466 million children eat at school worldwide — but quality, cost and politics differ wildly. An audit of eight national systems and who profits from feeding them.
The Trillion-Meal Question
What the world spends to feed its schoolchildren
School lunch is no longer a peripheral welfare programme. In 2024 the World Food Programme counted [1] 466 million children receiving a daily meal at school worldwide, funded by a combined public bill of US$84 billion — double the figure four years earlier ✓ Established. Ninety-nine per cent of that money comes from national budgets, not aid.
The scale is easy to miss. A trillion meals a year flow through the world's school kitchens, prepared by an estimated 7.4 million cooks [1], and the political economy of those meals — what is on the tray, who grew it, who delivered it and who paid — varies more sharply than almost any other public service. Two countries spending the same per-pupil sum can produce a dignified three-course meal in one school and a frozen pizza wrapped in plastic in another. The variance is not accidental. It reflects choices about procurement, regulation, lobbying and politics that this report sets out to read off the menus.
The WFP's headline claim — a US$9 return on every US$1 spent [1] — is unusually robust for the development economics literature ◈ Strong Evidence. The Inter-American Development Bank's joint modelling places the benefit-to-cost ratio in a similar range, with gains accruing across child health, schooling years completed, smallholder farm income and women's labour-force participation. A 2021 estimate of the US National School Lunch Program found US$40 billion in annual health and economic benefits against a US$18.7 billion programme cost — a benefit-cost ratio close to 2:1 [6], lower than the WFP global figure because rich-country children gain less marginal nutrition per meal than children at greater risk of stunting.
Coverage expansion has been driven from the South. Between 2020 and 2024, sub-Saharan Africa added 20 million children to government-led programmes, with Kenya, Madagascar, Ethiopia and Rwanda leading the surge [1]. The story Europeans tell themselves — that school feeding is an invention of postwar Nordic welfare states — has been outpaced by domestic budgeting in countries where the per-pupil sum looks tiny on a global accounts ledger but covers more than 90% of enrolled pupils. The result is a redistribution of the school-meals map: the largest single national programme is no longer European or American but Indian, at 118 million children a day [2].
National choices then layer politics on top of arithmetic. Finland and Sweden treat school lunch as a citizenship institution [9]; France and Italy treat it as part of national food culture [5]; Japan treats it as a curriculum subject [4]; the United Kingdom and United States treat it as a contested welfare line [7][10]; India and Brazil treat it as social development infrastructure [2][8]. None of these framings is neutral. Each produces a different menu, a different supply chain, a different industrial base, and a different long-run outcome on adult height, lifetime income and university entry. The chapters that follow walk through eight of these systems in turn, comparing menus and budgets but also the political and corporate interests that shape what ends up on the tray.
The WFP's 2024 audit documents school meals funding doubling from US$43 billion in 2020 to US$84 billion in 2024, with 99% coming from national rather than donor budgets [1]. The same dataset estimates 7.4 million cooking jobs supported and a US$9 multiplier per dollar spent — figures that place school feeding squarely inside national agriculture and labour policy, not the foreign aid envelope where it sits in many ministry budgets.
What the comparison clarifies is that the binary debate visible in Washington and Westminster — universal versus means-tested — is not the central question elsewhere in the world. Most systems that work, work because they integrate procurement, nutrition standards and labour into one accountable structure under a single act of national legislation. Japan codified its programme in 1954 and revised it in 2009 [4]. Finland legislated in 1948 [9]. Brazil's PNAE was federalised in 2009 [8]. India's PM POSHAN consolidates earlier state-level mid-day-meal schemes under a single 2021 framework [2]. The countries struggling are those where school feeding sits across multiple departments with multiple competing constituencies and no single piece of governing law.
Japan — Lunch as Curriculum
Kyushoku and the civic meal
Japan treats school lunch — kyushoku — as a subject taught daily for nine years. Pupils serve each other, eat together in the classroom, and learn nutrition and food culture as a regular part of the school day [4]. By 2025, roughly 30% of municipalities had abolished parental fees; a nationwide free elementary-school lunch programme is scheduled to begin in 2026 ✓ Established.
