High resolution geographical and sub-population projections are essential for the planning and delivery of services and urban infrastructure developments. SPENSER (a synthetic population estimation and projection model) uses dynamic microsimulation to produce projections under different, user defined scenarios.
SPENSER will make high resolution demographic forecasting accessible to stakeholders across a range of application areas, from physical infrastructure planning to health and social care spending, enabling users to run ‘what if’ scenarios and facilitating evidence based planning decisions.
Explaining the science
At the core of the model is an individual level synthetic population which can be simulated through time. This synthetic population is created using spatial microsimulation. The detailed attributes from a sample of people represented in survey and commercial datasets are merged with a complete, but less attribute rich, population from a census to create a synthetic dataset of individuals. This synthetic population is attribute rich and also has a spatial identifier, allowing for small area estimates and projections to be produced. Additional sources of data can be used to build custom populations for different application areas, for example health survey data to create a disease model or expenditure survey data to create a consumer demand model.
Individuals from the synthetic population are introduced in to a dynamic microsimulation model which is used to age those individuals over time and to estimate the occurrence of events throughout their life using transition probabilities in a Monte-Carlo simulation. These transitions are drawn from longitudinal (usually survey) datasets and can represent a wide range of demographic, economic, social and health events. For example the likelihood of having children, developing chronic health conditions, changing job or claiming a pension.
Results can be aggregated to different geographies or sub-groups of the population to investigate the impact that infrastructure developments or policy interventions might have on the population across a range of scales. This allows users to produce ‘what if’ scenarios for a range of application areas within a single framework.
Estimating and projecting populations for small areas and for specific sub-groups of the population requires careful model development and is computationally intensive. As scenarios are added, based on assumptions about variables which could change in the future (e.g. economic, political, health or educational attainment), the challenge increases substantially. These scenarios are very important for the utility of population projections, as they allow users to test a range of possible future scenarios and the impact that they have on the phenomenon of interest.
Providing quantification of the uncertainty surrounding projections helps planners to make informed decisions. Researchers usually rely on existing projections (e.g. those produced by National Statistical Agencies), which provide no scope for customisation and are generally not of high enough resolution for many applications. SPENSER will make demographic forecasting accessible to a wide range of stakeholders.
Development of SPENSER is threefold:
- Design of a user-friendly interface for the model.
- Implementation of a dynamic microsimulation model which translates user input and underlying data into a scenario projection.
- Experimentation with innovative visualisation of results, to include interactive maps and plots.
SPENSER will provide a robust framework for testing a wide range of socio-demographic scenarios and their impact on population change. Outputs will be of use to other Turing partners, business (in relation to demand modelling) and government (in relation to resource allocation).
High spatial resolution demographic projections are needed to inform the development and delivery of urban infrastructure projects, for example housing, road, rail, digital communications, water and others which are essential for business and society.
Simulations of individuals over time can be used in the planning of public and private service delivery and in the experimentation of policy interventions. For example, when estimating the demand for health care provision or for pensions, and testing how different policies might impact on that demand in the future.
Consumer demand can also be modelled, for example predicting future expenditure and consumption patterns given projected demographic change.
Nik Lomax – CDRC, University of Leeds and the Alan Turing Institute.