The Role of Extension and Sustainable Soil Management in Smallholder Agriculture - Evidence from Ethiopia
Rising demand for agricultural commodities coupled with population growth, climate change, declining soil fertility, environmental degradation and rural poverty in the develop-ing world call for strategies to sustainably intensify agricultural production. Sustainable in-tensification refers to increasing production from the same area of land while reducing its negative environmental consequences. Most of the adverse conditions are particularly preva-lent in Sub-Saharan Africa (SSA), where rates of undernutrition are the highest worldwide, while agricultural productivity is still far below global averages. An important factor in ex-plaining productivity deficits among smallholders in SSA is the slow adoption of new agri-cultural technologies. Recently, governments and international donors especially concentrate on the promotion of 'system technologies', i.e. packages of technologies that should be ap-plied jointly due to synergistic effects. Yet, evidence shows that farmers delay in particular the uptake of system technologies, and tend to scatter practices across plots instead of com-bining them on the same plot. Hence, analyzing how to effectively enhance the adoption of technology packages is crucial, but still understudied. In addition, comprehensive studies on the plot- and household level effects of system technologies that use micro data from farmer surveys are still scarce when it comes to impacts beyond traditional outcomes, such as crop yields and income, but important to understand the consequences of adoption for farmers. This dissertation addresses these gaps by studying the adoption and effects of 'Integrated Soil Fertility Management' (ISFM). ISFM is a system technology comprised of a set of site-specific soil fertility practices which should be applied in combination. Its core is the inte-grated use of organic and inorganic fertilizers with improved seeds. Practices should be adapted to local conditions, accompanied by a general improvement of agronomic tech-niques and, depending on the context, by other technologies such as crop rotation, agroforest-ry or reduced tillage. The general aim of ISFM is an improvement of the soil's fertility by replenishing its nutrient stocks and organic matter level. Enhanced soil fertility is likely to improve food security, incomes, and ultimately, livelihoods of the rural population depend-ing on small-scale agriculture. In addition, healthier and more fertile soils can contribute to restoring and conserving natural resources by providing crucial ecosystem services, such as the storage of soil carbon, erosion control and the prevention of further deforestation. Thus, they can make an important contribution to the sustainable intensification of smallholder agricultural systems. However, ISFM commonly also goes along with increased demand for capital and labor, which often prevents smallholders from adopting it. In addition, ISFM is considered knowledge-intensive, as combining several practices and adapting them to local conditions requires at least a basic understanding of biological processes. Against this background, the dissertation addresses two broad research objectives: Firstly, to assess the role of 'farmer-to-farmer' and non-traditional forms of agricultural extension to enhance knowledge and adoption of ISFM as a pathway to sustainable intensification. And secondly, to assess the productivity and welfare implications of adopting ISFM practices at the plot and household level. The thesis comprises three essays. The first essay concentrates on knowledge and adoption of ISFM as a complex agricultural technology, while the second and third essay analyze the effects of ISFM at the plot, respectively household level. All three essays build on primary data collected among 2,382 farm households in the three Ethi-opian regions Amhara, Oromia and Tigray. The research was carried out in cooperation with the 'Integrated Soil Fertility Management Project' (ISFM+ project) of the German Agency for International Cooperation (GIZ), launched in 2015 in 18 districts in the three highland regions. The first essay focusses on the role of agricultural extension in the dissemination of ISFM. In recent decades, decentralized and participatory extension models have become dominant in SSA. In these 'famer-to-farmer' approaches, only a few 'model farmers' are trained directly by extension agents and should then train other farmers, often organized in groups. From there, information should trickle down to all other households in a community. Yet, evidence suggests that information diffusion is a complex process and does not automatically reach all farmers. On the contrary, knowledge is likely to be transmitted incompletely from model farmers to extension group members and from there to 'ordinary' farmers. This applies in particular to complex system technologies, where farmers have to learn about each individu-al practice as well as the necessity of applying them jointly. In this article, we assess the ef-fects of a farmer-to-farmer extension model and an additional intervention in form of a video on farmers' knowledge and adoption of ISFM. We implemented a cluster randomized con-trolled trial, using 161 microwatersheds (mws) as primary units of randomization. 72 mws received the farmer-to-farmer extension treatment, with model farmers who maintain ISFM demonstration plots and train so-called 'farmer research and extension groups' as core ele-ments. 36 out of these treatment mws received an additional video intervention, explaining the underlying reasons for adopting the ISFM package, and featuring documentaries on suc-cessful ISFM adoption. 89 mws did not receive any intervention and serve as control group. In each of the three groups, 15 households per mws were randomly selected to be included in the sample. Findings show that farmer-to-farmer extension, both alone and in combination with video, increases ISFM adoption, both of its individual components as well as their com-bined adoption on the same plot. Effects are stronger for farmers who are involved in group-based extension activities, but exist to a weaker extent also for farmers in the same commu-nities who are not involved. On average, we find no significant additional effect of the video intervention on adoption. However, the video does show a significant additional effect for farmers in treatment mws who are not members of extension groups, in particular when it comes to the integrated use of the practices on the same plot. Further, while both farmer-to-farmer extension alone and in combination with the video induce gains in ISFM knowledge, effects are significantly stronger for the combined treatment. A causal mediation