Professor, Department of Agricultural Economics, Kansas State University
Prior to the mid-1960's, active research collaboration between technical agricultural scientists (i.e., mainly working on experiment stations), agricultural economists (i.e., mostly in planning units) and anthropologists/rural sociologists (i.e., generally in academia), was limited. By the mid-1960s, the Green Revolution was beginning to have a major impact on crop production in parts of Asia and Latin America through the introduction of fertilizer-responsive, high-yielding varieties of rice, wheat, and maize in favorable and relatively homogeneous production environments where there was assured soil moisture, good soils, ready access to cheap fertilizer, and relatively efficient output markets. However such conditions did not exist in most of Sub-Saharan Africa and in certain parts of Latin America and Asia, and as a result, these areas were bypassed.
The reductionist approach failed in terms of developing technologies for resource-poor farmers in less favorable heterogeneous production environments or agricultural areas. This led to the incorporation of a systems perspective in the identification, development, and evaluation of relevant improved technologies. Hence in the mid to late 1970s, the farming systems research (FSR) approach evolved, a basic principle of which was the need to create new types of partnerships between farmers and technical and social scientists.
FSR thus became very popular with donor agencies, to the extent that, by the mid 1980s, about 250 medium- and long-term externally funded (i.e., in addition to those domestically funded) projects worldwide were implementing FSR-type activities. Between 1978 and 1988, USAID2 alone had funded 76 bilateral, regional, and centrally funded projects containing a farming systems orientation. Forty-five of these were in Africa. Most of these projects supported the establishment of separate FSR units, which often were poorly integrated into, or poorly linked to, mainstream technology development activities. Although it is probably true to conclude that few of these projects succeeded in producing new technologies that were widely adopted, the approach of looking at farmers’ constraints and needs for technical change from within was eventually mainstreamed into most national and international agricultural research programs by the late 1980s. Therefore although donor support for supporting explicit FSR activities dwindled towards the end of the 1980s, most national agricultural research systems (NARS) had adopted major components of the FSR philosophy and approach, and the spirit of the FSR approach lived on.
Since then there has been considerable evolution in the methodologies employed (e.g., new farmer participatory research (FPR) techniques, gender analysis, environmental impact analysis, and statistical techniques adapted to on-farm research). Also participation has been broadened to include a wider set of agricultural stakeholders, including extension,3 development, and sometimes even planning/policy staff.4 Perhaps even more significant has been incorporating the underlying principles of the farming systems approach into the priorities of donors and nationally based agricultural programs. These include increasing emphasis on participatory approaches and empowerment of farmers and their families and a new focus on ecological sustainability and sustainable livelihoods. Although appropriate technologies still are viewed as important catalysts for improving farmers’ welfare, the criteria for relevancy have become more clearly defined and specific.
In this paper I summarise how the farming systems approach that has evolved over the last 30 years with a very brief indication of the factors that contributed to bringing about those changes.5 A key dimension of that evolution has been the way the scope or inclusiveness of a systems perspective has been expanded systematically over time. “The way in which the systems perspective is implemented, in particular the scope or inclusiveness of the systems analysis, depends on how, for any given problem, researchers define the ratio of variables to parameters; or put another way, which factors are considered endogenously determined and thus subject to analysis and modification, and which are taken as exogenously determined constants. Because of the analytical difficulties of simultaneously handling large numbers of variables, most of the early FSR programs took only incremental steps away from traditional reductionist approaches by limiting the number of variables they studied and by regarding the other factors that influence the farming system as parameters or constants. As analytical methods have grown more sophisticated, and particularly as farmers have become active partners in the analysis, the ratio of variables to parameters has increased, and the analytical domain has expanded considerably” [Norman and Matlon, 2000].
