have had made substantive livelihood impacts and reduced environmental footprints (Gan et al., 2011; Liu et al., 2016). A description of some of the CGIAR-NARS wheat innovations is provided as the online supplementary text. Despite the rapid technological change, several bottlenecks in innovation diffusion continue to undermine the livelihood impacts of R&D in wheat (Yigezu et al., 2018). Social scientists can help foster a deeper understanding of the reasons behind less than expected adoption and impacts of technologies in the farmers’ fields. Their contribution is potentially significant in designing and disseminating temporally and spatially relevant technologies that could benefit the poor and the vulnerable as quickly and effectively as possible. This calls for relevant and credible evidence on the adoption and impacts of agricultural innovations, and a body of literature has evolved. However, generic questions have been raised about the adequacy of the conceptual and empirical approaches employed for the evaluation and analytical rigor of some of these studies (beyond wheat per se). The thematic and topical coverage also may not be comprehensive enough to provide the desired level of confidence to researchers, development practitioners, policymakers, and donors (Glover et al., 2016; Leeuwis et al., 2018).

The present study identifies and reviews 47 empirical adoption-impact papers in wheat agri-food systems of the Global South over the period 2008–2017. We evaluate their approaches, methods, analysis, and thematic and topical coverage. The study thereby aims to contribute toward development of a better strategy for assessing technological change in agri-food systems in the Global South in general, and wheat systems in particular. The selection criteria used by the study are provided in the next section. In Section 3, we critically evaluate the reviewed papers based on their conceptual and empirical framework. In Section 4, a set of approaches to further enhance the scope of adoption-impact studies in wheat are described. The last section concludes.

Review methodology

The present paper reviews recent adoption and impact studies conducted in wheat systems of the Global South over the period 2008–2017. We focused on the adequacy of methodologies and approaches for empirical analysis of technology adoption and impacts. To ensure quality and comprehensiveness of the literature search, we followed the guidelines of the Peer Review of Electronic Search Strategies (PRESS; McGowan et al., 2016) and Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA; Moher et al., 2009). Studies for inclusion in the present review were identified through keyword searches in the Web of Knowledge, Scopus, AgEcon, Sciencedirect, Agris, and EconBiz. Our selection was restricted to the empirical studies published in peer-reviewed journals during the study period and to manuscripts written in English. First, titles and abstracts were examined to understand their relevance, followed by short-listing, and accessing the full-length papers of the studies. The reference sections of the selected studies were also scanned to reduce the risk of omission of relevant papers. The number of studies selected for inclusion in adoption and impact analyses is shown by various stages of selection in the flow diagram in Figure S1 of the online supplementary materials.

The approach yielded 47 studies conducted in the specified period in wheat systems, of which 23 focused on technology adoption, 19 on outcomes of adoption only, and 5 studies examined both. The complete list of studies is given in Table S1 (adoption studies) and Table S2 (impact studies) of the online supplementary material. Most of the studies were conducted in South Asia (70%), focusing mainly on India. The rest of the studies were distributed across the wheat systems of Ethiopia (n = 7), Syria (n = 3), and Iran (n = 1). Only a few studies used data from more than one country. The volume of identified adoption-impact studies in wheat is modest, compared, for instance, to the 137 adoption studies identified in maize during the same period.

A large share of the studies used cross-sectional farm-household datasets. The sample size ranged from 40 to 2069 in the cross-sectional studies, and one study (Abay et al., 2017) used a panel dataset with 7500 observations. Only three studies used data sources other than the traditional sample surveys such as census (Mottaleb et al., 2016), randomized control trials (Magnan et al., 2015), and remote sensing (Lobell et al., 2013). A variety of statistical and econometric tools were employed for the analyses.

Methodological approaches in the adoption-impact studies

Analyzing adoption

The identified recent adoption studies in wheat agri-food systems in the Global South are listed in Table S1 of the online supplementary material. The majority of these studies assessed the adoption of conservation agriculture or its components (n = 10), followed by the adoption of improved wheat varieties (n = 4) and irrigation (n = 3), each documenting varying levels of adoption of the technology under consideration. Six studies depended on descriptive statistics to determine the level and explain the process of adoption, and five studies used simple logit or probit framework to model adoption as a dichotomous variable. Eight studies used bivariate or multinomial or multivariate probit/logit models where adoption of more than one technology, again measured as a dichotomous variable, was assessed for technology inter-dependence. In only three studies were adoption intensities assessed, out of which one study used the Tobit model, while the other two used the Ordinary Least Squares regression analysis. A meta-analysis on spatial and temporal variation in technology adoption in wheat is a research topic that merits a separate analysis. On the other hand, the focus of our paper is on examination of the strengths and limitations of the empirical approaches employed in the analysis of adoption to contribute toward the development of better research strategies.

