Résumé
This paper analyses the impact of several avoided deforestation policies within a patchy forested landscape. Central is the idea that one neighbour's deforestation actions may impact the returns to deforestation in nearby patches. We determine the impact of each policy in terms of avoided deforestation and leakage levels at the landscape scale through modelling and simulations. Avoided deforestation policies at a landscape level are respectively: two Payment for Environmental Services (PES) policies, one focused on deforestation hotspots, the second being equally available to all agents; a conservation area; and, an agglomeration bonus. Because our model accommodates spatial interactions in the absence of a deforestation policy, it is possible that a spatial policy can affect both within-intervention areas and outside-intervention spatial spillovers in terms of leakage across different landowners’ forest patches. These two different elements of the total extent of displacement across the full landscape have not been considered before. Our contribution is twofold. In terms of methodology, we expand the concept of leakage in accounting for direct impacts to adjacent patches and spatial spillovers over the landscape, and we provide a measure of leakage in a dynamic manner for policy assessment. From our analytical model and simulations, we show that leakage is sensitive to the spatial distribution of forest patch types. The two PES policies are the most cost-effective policies regarding avoided deforestation. The agglomeration bonus policy is efficient at the expense of a higher cost, whilst the conservation area policy is efficient when patches with similar characteristics are gathered.