This evolutionary process is operationally summarized in four phases in Figure 1 [Norman and Lightfoot; Norman and Matlon, 2000]. The important point to bear in mind is that moving progressively through the four phases means dealing with a progressively higher ratio of variables to parameters. This has become feasible through the development of analytical methods that can handle increasingly complex situations and have become particularly significant in the last decade. Formal modeling techniques have often proved to be of limited value in addressing such situations simply because of the variation in, and the complexity of, potentially important relationships and the degree of understanding that is required initially to develop realistic models. In essence what has occurred is the ability to use less formal modeling approaches through farmer-participatory techniques that empower farmers to drive the modeling process and use farmers' minds as computers. This empowerment has been greatly helped as a result of the evolution of techniques for improving/systematizing collegial, interactive, and meaningful dialogue between farmers and other developmental actors.6
Before looking at the four phases I briefly look at the role of farm management that was the precursor to this evolutionary process. I do this because many involved in the early days of the farming systems approach were agricultural economists trained in the neoclassical tradition. Unfortunately in those days many of us had little appreciation of what other social scientists (i.e., sociologists and anthropologists) had to offer. Fortunately our early experiences in the field helped educate us in this regard!
1960’s TO EARLY 1970’s – Farm Management
The farm management approach of the early 1900s was in many ways analogous to what the farming systems approach had become in the latter part of the same century. The farm management of the early 1990s was multidisciplinary and holistic whereas by the middle of the century it, at least in the US, it had become narrower and more reductionist in perspective and increasingly focused on production economics. Emphasis was on normative and prescriptive issues through application of techniques such as budgeting, linear programming, and other tools for applied decision analysis [Johnson, 1981]. Thus agricultural economists armed with these analytical tools and with a strictly neoclassical orientation dominated the farm management-type studies undertaken in the low-income countries during the 1960s. Although useful studies using a more positivist orientation were conducted by sociologists and anthropologists, many of whom predated the work of the agricultural economists, they did not have a major influence on the initial development of the farming systems approach, at least in anglophone countries. They appeared to have had somewhat more influence in francophone countries.
Many of us agricultural economists, associated with development/government organizations, academic institutions, and/or occasionally agricultural research institutions working independently, and spent the 1960s using formal, structured, cost-route, farm-management-type sample survey techniques, to describe farm-level resource allocation patterns and productivity among resource-limited farmers in Asia, Latin America, and Africa.
These studies produced a great deal of quantitative information describing cropping systems (and to a lesser extent farming systems) and their major socio-economic production constraints. They also described how households allocated their resources, and provided estimates of factor returns. Such studies were particularly common in Africa and Asia [e.g., Collinson 1972; Walker and Ryan, 1990; Spencer, Byerlee and Franzel, 1976; Stevens,1977; IRRI, 1979], but less common in Latin America, where economists generally had greater preoccupation with institutional or policy-related issues. These studies showed overwhelmingly that limited-resource farmers have an intimate understanding of their spatially variable and temporally risky production environments within which complex (i.e., combining crop, livestock, and off-farm enterprises) but fundamentally sound and sustainable farming systems had evolved over time. Given these very positive findings about the rationality of limited-resource farmers and the farming systems they practiced, the focus of questions soon shifted to why formally recommended technologies were adopted so rarely [Matlon, 1987].
Thus many agricultural economists, particularly those associated with research stations in Africa, Asia, and Latin America, began to evaluate recommended technologies (i.e., usually packages and typically crop-oriented). Prior the mid-1960s, very few station-based experiments were subjected to any economic analysis, and therefore it was not surprising that the conclusion that often emerged was that many existing recommendations were poorly designed or irrelevant, especially when criteria relevant to farmers were applied. In addition three other significant insights emerged:
Contrary to expectations of many, farmers were found to be natural experimenters [Biggs and Clay, 1981], using informal methods and consequently it was wrong to conclude they were conservative and averse to change.
Farmers’ production environments were found to be much more heterogeneous than had been thought, and consequently there was a need to develop technological components that could be adjusted easily and combined variously to better respond to location-specific needs, rather than relying on a few technological packages (i.e., the one size fits all syndrome!).