The Japanese School Lunch Program Act, first passed in 1954 and substantially revised in April 2009, defines seven statutory objectives for kyushoku [4]. Three are nutritional — adequate intake, dietary habits, food safety. Four are explicitly civic — promoting social cooperation, respect for life and nature, understanding regional food cultures, and learning the role of food in everyday life. The Act elevates school lunch from a welfare service to an education subject, with dedicated diet-and-nutrition teachers (eiyō kyōyu) employed inside schools to design menus, supervise meals and teach short nutrition classes alongside the meal itself.
Each lunch is calibrated to provide approximately one-third of a child's recommended daily nutritional intake, with protein 13–20% of energy and fat 20–30% [4]. The structure is standardised: a staple (rice or bread), a main dish, a side dish, milk and fruit. Menus rotate through regional cuisines and seasonal produce, with the diet-and-nutrition teacher in each school responsible for verifying that statutory standards are met before the menu is published to parents. The price to parents covers ingredients only — typically ¥4,500 to ¥5,000 per month at elementary level, around ¥250 to ¥300 per meal ✓ Established Fact. Labour, facilities and equipment are public.
The 2009 revised School Lunch Program Act sets out seven statutory objectives, four of them civic rather than nutritional, and provides for diet-and-nutrition teachers (eiyō kyōyu) embedded in schools to deliver the programme [4]. Each meal must cover roughly one-third of a child's daily nutritional needs, with menus designed by a qualified nutritionist before publication. Parents pay for ingredients only; labour, facilities and equipment are public, removing the procurement asymmetry that drives much of the per-meal cost variation visible in Anglo-American systems.
The civic side of kyushoku is what tourists usually film. Pupils take it in turns — the tōban roster — to don white smocks and caps, collect the food trolleys, ladle out portions for classmates and clear the trays after the meal. The lunch is eaten in the classroom rather than a cafeteria; the teacher eats the same meal with the children; pupils on duty say a short pre-meal phrase (itadakimasu) and a post-meal phrase (gochisōsama). None of this is decorative. Diet-and-nutrition teachers report that the cooperative service routine is the principal pedagogical lever for behaviours that later show up in cafeteria-free workplaces — clearing one's own dishes, respecting kitchen labour, and treating shared food as a collective resource.
The 2025 inflection is fiscal rather than pedagogical. Tokyo's 23 wards made kyushoku free for all public elementary and junior-high pupils in 2024 [4], and by mid-2025 roughly 30% of Japanese municipalities had followed. The national government has committed to a universal free elementary-school lunch from 2026, financed jointly by central and prefectural budgets. The political driver is demographic: with Japan's working-age population shrinking and child-rearing costs cited as a primary disincentive to first births, free kyushoku has emerged as an unusually cost-effective family-support measure — visible, daily and impossible for parents to overlook.
Critics inside Japan focus on supply chains rather than pedagogy. The 2024 surge in rice prices, and the chronic labour shortage in school kitchens, raise questions about how the universal model will be financed against domestic food-price volatility [4]. There is also a quiet debate about the rigidity of nutritional standards in a country where dietary diversity has narrowed since the 1990s, and about the place of school lunch in private schools and senior-high schools, which the Act does not cover. None of these arguments, however, contest the basic architecture. Kyushoku is treated as a public good; the disagreement is about scope, not principle.
The lesson exportable from Japan is structural, not culinary. Other countries copy the menu and the trolleys; few copy the underlying legal status. By placing school lunch inside the Education Act and staffing it with credentialed teachers rather than caterers, Japan removes the question of whether the service exists from annual budget negotiations and shifts the question to how it is delivered. Anglo-American systems do the reverse: they begin every budget cycle by re-litigating eligibility, then leave delivery to whichever contractor wins the procurement [6][10]. The Japanese model is more expensive in the short run and dramatically cheaper in the long run, because it is not exposed to political cycles.
The American Tray — Debt, Pizza and Politics
Why the world's largest budget produces the world's messiest lunch
The National School Lunch Program serves 29.7 million children a day across 95,000 schools at an annual cost of US$17.7 billion [6] ✓ Established. It also runs US$25.3 million in reported parental meal debt [7], preserves a 2011 regulation under which two tablespoons of tomato paste on a slice of pizza count as a vegetable serving [12], and operates without national universal coverage in 41 of 50 states.