Over the last two decades, technological change has been depicted in most studies as a relatively simple farmer choice that can be represented by a dichotomous variable. These studies overlooked the importance of farmers’ learning process that is shaped by their socioeconomic, cultural, institutional, and demographic factors. Technological change is assumed in these studies as a complete substitution of old inferior practices with new superior ones, thereby making them fundamentally incapable of addressing the processes of adaptation, creolization, hybridization, and incorporation (Douthwaite et al., 2003; Glover et al., 2016). This review shows that the empirical approach and analysis of adoption had not changed significantly from what was noted in the seminal review by Feder et al. (1985). Most of the reviewed studies, constrained by small sample sizes and short reference periods, were unable to address the complexities of farmer decision-making. Stevenson et al. (2018; p.26) recommended a “move from many small-scale one-shot surveys to fewer, well-designed, and representative longitudinal surveys” in this connection.

Over the last decade, only a few studies addressed the adoption of seed-based technologies in wheat production systems of the Global South. One explanation may be that scholars find the topic to be “over-researched,” especially during the Green Revolution period, with only limited novelty and publication potential at present. This perception could be wrong; because the recent varietal change in many parts of the Global South does not resemble that in the Green Revolution era. Varietal change no longer entails a transformational shift from landraces to modern varieties, but an incremental shift from one modern variety released earlier to another released more recently (Krishna et al., 2016). Besides, more and more varieties are coming up with attributes other than yield enhancement (e.g., stress tolerance, micronutrient enrichment). Varietal adoption decisions thereby go beyond predicted yield and also need to consider complementary input requirements and consumption utilities (Nazli and Smale, 2016).

Farmer adoption of sustainable intensification practices has been featured more frequently in the recent wheat literature. The main research questions addressed varied widely, from farmers’ dis-adoption of technologies to inter-dependence in the adoption of different technology components. They have also provided valuable insights into carrying out quick but effective monitoring of the diffusion process. Magnan et al. (2015) is a study noteworthy in terms of its analytical rigor. It used a randomized control trial (RCT) to examine the diffusion of a resource-conserving technology, the effects of which were found to be heterogeneous in the target community. An array of tools is currently available to study microlevel technology adoption, ranging from the elicitation of expert opinions to moderate- and high-resolution satellite imageries and DNA fingerprinting, and these tools can complement or even substitute household surveys (Kosmowski et al., 2018; Kubitza et al., 2020). For example, Lobell et al. (2013) measured the trend in wheat sowing dates throughout the Indo-Gangetic Plains using satellite data.

Analyzing Impacts

The 24 identified recent impact studies in wheat agri-food systems in the Global South are listed in Table S2 of the online supplementary material. Eight studies focused on the impacts of conservation agriculture, while six dealt with irrigation and water conservation technologies. Varietal change in wheat was examined rarely.² The key findings of the impact studies are presented in Table S3 of the online supplementary material. With a single exception, these studies showed a positive and significant effect of adoption on several outcome variables. A quick look reveals that a large number of studies (15 out of 24) had attempted only to establish association and not causality between technology adoption and the outcome indicators. Five studies used the Matching method to control for the observed heterogeneity, while two studies used Endogenous Switching Regression (ESR) in addition to Matching. Both observed and unobserved heterogeneity can be controlled in the ESR framework, given that the selection instrument is valid (Di Falco et al., 2011). One study controlled for fixed effects of time-invariant heterogeneity using panel data, while another used within-farm variation to cancel out household-level heterogeneity (Table S2 of the online supplementary material).

Here we again focus on the conceptual and methodological rigor. Some of these studies displayed commendable scholarship to track impact pathways of wheat R&D, while some weaknesses are pervasive in most others. First, the number of studies is highly inadequate to cover the growing number of R&D efforts in wheat across the Global South, with many promising technologies and wheat-producing regions left uncovered. Similar to the adoption studies, there is a strong geographical bias toward South Asia, particularly India. Second, not many studies covered the effects of varietal technologies. Third, the impact studies were focusing mostly on changes in land productivity, while labor productivity and livelihood impacts are not adequately covered. Fourth, although productivity-enhancing innovations can have adverse effects on laborers and non-adopters (depending on the type of economy; Gollin et al., 2018b), none of these studies addressed technology spillovers. Fifth, most studies focused on estimating the average treatment effect while neglecting the distributional implications. Sixth, many studies could not establish causality and rely on strong assumptions about a counterfactual, thus weakened the reliability of the estimated impact parameters. Some of these weaknesses warrant a more detailed discussion.

Addressing impacts beyond land productivity

The initial step in impact assessment exercises is identifying and quantifying the changes in the production process, including input use, production cost, and yield per unit of land, associated with the intervention(s). In many instances, however, increasing labor productivity has more significant impacts on rural poverty and livelihoods than increasing land productivity (Stevenson et al., 2018). However, there is no evidence on the labor productivity impacts of wheat technological innovations. Another limitation of the current literature is that most of the studies assessed only the short-term impacts of the technology on yield and profitability, thereby implicitly assuming improvements in the livelihood status of farm households. In other words, heterogeneities in farmer strategies, constraints, and opportunities to improve their livelihood status by employing the additional profit from crop production were overlooked. There is a considerable debate on the extent to which crop yield gains translate into livelihood enhancement, including nutritional status. The livelihood impacts might become more evident over longer periods, as observed in the case of the adoption of high-yielding varieties during the Green Revolution period (Gollin et al., 2018a). While some of the studies have attempted to address the livelihood effects of technology adoption in the wheat production systems (Shiferaw et al., 2014), they have not elaborated whether the time gap between technology adoption and observation is sufficient to manifest these effects fully. The presence of long and complex impact pathways between agricultural R&D and welfare enhancement necessitates the use of a combination of multiple methodologies, ranging from qualitative case studies to RCTs (Gollin et al., 2018b).

