Abstract:
The use of spatial microsimulation by researchers and policy analysts around the world has increased with many viewing it as a viable way to work around difficulties in gaining access to data for small study areas, as well as in those situations when no public domain data is available for analysis. For example, some studies have simulated income from other information related to the variable that is being synthesised because no income data was available. Advances in computer processing power have meant that the option of building a spatial microsimulator to create synthetic populations and other datasets has become more feasible. However, validation of the results from this technique has proven challenging in terms of assessing the accuracy of results. Therefore, the synthetic data generated by microsimulation could have limited utility as a source of reliable information. My research has two main objectives. The first is to assess the accuracy of outputs from spatial microsimulation by simulating median household incomes for the census area units (CAU) on the Auckland isthmus, and comparing the synthetic data with the actual median household information referenced from the 2001 New Zealand Census of Population and Dwellings. My second goal is to perform a proof of concept study. I demonstrate this by utilising the simulated population, i.e. person and household information of members of the synthetic population, in order to derive a measure of deprivation for the CAUs. This output is then can be compared with the NZDep indices of deprivation for matching accuracy. The results from testing the accuracy of synthetic incomes highlight that there are certain factors that can impact on how well the estimates reflect the actual data. My assessment indicates that it is most important that the microdata be as representative of the population as possible in order to successfully provide accurate synthetic income. In assessing whether spatial microsimulation can provide more in-depth information about deprivation, this initial study has demonstrated that with further development, the technique can provide more information about what particular combination of deprivation measures distinguish areas which have the same NZDep decile score.