Modelling farm behaviour on soil fertility with the policy variable: a case from Africa Gian L. Nicolay & Louis Chikopela & Boubacar Diarra & Andreas Fliessbach Received: 31 March 2020 /Accepted: 15 April 2020 # Springer Nature B.V. 2020 Abstract This paper describes a new approach to make predictions on the behaviour of farms in Africa related to soil fertility management. Not only sectorial factors, but also the larger socio-economic context including policies influence the behaviour of small-scale farms. Science does not yet understand this context due to its vast heterogeneity, contingencies and complexity. We collected and processed qualitative data and sociological parameters to transform them into numbers. The model we constructed is based on probabilistic estimates of behaviour of farm classes under two scenarios in Mali (Mafèya) and Zambia (Chipata). We propose seven distinct farm classes to simulate the likeliness for a change from one class to another under defined policy regimes and other social conditions. In real life, network of actors, institutions and other social formations couple and decouple farmer’s identities and farms in highly dynamic social and ecological processes. We construct- ed a simplified model based on selected social theories, interpretative sociological inquiry and Markov chains in order to allow simulations under the two policy scenarios “as-is” and “to-be”. Our simulations result in significantly different outcomes per locality and scenar- io. This approach allows practical simulations of farm, food and agriculture systems and comparative research. We expect a better understanding of the dynamics of farms, faster adoption of innovations and a better base for a research-led dialogue of practitioners and policy makers. The paper demonstrates the primordial role of policies, influencing directly farmers’ behaviour, rural and labour markets as well as food systems and rural development. Keywords Soil fertility management . Smallholder farm behaviour . Social networks . Policies . Africa . Research technology . Modelling . Predictions Introduction and context Our aims are to improve the predictions of farm and farmer behaviour and to address issues of soil fertility management, climate change adaptation and food secu- rity. Scientific predictions are causal or explanatory statements about a certain set of variables (Merton and Nisbet 1976, p. 739). Our scientific stance is sociology, meaning that we observe the phenomena mainly through the lenses and with the concepts of this partic- ular scientific discipline and epistemology. The socio- logical method here applied treats its particular phenom- ena like mind, mind formations, knowledge, social for- mations as in principle deterministic realities, which science can explain as any other natural or mental Org. Agr. https://doi.org/10.1007/s13165-020-00293-4 G. L. Nicolay (*) : A. Fliessbach Research Institute of Organic Agriculture (FiBL), Frick, Switzerland e-mail: gian.nicolay@fibl.org L. Chikopela Department of Agricultural Economics & Extension, University of Zambia, Lusaka, Zambia B. Diarra Institut supérieur de formation et de recherche appliquée (IFRA), Katibougou, Mali