Heterogeneity and relevance of context

Most of the impact studies in wheat systems focused on the mean effect of technology intervention, despite having a large variability in their estimates. However, the effect of agricultural growth on rural livelihoods depends on endowments of production resources, especially land (Thirtle et al., 2003). Large farmers, often being early adopters, benefit the most from technological change (Page et al., 2009). Depending on the institutional and agro-climatic conditions, the “treatment effect” of interventions would be different for different groups of farm households. Variation in the treatment effects across different social groups is noted in several studies carried out in the non-wheat systems. Discrimination on the grounds of community characteristics such as ethnicity and gender has long drawn the interest of social scientists (Banerjee et al., 2005; Jurajda, 2005). In the field of political economy, social divisions undermining economic progress form one of the most relevant research hypotheses (Banerjee et al., 2005). However, there exists only limited empirical evidence on social segregation shaping agrarian change and rural development. Heterogeneous access to technologies and differential impacts have only received limited attention in wheat research so far. Gender is one of the most relevant sources of social segregation, making explicit integration of gender essential in many R&D activities (Badstue et al., 2017). While women make a vital contribution to food production, they face discrimination in terms of access to productive capital and technology and have more restrictions in agricultural production (Theriault et al., 2017; WB, FAO, & IFAD, 2008). Taking lessons from gender studies, we can quantify the unequal effect of technological interventions on economically marginalized groups and populations.

Establishing causality and other methodological challenges

The main challenge in impact analysis is establishing causality, and this hurdle appears in three different but interrelated forms. The first and most crucial challenge is establishing a viable counterfactual to predict the magnitude and direction of impact variables had there been no intervention. Second is attributing the impact of an intervention when multiple changes are taking place in the region. The third challenge relates to the time lag between technological interventions and the realization of impacts. Several study designs and analytical methods have been developed to address these problems in the impact evaluation literature. The design options used are experimental approaches, quasi-experimental approaches, and observational data approaches (Angrist and Pischke, 2008). The observational data approach, which lacks prior planning of the experiment and/or systematic administration of the treatment, includes longitudinal comparisons of participants, cross-sectional comparisons of participants versus nonparticipants, and panel data approaches, which is the combination of longitudinal and cross-sectional comparisons. In terms of analytical approaches, various econometrics methods have been used including the Double-Difference Estimator, Instrumental Variables regression, and ESR and other Treatment-Effect models (for mitigating selection bias on both observables and un-observables), and Matching for mitigating selection bias on observables only (Khandker et al., 2009). Despite these methodological advances in the impact literature, many of the issues related to causality are not adequately addressed in wheat adoption-impact studies.

Improving adoption-impact research in wheat systems

Obtaining quality information quickly and at lower costs

Against the backdrop of emerging biotic and abiotic stresses and technological transformation in the wheat production systems, there is a pronounced research gap in the existing literature on technology adoption and impact in wheat, especially in terms of thematic coverage and depth as well as analytical rigor. The need for timely and high-quality data for adoption and impact analyses is becoming increasingly evident. A substantial lag between data collection and publication of findings can reduce the value of socioeconomic research for monitoring the technology dynamics and impacts. There are attempts to reduce this time lag; Erenstein (2010) explored the use of village survey data as a rapid and less resource-intensive complement to household censuses and surveys. Some studies in wheat have also explored the possibility of using existing secondary datasets. While these alternatives have considerably addressed the issue of timeliness, the quality and availability of data on desired variables remain a challenge. Several recent technological developments offer prospects to improve the precision and quality of the data, including the use of satellite datasets to capture the adoption of certain agronomic practices (Kubitza et al., 2020). However, there are barely any recommendations for how to conduct data triangulation in order to determine the technological change in agriculture and how to cope with possible data conflicts. At present, researchers shall come up with the best practice in a case-by-case approach.

