Basin-Scale Econometric Modeling III: local interactions in deforestation decisions
Juan
Robalino, Columbia University, jar101@columbia.edu
Alexander
Pfaff, Columbia University, ap196@columbia.edu
(Presenting)
Claudio
Bohrer, Universidade Federal Fluminense, bohrer@vm.uff.br
Like other activities, an individual choice to deforest may be influenced by the choices of others. Many people will choose, for instance, to go to less crowded shopping centers. In forest clearing, individuals may be more likely to deforest their land, or less, if their neighbors are also clearing.
Such potential spatial interactions can affect the quantity and the spatial pattern of forest and thus affect the ecological services from forest (sequestration, habitat). Further, these external effects affect the social efficiency of land use, including via the potential for an inferior land equilibrium.
This presentation is the third one to report upon the earliest basin-scale research within the project “A Basin-Scale Econometric Model for Projecting Future Amazonian Landscapes”. This talk will focus on the empirical identification of such spatial interactions from the observed forest clearing. The crucial issue for identification is controls for other reasons for spatial correlation in clearing. Given such controls, we find evidence that deforestation decisions locally reinforce each other.
While such results have implications for policy, at this point we are demonstrating an approach. Care in identification of interactions is required given the likelihood that both spatially correlated unobservable drivers of deforestation and direct effects of neighboring characteristics are present. Both of these provide explanations other than interactions for spatial correlation in deforestation. However, appropriate instrumentation for neighboring clearing choices eliminates these stories. We discuss appropriate instrument choice and provide robust evidence of positive interactions.