Econometric models for spatio-temporal count data

To contribute to a better understanding of the fundamental process behind the spatial and temporal correlation as well as to describe the resulted dynamics, sometimes still less reflected in econometric models, is the aim of this study. It is intended to improve the development of the necessary economic analysis which allow to optimize management policies in the most diverse areas of activity, be it hospital, road or other. This work develops and applies econometric models for count data with dependencies in space and time. The existing models are often based on the Gaussian assumption, which is sometimes inadequate. It is interesting to extend it to other types of distributions, generalizing the applicability of the available models and accompanying this development with estimation methods that make them useful. Bayesian spatial autoregressive and hierarchical models are considered as alternatives to the aforementioned models, since they are a valid and flexible alternative in the modeling of spatial effects. Spatial and spatio-temporal versions of autoregressive Bayesian models are proposed, establishing the same mathematical framework for autoregressive and hierarchical models for counting data. This is an area still underdeveloped within econometrics, given the associated but necessary complexity, and it is essential to quantify the advantages and disadvantages of its use. For the proposed methodologies, it is considered its application and implementation, in several areas of activity with scientific and technological interest, namely in the health area. In this context, a study of hospital management data is carried out, specifically the calls for the national health care line, Saúde24, in order to the development of indicators for decision support, evaluation and implementation of management and government policies, as well as to the prediction of future behavior under different scenarios. Another application in the area of road safety is also considered