Agent‐based modeling for ex ante assessment of tree crop innovations: litchis in northern Thailand

This study uses an agent‐based model for ex ante assessment of agricultural innovations. The model builds on whole farm mathematical programming (MP) and extends the methodology with a spatial representation of the system, the heterogeneity of farm households and landscapes, and the interaction between farm households. We apply the model in a northern Thai watershed to study the potential of four innovations to increase the profitability of litchi orchards. Cost‐benefit analysis shows that each innovation would increase the profitability of litchi growing; however, the results of the agent‐based model show that at current price levels these innovations alone would not be enough to stem the decline in the area under litchis. The model was validated and the sensitivity of the results tested for variations in the irrigated water supply and liquidity. We report on how farmers responded to these results and discuss the implications for other areas in northern Thailand.     

Agricultural pesticides and land use intensification in high,middle and low income countries

We study levels and trends in agricultural pesticide use for a large cross-section of countries using FAO data for the period 1990–2009. Our analysis shows that a 1% increase in crop output per hectare is associated with a 1.8% increase in pesticide use per hectare but that the growth in intensity of pesticide use levels off as countries reach a higher level of economic development. However, very few high income countries have managed to significantly reduce the level of intensity of their pesticide use, because decreases in insecticide use at higher income levels are largely offset by increases in herbicide and fungicide use. The results also show very rapid growth in the intensity of pesticide use for several middle income countries such as Brazil, Mexico, Uruguay, Cameroon, Malaysia and Thailand. Complementing our analysis with data from the Rotterdam Convention on Prior Informed Consent (PIC), we show that hazardous pesticides covered in the PIC procedure are more weakly regulated in lower than in higher income countries. We discuss the policy challenges facing developing countries with a rapid growth in pesticide use and recommend a four-pronged strategy, including an environmental tax on pesticides with revenues allocated to long-term investments in awareness building, the development of integrated crop management methods and the setting of food safety standards. The interactions between these measures should help contribute to the effectiveness of the overall strategy package.     

International research on vegetable improvement in East and Southern Africa: adoption,impact, and returns

There is a lack of evidence for impact at scale of vegetable research and development, although the importance of vegetables for human nutrition and smallholder incomes is generally understood. We therefore study adoption and impact of improved tomato and African eggplant varieties developed through international agricultural research, released by national agricultural research and extension systems, and supplied to farmers by private seed companies in East and Southern Africa from 1990 to 2014. The study finds that in 2014, varieties developed by the World Vegetable Center accounted for 50% of tomato and 98% of African eggplant commercial seed production in East and Southern Africa. For Tanzania alone, investment in crop improvement generated economic gains of US$ 255 million for tomato and US$ 5 million for African eggplant up to 2014. The internal rate of return is 26% for tomato and 12% for African eggplant, though we project the latter to increase to 26% by 2024 as the variety was released only in 2007. These findings support the view that agricultural policy and investment reoriented towards contemporary nutritional challenges will give high returns to investment.     

Agricultural commercialization of Karen Hill tribes in northern Thailand

Based on a random sample of 240 farm households in Chiang Mai province, Thailand, this study shows that—contrary to widespread belief—Karen farm households are well-integrated into markets. Average levels of market integration are 31% for gross farm output, 35% for variable inputs, 49% for food consumption, and 80% for net family income. By estimating a two-stage least squares (2SLS) regression model, this study finds that integration into output markets is positively associated with a diversification of land use away from rice monoculture, more intense contact with nearby urban centers, and a greater number of roads connecting the village to the outside world. Controlling for these factors, the distance to urban centers does not impede market integration; distant villages are equally well integrated into output markets. The study further finds that integration into output markets improves farm productivity and net per capita income. Concerns about market integration are discussed. Results have implications for Thai policy makers who have recently placed increasing emphasis on the concept of "sufficiency economy" in order to promote the well-being of rural people.     

The Diffusion of Greenhouse Agriculture in Northern Thailand: Combining Econometrics and Agent-Based Modeling

