| Most previous cross-national studies of deforestation have been criticized for being largely atheoretical. While these studies provide some initial insights into deforestation, the absence of theory is problematic because the choice of variables for models remains unguided in that variables are included according to data availability or other ad hoc reasons. I address this concern by conducting an empirical analysis of deforestation informed by five different perspectives using the stochastic impacts (STI) by regression ( R) on population (P), affluence (A), and technology (T) or STIRPAT analytical framework. In doing so, I include variables not taken into account in previous research but theoretically relevant to any study of deforestation. These measures include democracy, international non-governmental organizations, and political protests. Initially, these variables do not explain a significant amount of variation in deforestation. However, subsequent analyses incorporating interaction terms suggest international non-governmental organizations and political protests reduce deforestation more in democratic nations than in repressive nations. Analyses also reveal that export partner concentration, commodity concentration, multinational corporate penetration, and International Monetary Fund conditionality increase deforestation more in repressive nations than in democratic nations. I increase the validity and reliability of the findings by estimating the models with three missing data techniques including listwise deletion, group mean substitution, and full information maximum likelihood estimation. Similarly, I use a variety of different model specifications. The findings remain stable and consistent regardless of the method for handling incomplete data and the indicators included in the models. |