And although the recommended technological packages were sometimes compatible with the biophysical environments within which farmers operated, farmers were often not able to adopt them because of their incompatibility with the socioeconomic environment within which they operated [Norman et al., 1982].
Questions started arising as to whether the current process for developing and evaluating technologies was relevant for resource-poor farmers operating in less favorable and highly variable environments. It also became apparent that standard conventional economic criteria didn’t ensure identification of a relevant technology (e.g., farmers and their households had goals other than profit maximisation, there were usually multiple market failures for capital, labor, land, and information, and risk and uncertainty were significant issues). As a result many of us came to the conclusion that:
The neoclassical economic paradigm was ineffective in dealing effectively with all the issues relating to small-scale farmers.
The approach was also too static and deterministic in its orientation rather than recognizing that farmers operate in a dynamic and often uncertain environment.
The approach of extracting data from farmers (i.e., treating them as objects) and analyzing it independently was much inferior to an approach that recognized the benefits of synergism as a result of interaction and active participation on the part of farmers themselves.
The approach was flawed in its ex post orientation of focusing on evaluating available technologies, rather than using one that encouraged an ex ante involvement of farmers in the technology design and development process itself.
As a result many argued for that the conventional research paradigm needed to be modified drastically and replaced by one that would involve farmers as stakeholders from the beginning of the technology design process and that would use an interdisciplinary approach. Thus momentum developed for the evolution of a new approach based on changing from a “top-down” (“supply-driven”) approach to farmers, to one characterized as being “bottom-up” (“demand-driven”) from farmers.
LATE 1970s To EARLY 1980s – Early Farming Systems Approaches
Given what was discussed in the preceding section it is not surprising that the newly christened FSR activities were focused primarily on technology development objectives [Gilbert et al., 1980]. In the anglophone countries, several of the international agricultural research institutes (IARCs) (i.e., especially IRRI and CIMMYT) associated with the Consultative Group on International Agricultural Research (CGIAR) played important roles in the early methodological development and popularisation of FSR. At the same time FSR was introduced, with donor support, in many nationally sponsored agricultural research institutes in low-income countries.
The FSR approach that evolved was based on the notion that: one had to begin with understanding the problems of farmers from the perspectives of farmers; and that solutions had to be based on a proper understanding of their objectives and their environments, including both biophysical and socioeconomic components. Also a central tenet of the new approach was that not only did farmers have a right to be involved in the technology development and evaluation process, but that their inputs were essential. Other significant features were its holistic perspective, the fact that scientists involved in the process should represent both technical and social scientists, and that the process was by nature iterative. Consequently, some of the characteristics of the early farm-management approach started to reappear, leading Johnson to observe, in commenting on the new FSR methods, “there has been much reinventing of the wheel in developmental thinking” [1981: p. 2-3].
Although there was a commitment in principle to include a broader set of farmer-based criteria, in its earliest days FSR continued to focus on how yields of particular crops could be increased. However, even though the new on-farm research approach did involve the inclusion of socioeconomic elements, and hence had a farming systems perspective, it was done generally with a predetermined focus that targeted the productivity of a particular commodity (Figure 1). Thus, this approach involved looking at one part of an enterprise or one specific enterprise and identifying improvements within that focus that were compatible with the whole farming system. For example, CIMMYT and IRRI worked on maize-, wheat-, and rice-based systems, which were compatible with their crop mandates. Undoubtedly they had a favorable impact in introducing more of a systems perspective to the influential commodity-based research programs that had a strong reductionist orientation. They believed that the predetermined focus approach was directly relevant to farming systems dominated by one crop, because improving the productivity of that enterprise would have the greatest impact on the productivity of the overall farming system. These two IARCs, in particular, played influential roles through networks and training programs in Africa and Asia, thereby exposing scientists to the principles of the farming systems approach.
