The determinants of global tropical deforestation

This thesis develops an econometric model of global tropical deforestation with special attention to problems in earlier studies related to hypothesizing the effect of explanatory variables, deforestation data, and estimation techniques. The issue of hypothesis testing is addressed by hypothesizing that different underlying causes used to develop a theoretical deforestation model may have both positive and negative effects on deforestation, the net effect of which will vary in different circumstances. The data problem is dealt with by categorizing the deforestation data. Multinomial logistic regression results, used to address some of the estimation issues, are compared with those from OLS and binary logistic regression methods. Multinomial logistic regression was found to be more informative than the other two approaches. While growth in population, agriculture, and road construction caused deforestation in high deforesting countries, debt service growth replaces population growth as the cause of deforestation in countries experiencing medium deforestation levels.