The American programme is the world's largest by budget and one of its most contested by design. The NSLP, created in 1946 under the National School Lunch Act, was conceived as a dual-purpose policy — to feed children and to absorb US agricultural surplus. The dual mandate has never been fully reconciled. The Act's procurement preference for domestically produced commodities pulls school menus towards whatever American agriculture has most of in any given year — milk, corn-fed beef, frozen vegetables, processed grains — and away from whatever produces best nutritional outcomes per dollar. Half of the political controversy that erupts around the programme every few years traces back to this structural tension ◈ Strong Evidence.
The participation numbers are large but uneven. Of the 29.7 million children served daily, 20.5 million receive their meals free under the eligibility framework set by federal poverty guidelines; a further tier receive reduced-price meals; the remainder pay full price [6]. In FY2024 the programme served 4.8 billion lunches at a total cost of US$17.7 billion — roughly US$3.69 per meal in federal outlay, below the average reported cost of producing a school lunch in most districts [6]. The shortfall is filled by paid meal revenues, à la carte sales, and — when households fall behind on payments — debt.
The Education Data Initiative documents US$25.3 million in reported school meal debt in 2025, up 25% from the prior year, with average reported debt per district rising 49% to US$39,329 [7]. Children whose families fall behind on payments are routinely served alternative "shame meals" — cold cheese sandwiches and milk — in violation of federal anti-discrimination guidance that most districts nonetheless interpret as advisory. No other rich-country system tracked in this report runs school lunch on a household-debt basis.
The pizza-as-vegetable rule is the most internationally famous artefact of US school-lunch politics, and it is more revealing than it sounds. In November 2011, Congress passed the Agriculture Appropriations Act with a rider blocking a proposed USDA reform that would have required half a cup of tomato paste to count as a vegetable serving; the previous standard — two tablespoons of paste — remained in force [12]. The practical effect was to preserve the accounting treatment under which a slice of school pizza, carrying roughly two tablespoons of paste, fulfils the meal's vegetable requirement without any actual vegetable on the tray.
PolitiFact's review of the 2011 Agriculture Appropriations Act establishes that the two-tablespoon tomato-paste-as-vegetable rule was preserved at the request of frozen pizza manufacturers, the salt industry and potato growers [12]. The rule is not a quirk of historical regulation — it is the operative law governing what counts as a vegetable on every NSLP tray fifteen years later, and it has survived multiple subsequent reform attempts. The political economy lesson is that small lobbying interventions on accounting rules outlast much larger nutritional reforms.
The state-level picture is moving faster than the federal one. Nine US states — California, Colorado, Maine, Massachusetts, Michigan, Minnesota, Nevada, New Mexico and Vermont — have legislated permanent universal free school meals [14], with New York becoming the ninth in its FY2025-26 budget. California, the first to act, used 2020 federal pandemic relief funds to extend free meals to every public-school pupil and then made the policy permanent in the 2022-23 school year. Michigan's 2023 universal programme is funded out of US$160 million in state appropriations. These nine states cover roughly 40% of US enrolment but no federal universal programme exists, and the 2024 expiration of pandemic-era waivers exposed the underlying patchwork.
The 2024 federal expansion took a different route. The Community Eligibility Provision threshold, lowered from 40% to 25% in October 2023, makes universal free meals administratively feasible in roughly 3,000 additional high-poverty school districts [6]. This is not the same as universal entitlement; it is a workaround that lets districts in the bottom quartile of household income offer free meals to all pupils without per-child means testing. The political reading is that the federal programme is moving towards de facto universal coverage at the bottom of the income distribution while continuing to means-test in the middle — the opposite of the Nordic model and a constraint on the universal model's headline benefits.
What the comparison sharpens is the contrast between budget and architecture. The United States spends more per fed child than India, France or Brazil but recovers less of the per-meal benefit, because the system is built around a dual-purpose mandate (feed children + absorb surplus), competing federal and state authorities, and a thirty-year accumulation of lobbying interventions that no longer reflect any coherent nutritional theory ◈ Strong Evidence. The same dollar buys a Finnish lunch and a US meal with markedly different content [9][6].
France — The Three-Course Sliding Scale
The cantine and the politics of taste
The French cantine scolaire serves a three-to-five course meal — first course, main, vegetables, dairy and dessert — to roughly six million pupils each school day [5]. Parents pay according to household income, from €0.13 to €7 per meal on the Paris grid in 2025, against a Cour des Comptes estimate of true cost around €7.30 ✓ Established.