A unique challenge to germplasm adoption-impacts studies is the difficulty of precisely identifying varieties grown on farmers’ fields. Errors and mismatches are common in such research, which long relied either on farmer recall of varieties or on expert opinion (Kosmowski et al., 2018; Yigezu et al., 2019). Precise crop variety identification through DNA fingerprinting has significant positive effects on germplasm adoption-impact studies in farmers’ fields (Dreisigacker et al., 2019; Poets et al., 2020). DNA fingerprinting is also a valuable tool to evaluate the genetic diversity of crops, a decline of which might reduce the plasticity of crops to respond to biotic and abiotic stresses (Manifesto et al., 2001). The cost of sequencing crop DNA has fallen sharply over the last decade (Stevenson et al., 2018), which has facilitated its use for several crops, mainly clonally propagated crops like sweet potato (Kosmowski et al., 2018) and cassava (Floro et al., 2018; Wossen et al., 2019). These studies have either shown extensive mismatching between DNA fingerprinting results and farmer self-reported data, questioning the underlying assumption of varietal adoption studies that farmers can identify crop varieties cultivated on their farms with a fair degree of accuracy. Although the DNA fingerprinting technology is promising, its wider roll-out in adoption studies still faces some challenges, including resource requirements, methodological rigor (reference library, sampling, DNA extraction, transportation and preservation of samples and DNA material), and the need to be compliant with the national biodiversity regulations.³ A recent handbook (Poets et al., 2020) attempts to standardize the procedure, building on lessons learned from successful applications in wheat (e.g., Ethiopia), and other crops conducted with funding from the CGIAR Standing Panel of Impact Assessment (SPIA).

Many emerging threats to wheat production require immediate intervention, necessitating rapid evaluations of their efficacy. Detailed household surveys may not serve this purpose. Hence, a rapid survey module, similar to the SWIFT surveys to capture global poverty (Elbers et al., 2003; World Bank, 2017) and the Rural Household Multi-Indicator Survey (RHoMIS; Hammond et al., 2017), may prove useful. No such instrument for adoption-impact studies exists. The feasibility and usefulness of carrying out such surveys, both through remotely supervised group-sourcing and the conventional data collection methods, need to be verified. Furthermore, the researchers may depend on new information sources such as satellite imagery and the use of Unmanned Aerial Vehicles (UAVs). Satellite data are increasingly used to detect farmer practices such as cropping intensity, tillage, and crop residue burning and estimate crop yield (Lobell et al., 2013; Zheng et al., 2014), but rarely used to estimate yield impacts of technologies (Kubitza et al., 2020).⁴ The UAVs have advantages such as lower cost of operation, higher picture resolution, and high flexibility in image acquisition programming (Zhang and Kovacs, 2012). A detailed study on the pros and cons of different methods of generating data and on developing protocols for cost- and time-efficient and high-quality data collection is highly warranted.

Better analytical framework

Analysis of empirical data on technological interventions in agriculture often relies on descriptive and regression analyses to test the hypothesized association between intervention and outcome. However, the existence of an association does not imply causation. Unless technology dissemination takes place in a completely randomized experiment, farmers themselves decide whether to adopt the technology or not, making adoption a non-random process. Directly comparing the outcomes between adopting and non-adopting farm-households may be misleading as these groups may differ systematically due to the self-selection of subjects, leading to selection bias. Even a multivariate regression model that contains numerous control variables, alongside the adoption variable, may not rectify this bias as there could still be unobserved heterogeneity.

RCTs are one of the widely accepted methods for providing credible estimates of impacts, where the counterfactual is artificially constructed by a random selection of technology recipients. This approach can ensure that both control and treatment groups are statistically similar in observable as well as unobservable characteristics, thus avoiding program placement and self-selection biases (Janvry et al., 2017). However, not many have so far used an experimental design to elicit technology impacts in cereal production systems. Among the studies listed in Table S2, none had used RCTs to identify the causal effect of innovations. In demand-driven systems in which people make their own choices on whether to participate or not, the experimental approach is less feasible. The random assignment often runs contrary to the essence of community-led development programs (Davis et al., 2012). There are also practical problems that arise from the experimental design that bias the estimates. The implementation of the experiment itself alters the framework within which the program operates, resulting in a “randomization bias” (Potash, 2018; Sianesi, 2017). Additionally, experiments are generally costly to perform and need close supervision to maintain successful administration. The potential for denying treatment can pose ethical questions that are politically and culturally sensitive. To sum up, the selection between experimental and non-experimental evidence depends on several variables, including the context. It is therefore also useful to recall that “relying on observational data analysis from within context produces treatment effect estimates with lower mean square error than relying on experimental estimates from another context” (Pritchett and Sandefur, 2015: 474).

Among the non-experimental approaches, the most effective statistical procedure to deal with endogeneity bias in the literature is the application of instrumental variables. Estimation using instrumental variables allows for identification of the impact of exogenous changes on technology adoption and eliminates the effect of reverse causation or simultaneity. However, the success in using instrumental variables squarely depends on the suitability of the instrumental variables. Suitable instrumental variables are not often readily available (Kubitza and Krishna, 2020). Because of this reason, many researchers argue that causality is a near-impossible task to establish. In the few studies that have used the ESR framework, identification could still be a problem, and caution should be exercised while interpreting the results. Among the studies listed in Table S2, only a few used the Matching methods, which, to a certain extent, control for (a share of) the observed heterogeneity. Most of the studies use cross-section datasets, although the use of panel data would have controlled for part of the unobserved heterogeneity that is time-invariant. There is an increased need to invest researcher effort in identifying unique and/or widely applicable instrumental variables from the literature.

