The evolution of innovation networks was analyzed looking at the development and diffusion of zero tillage (ZT) in Brazil. The increasing complexity of technology development and adoption is rapidly changing the effectiveness of scientific and technological policies. Complex technologies are developed and disseminated by networks of agents. The impact of these networks depends on the assets they command, their learning routines, the socio-economic environment in which they operate and their history. In this new environment, scientific and technology policies should: (1) foster interactions among agents (whether public or private), (2) increase the effectiveness of public research, extension and funding institutions, (3) give sufficient freedom to researchers to set their research programs, and (4) monitor the quality of research.Zero Tillage is the most important agricultural technology adopted in Brazil in the last 50 years. The study of the Brazilian experience sheds light on several research policy issues. First, Brazil’s experience with ZT cannot be explained by simple theories of technical change, a systemic approach is needed. The early development of modern ZT was supply-pushed, i.e. an agrochemical company (ICI) saw the market potential of a new idea and funded research to develop it before most farmers realized the need to conserve the soil.
The first ZT package was adapted to local conditions through the combined efforts of input suppliers, individual researchers, public institutions (IAPAR and EMBRAPA-CNPT), foreign aid agencies and pioneer farmers. After these first results, ZT evolved by the interaction of both supply and demand for technology: after the package matured in the late 1970s, technology development was organized by the Associations of Zero Tillage Farmers (AZTFs) and suppliers of critical inputs like herbicides and planters. This shows that public research programs based only on technological demands may be too narrow, because researchers have the best understanding of potential uses for their research outputs. Establishment of research priorities should consider supply as well as demand signals.
Second, research agendas cannot be dictated by short-term considerations. Development of the ZT package took more than 12 years. For most of this period, short-term production costs under CT were equal or lower than under ZT. New policies introduced in the early 1990s increased substantially the price of grains relative to the price of agrochemical products. At the same time, Monsanto reduced the price of a key input. Suddenly, ZT became efficient, both from the agricultural and economic points of view.
Third, this particular experience cannot be explained by a linear model of science. Even though development of the first product that enabled ZT (the herbicide) fits this model, the other components of the package were technology developments without an understanding of the scientific processes behind them. For example, even today the changes in soil structure, soil flora and fauna and nutrient mobilization under ZT in different regions are barely understood.
Fourth, the previous two points suggest that research impacts should not be used for research evaluation. If technology development and adoption occurs within a complex innovation system, the impacts cannot be allocated to any single agent but to the whole set; in other words, any impact (or lack of) is not the exclusive result of research. Also, since the timing of the impacts cannot be foreseen, lack of impact may result from technical problems or just from insufficient time for the technology to diffuse.
Fifth, three alternative technologies were developed and tried before soil erosion could be eliminated while maintaining the profitability of agriculture. A prior rejection of duplicate research efforts could have resulted in the adoption of an inferior technology—since alternatives would not have been developed. The uncertainty about the benefits of particular solutions indicates that a certain duplication of research efforts is necessary.
Sixth, the government twice chose to promote inferior technologies with supervised credits. A more flexible technology extension approach would have been more efficient both for farmers and the government.
Article extracted from Research and technology policies in innovation systems:zero tillage in Brazil by Javier M. Ekboir published in Research Policy 32 (2003) 573–586.