Agent-Based Assessment of Future Demographics and Impact on Infrastructure Transition Needs

Abstract

Most World countries are currently facing an important transformation in demographic structure due to the overall ageing of the population, additionally in the recent years an extreme increase of mobility, both physical and economical, has been observed, drastically increasing migration flows and accelerating the transformation of demographics. The changes in demographics are one of the primary sources of uncertainty for infrastructure developers and policy makers; different demographics have different behaviours and requirements, affecting the demand for infrastructure, services, housing, and impacting differently the social security system. Thus it is essential to understand the mechanisms and quantitatively assess the future evolution of demographics, accounting for the sensitivity to multiple interdependent factors in a holistic way. Typically the impact of demographics on infrastructure is assessed based on projections of past trends. This approach is not sufficient to predict effects of demographic change on infrastructure, since not accounting for changes in behaviour, due for example to ageing or changes in household structure. Recently, agent-based simulations are being used, but being computationally demanding are often limited to the observation of individual elements of behaviour at small scale. In the course of this work for the first time ever, a fully-coupled, high-resolution (1 meter), continental-scale agent-based model for population, dwellings, and jobs has been developed. The interlinked models are designed to run on a GPU framework, allowing for fast computations, making it possible to simulate the whole EU in less than 1 hour per simulated year. The results show how in Switzerland and EU migration is the primary driver for changes in demographics. Difference in demographics strongly affect development of housing within urban agglomerations, while young immigrant population contributes to the support of social security for Switzerland: if the annual immigration rates are limited to 50% of today’s rate, the social security system will become insolvent by 2027. Impact of different population densities are also observed in relation to public transport and disease transmission, the studies show also the effect of densification on commuting patterns and time, increasing up to 20% where new urban centres are established. Finally the simulations of the whole European population in the context of Brexit shows the re-distribution of migration flows towards UK, France, and Germany, while simultaneously future migration will decrease by 38%, posing important challenges in terms of infrastructure, support for social security and availability of workforce. In general, it is clear why today’s concerns related to immigration revolves towards overcrowding, but in future migrants will become a valuable asset to provide for social security support and workforce for most European countries. Thus, it is important to focus on the typology of migrants and favour the creation and growth of new families, to make the future more sustainable with respect to demographic ageing. In conclusion, it has been demonstrated how agent-based models can be an effective, reliable, and sensitive tool for the assessment of multi-scale infrastructure and policies performances. The flexibility and consistency of the outputs required by the holistic framework allow for extensive case studies, while the GPU-acceleration broke the traditional limits of these models, allowing for detailed continental-scale simulations.