French school food has been a question of national identity since at least the 1880s, when the Third Republic's Jules Ferry laws bound republican education to the everyday meal as a leveller of class. The contemporary cantine inherits this premise. Lunch is treated as a cultural inheritance, not a nutritional minimum: the structure is multi-course; pupils sit at tables, are served from communal dishes, and are expected to eat the same menu rather than picking from a counter. The model is comparatively expensive in operational terms — labour-intensive, slow, and dependent on actual cooking on site — and it is defended in those exact terms.
The 2018 Egalim law tightened the architecture. From 2022, public canteens were required to serve at least 50% "quality and sustainable" food (organic, label rouge, AOC and equivalent), 20% of it certified organic, with a minimum of one vegetarian meal per week [5]. Use of disposable plastics is banned in school canteens, and food-waste targets are statutory rather than aspirational. The School Meals Coalition's 2025 country progress review of France records compliance still patchy — many municipalities are below the 50% sustainable threshold — but the legal baseline shifted, and procurement contracts have followed.
The school cantine is the place where the Republic teaches what is normal to eat. If the menu becomes industrial, the lesson becomes industrial. The fight is not nostalgic; it is about what kind of citizen the school produces.
— Christophe Hébert, President, Agores (French municipal caterers' association), 2024The pricing structure is the second distinctive feature. French municipalities use a sliding-scale tariff based on the household's quotient familial — taxable income divided by family size — with brackets typically ranging from a token €0.13 or €0.20 for the lowest-income families to €6 or €7 for the highest. Paris's 2025 grid publishes ten bands; most families pay between €1 and €4 [5]. The full cost of producing the meal is roughly €7.30 according to the Cour des Comptes 2020 audit — €2.45 in raw ingredients, €2.70 in supervision labour over the lunch break, the remainder in kitchen labour, fluids and capital. The municipal subsidy is therefore typically around €4–5 per meal even for the highest-income family, and approaches €7 for the lowest.
The political controversy that runs through French school food is not whether to feed children — that is settled — but what to feed them. The 2018-2022 fight over the mandatory vegetarian day pitted environmental and animal-welfare advocates against meat industry lobbies and against some traditionalist mayors who argued that displacing a meat course from the weekly menu erodes a cultural inheritance. The current centre-right argument concerns halal and other religious-observance menus, which most municipalities decline to offer, defending the principle of a single republican meal. The progressive counter-argument is that uniformity in the name of secularism functionally excludes a generation of Muslim and Jewish pupils from the cantine altogether — pushing them to packed lunches and undoing the leveller logic the system was built on ⚖ Contested.
Quality remains comparatively high by international standards. The 2023-25 academic literature on school-lunch attendance and food intake in France finds that pupils who eat at the cantine consume significantly more legumes, fish, fruit and dairy than pupils bringing a packed lunch, with measurable differences in micronutrient adequacy [5]. The structural cost is that French school meals are among the most expensive per child in Europe to produce, and that municipal kitchens are increasingly outsourced to large caterers — Elior, Sodexo, Compass — which compress labour costs and centralise procurement away from the local supply chains the Egalim law aims to support. The tension between Egalim's sustainability targets and the operating economics of outsourced delivery is one of the active political battlegrounds.
What the French system illustrates is that high-quality school lunch is achievable in a large rich country, but it costs roughly €7 per meal in real money and it requires a municipal kitchen base that survives outsourcing pressure [5]. The cross-subsidy from richer to poorer families is structural rather than apologetic, and the legal architecture (Egalim plus the older national nutritional guidelines) keeps the question of what counts as food away from individual school principals. The country's challenge is no longer to defend the principle of the cantine but to defend its operating margin against the same contract-caterer consolidation that has reshaped its US and UK peers.
India — Feeding a Continent at ₹7 a Plate
PM POSHAN and the largest school meal on Earth
India's PM POSHAN (Pradhan Mantri Poshan Shakti Nirman) scheme feeds 118 million children across 1.12 million schools every school day [2] ✓ Established. The cooking cost per child was revised from 1 May 2025 to ₹6.78 in primary schools and ₹10.17 in upper primary — under US$0.10 per plate at the bottom of the band — yet each meal must deliver 450 to 700 calories and 12 to 20 grams of protein [3].
