Stata in Space: Econometric Analysis of Spatially Explicit Raster Data

Abstract

Realizing the importance of location, economists are increasingly adopting spatial analytical and spatial econometric perspectives to study questions such as the geographical targeting of policy interventions, regional agglomeration effects, the diffusion of technologies across space, or causes and consequences of land-cover change. Explicitly accounting for location in econometric estimations can be of great benefit for researchers working at the interface of economics or environmental sciences and geography. The objective of this article is to demonstrate how spatially explicit raster data derived from standard geographical information system (GIS) software can be used within Stata. Three programs implemented as ado-files are presented. These import geographic raster data into Stata (ras2dta), draw systematic spatial samples within Stata (spatsam), and export data and estimation results in a form usable by standard GIS software (dta2ras). A numerical example is presented to estimate the determinants of forest cover with a spatially explicit logit model, calculate predicted probabilities, and map the predictions with GIS software.