Résumé
Changes to the global agro-food-energy system (e.g. changing consumption patterns in the North (SNF, 2012), Europe's Climate and biofuel policies, etc.) over the past few years have led to a renewed interest in agriculture and a rush to acquire land (Cotula, 2012; Anseeuw et al, 2013). The impacts of this rush on sustainability are not always evident as its assessments focus on the short-term and generally remain at a case study level, without considering the broader agrarian and socio-economic transformations it entails (Borras et al. 2012). If a consensus emerges regarding the necessity of additional investment into agriculture (FAO, 2010), it is less evident whether large-scale agricultural investments (LAI) are a vector for broader agrarian and socio-economic transformations in a sustainable manner (Borras et al. 2012, Deininger and Byerlee 2011; Collier and Dercon 2014). Despite a growing literature (World Bank, 2010; White et al., 2012, Cotula 2014 etc.), most assessments of LAI impacts tend to remain local, in the form of specific case-studies, and are often short term without broader contextualization (Fairhead et al., 2012). Efforts to overcome these limitations through different types of meta-analyis have been undertaken (Oberlack et al., 2015, Schoneveld 2014, Schoneveld 2017, Dell'Angelo et al. (2017). However, a more empirical understanding of the diverse changes and impacts at various levels is necessary for reflecting on visions for the planetary land system. Against this backdrop, this paper presents the results of a study aiming, on one hand, at assessing the changes and impacts of LAIs at various (individual, household, regional) levels within target regions, and on the other hand, at a nuanced account of how and why LAIs subsequently induce diverse regional development trajectories in these regions. We focus on LAIs in Kenya, Madagascar and Mozambique. Specifically, this study provides a cross-national comparative analysis of business models, land-use changes, governance dynamics of LAIs and their socio-economic, food security, and environmental impacts in Kenya, Madagascar and Mozambique. It brings together the individual results on these aspects, which were generated in the Afgroland project (www.afgroland.net). The following research question guides this analysis: How do contextual and institutional nuances of large-scale agricultural investments impact on land-use changes, the organization of production and investment processes, socio-economic outcomes, food security, and the environment in LAI target regions in Kenya, Madagascar and Mozambique? Methodologically, this study utilizes a set-theoretic methodology for a case-based comparative analysis. It responds to calls for the use of robust empirical methodologies to provide reliable evidence on the impacts of LAIs and to expand the use of comparative methods to attribute LAI impacts to causal factors. Data were collected in six study areas in the three countries by means of household surveys with more than 1500 households, more than 200 key-informant and in-depth interviews with business managers, policymakers, households, development agencies, and NGOs; remotely sensed data between 2016 and 2018, and complemented with document analysis. Data analysis involved mixed qualitative and quantitative techniques. A first set of tentative results, more conceptual in nature, show that LAIs induce regional development trajectories with sustainability impact patterns that can be characterized as conflictual sustainability trade-offs; employment vs. land access and environment trade-offs; widespread hostility; or moderate impacts. The set-theoretic analysis shows that the operational farm size, labour intensity, experience in local agriculture or domestic origin of investors, and prior land uses have the most significant impact on land-use changes, evolution of business models and adaptation of governance systems:. These transformation patterns are described in detail in the paper. A second set of results shows how the same international drivers can have divergent impacts, with local-level outcomes which can differ significantly in terms of land use change, ecological impacts, food security, and livelihoods. These divergences are determined by national politics and policy frameworks, land tenure rights, business models, land and water resource endowments, and path-dependencies regarding investment and business practices. As such, in Kenya, and more particularly in the Nanyuki region characterised by longstanding LAIs, an agrarian normalisation process has established, based on labour intense production systems mainly in the flower and horticultural sectors. Better established labour rights, technology transfer and an agrarian sector that has developed over time leads presently to a relatively dynamic local economy with subsequent livelihood opportunities. In Mozambique, these regional dynamics are minimal however, albeit indirect through basic infrastructural and service development. On contrary, through land loss and increased land pressures, labour extensive crops and production models, and not well developed labour rights, major fractions of the local populations tend to be affected negatively. Lastly, in Madagascar, for the few investments that are still operational, they tend to function on an enclave model, with very little – if any – interactions and impacts on the regional economy and local populations. The paper concludes by repositioning these results in the broader framework of interactions among sustainable development goals (SDGs), representing a critical, but mostly overlooked aspect in the debate on LAIs. In policy debates, LAIs are frequently justified with the argument that LAIs would create new flows of investments to capital-poor regions; create new employment; enhance agricultural productivity. In other words, this narratives relates LAIs positively to SDG10.B and SDG17.3 (investment flows), SDG8.5 (employment), and SDG2.3 (agricultural productivity), among others. By contrast, a recent review finds that LAIs can affect 14 of the 17 Sustainable Development Goals of the UN 2030 Agenda for Sustainable Development in adverse ways. The results of this study and the analyses in terms of diverging development trajectories induced by LAIs allows to assess how LAIs shapes the interaction between multiple SDGs. These interactions among SDGs in the framework of LAIs can take the forms of trade-offs, co-benefits, and co-damage.