Originally envisioned by the American economist Guy Orcutt in 1957, microsimulation models have become widely used within economics, and are increasingly used in public and population health and other fields to model the outcomes of government policy.
While more complex and requiring more data than simpler arithmetic and representative agent models, microsimulation models use individual-level data to construct a group of individuals to represent a target population and their behaviours. Then the model simulates, using known probabilities, changes in individual’s characteristics over time.
Microsimulation is extremely useful for modelling the impact of counterfactuals i.e. ‘alternative versions of the future’, such as the impact of proposed government policies or impacts of new medical technologies (e.g. statins), upon the population, and examining the impacts on different subgroups of the population, especially when these groups are heterogenous and difficult to partition into smaller homogenous groups.
Microsimulation can be ‘static’, i.e. a snapshot at a single point in time, or track the changing behaviour inside a dynamic model as the environment and individual change over time. This is particularly important when modelling chronic disease, where a person’s history of a risk factor matters for their future likelihood of disease.
Over the past few years the use of microsimulation in health has continued to expand beyond helping shape government policy into commercial applications in the private sector. For example, applying microsimulation modelling in cost-effectiveness studies associated with clinical trials, to simulate the outcomes. This has the potential to assess the clinical and economic impact of prevention strategies or treatments before expensive, time-consuming implementation in the “real world”.