PM POSHAN is the world's largest school feeding programme by enrolment, by some distance. The scheme, renamed in September 2021 and consolidating earlier state-level mid-day-meal initiatives, is anchored in the 2013 National Food Security Act, which sets the nutritional floor for each meal and obligates the central government and states to share its costs (60:40 for general states, 90:10 for special-category states and union territories) [2]. The five-year central financial outlay for 2021-22 to 2025-26 is ₹54,061.73 crore (US$6.34 billion), with an additional ₹31,733.17 crore (US$3.72 billion) from states and union territories. For 2024-25 alone, ₹12,467.39 crore was allocated.
The economics are extraordinary. A cooking cost of ₹6.78 per child is roughly US$0.08 at May 2025 exchange rates — an order of magnitude below the US$3.69 per-meal federal outlay in the US National School Lunch Program [6], and two orders of magnitude below the €7.30 true cost of a French cantine meal [5]. The figure is achievable because PM POSHAN is built on a different logic from rich-country systems: it leans on Public Distribution System grain (rice and wheat allocated below market cost), village-level kitchens, women's self-help groups for the cooking labour, and a procurement chain that bypasses the contract-caterer middle layer entirely.
The nutritional architecture is set by statute. Each primary-school meal must deliver 450 calories and 12 grams of protein; each upper-primary meal must deliver 700 calories and 20 grams of protein [3] ✓ Established. The standard is enforced through district education officers and audited periodically by the central scheme management cell. The 2025 cost revision was triggered by sustained food-price inflation between 2022 and 2024, which had eroded the operational margin in many states; the central government also issued advisories instructing states to cut cooking oil usage by 10% on health and budgetary grounds, an unusually granular intervention.
PM POSHAN's per-meal economics are unmatched globally, but the trade-off is uneven quality across states. UDISE+ 2024-25 data show wide variation in kitchen infrastructure, fuel sources, and protein delivery, with stronger performance in Tamil Nadu and Kerala than in several Hindi-belt states [2]. The scheme has nonetheless delivered measurable gains in primary-school enrolment, attendance and child anthropometrics over its three-decade history — gains that few rich-country school meal programmes can document with comparable causal clarity.
The scheme's earlier history clarifies what is being measured. The Mid-Day Meal Scheme was introduced in Tamil Nadu in 1962 under chief minister K. Kamaraj, then taken national in 1995 as a centrally-sponsored programme. Tamil Nadu's nutritional model was adopted as the template, and the scheme's documented impacts on primary-school enrolment — particularly for girls — were among the evidence cited when the National Food Security Act was passed in 2013. The 2021 rebranding to PM POSHAN added pre-primary (Bal Vatika) classes to the eligible cohort and tightened nutritional standards, expanding daily reach by roughly 15 million children.
The current debate inside India is about quality, not coverage. UDISE+ data show wide inter-state variation in kitchen infrastructure, fuel use and protein adequacy. The southern states — particularly Tamil Nadu and Kerala — outperform the Hindi-belt average on both delivery and outcomes. Periodic controversies about contaminated meals, often traced to single districts, drive parliamentary scrutiny without changing the underlying architecture. The scheme survives political alternation in Delhi because no political party can credibly oppose it; the question is execution, not principle. That is itself an unusual political achievement ◈ Strong Evidence.
The lesson PM POSHAN exports is that a continental-scale school feeding programme is fiscally feasible at very low income levels if it is built around local cooking, basic grain procurement and statutory nutritional minima. The US$0.08 cooking-cost figure is not a target for richer systems to copy; it is a demonstration that the absence of large-scale school feeding in some lower-middle-income countries reflects political choice, not budgetary impossibility. A country that can feed 118 million children at this unit cost cannot credibly be told the model is too expensive.
Brazil — The Family-Farm Mandate
PNAE and the school as agricultural policy
Brazil's Programa Nacional de Alimentação Escolar (PNAE) reaches all 40 million pupils in basic education [8] ✓ Established. Since 2009, federal law requires at least 30% of school food spending to be directed to family-farm produce; from 1 January 2026, the floor rises to 45%. PNAE is the largest single procurement channel for Brazilian smallholders, transferring R$1.6 billion (US$284 million) to family farmers in 2024 alone.
Brazil's PNAE traces its origins to a 1955 federal programme, but the modern architecture dates from Law 11.947, enacted on 16 June 2009 under the Lula government's first-term Zero Hunger framework. The 2009 law did two things that set Brazil apart from comparable rich-country systems. First, it federated universal coverage — all 40 million pupils in basic education are entitled to at least one school meal regardless of household income [8]. Second, it imposed a 30% minimum on family-farm procurement, with the threshold falling on every municipality that receives federal school-meal transfers.