analysis reveals that increases in knowledge explain part of the treatment effects on adoption. Over-all, these results suggest that farmer-to-farmer extension can effectively foster technology adoption; both among extension group members as well as among non-members residing in the same communities, probably a sign of information spillovers. Yet, for the non-members, providing complementary information via video seems a valuable method to counterbalance incomplete information diffusion and ultimately, foster the adoption of complex system technologies such as ISFM. Essay two analyzes of the effects of different combinations of ISFM practices on land productivity, net crop value, labor demand, labor productivity and financial returns to unpaid labor at the plot level. To date, evidence on the profitability of ISFM in smallholder settings is scarce, in particular when it comes to labor investments. The study differs from previous research by looking into a broader range of outcome indicators, and into the effects of dis-tinct combinations of inorganic fertilizer, organic fertilizer and improved seeds. We employ a multinomial endogenous switching model to account for endogeneity, and data from over 6,000 teff, wheat and maize plots. Results show that both partial and complete ISFM adop-tion lead to significant increases in land productivity and net crop value, in particular when improved seeds are used. On average, the largest effect on land productivity stems from adopting complete ISFM, i.e. improved varieties with inorganic fertilizer and organic ferti-lizer, followed by the combinations containing only one fertilizer type. Analyses for two different agroecological zones suggest that in moister regions, complementing improved varieties with inorganic fertilizer is most important, while in drier regions, enhancing it with organic fertilizer is crucial, most probably due to its water-retaining effect. Regarding net crop value, average effects of combining improved seeds with either one or both fertilizer types are similar, despite the larger effect of the complete package on land productivity; probably due to reduced input costs when only one of the two fertilizer types is used. Further, as expected, ISFM is related to higher labor demand, but also significantly increases labor productivity and financial returns to labor. Hence, despite the additional demand for labor and capital, results suggest that ISFM can be a profitable technology for smallholders, at least when assessed at the plot level. The third essay complements the picture on ISFM effects by analyzing its impacts at the household level. This is important since additional demand for resources associated with a technology (package) may imply a reallocation of labor from one income-generating activity to another, leaving net effects for a household uncertain. Therefore, we study whether adopt-ing ISFM on at least one teff, wheat or maize plot increases income obtained from these crops, as well as total household income and household labor demand, and whether ISFM adoption is related to the probability of pursuing other economic activities. In addition, we assess impacts on food security, measured by self-reported incidences of food deprivation. Further, the essay analyzes effects on children's education as indicator for longer-term wel-fare, assessed by the enrollment rate of children in primary school age, the average number of absent school days and average educational expenditure. On the one hand, additional labor requirements may increase the work burden for children, with possible negative effects for their education. On the other hand, if ISFM is related to income gains, it might also lead to additional investments in education. We apply the inverse probability weighting regression adjustment method to account for selection bias, with propensity score matching as robust-ness check, and account for dissimilar agroecological potential by running disaggregated analyses for moist and dry regions. Results show that ISFM adoption for main cereal crops is related to increased income per capita obtained from these crops in both agroecological zones. Effects sizes of a rather lax definition of ISFM – having used improved seeds in com-bination with at least either organic or inorganic fertilizer – and a stricter definition, which comprises both fertilizer types, are very similar. A reason for that might be the additional costs associated with using two instead of only one fertilizer type; or because the synergistic potential of their joint use does not materialize immediately. Yet, only in the moister re-gions, higher crop income seems to translate into higher household income per capita, while it does not in the dry region. This might be because the share of income from these crops in total household income is not important enough in the latter subsample. Yet, in the dry re-gion, ISFM adoption for main cereals also leads to a lower probability of achieving income from other crops and off-farm activities, probably an effect of resource reallocation (in par-ticular labor). Moreover, we find a food security-enhancing effect of ISFM only for the moister areas, but not for the dry region. In both subsamples, ISFM adoption is related to increased demand for household labor. Yet, despite the higher labor demand, we find no in-dication for increased school absenteeism or even reduced enrollment rates of children, and no effects on educational expenditure. By contrast, ISFM adoption is associated with higher primary school enrollment the moist agroecology. Hence, only for areas where ISFM adop-tion is related to gains in overall household income, we also find positive effects on other welfare indicators, such as food security and education. All in all, these results suggest that broader welfare effects of agricultural innovations have to be evaluated within the complex system of households' income diversification strategies. Overall, this dissertation contributes to the state of research by drawing a more comprehen-sive picture of the effects of ISFM in resource-constrained and diversified smallholder sys-tems, as well as of interventions to foster the adoption of ISFM, or system technologies in general. Firstly, results imply that farmer-to-farmer and other, not traditional forms of agri-cultural extension have the potential to increase knowledge and adoption of complex innova-tions. Yet, extension systems still have to overcome shortcomings and find ways to be more inclusive, probably by means of an effective and creative mix of interventions. And second-ly, findings suggest that ISFM can be a profitable technology for farmers, but also requires more resources. When evaluating broader impacts of its adoption, it is important to account for heterogeneous conditions and contexts.
