On Tuesday 15 September the three children of Palmoli, a village of about 800 people in Abruzzo, central Italy, were due to spend a few hours with their parents in a house the town council had lent the family. They arrived with the social workers, the carabinieri and the mayor. The meeting did not happen. The mayor, Giuseppe Masciulli, said the conditions for safety were lacking. The next day the children started at a state school in Vasto, forty kilometres of mountain road from the house. They have been out of their family since 20 November 2025, when the juvenile court of L’Aquila suspended their parents’ responsibility. A ruling filed on 1 September orders their return, but in stages: daytime visits first, overnight weekends from October, with the social services certifying each step.
We have not read the court files, and this article does not judge that decision. It is about something else: the system the three children entered, what the numbers say about it, and what research says about children who are removed and children who stay.
The numbers are kept by the Ministero del Lavoro e delle Politiche Sociali (the Italian Ministry of Labour and Social Policy) through SIOSS, the social services information system, which since 2017 has collected data from every local social district. The latest report, published on 5 December 2025, is a snapshot of 31 December 2024. On that date 345,083 minors were on the books of the social services, forty in every thousand residents under eighteen. The great majority live at home: the service follows them, it does not move them. Out of their family of origin there were 46,107, including unaccompanied foreign minors. Setting those aside, since they have no family in Italy to be removed from, 15,075 children were in foster care and 20,592 in residential services, that is, children’s homes.
The ratio between the two is the figure that matters. Italian law 149 of 2001 says that placement with a family is the preferred option and a residential home the residual one. The data say the opposite: for every three children with a foster family there are four in a home, and the gap is widening. On 31 December 2022 residential services held 18,081 minors; two years later, 20,592. Foster care over the same period stood still, from 15,218 to 15,075. The system is growing on the side the law meant to be residual.
Minors in foster care and in residential services in Italy at 31 December 2022 and 31 December 2024, excluding unaccompanied foreign minors. Chart produced by ScienceOnline from data in Ministero del Lavoro e delle Politiche Sociali, Quaderni della ricerca sociale n. 60 (SIOSS data 2022) and n. 66 (SIOSS data 2024). Chart produced in-house: numerical data are not covered by copyright.
What it costs, in the Palmoli case, the mayor has said: the town council pays 7,500 euros a month for the three children’s placement. It is a figure stated by a party to the dispute, not a statistic, but it is consistent with the daily rates of Italian children’s homes, which run to tens of euros per child per day.
Then the children turn eighteen. The ministry’s report counts 26,053 young adults still on the books of the social services at the end of 2024, and 3,112 still housed in residential facilities. Since 2017 Italian law has used an English term for them, care leavers, and has set up a fund to support them to the age of twenty-one. The national pilot that followed, run by the Istituto degli Innocenti in Florence, registered 766 young people between 2018 and 2023 in 66 local districts across 17 regions; for 636 of them an individual plan was started. It is the only systematic follow-up Italy does of anyone leaving the system. Nobody knows, from a national figure, how the young people who grew up in Italian children’s homes are doing at twenty-five: whether they work, whether they have a home, whether they have gone back into a public institution of another kind. The ministry counts who goes in. Who comes out is not counted.
To know what happens next one has to look abroad. A systematic review published in 2017 in Children and Youth Services Review by Laura Gypen and colleagues at the Vrije Universiteit Brussel brought together thirty-two quantitative studies of young people who grew up in foster care. Their outcomes are worse than the general population’s in every domain measured: education, employment, income, mental health, housing. This holds in countries whose child protection systems are built around control, such as the United States and the United Kingdom, and in those built around family support, such as the Nordic countries. The two protective factors that emerge most consistently are a stable base, meaning not changing home and not changing school, and one adult who stays.
These are associations, and must be read as such. Children who enter care carry with them the harm they are taken from: maltreatment, neglect, poverty, parental illness. A comparison with the general population measures everything at once, the effect of the birth family and the effect of the system. It does not say whether removal helped or made things worse.
Separating the two takes an experiment, and nobody runs experiments on children. Joseph Doyle, an economist at the Massachusetts Institute of Technology, found a way round. In Illinois, reports of maltreatment are assigned in rotation to child protection investigators, and investigators differ widely in how readily they remove children from home. For a family, getting one investigator or another is close to a lottery. Doyle used that lottery as a natural experiment, tracking Illinois children for years through school, court and employment records. The result, published in 2007 in the American Economic Review, concerns the marginal cases, those where one investigator would have removed and another would not: those children, left at home, had fewer juvenile arrests, fewer teenage pregnancies and more employment as adults. The effect was stronger for older children. A later study by the same author, in the Journal of Political Economy in 2008, found the same sign for adult arrests and convictions.
