Analyzing spatial hierarchies in remotely sensed data: Insights from a multilevel model of tropical deforestation

This paper advances an empirical model assessing how changing economic and ecological conditions at different spatial scales affect land conversion decisions. We apply a multilevel econometric model to explore the implications for parameter estimates and their standard errors of ignoring hierarchical groupings in the data. The paper draws on a panel of agricultural-household data collected from a survey of Mexican farmers. A comparison of results obtained from a standard single level model reveals several stark distinctions in the estimated effects, some of which have immediate relevance for conservation policy. We conclude that the multilevel specification is warranted for alleviating issues associated with error structures inherent to spatial data.