Who burdens the welfare state? Migration and ageing in housing, education, and healthcare demand

Abstract

Political discourse attributes the pressure on European welfare systems to foreign nationals. Yet projections of service demand rarely disaggregate service demand by citizenship status. We develop a structural demographic model and project healthcare, education, and housing demand in Austria through 2050, disaggregated by citizenship status and regions across migration scenarios. We find that migration, ageing, and fertility shape each sector differently. In healthcare, the ageing of Austrian nationals contributes 4.7 times more to demand growth than immigration, with the most acute pressures in rural, low-migration regions. In housing, migration accounts for the entire net growth in demand, concentrated in metropolitan hubs. In education, aggregate demand contracts regardless of migration assumptions, whereas future needs are driven more by the births of foreigners in Austria than by new arrivals. Foreign nationals consume services in proportion to their demographic weight, with deviations explained by age structure rather than over-utilisation. These results show that the drivers of service demand are sector-specific: migration restrictions could ease housing pressure, but would not address ageing-driven healthcare demand and may accelerate contraction in the education system.

Figures (27)

• Figure 1: Modelling framework applied to the education sector. The figure illustrates how total teacher demand is derived by combining disaggregated population structures (left panel) with cohort-specific per-capita utilisation rates (middle panel), yielding age-specific demand contributions by citizenship (right panel). In each panel, Austrian nationals are shown in green and foreign nationals in pink; solid lines represent 2025, and dashed lines represent 2050. The left panel shows the population pyramid for both groups, reflecting the older age structure of Austrian nationals and the younger, growing foreign-national population. The middle panel shows the per-capita teacher demand rate by age, which is concentrated in school-age cohorts. The student enrollment change from 2025 to 2050 is small. The right panel shows the resulting teacher demand by age group, obtained as the product of the left and middle panels. The shift from solid to dashed lines across all three panels captures the combined effect of demographic change and evolving enrolment behaviour between 2025 and 2050.
• Figure 2: Sectoral demand projections and driver decomposition (2025--2050). Columns show results for housing (left), education (middle), and healthcare (right). (a) Projected evolution of total service demand (black line), disaggregated into demand from Austrian nationals (green) and foreign nationals (pink). Shaded regions denote the sensitivity of the results to the migration scenario ensemble. (b) Cumulative absolute contribution of specific drivers to changes in demand ($\Delta$) relative to the 2025 baseline for the median migration scenario, shown separately for Austrian nationals (left, green) and foreign nationals (right, pink). Drivers are classified as volume (Vol., changes in total population size), ageing (Age., shifts in population age structure), and behaviour (Beh., changes in per-capita utilisation intensity). The healthcare model assumes constant utilisation rates, thereby excluding behavioural effects. (c) Summary tables reporting baseline demand in 2025, the percentage contribution of each driver decomposed by citizenship, and total projected demand for 2050.
• Figure 3: Service demand shares and representation index by sector and citizenship (2025--2050). Columns show results for housing (left), education (middle), and healthcare (right). (a) Share of total demand generated by Austrian nationals (green) and foreign nationals (pink). (b) Representation index, defined as each group's demand share divided by its population share; a value of $1.0$ indicates that a group consumes services in exact proportion to its demographic weight, values above $1.0$ indicate over-representation, and values below $1.0$ indicate under-representation. Shaded regions represent the sensitivity to the migration scenario ensemble.
• Figure 4: Spatial concentration and growth elasticity of service demand to foreign population (2026--2050). (a) Temporal evolution of the spatial concentration coefficient ($\beta^{\text{conc}}$), which measures whether cumulative demand growth localises in regions with higher foreign population shares. A positive and rising coefficient indicates increasing spatial concentration; a coefficient indistinguishable from zero indicates that demand growth is spatially independent of foreign settlement, pointing instead to universally operating drivers such as population ageing. The inset scatter plot shows the underlying regional regression at 2050, illustrating the cross-sectional relationship between foreign population share ($\psi_{r,2050}$) and cumulative demand growth ($g_{r,2050}$) across regions. (b) Temporal evolution of the foreign population growth elasticity ($\beta^{\text{growth}}$), which measures the sensitivity of demand growth to the growth of the foreign population. A declining coefficient indicates decoupling, in which demand is increasingly governed by the demographic momentum of the settled population rather than by new arrivals. The inset scatter plot shows the underlying regional regression at 2050, illustrating the cross-sectional relationship between cumulative foreign population growth ($f_{r,2050}$) and cumulative demand growth ($g_{r,2050}$). Shaded areas represent 95% confidence intervals; coefficients are statistically significant where the interval does not cross zero. Values are based on the median across the migration scenario ensemble. Housing and education use $N=35$ NUTS 3 regions; healthcare uses $N=9$ NUTS 2 regions.
• Figure 5: Spatial distribution and projected change in service demand by sector and citizenship (2025--2050). Columns show results for housing (left), education (middle), and healthcare (right). The top row shows the baseline foreign national share of total sectoral demand in 2025. The middle row shows the absolute change in demand ($\Delta$) from 2025 to 2050, decomposed into the foreign national contribution (left larger map) and the total change (right smaller map); orange indicates growth and blue indicates contraction. The bottom row shows the projected foreign national share in 2050, enabling direct comparison with the 2025 baseline. Selected regions are annotated with their foreign share values, highlighting the regions with the highest and lowest foreign national shares in 2025. Spatial units correspond to NUTS 2 regions for healthcare and NUTS 3 regions for housing and education. Absolute change values are capped at the 95th percentile to mitigate the influence of outliers. All projections are reported as median values across the migration scenario ensemble.