COMPASS Research Centre has worked on a number of projects involving microsimulation over the last decade. There have been evolutions in the data sets available, the population groups covered, the outcomes modelled, and the methods and software used to present and share the outputs.

Microsimulation is about seeing how an entire population might change over time, using a baseline sample and transition probabilities based on repeated observations from the same population. These data are used to construct models to describe how that population changes, and the main aim is to enable the testing of different scenarios: what happens if we change the underlying population in some way, or if we change the rules in those models?

We have used these methods to examine how the health system might be affected by the ageing of our population, how we might improve outcomes in the early life-course, and to understand more broadly how the New Zealand population progressed from 1981 to 2006.

Over the years we have benefited from increased openness and sharing of data, and different data sets have enabled us to specialise our analyses, for instance on health and wellbeing outcomes, and to generalise, especially with the advent of the New Zealand Longitudinal Census.

One of the most important evolutions has been in technology, and the software that has been available for us to use to share our outputs. We started with static simulation products constructed using SAS, and scenarios tested in-house with complicated setup. We decided to pioneer our own interactive solution, which was through a pioneering link between Java and R, but still required our own piece of software to use; and then this was superseded by R Shiny, an even more user-friendly visualisation tool with the added benefit of simple web-based delivery. End-users can see how our models work and can change the rules to test their own scenarios, without any need for a login or special software. We have come a long way.