Date & time
John Gibson, Susan Olivia and Bonggeun Kim
University of Waikato, University of Melbourne and Seoul National University
Many household choices may be correlated with choices made by nearby households, especially in developing countries where residential location strongly affects economic activity. Such effects could reflect either spatial errors which arise because nearby locations have unobserved factors in common (e.g., infrastructure quality) or spatial lags because of interactions between households and their neighbours (e.g., learning about new technology, coordination problems when switching from farm to non-farm production). Survey data usually do not give precise measures of location so it is not possible to fully account for this spatial autocorrelation. In this paper we use data from Indonesia that allow exact distances between each household to be measured. These distances are used to estimate spatial effects in equations for the non-farm enterprise share of household incomes so that any bias and inferential errors from ignoring such effects can be assessed.
Keywords: Autocorrelation, Non-farm enterprises, Spatial statistics, Indonesia