Then the same lens was turned on younger children, and the sign changed. Anthony Bald, Eric Chyn, Justine Hastings and Margarita Machelett repeated the method in Rhode Island on children investigated before the age of six. Their study, published in 2022 in the same Journal of Political Economy, finds that removal raises test scores and cuts grade repetition for girls, and has no measurable effect on boys. The authors find no explanation in the families’ characteristics or in the kind of placement.
Taken together, the two studies say something precise. Removal is neither good nor bad in itself. It pays when the harm at home is serious and the child is young, and it does not pay when the case is marginal and the child is older. And the marginal cases are exactly the ones investigators disagree about: the ones where the decision depends on who makes it.
There is a structural reason why the system tends to err in one direction only, and Doyle set it out with Anna Aizer in a 2018 review in the Annual Review of Economics. If a caseworker leaves a child at home and the child is then harmed, the responsibility is theirs and it arrives at once. If they remove a child who could have stayed, the harm appears years later and nobody will trace it back to them. The incentives push towards protection and away from the family, and they do so even when the caseworker acts in good faith. No malice is needed. It is enough that the error in one direction has a name and the error in the other does not.
The Italian state has already written that principle into law, elsewhere. A convicted mother may keep her child with her in prison up to the age of three in the nursery wings, and up to six in the attenuated-custody institutions for mothers created by law 62 of 2011. On 31 August 2026, according to the Ministero della Giustizia (the Ministry of Justice), 37 children were in prison with their mothers, 32 of them. At the start of 2025 there were twelve, at the end of August 2025 twenty-one, at the start of 2026 twenty-six: the number has tripled in twenty months. The rise comes from law 80 of 2025, which made the deferral of sentence for pregnant women and mothers of children under one optional rather than mandatory. Whether that is right can be argued. But it is worth setting the two houses side by side. A prison, even an attenuated-custody one, has bars on the windows, imposed hours, officers in the corridor and no room where a mother is alone with her child. As an environment it ranks below a stone house with no electricity. As a family it is a mother alone, with a conviction. As for privacy, it has none. The Palmoli meeting of 15 September was called off, the mayor said, for lack of safety and privacy. In its prison nurseries the state keeps 37 children with neither, and it does so for a reason it has written into law: between separating a child from his mother and keeping him with her in a worse place, the bond weighs more than the environment. It is the same criterion research suggests for removal.
The state applies it to those with a conviction, and not to those with a house in the woods.
Finally there is a conflict of interest that is built into the structure, not into people. In many local districts the service that writes the report to the court and proposes the placement belongs to the same body, public or cooperative, that runs the residential homes and collects their fees. The one who assesses is often the one who houses. The best-known Italian case, Bibbiano, has been of little help in understanding this: prosecutors in Reggio Emilia alleged an organised system of unlawful placements, and on 9 July 2025 the court acquitted eleven of fourteen defendants and ruled that no such system existed. Three suspended sentences remained, one of them against a social worker for an untruthful report on a minor’s condition. The real conflict does not lie in the bad faith of one office. It lies in a design that puts assessor and provider under the same roof, and that no Italian law today separates.
The last question remains: if a child is removed, where to. Here the answer is the only one that comes from a real experiment. In 2000 an American team led by Charles Nelson, Nathan Fox and Charles Zeanah obtained the Romanian government’s consent to assign 136 children abandoned in Bucharest institutions at random, half to remain in institutional care and half to move to foster families set up for the study. The results, published in Science in 2007 and followed into adulthood, are stark: the children who stayed in institutions had a far lower IQ than both children never institutionalised and children moved to families, and recovery was greater the earlier the move, with a window that closed at around two years of age. Romanian institutions were worse than an Italian children’s home, and the generalisation must be made with care. But the direction is clear, and Italy has four children in homes for every three with a foster family.
The principle that comes out of all this fits in a line. Separation is a certain and immediate harm; the harm of the environment is a harm to be proved. When it is proved and serious, and the child is young, removal pays. When it is not, the child who stays at home does better. And if a child is removed, a foster family beats an institution, always.
The court of L’Aquila has judged the measure it took in November to be reversible. That decision can be argued over, and everyone is arguing over it. The piece of data missing from the argument is another one: how many of the 20,592 children in Italian residential homes are marginal cases, and who decided that. Nobody knows, because nobody counts. On Wednesday 16 September the three children of Palmoli started the second and fourth year of primary school.
