Browsing by Subject "MPMAS"

Now showing 1 - 1 of 1

Smallholder adaptation through agroforestry : agent-based simulation of climate and price variability in Ethiopia
(2021) Yismaw, Habtamu Demilew; Berger, Thomas
Climate variability has been posing formidable policy challenges in Ethiopia by deteriorating rural livelihoods. Climate variability-induced shocks have a profound impact on smallholder farmers welfare both in the short run via reducing production and increasing output prices and in the long run by depleting productive farm assets and leading them to a poverty trap. However, the impact of shocks on smallholders welfare depends on their choice of adaptation and coping measures to deal with them, which are in turn farmer-specific. This thesis applied an integrated econometric analysis and farm-level simulations to assess smallholder farmers adaptation and coping measures under extreme climate and price variability in Ethiopia. The thesis provides a special focus on the role of investment in small-scale agroforestry to curb the adverse effects of shocks. Drought, hailstorms, pests, and crop diseases are identified as the most frequent and intense climate variability-induced shocks in Ethiopia. Smallholders dominant adaptation and coping measures for each shock are identified using logistic principal component analysis (LPCA). The application of dimensionality reduction of binary data using LPCA to select smallholder farmers dominant adaptation and coping measures is unique to this study. Results show that planting stress-resistant crops and varieties, early planting, increasing seed rate, and soil and water conservation practices are the dominant ex-ante adaptation measures. Whereas selling livestock, selling assets, reducing consumption, borrowing and replanting are the dominant ex-post coping measures. In what follows, household-specific drivers of smallholder farmers choice of adaptation and coping measures are disentangled using multivariate probit regression (MVP) for each shock. Gender, knowledge and experience, participation in rural institutions, social networks, resource endowments, and their shock experience and expectation are the significant drivers of farmers choices of ex-ante adaptation and ex-post coping measures. Results suggest that smallholder farmers choice of measures to deal with climate variability-induced shocks is highly distinctive and depends on their socioeconomic settings, experience and knowledge, and their interactions with the environment. Correlation analysis both in LPCA and MVP results show that most of the measures farmers choose are complementary, which implies that no single measure is robust and works best for all farmers. Farmers invest more on ex-post measures than ex-ante measures. Those who invest on tree perennials as ex-ante drought measures are less likely to use severe measures such as selling livestock and other assets in the aftershock. The econometric analysis in the first part of the thesis establishes a descriptive analysis of smallholder farmers behavioral responses to climate variability-induced covariate shocks. This captures the behavior of farmers in the status quo. However, it does not tell us much about how farmers would behave in different future circumstances, especially with extreme climate and price variability. This requires a prescriptive and descriptive approach with a detailed investigation of farmers behavior down to the plot level. To achieve this objective, the second part of the thesis applies household-level micro-simulation to analyze ex-ante planning and ex-post responses to future climate and price variability, focusing on the role of smallholder farmers’ investment in woodlot perennials to their livelihoods. The agent-based simulation package - Mathematical Programming-based Multi-Agent Systems (MPMAS) is used for this purpose to capture investment, production, and consumption decisions at the farm household level. A farm decision model representing smallholder farmers in the Upper Nile Basin in Ethiopia is developed accordingly. The farmers in the area are known for their integrated crop-livestock system, and a unique Acacia Dicurrens based emph{taungya} system in the country. This thesis shows a first-time use of an agent-based modeling approach representing the acacia-based emph{taungya} system in the Upper Nile Basin in Ethiopia, which is another contribution of this study. The farm decision model is validated using empirical data and interactive sessions with experts. Another methodological novelty of this study is developing and using interactive web applications to validate the farm decision model with experts remotely due to the COVID-19 pandemic. Interactive model validation is not new in agent-based modeling using MPMAS. However, the web-based R Shiny app and an integrated web-based expert survey questionnaire app is an original contribution of this study. The study also uses Zoom virtual conferencing to record all interactive validation sessions with experts. Two simulation experiments were designed to quantify the effects of shocks and price variability on agents livelihoods. The High-Performance Computing platform in Baden Wurttemberg (bwHPC) is used to run simulation experiments for this study. The first simulation experiment aims at measuring the effects of shocks on agents livelihoods and the effectiveness of ex-ante planning to curb the adverse effects of these shocks. The application of model features on agents ex-ante preparation for the possible occurrence of shocks within the farm decision model is a new development in this study. Accordingly, the frequent and intense crop and tree diseases in the area - potato late blight and acacia seedling disease are introduced as shocks in the model. Simulation results show that both potato late blight and acacia seedling disease reduce annual per capita discretionary income significantly and forcing some poor resource agents to fail to fulfill minimum non-food expenditure. The trade-off in agents land-use decisions between trees and crops by agents shows that they prefer to plant trees than crops as an ex-ante planning strategy for shocks. The second simulation experiment aims at examining the effect of long-run expected price changes, mainly on land-use decisions of agents in the model. Four future price scenarios are designed, and the results are compared with the baseline to examine the effect on agents discretionary income and land-use decisions. The purpose of this simulation experiment is to see if there is a deviance from croplands to woodlots and vise versa based on long-run changes in expected prices. Simulation results show that agents are highly responsive to changes in expected prices in the long run, except for the expected price of bamboo. In cases where there is a decrease in the expected price of acacia charcoal or an increase in the expected price of crops or both, results show that agents will go back to potatoes and wheat-dominated production systems instead of the acacia-dominated production system . This study suggests that supporting farmer adaptation to climate variability-induced shocks should focus on policy interventions related to crop and land management activities. Policy interventions should also focus on building the household asset base to boost farmers coping ability and resilience to shocks. Moreover, results suggest that robust climate adaptation and mitigation interventions should take the heterogeneity of farmers into account. Furthermore, both econometrics and farm-level simulation analyses show the importance of planting trees as a crucial adaptation strategy. Findings suggest that investment in woodlot perennials is an essential adaptation strategy for smallholder farmers with scarce resource settings and should be promoted and scaled to a broader area in the region.