LC-24  Abstract

A Basin-Scale Econometric Model for Projecting Future Amazonian Landscapes

Eustáquio J Reis, IPEA/DIMAC (SA-PI)
Robert T. Walker, Michigan State University (US-PI)

The proposed project seeks to develop a basin-scale econometric model for use in predicting land cover change scenarios in the Amazon basin.  It will combine the region-scale of the GIS-based approach, with the behavioral rigor of the micro-level studies, in constructing a model that can predict land cover change over large areas with theoretical grounding, and empirical testing.

Three key issues will be addressed in the research in order to improve upon existing aggregate models for the Amazon.  First, the model will not only incorporate economic theory for individual decisions, but also theory related to macro-level effects on land cover, as generated by state policies. Second, the model will account for spatial heterogeneity in the determinants of land use across the region by drawing on empirical research from multiple sites with contrasting settlement histories. And third, in both estimation of the effects of land-cover drivers and landscape projections, “endogenous” road construction following initial deforestation along major roads will be accounted for.  These three innovations will allow an improved econometric model from which coefficients used in projections are based on observed historical deforestation in the basin. This empirical approach also facilitates the evaluation of confidence intervals for coefficients and comparison of model predictions with out-of-sample new data.

The proposed research thus addresses the LBA-ECO theme “land cover and land use change.”

By focusing on construction of an econometric statistical model of deforestation for the entire Amazon basin, this analysis seeks to help answer the second-tier NASA research question (F2):

“What changes are occurring in the global land cover and land use, and what are their causes?”

It also seeks to address the related question (LC-Q1):  “What are the rates and mechanisms of forest conversion to agricultural land uses, and what is the relative importance of these land uses?”  In addressing such questions, the proposed project will provide projections of land use, with error bounds, contributing to understanding of future land-use and vegetative-cover scenarios for the Amazon basin. That could also prove of value for second-tier issues such as (P5) regarding the “reliability of models predicting future atmospheric concentrations of carbon dioxide and methane.”  The research will involve key informant interviews at several field sites to address the issue of road endogeneity, satellite image interpretation to provide data for dependent variables that will be used in statistical modeling, and specification and estimation of an econometric model.  In addition, an explicitly spatial model will be developed for implementation in the Santarém/Tapajos sub-region.