Childcare as a Magnet: New Nurseries Briefly Draw Families Back to Italy’s Fading Interior

Across much of the developed world, depopulation has quietly become one of the defining demographic challenges of the era, and nowhere is the pattern more stubborn than in remote rural regions where jobs, schools and hospitals are increasingly out of reach. A new study published in Social Indicators Research by Viviana Celli of Sapienza University of Rome and Roberta Di Stefano of the University of Molise offers the first causal evidence that a surprisingly modest piece of social infrastructure — the opening of a single nursery for children under three — can measurably shift where families choose to live. Using administrative data on every Italian municipality between 2014 and 2022, the researchers find that when a municipality that previously had no early childhood education and care (ECEC) facility opens one, the effect on internal migration is strikingly uneven across space. In well-connected towns the effect is essentially zero. In Italy’s so-called inner areas, however, the first year after opening brings a significant rise in the net migration balance, a result the authors interpret as evidence that childcare can support, but not by itself sustain, long-term demographic revitalization.

The stakes of the question are considerable. Eurostat projects that the European Union’s population will decline by 1.8 percent between January 2023 and January 2051, with predominantly rural regions expected to contract steadily in every five-year interval, at an average annual rate of roughly 3.1 to 3.6 per thousand. Urban regions, by contrast, are projected to keep growing, if more slowly. Depopulation in remote areas is compounded by limited access to essential services and scarce employment opportunities, which makes those territories less attractive to families with children; shrinking population density then undermines the financial sustainability of the very services that remain, creating a self-reinforcing feedback loop that deters new residents and investment. Italy is a particularly stark case: the country has lost almost 1.9 million people, births have fallen for sixteen consecutive years, and the 2024 fertility rate of 1.18 children per woman is the lowest ever recorded there, well below both the EU average of 1.38 and the replacement threshold of 2.1.

The theoretical backdrop for the study is Tiebout’s classic 1956 idea of voting with their feet: people relocate to jurisdictions whose bundles of public goods best match their preferences. Schools, healthcare, transport and social care jointly determine a territory’s attractiveness, and prior research has shown how powerfully education infrastructure shapes residential choices. Danish quasi-experimental evidence on the closure of eight village schools in a peripheral municipality documented a subsequent population decline of 7.6 percentage points over the following decade, while work on Italian primary school closures after a national network-rationalization reform found significant reductions in both population and income, with the damage concentrated in peripheral areas. What had been missing, the authors argue, is causal evidence on whether the reverse move — introducing childcare where none existed — can actively attract or retain residents, rather than merely failing to prevent decline.

Childcare has traditionally been studied through the lens of family policy, as a tool for reconciling work and parenthood and for supporting fertility. Studies from Norway and Germany have linked expansions of public childcare to higher birth rates, and recent Italian evidence finds a positive and substantial effect of ECEC availability on the number of births at the municipal level. Drawing on Amartya Sen’s capability approach, Celli and Di Stefano reframe ECEC as a territorially embedded social infrastructure: its absence represents a capability deficit that constrains family choices and can trigger out-migration well before the quality of primary schools becomes the decisive factor. At the same time, comparative assessments of European cohesion and family policy caution that childcare alone rarely compensates for deficits in transport, housing, digital connectivity or employment, a warning that foreshadows the study’s own conclusions.

Identifying a genuine causal effect in this setting is methodologically demanding, and the paper’s technical apparatus is central to its credibility. The treatment is defined narrowly as the first opening of an ECEC facility in a municipality that previously had none, between 2016 and 2018 — the extensive margin of provision, rather than an expansion of existing capacity. The outcome is the internal migration balance: in-migrants from other municipalities minus out-migrants to other municipalities. Because municipalities adopted the facilities at different times, the authors employ a non-parametric generalization of the difference-in-differences estimator developed by Imai and colleagues, which sidesteps the weighting pitfalls of conventional two-way fixed effects models under heterogeneous treatment effects. For each treated municipality, the method constructs a matched set of controls that share the same treatment history and similar pre-treatment trajectories in population, age structure, housing prices, income, employment, births and prior ECEC spending, matched exactly within the same broad geographical area and refined via propensity score matching to the five closest controls.