The mechanism is straightforward. Each municipality receives a per-pupil federal transfer based on enrolment, then composes its own menu and procurement plan within the statutory minimum and the nutritional guidelines set by the FNDE. A nationwide network of roughly 8,000 nutritionists is employed by municipalities to design menus that reflect regional food cultures while meeting macronutrient and micronutrient targets. The family-farm minimum forces those nutritionists to design menus that local agriculture can actually supply, rather than menus assembled from centralised industrial procurement.
The 2024 fiscal picture is large enough to matter to Brazilian agriculture but small enough relative to the country's GDP to fit within ordinary federal budgeting. Total federal PNAE spending of R$5.5 billion (~US$976 million) [8] amounts to roughly 0.05% of Brazilian GDP. The R$1.6 billion family-farm directive — US$284 million — is the largest single procurement channel for smallholder agriculture in the country, and PNAE is now studied by FAO and IFAD as a structural model for linking school feeding to rural livelihoods.
Implementation is uneven. The most-cited finding from the academic literature on PNAE is that 50% of municipalities do not meet the 30% statutory minimum [8] ⚖ Contested, with strongest compliance in the South (95.5% of municipalities buying from family farms) and weakest in the Central-West (67.9%). The forthcoming rise to 45% will sharpen this divergence rather than resolve it. Federal enforcement is administrative rather than punitive — non-compliant municipalities face reporting requirements and corrective plans rather than financial penalties — and several Northern states continue to under-purchase from local family agriculture despite formal compliance with federal procurement protocols.
The Brazilian system illustrates a different theory of school feeding. The British, American and (increasingly) French systems treat school food as a downstream procurement question — what can we buy cheaply that meets nutritional minima? The Brazilian system treats school food as an upstream agricultural policy question — what can we ask family farms to grow, knowing the school system is a guaranteed buyer? The downstream theory delivers cheap industrial menus; the upstream theory delivers more expensive meals that double as rural development. Neither is intrinsically right; the choice is a national one about what the school system is for.
The cross-cutting evidence is that PNAE's family-farm channel has measurably altered Brazilian rural economies. The R$284 million directed annually to family farmers represents a guaranteed off-take channel that has stabilised production decisions and shortened supply chains in the regions where compliance is strongest. The pending rise to 45% is the most ambitious procurement-reform commitment of any large school feeding programme worldwide and will be closely watched. The question is no longer whether the model works but how robustly it can be scaled against entrenched centralised procurement interests.
Who Cooks the Meal — Corporate Versus Local
The contract caterers and the cost of consolidation
Three multinational caterers — Compass Group, Sodexo and Aramark — booked combined revenue of US$70.9 billion in FY2024 [13] ✓ Established. Education is the largest single Compass vertical in North America at 21% of revenue. The political and nutritional consequences of who actually cooks the school meal are now as important as the eligibility debate that dominates national politics.
The three companies that dominate global contract catering are not household names outside the industry, but they prepare an estimated 5.5 billion school meals a year between them. Compass Group reported US$28.58 billion in FY2024 revenue, up 10.5% organically, with the United States contributing 65% of the total and education accounting for 21% of North American revenue [13]. Sodexo reported US$24.9 billion, up 7.9%, with North America at US$11.65 billion. Aramark posted US$17.4 billion, up 10%, with US revenue around US$12.58 billion. Together the three account for roughly 30–35% of the global contract-catering market.
The arguments for outsourcing school catering are economic. A central kitchen serving twenty schools achieves procurement scale that no single municipal kitchen can match; standardised menus reduce labour costs and waste; the contract caterer absorbs the legal and HR overhead of a unionised kitchen workforce. UK procurement officers routinely cite per-meal savings of 10–25% relative to in-house provision. The Compass-Sodexo-Aramark model functions, in its preferred framing, as a public-private partnership that lets municipal authorities focus on education rather than food preparation.
When the same three caterers run school kitchens, prison kitchens, hospital kitchens and military kitchens across four continents, the question of what counts as food becomes a procurement question, not an educational one. That is the structural change of the last two decades.