This paper studies the diffusion of greenhouse agriculture in a watershed in the northern uplands of Thailand by applying econometrics and agent‐based modeling in combination. Adoption has been rapid by farmers in the central valley of the watershed, while farmers at higher altitudes, lacking transferable land titles that could serve as mortgage collateral, have been unable to obtain loans for greenhouse investment. The objectives of the paper are both methodological and empirical. On the methodological side, it shows that econometrically estimated models of farm household behavior are useful to design and to parameterize an agent‐based model. On the empirical side, simulation results show that if mortgage collateral would not be required, then adoption in the upper part of the watershed could reach nearly 77% of farm households by 2020, as compared to about 36% under current conditions. Furthermore results suggest a significant increase in incomes related to the innovation and a substantially greater irrigation water use, especially in the central part. As bell pepper under greenhouses has replaced pesticide‐intensive chrysanthemum, it has declined average levels of pesticide use. Nevertheless, pesticide use is high and farmers are struggling to control pests, which raises questions about the long‐term sustainability of the innovation. Dans le présent article, nous avons analysé, à l'aide d'un modèle économétrique et d'un modèle multi‐agent, l'expansion de la culture en serre dans un bassin versant des hautes terres du Nord de la Thaïlande. Les agriculteurs de la vallée centrale du bassin versant ont adopté rapidement cette forme d'agriculture, tandis que les agriculteurs installés dans les hautes altitudes n’ont pu, faute de titres fonciers transférables pouvant servir de garantie, obtenir de prêts pour construire des serres. Les objectifs du présent article étaient à la fois méthodologiques et empiriques. Sur le plan méthodologique, notre étude a montré que les modèles de comportement des ménages agricoles estimés économétriquement sont utiles pour concevoir et paramétrer un modèle multi‐agent. Sur le plan empirique, les résultats de simulation ont montré que, si des garanties de prêt n’étaient pas exigées, 77 p. 100 des ménages agricoles adopteraient la culture en serre dans les hautes terres du bassin versant d'ici 2020, comparativement à environ 36 p. 100 dans les conditions actuelles. De nouveaux résultats ont indiqué que cette innovation ainsi qu’un usage accru de l'eau pour l'irrigation, particulièrement dans la partie centrale, pourraient générer une hausse substantielle des revenus. Depuis que la culture en serre du poivron vert a remplacé la culture du chrysanthème exigeante en pesticides, l'usage des pesticides a beaucoup diminué, mais demeure tout de même élevé. Les agriculteurs ont de la difficultéà lutter contre les ravageurs, ce qui soulève des questions sur la viabilitéà long terme de l'innovation.     

Participatory computer simulation to support collective decision-making: Potential and limits of stakeholder involvement

System models in agriculture and natural resource management are usually developed by researchers to advise policy makers on the likely outcomes of alternative scenarios. Except for data collection, stakeholders—like farm households—are rarely involved in the research process. Companion modeling (ComMod) has been developed as a modeling approach to include various stakeholders in a continuous feedback loop between researchers and stakeholders. Whereas other ComMod approaches have used role playing games as an interface between researchers and stakeholders—assuming that stakeholders have difficulties understanding a computer model—this paper explores the potential of a participatory modeling approach in which stakeholders are directly confronted with the model by assessing its assumptions, interpreting simulation results, and suggesting scenarios.The approach is applied to two villages in a watershed in northern Thailand. One lowland village, populated by farmers of mostly Thai ethnic origin, depends partially on an upland village inhabited by farmers of Hmong ethnic origin for its water supply. Water scarcity has led to conflicts between these two villages in the past.In three rounds of participatory simulation sessions the potential of the ComMod model was tested. Our findings confirm that researchers face particular challenges in making local stakeholders understand the model as a reproduction of reality and not as reality itself. The idea of a scenario as a hypothetical situation was also difficult to convey. An ex-post analysis among participants showed that farmers who attended several sessions had a clear understanding of the model and one-third of the participants thought it useful to exchange points of view with the other community and to define new collective rules for water sharing. Applying ComMod in a context that is characterized by social tensions and power differentials, however, needs careful consideration of the potential implications on less powerful actors.     

Simulating soil fertility and poverty dynamics in Uganda: A bio-economic multi-agent systems approach

Declining soil fertility and increasing rural poverty are major problems facing sub-Saharan agriculture. Bio-economic modeling has been used to analyze the complex interaction between ecological sustainability and rural poverty as well as to explore policy options promoting sustainable development. This paper shows that these models can be further advanced by adopting an agent-based modeling approach. This gives a more realistic representation of diversity in socioeconomic and biophysical terms, allows local interaction between households, and can yield an ex-ante assessment of the distributional consequences of policy intervention. This paper describes the modeling approach and illustrates it with an empirical application to two village communities in the Lake Victoria Crescent of Uganda. It is shown how the model system can be calibrated with and validated against empirical data. The model is used to analyze the potential effect of short-term credit, mineral fertilizer, and improved maize seed on poverty and sustainability. Simulation results suggest substantial reductions in poverty although the incidence of poverty would remain high and these innovations alone would have little effect on the long-term ecological sustainability of the system.