Shifting the research focus

Apart from enhancing the precision in data collection and ensuring the robustness of estimation, more research focus is warranted on socially relevant aspects of technological changes, such as social inclusion in the diffusion process, institutions facilitating access to complementary production resources and information, and different social and economic inequalities. The notion of inclusive development is becoming increasingly popular in both academic and policy literature, and with Agenda 21 and the Sustainable Development Goals of the United Nations (SDGs), it has emerged as of global importance (United Nations, 2017). In principle, a development process can be considered as inclusive only if it enhanced welfare of a society by advancing equality of opportunities for every member, with particular focus on the poorest, vulnerable, and most marginalized and disadvantaged members and sectors that are typically excluded from the conventional perspective of economic growth (Kozuka, 2014). Existing resource inequalities determine the diffusion pattern generally favoring large and wealthy farmers (Innovativeness-Needs Paradox, proposed by Rogers (2010)), whereas the diffusion process can further exacerbate these inequalities if not accompanied with effective targeting and priority setting.

Social inclusion studies are historically oriented toward the gender component while examining the technological changes in agriculture. Empowerment of rural women is considered as a necessary prerequisite to attain food security and alleviate poverty in the Global South (Diiro et al., 2018; Malapit and Quisumbing, 2015). While several studies address women’s empowerment as a developmental outcome (Akter et al., 2017), the crucial role of rural women empowerment on agrarian development and gender transformative approaches have not yet received significant research focus. On the one hand, the quantitative empirical studies addressing technological change often limit the gender dimension to the sex of the household head, overlooking the key roles and responsibilities of household members in farming. On the other hand, in-depth qualitative gender case studies are often not sufficiently broad to allow for generalization, due to their small sample size. In this context, more mixed-effect studies that combine information from quantitative household surveys and qualitative case studies are needed to generate insights on women’s involvement in agricultural decision-making and its implications.

Contextualization or understanding the adoption problem from an agri-food system perspective is also warranted. As observed by Glover et al. (2016) for technological change in African agriculture, adoption studies might present an inaccurate and misleading picture for the policymakers and evaluators. Oversimplification of research problems to making it amenable for econometric analysis is the reason. For example, wheat farmers who adopt certain agronomic practices might also be interested in certain varieties, which are compatible with their preferred management practices. In some instances, the popularity of less sustainable alternatives might be posing an impediment to the diffusion of new technologies (Paudel et al., 2020). While there is no simple procedure to capture the multitude of determinants and direct and indirect effects of technological change, systematically sketching out the Theory of Change and the impact pathways would be an ideal starting point. Establishing longer term data collection activities in the target regions will be highly valuable for rigorous assessment of technology diffusion patterns. Mixed-method approaches and collaboration between economists and other social researchers across centers and regions might also be helpful to explore some of these topics in depth.

Finally, as we observed in the previous section, the existing adoption-impact literature in wheat covered only farming community, albeit that many interventions could have far-reaching implications. First, consumer effects of wheat innovations (e.g., price reduction in the output markets due to productivity enhancement and grain quality improvements) are hardly examined, although its implications for poverty reduction could be highly significant. Wheat innovation effects beyond the farm in the agri-food system (e.g., changes in costs and revenues from wheat agro-processing) have also not been covered in the recent literature. Second, the cultivable land (and the other factors of production) saved due to the yield enhancement and the associated ecological benefits are rarely examined. Third, the indirect effects of CGIAR research in agricultural systems of the Global South through building communities of knowledge may need a detailed examination. Documenting the capacity building activities in CGIAR centers and identification of their contributions are the first steps toward capturing these indirect impacts (Huang et al., 2014). However, even if we decide to focus on contributions alone, the challenge of obtaining viable counterfactuals remains. Multidisciplinary research groups are necessary to develop a methodological framework for assessing the indirect impacts of technological changes.

Conclusion

Public R&D often is under pressure to demonstrate to governments and funders that investments in agricultural research represent money well-spent (Renkow and Byerlee, 2010). Moreover, adapting agricultural research to address emerging challenges requires effective strategies for learning from experience, documenting the levels and patterns of outcomes, and identifying the factors affecting the adoption and impacts of agricultural interventions. As a result, institutionalizing adoption and impact assessment to accurately and rigorously document the outcomes of R&D investments has become a strategic priority for the CGIAR and NARS institutions. Here we have reviewed wheat adoption-impact studies over 2008–2017, and identified several aspects for enhancing the utility of such research. The approaches and methodologies for empirical adoption-impact analyses have not changed much over the last decade, and do not match the inherent potential to facilitate agrarian development. For example, most adoption studies have oversimplified and decontextualized technological change in wheat. Moreover, the impact studies are not only inadequate in number to capture the technological change in wheat, but several of them have also relied on strong and sometimes unfounded assumptions about counterfactuals. However, these are not unique challenges to wheat studies; such limitations long pervade across crops and research institutions (CGIAR Science Council, 2009).