— Nick Hewson, School Food Matters policy review, 2024The counter-arguments are nutritional, agricultural and political. The same scale economies that lower per-meal costs encourage centralised industrial menus — chilled, transported, reheated — that crowd out fresh local produce ⚖ Contested. The Brazilian family-farm mandate, by design, makes large-scale contract catering harder to operate [8]: a caterer required to source 30–45% of its school food from local smallholders cannot run a single centralised kitchen for an entire state. The French Egalim law has the same effect, more weakly enforced. The American system, by contrast, places no procurement floor on local sourcing and consolidates accordingly: districts that outsource to Aramark or Sodexo typically draw 80%+ of their menus from national wholesale supply chains.
The Case For Contract Catering
Central buying reduces per-meal ingredient cost by 10–25% relative to fragmented municipal kitchens.
Caterer absorbs kitchen HR, training and statutory compliance, freeing schools to focus on education.
Standardised menus designed by registered dietitians produce predictable macronutrient profiles.
Caterer invests in central kitchens and logistics; municipality avoids capital outlay on refurbishing school kitchens.
Food-safety, supply-chain and labour-shortage risks are contractually shifted to the caterer.
The Case Against Contract Catering
Central kitchens favour chilled, transported, reheated dishes that crowd out fresh local produce.
National wholesale procurement diverts spending out of the school's local agricultural and labour market.
Standardised menus erode regional food cultures and local culinary knowledge.
The municipality becomes a contract monitor rather than a kitchen operator, losing the daily quality signal.
Once a municipality outsources, returning to in-house provision is fiscally and politically difficult; renegotiation power shrinks with each cycle.
The risk profile of the contract-catering model is now well-documented. School Food Matters' UK review and parallel French municipal audits identify five recurrent risks: a thinning of local food culture, a procurement-monitoring deficit, the cumulative loss of in-house kitchen capacity, a transfer of nutritional decision-making to private contractors, and a political-economy risk in which the same three firms dominate education, health and prison food procurement ⚖ Contested. Each risk is real but variably severe; the risk-table below grades them on the published evidence.
| Risk | Severity | Assessment |
|---|---|---|
| Cultural and culinary homogenisation | Central kitchens converge on a narrow menu set engineered for low-cost transport and reheating; regional and seasonal variation erodes within a decade of outsourcing. | |
| Loss of in-house kitchen capacity | Outsourcing dismantles municipal kitchen labour and infrastructure; reversing the decision typically requires capital outlay and 3–5 years to rebuild operational capacity. | |
| Procurement-monitoring deficit | Municipalities become contract monitors rather than operators, losing the daily signal that in-house kitchens provide on quality, waste and pupil acceptance. | |
| Local economy leakage | Without a statutory local-sourcing floor (Brazil) or sustainability target (France), national wholesale procurement diverts school food spending out of the school's locality. | |
| Cross-sector market concentration | The same three caterers operate across schools, hospitals, prisons and military bases, narrowing the pool of credible alternative suppliers and weakening rebid competition. |
The Jamie Oliver reform of 2005 is instructive on the lock-in risk. Oliver's "Feed Me Better" campaign in the UK secured Tony Blair's government's commitment to ring-fenced school-food funding and a tightening of nutritional standards. Within two years, however, school-meal participation collapsed from 42% to 32% of pupils as parents — fronted by tabloid coverage — switched to packed lunches ◈ Strong Evidence. The lesson read off by procurement directors was not that nutritional standards failed but that contract caterers had built fragile demand: the moment the menu changed, the £10 million revenue loss within six months of the broadcast became a contractual problem rather than a public-health success.
The structural reading is that the contract-catering market is now large enough to function as a quiet veto on school food reform. A national caterer with multi-year contracts in 200 districts has more lobbying weight than any single municipality; its preferred menu architecture — frozen, central, low-margin per item — survives political alternation. Countries with a statutory counterweight (Brazil's family-farm floor, France's Egalim sustainability target, Japan's in-school nutritionist) preserve more menu diversity. Countries without such a counterweight (UK and US, at federal level) converge on the contract-caterer template by default.
What the Evidence Tells Us
Returns to feeding the next generation
The strongest single piece of long-run causal evidence on school feeding — Lundborg, Rooth and Alex-Petersen's study of Sweden's staggered postwar rollout — finds a 3% lifetime-income premium for pupils exposed to nine full years of the universal programme [11] ◈ Strong Evidence. Add the WFP's US$9-per-dollar global multiplier, China's documented test-score gains, and the UK's 14% absence reduction post-2005, and the case for universal school feeding is now empirically settled.