References
Ministero del Lavoro e delle Politiche Sociali, «I minorenni e neomaggiorenni in carico ai servizi sociali, in affidamento familiare e accolti nei servizi residenziali. Dati SIOSS 2024», Quaderni della ricerca sociale, n. 66, Rome, 5 December 2025. https://www.lavoro.gov.it/temi-e-priorita-infanzia-e-adolescenza/studi-e-statistiche/qrs-66-report-2024
Ministero del Lavoro e delle Politiche Sociali, «I minorenni in affidamento familiare e nei servizi residenziali attraverso i dati SIOSS. Anno 2022», Quaderni della ricerca sociale, n. 60, Rome, 2024. https://www.lavoro.gov.it/notizie/pagine/dati-sui-bambini-e-gli-adolescenti-fuori-dalla-famiglia-di-origine
Istituto degli Innocenti, «Report Sperimentazione Care leavers 2018-2023», Quaderni della ricerca sociale, n. 59, Ministero del Lavoro e delle Politiche Sociali, Florence, 2024. https://www.lavoro.gov.it/documenti-e-norme/studi-e-statistiche/qrs-59-report-valutazione-care-leavers-2018-2023.pdf
Gypen L., Vanderfaeillie J., De Maeyer S., Belenger L., Van Holen F., «Outcomes of children who grew up in foster care: Systematic-review», Children and Youth Services Review, 76, 74-83, May 2017. doi: 10.1016/j.childyouth.2017.02.035
Doyle J.J. Jr., «Child Protection and Child Outcomes: Measuring the Effects of Foster Care», American Economic Review, 97, 5, 1583-1610, December 2007. doi: 10.1257/aer.97.5.1583
Doyle J.J. Jr., «Child Protection and Adult Crime: Using Investigator Assignment to Estimate Causal Effects of Foster Care», Journal of Political Economy, 116, 4, 746-770, August 2008. doi: 10.1086/590216
Bald A., Chyn E., Hastings J., Machelett M., «The Causal Impact of Removing Children from Abusive and Neglectful Homes», Journal of Political Economy, 130, 7, 1919-1962, July 2022. doi: 10.1086/719856
Doyle J.J. Jr., Aizer A., «Economics of Child Protection: Maltreatment, Foster Care, and Intimate Partner Violence», Annual Review of Economics, 10, 87-108, August 2018. doi: 10.1146/annurev-economics-080217-053237
Nelson C.A. III, Zeanah C.H., Fox N.A., Marshall P.J., Smyke A.T., Guthrie D., «Cognitive Recovery in Socially Deprived Young Children: The Bucharest Early Intervention Project», Science, 318, 5858, 1937-1940, 21 December 2007. doi: 10.1126/science.1143921
Legge 28 marzo 2001, n. 149, «Modifiche alla legge 4 maggio 1983, n. 184, recante Disciplina dell’adozione e dell’affidamento dei minori» (Italian law on adoption and foster care).
Legge 27 dicembre 2017, n. 205, article 1, paragraph 250 (Italian care leavers fund).
Ministero della Giustizia, Dipartimento dell’amministrazione penitenziaria, «Detenute madri con figli al seguito – 31 agosto 2026», monthly statistics. https://www.giustizia.it/giustizia/it/mg_1_14.page?facetNode_1=0_2&selectedNode=0_2_1
Garante delle persone sottoposte a misure restrittive della libertà personale della Regione Lazio (Lazio regional ombudsman for detainees), «Sovraffollamento: un’emergenza senza precedenti nell’ultimo decennio», 4 September 2026 (DAP data at 31 August 2026). https://www.garantedetenutilazio.it/sovraffollamento-unemergenza-senza-precedenti-nellultimo-decennio/
Garante delle persone sottoposte a misure restrittive della libertà personale della Regione Lazio, «Il Dap al 31 agosto: 63.167 persone detenute in Italia, 6827 nel Lazio», 4 September 2025 (DAP data at 31 August 2025). https://www.garantedetenutilazio.it/ll-dap-al-31-agosto-63-167-persone-detenute-in-italia-6827-nel-lazio/
Legge 21 aprile 2011, n. 62 (Italian law on the relationship between imprisoned mothers and their minor children).
Legge 9 giugno 2025, n. 80, converting decree-law 11 April 2025, n. 48 (public security provisions), art. 15.
Tribunale di Reggio Emilia, first-instance judgment in the «Angeli e Demoni» trial, 9 July 2025, as reported by Il Fatto Quotidiano and Sky TG24, 9-10 July 2025.
ANSA, «Mancano sicurezza e privacy, salta l’incontro della famiglia del bosco», 15 September 2026.
Previous articles
Donati G., «Twenty thousand boys alone. Not a colour, an age», ScienceOnline, 9 September 2026.
Donati G., «Ventimila ragazzi soli. Non è un colore, è un’età», Scienzaonline, Attualità, 9 September 2026.
