Understanding the adoption of systemic innovations in smallholder agriculture: the System of Rice Intensification (SRI) in Timor Leste

The latest turmoil of production and price volatility in the global food sector has put agriculture back to the top of the development agenda. Population growth, changing consumer preferences, bioenergy demand and climate change are some of the huge challenges for agricultural production today and in the future. In the last decades, productivity has been constantly improved through the introduction of improved crop varieties and the greater use of mechanization, irrigation, chemical fertilizer and pesticides. However, such input-intensive strategies do not always correspond to the livelihoods and capacities of millions of smallholders, who contribute substantially to global agricultural output, but are also strongly affected by persistent poverty and growing agro-environmental challenges. Moreover, recently farmers have experienced a downturn of productivity growth which in some cases is associated with environmental degradation and depletion of natural resources. This holds true in particular for rice, one of the world's most important food crops. In the course of growing agricultural challenges, it is widely recognized that innovative strategies are needed to improve human well-being and future food security. Natural resource management (NRM) practices are one stream of innovations that have been proposed to improve the efficiency of cropping systems in a systemic way. Prominent approaches are conservation agriculture, agroforestry and organic farming, which raised considerable attention within the last decades. Such NRM technologies are integrated innovations to improve agricultural productivity and agroecosystem resilience, involving different agronomic and management components with often synergistic relationships. Therefore, the term system technologies is also used here. Studies found that smallholder farmers often face difficulties with the adoption of complex system technologies. Some of the benefits also remain highly debated. In the rice sector, the so-called System of Rice Intensification (SRI) has been proposed as a promising technology to increase productivity at affordable costs for resource-poor producers. The principles of SRI focus on neglected potentials to raise yields by changing farmers' agronomic practices towards a more efficient use of natural resources. The innovativeness is based on a set of modified management practices concerning irrigation, plot preparation, transplanting, nursery and fertilization. Even though SRI has been widely promoted in some countries, partial adoption and discontinuance are common and the impacts are often found to be context-specific. However, most of the available literature is based on agronomic studies. There is limited evidence in terms of socioeconomic aspects, which is considered a drawback, as system technologies such as SRI may affect farming systems as a whole. In order to explore opportunities and constraints of technological innovations in smallholder farming, studies have to account for the observed variability of resource endowments and farm management options. This study aims to contribute to this research direction by analyzing the linkages between SRI adoption, rice yields, household income and poverty. Investigating the case of SRI may allow us to draw wider conclusions towards the nature of system technologies in general. The results may help researchers and policy makers to understand socioeconomic constraints to farmer technology adoption and integrate this knowledge into the formulation of rural development strategies. This study uses household and plot level data from small-scale rice farmers in Timor Leste. Assessing the role of improved rice management practices in Timor Leste is highly relevant from a development perspective. First, this young nation state remains one of Asia's poorest countries in terms of income and food security measures. Second, rice is the main staple food for the majority of the population, but domestic production faces severe technical and environmental challenges such as low levels of mechanization, water scarcity and limited access to agricultural technologies. Since 2007, SRI has been introduced by the Second Rural Development Programme (RDPII). Jointly implemented by the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) and the Timorese Ministry of Agriculture and Fisheries (MAF), the extension program aims to improve the productivity of rice production systems in the research area. As part of this study, a farm survey was conducted between August and December 2009. Stratified random sampling was used to select 400 households from participants and non-participants of SRI training programs. These households were interviewed. In addition, plot level data from 475 paddy fields owned by these sample households were collected. We begin our analysis by identifying SRI adoption patterns and differences between SRI and non-SRI farmers. We show that adoption patterns vary substantially, and partial adoption is commonplace. Whereas some technology components are widely applied by households in the research area, others lack widespread acceptance. The highest SRI adoption rates are recorded for the group of training participants. However, the descriptive analysis also reveals that land and household characteristics seem to play a role in the adoption decision. For example, owners of larger farms are more likely to adopt SRI. The outcomes point at substantial heterogeneity among and between adopters and non-adopters, which has to be considered in the econometric analysis of adoption determinants and impacts. For the econometric analysis of adoption determinants, different decisions points are identified. A double-hurdle adoption model at the household level shows that variables such as farm size, availability of family labor and participation in extension training determine the initial adoption decision and the share of rice acreage under SRI. However, household level characteristics alone are insufficient to explain adoption. Therefore, an additional double-hurdle model is estimated at the plot level. Several plot level determinants have a significant effect on SRI adoption and the number of technology components used. For example, the availability of an irrigation system, which can be individually controlled by the farmer, is an important determinant for SRI adoption on a particular plot. However, understanding the adoption determinants alone is insufficient to determine whether or not wider adoption is actually desirable. To analyze this, the third part of the analysis explores the impacts of SRI in terms of yields, household income and poverty. In order to account for the differences and variability among household and plot level parameters, the study accounts for differential technology impacts between the adopters and non-adopters of SRI, using an endogenous switching regression approach. Simple comparison of yield and incomes between adopters and non-adopters does not reveal significant differences, however, we find negative selection bias, meaning that SRI is adopted on plots and by farmers that would have below average yields without adoption. Controlling for external factors and selection bias, it is estimated that SRI is increasing yields by 46% against the counterfactual outcome of non-adoption. We also find a small but significant positive household income effect. Both poor and non-poor households benefit from SRI adoption. Especially smaller and more specialized farms realize high returns from adoption due to lower opportunity costs of investment. Moreover, SRI farmers also use lower amounts of inputs such as water, seeds and pesticides. Yet, we also find that the gains from adoption depend on plot and farmer heterogeneity. That is, assuming that the same gains were to occur for the non-adopters would they decide to adopt is too simplistic. To conclude, we have shown that farmers can benefit from the introduction of the system technology SRI. Therefore, SRI adoption presents a potential pathway towards food security, poverty reduction and rural development. However, we have also identified several constraints that hinder the adoption of SRI. Not all farmers can easily implement each component at any given plot, and the gains of adoption depend on the reference system. This is an important outcome with regard to extension services and development agencies highlighting that location-specific factors are relevant with regard to adoption and impacts of system technologies. Moreover, improved rural infrastructure and irrigation systems can further increase adoption rates and adaptation capacity. These challenges need to be overcome, in order to fully harness the potential of promising system technologies in smallholder agriculture.