Résumé
A large number of countries have initiated similar sets of policies aiming to increase forest cover. These have usually included large- scale afforestation campaigns and the devolution of land property rights to households. Most of the research works that have analysed the link between state policies and land-use change have hitherto been restricted either to qualitative local level studies or to quantitative macro-scale analyses. The former have offered an in-depth understanding of the drivers for farmers decisions but their applicability to different local contexts is questionable. The latter have identified general trends and proximate causes for reforestation but often without being able to explicitly separate the effect of a particular policy or to ascertain the causal mechanisms that link policy and land-use change. Using the Institutional Analysis and Development (IAD) framework as a unifying canvass, our assessment of the impact of forest policies on reforestation in Vietnam investigates several governance levels and uses both quantitative and qualitative approaches. We started from the analysis of farmers land use decisions at the local level relying on ethnographical methods and institutional analysis. Then, we used these findings to build models of forest cover change, which were tested at the meso-scale level using remotely-sensed data and spatial regression analysis. This quantitative study was complemented by an institutional and political economy approach that explored the underlying drivers for reforestation and policy implementation at the provincial level. Finally, a discursive and political ecology perspective allowed us to analyse the role of the prevailing narratives and beliefs in policy design at the national level. We discuss in this paper why this multi-level and holistic methodology is particularly effective for identifying the links between policies and forest cover change and for understanding the discrepancies that exist between policy intentions and observed outcomes. Lastly, we argue that this approach is also particularly well suited for designing and effectively disseminating appropriate policy recommendations.