Two subtle identification threats receive particular attention. The first is interference: if a new nursery in one town draws families who would otherwise have settled in a neighboring municipality, including that neighbor in the control group would contaminate the counterfactual. The second is contamination: controls that themselves open a facility shortly after the treated unit would begin to reflect treatment effects during the evaluation window. To address both, the authors exclude all untreated municipalities in the same Ambito Territoriale Sociale — the inter-municipal districts of roughly six hundred units through which Italian social services are planned — reasoning that spillovers operate through local service basins rather than simple straight-line distance, and they require that every control unit remain untreated throughout the treated unit’s entire post-treatment horizon of up to four years. Covariate balance diagnostics confirm that treated and matched controls follow similar pre-treatment trajectories, though the authors acknowledge the two-period pre-treatment window limits the power of trend tests and cannot rule out unobserved time-varying shocks.

The sample comprises 300 treated municipalities, 163 of them classified as inner areas under Italy’s official National Strategy for Inner Areas, which defines inner territories not only by low density and remoteness but by limited access to essential services. Descriptively, treated municipalities are small — averaging 4,378 inhabitants against a national mean of 7,633 — demographically weaker, with roughly half the national averages of young children and women of childbearing age, and carrying a negative pre-treatment migration balance of about minus 5.3 people per year. Kernel density comparisons show inner and non-inner treated municipalities are demographically similar but differ sharply in economic conditions, with non-inner areas exhibiting higher income, employment and housing prices, suggesting the structural divide is primarily economic rather than demographic.

The headline result is one of pronounced spatial heterogeneity. Nationally, opening a first ECEC facility produces weakly positive but statistically indistinguishable-from-zero effects on the migration balance. Splitting the estimates by territory reveals the real story: one year after opening, treated inner-area municipalities record a migration balance roughly 7.5 individuals higher than their counterfactual — equivalent to about 2.7 additional net residents per thousand inhabitants, sizeable for towns averaging 4,400 people that were previously shedding population. A decomposition of the flows shows the gain comes from both directions: inflows rise by about 4.2 people while outflows fall by about 3.3, indicating that the new service both attracts newcomers and persuades existing residents to stay. In non-inner municipalities the effect hovers near zero at every horizon. The timing also rules out a construction-works explanation, since any building stimulus would appear at or before opening, whereas the largest response arrives precisely when the service becomes available to families.

Why does the effect fade? Because the outcome is an annual flow, the fading does not mean the new residents leave; rather, the additional net inflow is concentrated in the year immediately after opening, after which treated municipalities return to migration trajectories similar to comparable untreated towns. The authors suggest that opening a facility in an inner area immediately relieves a binding constraint for families and signals renewed institutional commitment to the community — a strategic act of confidence by small administrations betting on their own future — but that once this initial adjustment has occurred, the service’s continued presence cannot by itself generate fresh inflows each year without complementary improvements in employment, housing and transport. Notably, the demographic response shows no systematic relationship with the initial coverage capacity of the new facility, implying that what matters is the introduction and local availability of the service, not its initial scale.

The policy implications are sobering but actionable. For municipalities that undertake the investment, betting on childcare appears to pay off at least locally, and robustness checks — varying the number of matched neighbors, applying covariate-balancing propensity score weights, and alternative bootstrap resampling schemes — leave the one-year inner-area effect intact. Yet whether local gains aggregate into net gains for inner regions as a whole, rather than simply reallocating population between neighbors, remains an open question requiring direct estimation of spatial spillovers. The authors’ message is that ECEC can act as a short-term game changer and a credible signal that a territory is worth staying in, but sustaining that impulse demands a broader place-based policy mix, coordinated through frameworks such as Cohesion Policy, that links social services to jobs, connectivity and institutional capacity. Childcare, in other words, can open the door to demographic revival — but something else must walk families through it.

Subject of Research: The causal effect of early childhood education and care facility openings on internal migration and depopulation in Italian inner areas

Article Title: Breaking the Cycle of Demographic Decline: could Early Childhood Education and Care services be a Path to Revitalization?

Article References: Celli, V., & Di Stefano, R. (2026). Breaking the Cycle of Demographic Decline: could Early Childhood Education and Care services be a Path to Revitalization?. Social Indicators Research, 184(3), Article 50. https://doi.org/10.1007/s11205-026-03939-0

Image Credits: AI Generated

DOI: 10.1007/s11205-026-03939-0

Keywords: depopulation, early childhood education and care, internal migration, inner areas, difference-in-differences, Italy, fertility, social infrastructure, rural decline, causal inference, place-based policy, demographic revitalization