Small-area estimation using a generalised regression procedure

Abstract

This thesis describes a process of small area estimation which has been used in Australia to develop estimates and projections of poverty and housing stress for small areas, and which has been linked to a Tax/Transfer microsimulation model to estimate the effects of a policy change for small areas. The thesis consists of a literature review in the area of spatial microsimulation; the five journal articles as they are published; and then a concluding chapter. The model is described in detail in one journal article, and a number of applications are presented in three other journal articles. These articles also show how the model has been validated. A fifth article shows how the model has been tested and further developed in a number of ways. The thesis concludes that the model developed is a useful addition to a microsimulation modelling toolkit, especially given the need to investigate social characteristics for small areas. The method can be used by Government's and researchers to derive not only small area estimates of a number of variables, but also small area effects of a policy change; and small area cross-tabulations for a number of variables. This flexibility of the spatial microsimulation means it is a very powerful approach to small area estimation, which can traditionally only provide point estimates for certain variables.

Date of Award	2011
Original language	English
Awarding Institution	University of Canberra
Supervisor	Ann Harding (Supervisor), Alan Duncan (Supervisor) & Paul E. Williamson (Supervisor)