The economic case is stronger than it was a decade ago. Lundborg, Rooth and Alex-Petersen's Review of Economic Studies paper exploits the staggered rollout of Sweden's 1946-66 free school meal reform across 265 municipalities to identify a causal effect on cohorts exposed for the full nine compulsory years [11]. The headline finding — a 3% lifetime-income premium — is supported by ancillary results: greater adult height (around 1 cm), higher university attendance, and improved long-run health outcomes. The identification strategy survives standard robustness checks and the effect size is consistent with parallel quasi-experimental work on China's Nutrition Improvement Program, which raised high-school enrolment by 8.2% over the long term.
The Cohen et al. 2024 systematic review in JAMA Network Open covers a broader set of studies on universal free school meal policies in higher-income contexts [14]. The pooled findings — higher meal participation, improved attendance, modest gains in academic achievement, particularly concentrated in lower-income pupils — are weaker than Sweden's lifetime-income result but consistent with it. The systematic review notes that test-score effects are smaller and less reliable than attendance and participation effects, and the long-run human-capital effects (Sweden, China) have not yet been replicated in modern rich-country settings, partly because universal free meal policies have only recently been introduced in the United States and parts of Europe.
What is settled: school feeding increases participation, improves attendance, and produces measurable child-health and human-capital gains, with the strongest causal evidence from Sweden, China and India. What is unsettled: whether universal coverage outperforms targeted means-tested coverage on a strict cost-benefit basis, and whether contract-caterer industrial menus deliver the same long-run gains as municipal or in-house cooking. The remaining debates are about delivery architecture, not about whether the next generation should be fed at school.
The political-economy reading is that the empirical case has now overtaken the political case in most rich countries. The expansion of universal free meals across nine US states [14], the UK's Universal Credit-linked expansion to an additional 500,000 children from 2026-27 [10], Japan's planned 2026 national rollout [4], and the Brazilian family-farm floor's rise to 45% [8] together describe a trajectory in which the question is no longer whether to expand school feeding but how to design its delivery and procurement. The political fight is now downstream of the principle, where it was upstream of the principle a decade ago.
The Lund University identification of a 3% lifetime-income premium for Swedish cohorts exposed nine years to free school meals [11], combined with the WFP's US$9 multiplier per US$1 spent globally [1] and the China Nutrition Improvement Program's documented 8.2% gain in high-school enrolment, places the long-run benefit-cost ratio of universal school feeding well above 5:1 in the published literature. The headline US$40 billion-versus-US$18.7 billion estimate for the US programme is conservative against this benchmark, because it does not include the lifetime-income effects [6].
The unsettled question is whether the political coalition for school feeding survives the inflationary period the WFP's 2024 report documented. Global food-price volatility between 2022 and 2025 pushed many low- and middle-income school feeding budgets into structural deficit, and rich-country programmes — particularly in the US and UK — became politically vulnerable as the cost of providing each meal rose against frozen federal reimbursement rates. The 2025 US per-meal subsidy of US$3.69 is below the average reported cost of producing a school lunch in most districts [6], and the gap is widening ◈ Strong Evidence. Whether legislative reimbursement keeps pace with food inflation is the unsettled fiscal question that determines whether the empirical evidence translates into delivery.
The structural reading across eight countries is that school feeding works best where it is legislated as a public good, embedded in the education act rather than the welfare code, and procured through a mechanism that ties feeding to local agriculture or local labour. Japan codifies it as curriculum [4]; Finland and Sweden as universal entitlement [9][11]; Brazil as agricultural policy [8]; India as social development [2]. The countries that struggle — the United States most visibly [6][7], the United Kingdom in its restricted form [10] — are those where school feeding sits across multiple departments and survives on annual appropriations rather than dedicated statute.
The conclusion the evidence supports, in the language of this publication, is that the political argument over school lunch is now downstream of an empirical case that is effectively settled. A US$84-billion-a-year global system that feeds 466 million children, returns US$9 for every US$1 spent, and raises lifetime incomes by 3% in the cohorts that receive it through its full duration [1][11] is no longer a question for cost-benefit analysis. It is a question of who profits, who is excluded, and which procurement structure best transfers the gains to the children rather than to the contract caterer. That is the question the next decade of policy reform will answer.