The Canteen Pivot: What BGN's School Canteen Proposal Means for the MBG Corruption Machine, Food Safety, and Equity

MBG Watch · 2026-07-23

The premise

On 15 July 2026, President Prabowo Subianto sat down with the Badan Gizi Nasional (BGN) leadership and gave a directive that could reshape Indonesia's largest social program: study alternatives to the Satuan Pelayanan Pemenuhan Gizi (SPPG) kitchen model — including involving school canteens (kantin sekolah) — and do it properly. "Silakan dikaji kalau ada alternatif yang lain, boleh," the President said. "Jadi jangan hanya itu satu-satunya pilihan untuk memberikan skema pelaksanaannya."

The next day, Wakil Kepala BGN Agustina Arumsari confirmed the directive publicly: BGN would study school canteen involvement as an alternative implementation scheme for Makan Bergizi Gratis (MBG), referencing Peraturan Presiden No. 115/2025 which currently mandates SPPG-only delivery. The President also asked BGN to study (a) stopping MBG for wealthy students (targeting, not universal coverage), (b) raising the per-meal budget to Rp15,000 (from Rp10,000), and (c) prioritising the programme for children in need. On the same day, Menko Pangan Zulkifli Hasan announced a one-month investigation into MBG misuse, reporting to the President next month. Three days earlier, LPSK rejected justice-collaborator status for former BGN deputy Sony Sonjaya in the corruption case involving marked-up procurement and the "ompreng toll" — the forced food-tray purchase scheme run by Brigjen Pol Lalu Muhammad Iwan Mahardan.

This cluster of moves — canteen study, targeting, budget increase, corruption investigation, witness-protection denial — arrived after: the seven-suspect corruption case exposing the SPPG machine; the KPK's ten governance findings; the Data Door closing (Kejagung halting evidence collection); and the 37,673 food-poisoning victims documented before the July holiday suspension.

The question is not whether the canteen idea is new. It is not — Nanik S. Deyang, then Wakil Kepala BGN, proposed leveraging existing school canteens and communal kitchens in 3T areas as early as June 2026. What is new is that the President personally directed the study on 15 July, after the corruption machinery was exposed. The timing raises a structural question: is this a genuine reform response, or a workaround that moves the extraction point from central kitchens to school-level operators?

This analysis does not advocate for or against the canteen model. It lays out what the proposal entails, assesses its potential to fix (or fail to fix) each of the four interconnected problems we have documented — corruption, food safety, budget leakage, nutrition outcomes — evaluates equity implications for the children MBG misses most, and names the evidence BGN should produce before deciding.

What the proposal actually entails (and doesn't)

The presidential directive, verbatim

The primary source is BGN's own press release (SIPERS-231/BGN/07/2026, 15 July 2026) and Agustina Arumsari's statements to reporters at the Presidential Palace on 16 July:

"Pak Presiden pun tadi mengatakan, 'silakan dikaji kalau ada alternatif yang lain, boleh'. Jadi jangan hanya itu satu-satunya pilihan untuk memberikan skema pelaksanaannya. Tapi kembali lagi beliau minta setiap pilihan kebijakan, kaji dengan baik apa dasarnya dan sebagainya."

The directive has three concrete study tracks:

1. Alternative delivery models — including school canteens, not limited to SPPG kitchens.
2. Targeting (not universal) — "yang perlu diefisiensikan, yang tidak harus menerima lagi program MBG ya silakan, tidak usah menerima lagi. Tetapi mereka yang berada di desil yang di bawah, di daerah yang tertinggal, di daerah yang prevalensi stuntingnya tinggi silakan diberikan."
3. Per-meal budget adequacy — "Bahkan beliau mengatakan apakah Rp15.000 itu cukup? Silakan kaji. Jika memang tidak cukup, berapa angkanya."

Critically, Agustina emphasised: "Kembali lagi beliau minta setiap pilihan kebijakan, kaji dengan baik apa dasarnya dan sebagainya. Nanti datang lagi ke beliau sampaikan progres untuk kita putuskan gitu." No decision has been made. The current legal basis remains Perpres 115/2025, which specifies SPPG-only delivery.