We have pointed out several potential resolutions, particularly concerning better data collection and analytical methods. However, one may not overlook the institutional bottlenecks that are the root cause of some of the abovementioned limitations. Many of these challenges—including excessive dependence on cross-sectional data, lack of data triangulation, and overlooking spillover effects—arise from constraints in time, human and financial resources to design and implement robust adoption-impact research properly. During the last two decades, the CGIAR landscape has become more competitive for third-party funding (Barrett, 2020), and this has also influenced the nature and quality of adoption-impact research. The CGIAR Science Council (2009) attributed the unrealistic expectations of some donors to perform tasks within impossible time restrictions, among other factors, to a reduction in quality of socioeconomic research: “unrealistic expectations of short-term impact draw CGIAR social science away from its areas of comparative advantage, leading to problems of deteriorating research quality and staff morale” (p. 2). While the CGIAR centers are undergoing drastic changes at present (Barrett, 2020), one remains hopeful about favorable institutional changes during the coming decade. Several project-level modifications could also be made to increase the efficiency and quality of adoption-impact studies within the budgetary constraints. Changes, such as the enforcement of mandatory statistically defendable experimental designs for all technology scaling interventions, the use of power analysis to determine the sample size required to ensure absolute minimum confidence and precision levels, and the use of methods that are potent in addressing fundamental issues in impact analysis (e.g., endogeneity and selection bias), can go a long way in improving the quality of adoption-impact studies. This paper places a step in this direction to contribute to developing effective strategies for standardization and enhancement of the quality and use value of adoption-impact literature in general, and for wheat studies in particular.

Supplemental material

Supplementary_Materials_Wheat - Assessing technological change in agri-food systems of the Global South: A review of adoption-impact studies in wheat

Supplementary_Materials_Wheat for Assessing technological change in agri-food systems of the Global South: A review of adoption-impact studies in wheat by Vijesh V Krishna, Yigezu A Yigezu, Aziz A Karimov and Olaf Erenstein in Outlook on Agriculture

Footnotes

Acknowledgement

We thank Hans-Joachim Braun and Victor Maurice Kommerell of CGIAR Research Program on Wheat agri-food systems (CRP WHEAT), Maria Boa Alvarado of CIMMYT, Nancy Johnson of the CGIAR Standing Panel on Impact Assessment (SPIA), and the anonymous reviewer of the journal for their valuable comments.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study received financial support from CRP WHEAT in the form of salary support to all authors in 2018. The funder had no other role in study design, preparation of the manuscript, and decision to publish. The views expressed here are those of the authors, and they do not necessarily reflect the views of the funders or the associated institutions.

ORCID iD

Vijesh V Krishna

Supplemental material

Supplemental material for this article is available online.

Notes

References

Abay

Berhane

Taffesse

, et al. (2017) Estimating input complementarities with unobserved heterogeneity: evidence from Ethiopia. Journal of Agricultural Economics 126(C):51.

Akter

Rutsaert

Luis

, et al. (2017) Women’s empowerment and gender equity in agriculture: a different perspective from Southeast Asia. Food Policy 69:270–279.

Angrist

Pischke

(2008) Mostly Harmless Econometrics: An Empiricist’s Companion. New Jersey: Princeton University Press.

Asseng

Ewert

Martre

, et al. (2015) Rising temperatures reduce global wheat production. Nature Climate Change 5(2):143–147.

Badstue

Petesch

Williams

, et al. (2017) Gender and Innovation Processes in Wheat-Based Systems: GENNOVATE Report to the CGIAR Research Program on Wheat. Mexico, DF: CIMMYT.

Banerjee

Iyer

Somanathan

(2005) History, social divisions, and public goods in rural India. Journal of the European Economic Association 3(2-3):639–647.

Barrett

(2020) On research strategy for the new one CGIAR: editor’s introduction. Food Policy 91:101844.

Bhavani

Hodson

Huerta-Espino

, et al. (2019) Progress in breeding for resistance to Ug99 and other races of the stem rust fungus in CIMMYT wheat germplasm. Frontiers of Agricultural Science and Engineering 6(3):210.

CGIAR Science Council (2009) Stripe Review of Social Sciences in the CGIAR. Rome: CGIAR Science Council.

10.

Cudjoe

Breisinger

Diao

(2010) Local impacts of a global crisis: food price transmission, consumer welfare and poverty in Ghana. Food Policy 35(4):294–302.

11.

Davis

Nkonya

Kato

, et al. (2012) Impact of farmer field schools on agricultural productivity and poverty in East Africa. World Development 40(2):402–413.

12.

de Janvry

Sadoulet

Suri

(2017) Field experiments in developing country agriculture. In: Banerjee

Duflo

(eds) Handbook of Field Experiments: Amsterdam, The Netherlands: Elsevier BV, pp. 323–388.

13.

Di Falco

Veronesi

Yesuf

(2011) Does adaptation to climate change provide food security? A micro-perspective from Ethiopia. American Journal of Agricultural Economics 93(3):829–846.

14.

Diiro

Seymour

Kassie

, et al. (2018) Women’s empowerment in agriculture and agricultural productivity: evidence from rural maize farmer households in western Kenya. Plos One 13(5):e0197995.

15.

Douthwaite

Kuby

van de Fliert

, et al. (2003) Impact pathway evaluation: an approach for achieving and attributing impact in complex systems. Agricultural Systems 78(2):243–265.

16.

