Unit level spatio-temporal models

This chapter is dedicated to the unit level spatio-temporal models, in the context of the empirical best linear unbiased predictors (EBLUP) estimation or small area model-based estimates. Spatial models consider a contiguity matrix to describe the neighborhood structure between small areas. In the field of unit level spatial models, the standard procedure is to incorporate spatial dependence in the random error structure through a simultaneous autoregressive (SAR) process. State space models are one of the most important tools for time series analysis. When there is a multiple set of surveys that are repeated in two or more time instants, the state space models analysis allow us to take advantage from the interpretation of the updating information process, collected by the repeated measurements on units in the small areas. The chapter also talks about a study where the Italian EU-SILC data was applied with the spatio-temporal unit level models.