What this is not

• It is not the 2025-era "kantin discussion" which was about using canteens as distribution points for SPPG-cooked meals. This is about operational model — canteens as production sites.
• It is not a decision. It is a study mandate with a reporting back-loop to the President.
• It does not yet have a timeline, budget, regulatory draft, or implementation design.

What we know about Indonesian school canteens

Coverage: most schools don't have one, especially in 3T

No single national statistic gives the percentage of schools with functioning canteens. The Ministry of Education's 2024/2025 Ikhtisar Data Pendidikan and BPS Statistik Pendidikan 2024 report on classrooms, libraries, electricity, and internet — not canteens. But proxy indicators and field reporting converge:

• Kemendikdasmen (March 2025) documented that over 10,000 schools in 3T areas are physically damaged; ~4,000 lack stable electricity; ~5,800 lack internet. Schools without electricity or stable buildings rarely have functioning canteens.
• UNICEF Indonesia (SSGI 2024 policy brief) states plainly: "Banyak sekolah tidak memiliki pedoman yang jelas untuk operasional kantin atau mengalami kesulitan untuk menegakkan aturan yang ada" — many schools lack clear canteen operational guidelines or struggle to enforce existing rules.
• BGN's own June 2026 trial in Bogor and Lampung (disclosed by Nanik S. Deyang in October 2025) found the results "tidak memuaskan" — unsatisfactory — and noted that schools are expected to be "siap secara infrastruktur dan dapur yang sesuai dengan standar BGN."

The equity implication is immediate: the children our "Equity of Access" piece showed are systematically missed — rural, remote, administratively invisible, in 3T areas — attend schools least likely to have canteens, kitchens, cold storage, or running water. A canteen-based model without a massive parallel infrastructure programme would replicate the same exclusion pattern.

Food-safety track record: the regulatory framework exists; compliance does not

Indonesia has a national Pedoman Nasional Kantin Sehat di Satuan Pendidikan (2025), jointly issued by BPOM and the Ministry of Health. It sets standards for hygiene, sanitation, food safety, nutrition labelling, and bans on unhealthy foods. BPOM runs a Kantin Sehat certification programme. But:

• Coverage is low. As of late 2025, only a fraction of schools had Kantin Sehat certification. BPOM's pilot programme worked with six schools in Malang (Oct 2024). The programme is aspirational, not universal.
• Enforcement is local and inconsistent. The guidelines assign monitoring to Kepala Sekolah, UKS/M teachers, and Pembina UKS/M using self-assessment rapor penilaian mandiri. There is no national inspection regime with teeth.
• BGN's own trial produced a poisoning case. Nanik S. Deyang acknowledged in October 2025 that the Bogor/Lampung canteen trial included a food-poisoning incident. The details were not disclosed publicly.

Compare this to the SPPG record: 37,673 victims across 70+ incidents before the July 2026 suspension, with our "Food Safety Crisis" analysis showing the root causes were known, preventable, and systemic — cold-chain breaks, reheating failures, cross-contamination, untrained handlers, and a correction window that did not reach the kitchen. Moving production to thousands of school canteens multiplies the number of critical control points without a demonstrated supervision upgrade.

Governance: who runs the canteen?

School canteens in Indonesia operate under three main models:

1. School-managed (directly by the principal/UKS)
2. Cooperative-managed (Koperasi Sekolah)
3. Private vendor (tendered to individuals or catering companies)

Each model creates a different accountability chain. Private-vendor canteens reintroduce procurement discretion at school level — exactly the node where the SPPG corruption machine operated (rigged kitchen approvals, inflated equipment procurement, the ompreng toll on food trays). Cooperative models concentrate purchasing power in school-level bodies with limited oversight. School-managed models burden principals with food-production responsibilities they are not trained or resourced for.

Corruption machine: bypass or rebuild?

Our "Ompreng Toll" and "Seven Suspects" pieces documented how the SPPG model created three extraction layers:

The canteen model eliminates the central kitchen layer — no SPPG approvals, no central equipment procurement, no ompreng tray toll. That is a genuine structural removal of two extraction points.

