Abstract:We revisit the long-standing hypothesis that the process of human development and land clearing in Amazonia follows a boom-and-bust (inverted U) pattern, where early clearing leads to a socioeconomic 'boom' which then turns to 'bust' after the deforestation process has matured. Although the hypothesis has found some empirical support in cross sectional data, a handful of longitudinal case studies have failed to identify incidences of 'busts.' We show that the cross sectional results are a spurious artifact of spatial correlation, driven primarily by the large, multifaceted (and unobserved) differences between municipalities in the states of Amazonas and Maranhão. Furthermore, using new panel data on the Human Development Index (HDI) and deforestation rates from 1991 to 2010 we find no evidence of such boom-bust patterns in the time series. Municipalities categorized as either 'post-frontier' or 'pre-frontier' in 2000 enjoyed equal increases in HDI over the subsequent decade as the rest of the Amazon. Panel data analysis with fixed effects (within estimation) robustly rejects the hypothesis that HDI and deforestation follow an inverted-U relationship.
Acknowledgements:We thank Toby Gardner, Daniel Phaneuf, Philippe Delacote, Simon Dietz, Lykke Andersen and two anonymous referees for useful comments and suggestions. We also thank Ana Rodrigues and her coauthors for sharing the original dataset of Rodrigues et al.