Dreisigacker

Sharma

Huttner

, et al. (2019) Tracking the adoption of bread wheat varieties in Afghanistan using DNA fingerprinting. BMC Genomics 20(1):660.

17.

Elbers

Lanjouw

(2003) Micro-level estimation of poverty and inequality. Econometrica 71(1):355–364.

18.

Erenstein

(2010) Village surveys for technology uptake monitoring: case of tillage dynamics in the trans-Gangetic plains. Experimental Agriculture 46(03):277–292.

19.

FAOSTAT (2020) Crops. Rome, Italy: Food and Agriculture Organization of the United Nations (FAO).

20.

Floro

Labarta

Becerra López-Lavalle

, et al. (2018) Household determinants of the adoption of improved cassava varieties using DNA fingerprinting to identify varieties in farmer fields: a case study in Colombia. Journal of Agricultural Economics 69(2):518–536.

21.

Gan

Liang

Hamel

, et al. (2011) Strategies for reducing the carbon footprint of field crops for semiarid areas. A review. Agronomy for Sustainable Development 31(4):643–656.

22.

George

(2014) Why crop yields in developing countries have not kept pace with advances in agronomy. Global Food Security 3(1):49–58.

23.

Glover

Sumberg

Andersson

(2016) The adoption problem: or why we still understand so little about technological change in African agriculture. Outlook on Agriculture 45(1):3–6.

24.

Gollin

Hansen

Wingender

(2018a) Two Blades of Grass: The Impact of the Green Revolution. Working Paper Series No. 24744. Cambridge, USA: National Bureau of Economic Research (NBER).

25.

Gollin

Probst

Brower

(2018b) Assessing Poverty Impacts of Agricultural Research: Methods and Challenges for CGIAR. Rome: Independent Science and Partnership Council (ISPC).

26.

Hammond

Fraval

van Etten

, et al. (2017) The rural household multi-indicator survey (Rhomis) for rapid Characterisation of households to inform climate smart agriculture interventions: description and applications in East Africa and Central America. Agricultural Systems 151:225–233.

27.

Huang

Xiang

Wang

(2014) Hidden Value of CGIAR Training Programs for National Research Capacity: a Case Study of Cimmyt’s Impact on China’s Wheat R&D Productivity: CGIAR Research Program on Wheat, Mexico: CDMX, p.36.

28.

Islam

Croll

Gladieux

, et al. (2016) Emergence of wheat blast in Bangladesh was caused by a south American lineage of Magnaporthe oryzae. BMC Biology 14(1):84.

29.

Jones

Kondylis

(2018) Does feedback matter? Evidence from agricultural services. Journal of Development Economics 131:28–41.

30.

Jurajda

(2005) Gender segregation and wage gap: an east-west comparison. Journal of the European Economic Association 3(2-3):598–607.

31.

Khandker

Koolwal

Samad

(2009) Handbook on Impact Evaluation: Quantitative Methods and Practices. Washington DC: World Bank.

32.

Kosmowski

Aragaw

Kilian

, et al. (2018) Varietal identification in household surveys: results from three household-based methods against the benchmark of DNA fingerprinting in southern Ethiopia. Experimental Agriculture 55(3):371–385.

33.

Kozuka

(2014) Inclusive Development: Definition and Principles for the post-2015 Development Agenda. Chapter 5, Perspectives on the Post-2015 Development Agenda, Tokyo: Japan International Cooperation Agency Research Institute (JICA-RI).

34.

Krishna

Spielman

Veettil

(2016) Exploring the supply and demand factors of varietal turnover in Indian wheat. The Journal of Agricultural Science 154(2):258–272.

35.

Kubitza

Krishna

Schulthess

, et al. (2020) Estimating adoption and impacts of agricultural management practices in developing countries using satellite data: a scoping review. Agronomy for Sustainable Development 40:16.

36.

Kubitza

Krishna

(2020) Instrumental variables and the claim of causality: Evidence from impact studies in maize systems. Global Food Security

37.

Leeuwis

Klerkx

Schut

(2018) Reforming the research policy and impact culture in the CGIAR: integrating science and systemic capacity development. Global Food Security 16:17–21.

38.

Liu

Cutforth

Chai

, et al. (2016) Farming tactics to reduce the carbon footprint of crop cultivation in semiarid areas. A review. Agronomy for Sustainable Development 36(4):290.

39.

Lobell

Ortiz-Monasterio

Sibley

, et al. (2013) Satellite detection of earlier wheat sowing in India and implications for yield trends. Agricultural Systems 115:137–143.

40.

Magnan

Spielman

Lybbert

, et al. (2015) Leveling with friends: social networks and Indian farmers’ demand for a technology with heterogeneous benefits. Journal of Development Economics 116:223–251.

41.

Malapit

HJL

Quisumbing

(2015) What dimensions of women’s empowerment in agriculture matter for nutrition in Ghana? Food Policy 52:54–63.

42.

Manifesto

Schlatter

Hopp

, et al. (2001) Quantitative evaluation of genetic diversity in wheat germplasm using molecular markers. Crop Science 41:682–690.

43.