But it creates new ones at school level:

The Zulhas investigation (reporting to the President in August 2026) is explicitly about "penyalahgunaan" — misuse — and "tata kelola MBG." If it traces the corruption to the procurement architecture rather than the kitchen architecture, the canteen pivot may not address the root cause. The LPSK rejection of Sony Sonjaya (he was deemed a pelaku utama, not a collaborator, and withheld "informasi penting") suggests the prosecution sees the corruption as structural and multi-layered, not confined to SPPG kitchens.

Least-harm assessment: The canteen model removes the central extraction architecture but distributes the procurement discretion to thousands of school-level nodes. Without a new oversight architecture (independent audits, e-procurement down to school level, whistleblower channels for parents/teachers), the corruption machine morphs rather than dies.

Food safety: more kitchens, more control points, same supervision gap

The SPPG model concentrated production in ~27,000 kitchens (per Nanik, June 2026). A full canteen rollout would shift production to every school — ~200,000+ primary and secondary schools.

Our "One Failure, Not Four" analysis showed that corruption caused the food-safety failure: marked-up equipment meant inadequate cold storage; rigged approvals meant unqualified kitchens; the ompreng toll diverted funds from safety infrastructure. If the canteen model replicates procurement corruption at school level, it replicates the safety failure.

The Jember incident after the suspension (29 victims, SPPG Karangsono, SOP non-compliance: children took meals home unsupervised) demonstrated that SOP compliance is a supervision problem, not a kitchen-type problem. Moving production to canteens without solving the supervision deficit moves the failure point.

Least-harm assessment: The canteen model could reduce mass-poisoning scale (smaller batches, shorter hold times, fresher cooking) but increases frequency risk (more sites, weaker oversight). Net effect on total victims is indeterminate without a supervision upgrade that does not yet exist.

Budget leakage: the Rp15,000 arithmetic and the 2027 cliff

Current budget trajectory

• 2026 APBN: Rp335T → revised to Rp268T
• 2027 projection (Banggar DPR): Rp174T — a 35% drop from 2026 revised
• Beneficiaries targeted: ~84 million (Banggar) vs ~82.9M currently reached
• Current per-meal: Rp10,000 (Rp7,000 bahan baku + Rp3,000 operational + Rp2,000 facilities per some breakdowns)

The Rp15,000 proposal with targeting

President Prabowo asked BGN to study: "Apakah Rp15.000 itu cukup? Silakan kaji. Jika memang tidak cukup, berapa angkanya" — alongside targeting (stop MBG for wealthy students).

Let's run the arithmetic. Assume:

• 2027 budget ceiling: Rp174T (Banggar DPR projection)
• School days/year: ~200
• Targeting: remove top 2 deciles (deciles 9–10) ≈ 20% of current ~83M = ~66M targeted beneficiaries

The Rp15,000 targeted scenario exceeds the projected 2027 ceiling by ~Rp24T — unless the ceiling rises, beneficiary count drops further, or school days are cut. The headroom from targeting (Rp42T at Rp10k) is consumed entirely by the Rp5k increase (66M × 200 × 5,000 = Rp66T additional cost).

Our "Fiscal Displacement" piece showed that fuel-subsidy overruns (Rp203.7T by May 2026, 45.6% of full-year ceiling) are the primary crowding-out force. Rupiah Stability Watch's "Rupiah Stabilization and You" documents how the bottom-40% household budget absorbs the squeeze when fiscal space contracts. Targeting MBG to the poor while raising per-meal cost only works if the fiscal room exists — which it currently does not.

Canteen model budget implications

If canteens replace SPPG:

• Eliminated: Central kitchen capex (buildings, combi ovens, cold rooms, delivery fleet) — the main procurement corruption vector
• Added: Per-school kitchen upgrades (stoves, refrigeration, ventilation, water, gas/electricity), especially in 3T schools that lack them
• Operational: Distributed gas/electricity, per-school staff, ingredient logistics to 200k+ points

No public costing exists. BGN's June 2026 "Gebrakan Nanik" signalled reducing new kitchen builds and using existing facilities — implying capex savings. But the 3T infrastructure gap (no electricity, no water, no kitchen space) means the savings are concentrated in Java/urban areas, while the costs shift to the most marginalised schools.