Although the IARCs undoubtedly played an important role in introducing and nurturing the farming systems perspective within national agricultural research systems (NARS), simultaneous independent efforts occurred in developing and promoting the approach, usually supported by donor funds. Notable examples were ICTA (Guatemala), Changmai (Thailand), Unite Experimentales, ISRA (Senegal), and the Institute of Agricultural Research, Ahmadu Bello University, Northern Nigeria. Because of the multi-commodity mandates of most NARSs, the farming systems efforts evolved quickly towards a more holistic orientation (i.e., what is labelled farming systems with a whole farm focus in Figure 1). This approach enabled focusing on constraints in any enterprise depending on farmers’ articulated needs. This evolution was further stimulated by two other factors, specifically:
A desire to encourage greater participation by farmers through addressing their specific needs rather than simply trying to “fit” technologies to specific enterprises that had been preselected by researchers/mandates.
The trend towards establishing separate area-based farming system teams in contrast to an on-farm testing component associated with each station-based commodity team.
The dominant disciplines in the early days of the farming systems approach were cropping systems agronomy and agricultural economics and the methodologies reflected this discipline mix. The trend away from treating farmers as “objects” to treating them as “people”, with whom useful interactive dialogue could be established, was helped greatly by increasing reliance on informal surveys or rapid rural appraisal (RRA) techniques. The results of these were sometimes verified by limited-visit formal surveys. The results were helpful in designing on-farm trials that were either managed by researchers or farmers -- but usually executed by the latter -- and were validated by conventional statistical techniques. Trials superimposed as separate plots on farmers’ operational fields also became commonplace.
Three very positive results from these early experiences with the farming systems approach:
Technical scientists were increasingly sensitised to the complexity and variability of farmers’ production environments (i.e., consisting of both physical and socioeconomic components) thereby helping reorient technology generation towards addressing needs of different types of farmers and emphasizing more flexible technological components rather than simply blanket-type package technologies.
The approach provided an opportunity for technical and social scientists to cooperate in the diagnosis of farmers’ situations and in the design, testing, and evaluation of new technologies, thus helping those concerned to better understand better the disciplinary perspectives and tools of other specialties.
Results also demonstrated the importance of complementary policy/support systems (e.g., input distribution systems and product markets) in determining the appropriateness of new technologies.7
However, a number of limitations or weaknesses became increasingly apparent. Five particularly significant weaknesses were:
Farmers’ participation was still limited largely to roles assigned by researchers and methodologies for obtaining and systematizing farmers’ knowledge and for analyzing the results of on-farm experiments were also poorly developed, often resulting in scepticism about how valid were researchers’ interpretation of information obtained from farmers.
Although the complexities of linkages between farm and household in influencing decision making and flows of resources and benefits were increasingly recognized, methodologies for incorporating such considerations into technology design and evaluation were inadequate and too often purely subjective and ad hoc in nature.
Although some linkages had been developed between the different disciplines (i.e., mainly agronomy and agricultural economics) there was still a lot of room for improvement. The application of the farming systems approach to livestock enterprises was generally particularly weak.
The most commonly used methodologies for data collection and analysis generally were based on the assumption of a monolithic household that could be described by a single objective function and with the household head at the center of the information nexus, in spite of increasing evidence to the contrary (e.g., multiple decision makers, differentiation in terms of distribution of benefits).
Factors relating to the policy/support system were treated as parameters within which the search for improved technologies took place. This was partly because the mandates of the technology-oriented institutions in which most farming-systems-related work was – and for that matter still is -- based did not include objectives of influencing the policy context and support systems. As a result this severely constrained the types of technologies that could be developed/evaluated.
Fortunately, there was increasing recognition of the above limitations as the decade of the 1980s progressed. The changes that occurred during the decade were supported by an increasing acceptance of a new developmental paradigm, which Korten  characterizes as a “learning process” (i.e. people centered) approach to the earlier “blueprint” (i.e. technology) approach.