The geographical distribution of the consumption expenditure in Ecuador: Estimation and mapping of the regression quantiles

The real consumption expenditure of families provides important information about the welfare of people residing in a given administrative area. State policies designed to alleviate poverty and funding management plans rely on the ability of statistical models to provide detailed and correct information. It can be argued that mean regression alone does not provide a satisfactory picture of the distribution of the response. We explore the use of nonparametric quantile regression for geographically referenced data. The motivating example pertains the distribution of the consumption expenditure in Ecuador, whose shape, conditional on some predictors, varies across the locations and reveals that the spatial heterogeneity has a very different impact on the quantiles of the response.