This proposal is concerned with two research lines: spatial econometrics and worst-case estimation. Spatial Statistics is a set of techniques for the analysis of spatially located data.
The main characteristic of spatial data is that the nearer observations tend to be more dependent. When these techniques are mainly concerned with economic problems, they are know as Spatial Econometrics. A large proportion of the Spatial Statistics work is appearing outside statistical journal (see e.g. Cresses, 1993 Preface). In fact, some inference techniques seem to be heuristic and could be improved. One of the aims of these research projects is to develop a wide range of statistical techniques for the analysis of spatial data, providing a formal mathematical basis for their use. This research project extends previous work of the applicant, supported by a Marie Curie Fellowship. The previous research focused on spatial data regularly sampled on a network. In this project we stress the estimation of random fields aggregated in small areas. An additional aim is the development of estimation techniques for econometric models with absence of empirical information about some of the involved variables. We propose a worst-case approach that is robust against the worst possible outcome of the unobserved variables. This methodology could be especially useful in management modelling, where firms often do not have available all the information required to estimate a decision model.
Call for proposal
See other projects for this call