McGowan

Sampson

Salzwedel

, et al. (2016) PRESS peer review of electronic search strategies: 2015 guideline statement. Journal of Clinical Epidemiology 75:40–46.

44.

Moher

Liberati

Tetzlaff

, et al. (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Plos Medicine 9(7):e1000097.

45.

Mottaleb

Krupnik

Erenstein

(2016) Factors associated with small-scale agricultural machinery adoption in Bangladesh: census findings. Journal of Rural Studies 46:155–168.

46.

Mottaleb

Singh

Sonder

, et al. (2019) Averting wheat blast by implementing a ‘wheat holiday’: in search of alternative crops in West Bengal, India. PloS One 14(2):e0211410.

47.

Müller

Elliott

Levermann

(2014) Fertilizing hidden hunger. Nature Climate Change 4(7):540–541.

48.

Nazli

Smale

(2016) Dynamics of variety change on wheat farms in Pakistan: a duration analysis. Food Policy 59:24–33.

49.

Orr

Ritchie

(2004) Learning from failure: smallholder farming systems and IPM in Malawi. Agricultural Systems 79(1):31–54.

50.

Page

SLJ

Baksh

Duveiller

, et al. (2009) Putting the poorest farmers in control of disseminating improved wheat seed: a strategy to accelerate technology adoption and alleviate poverty in Bangladesh. Food Security 1(1):99–109.

51.

Paudel

Krishna

McDonald

(2020) Apparent gains, hidden costs: examining adoption drivers, yield, and profitability outcomes of rotavator tillage in the wheat systems of Nepal. Journal of Agricultural Economics 71(1):199–218.

52.

Poets

Silverstein

Pardey

, et al. (2020) DNA Fingerprinting for Crop Varietal Identification: Fit-for-Purpose Protocols and Their Cost and Analytical Implications. Rome: Standing Panel on Impact Assessment (SPIA).

53.

Potash

(2018) Randomization bias in field trials to evaluate targeting methods. Economics Letters 167:131–135.

54.

Pritchett

Sandefur

(2015) Learning from experiments when context matters. American Economic Review 105(5):471–475.

55.

Renkow

Byerlee

(2010) The impacts of CGIAR research: a review of recent evidence. Food Policy 35(5):391–402.

56.

Rogers

(2010) Diffusion of Innovations. New York, USA: Free Press.

57.

Shiferaw

Kassie

Jaleta

, et al. (2014) Adoption of improved wheat varieties and impacts on household food security in Ethiopia. Food Policy 44:272–284.

58.

Shiferaw

Smale

Braun

H-J

, et al. (2013) Crops that feed the world 10. Past successes and future challenges to the role played by wheat in global food security. Food Security 5(3):291–317.

59.

Sianesi

(2017) Evidence of randomisation bias in a large-scale social experiment: the case of ERA. Journal of Econometrics 198(1):41–64.

60.

Stevenson

Macours

Gollin

(2018) The Rigor Revolution in Impact Assessment: Implications for CGIAR. Rome: Independent Science and Partnership Council (ISPC).

61.

Theriault

Smale

Haider

(2017) How does gender affect sustainable intensification of cereal production in the west African Sahel? Evidence from Burkina Faso. World Development 92:177–191.

62.

Thirtle

Lin

Piesse

(2003) The impact of research-led agricultural productivity growth on poverty reduction in Africa, Asia and Latin America. World Development 31(12):1959–1975.

63.

United Nations (2017) The sustainable development goals report. Available at: https://unstats.un.org/sdgs/files/report/2017/TheSustainableDevelopmentGoalsReport2017.pdf (accessed 6 February 2018).

64.

Velu

Ortiz-Monasterio

Cakmak

, et al. (2014) Biofortification strategies to increase grain zinc and iron concentrations in wheat. Journal of Cereal Science 59(3):365–372.

65.

WB, FAO, & IFAD (2008) Gender in Agriculture Sourcebook. Washington DC: World Bank, Food and Agriculture Organization (FAO), International Funds for Agricultural Development (IFAD).

66.

World Bank (2017) Monitoring Global Poverty: Report of the Commission on Global Poverty. Washington, DC: World Bank.

67.

Wossen

Abdoulaye

Alene

, et al. (2019) Estimating the productivity impacts of technology adoption in the presence of misclassification. American Journal of Agricultural Economics 101(1):1–16.

68.

Yigezu

Alwang

Rahman

, et al. (2019) Is DNA fingerprinting the gold standard for estimation of adoption and impacts of improved lentil varieties? Food Policy 83:48–59.

69.

Yigezu

Mugera

El-Shater

, et al. (2018) Enhancing adoption of agricultural technologies requiring high initial investment among smallholders. Technological Forecasting and Social Change 134:199–206.

70.

Zhang

Kovacs

(2012) The application of small unmanned aerial systems for precision agriculture: a review. Precision Agriculture 13(6):693–712.

71.

Zheng

Campbell

Serbin

, et al. (2014) Remote sensing of crop residue and tillage practices: present capabilities and future prospects. Soil and Tillage Research 138:26–34.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.53 MB