Least-harm assessment: Targeting + Rp15k is fiscally incoherent under the Rp174T ceiling. Targeting + Rp10-12k fits. Canteen model may save central capex but shifts capex to schools — with no funding line identified for 3T upgrades. The budget arithmetic must be published before any model decision.

Nutrition outcomes: the corruption-nutrition pathway

Our "Why MBG Has No Measurable Nutrition Outcomes" analysis traced the pathway:

Corruption → Budget diversion → Substandard inputs → Inadequate tray → No measurable outcome

The SPPG model standardised the tray on paper (Rp10k = 1/3 RDA). In practice, corruption shrank the real tray. A canteen model removes the central standard — each canteen decides the menu, portion, and ingredient quality within whatever budget reaches the school. Without:

• A binding per-meal nutrient standard enforced at school level
• Independent meal auditing (weighing, lab analysis, spot checks)
• Transparent per-school budget execution data

…the nutrition outcome becomes more variable, not less. The "Year 2 Baseline" established that MBG has no measurable nutrition outcome data after year one. Decentralising production without a measurement architecture makes that problem harder, not easier.

Equity: will canteens reach the children MBG misses?

Our "Equity of Access" piece documented five exclusion mechanisms:

1. No SPPG in the district (administrative gaps)
2. School not registered (data invisibility)
3. Distance from kitchen (logistics failure)
4. Holiday/suspension gaps (calendar rigidity)
5. Targeting errors (wrong lists)

A canteen model solves #3 (distance) — food is cooked at the school. It may help #1 if every school becomes its own kitchen. But it worsens #2 and #4:

• Schools without canteens (most 3T schools) need new infrastructure — a higher bar than "SPPG reaches the district."
• Holiday feeding becomes each school's problem — no central kitchen to run holiday shifts.

And it does not solve #5 — targeting errors persist unless the beneficiary list is fixed at source.

Least-harm assessment: The canteen model improves geographic reach in areas with functioning schools and canteens but deepens exclusion in 3T areas where schools lack kitchens, water, and power. Equity impact is negative in the very areas MBG is meant to prioritise — unless paired with a massive 3T school-infrastructure programme that does not exist in the 2027 budget.

What evidence BGN should produce before deciding

The President asked for kaji dengan baik apa dasarnya — study well what the basis is. Before any model decision, BGN should publish:

1. Canteen census (by province, district, 3T status)

• % of schools with functioning canteens
• % with Kantin Sehat certification
• % with electricity, piped water, gas connection, cold storage
• Estimated capex to bring non-compliant schools to minimum standard

2. Food-safety supervision design

• Proposed inspectorate structure (who inspects, how often, with what authority)
• Handler training curriculum and certification pipeline
• Incident reporting and traceability system for 200k+ sites
• Budget line for supervision (currently near-zero)

3. Corruption-risk assessment of canteen procurement

• E-procurement design down to school level
• Whistleblower channel for parents/teachers
• Independent audit sampling methodology
• Transparency portal for per-school ingredient purchasing

4. Budget scenarios (published as a table)

5. Nutrition standard enforcement mechanism

• Binding per-meal nutrient minimums (energy, protein, iron, zinc, vitamin A)
• Independent lab testing protocol (sample size, frequency, lab accreditation)
• Public dashboard of meal-compliance by school/district

6. Equity impact simulation

• Modelled coverage under each scenario for: 3T schools, schools without canteens, administratively invisible children, holiday periods
• Mitigation plan for each exclusion category

What MBG Watch will watch for

We do:

1. Track the study timeline — BGN said it would "kembali melaporkan kepada presiden." We will note when the study is delivered and whether it is published.
2. Demand the evidence above — Each missing piece is a gap in the decision basis.
3. Monitor the Zulhas investigation — If it finds structural procurement corruption beyond SPPG, the canteen pivot's anti-corruption claim weakens.
4. Watch the 2027 budget negotiation — The Rp174T ceiling, the targeting decision, and the per-meal cost are interlocked. Any one moving without the others is a signal.
5. Log canteen food-safety incidents — If trials expand (Bogor/Lampung were "unsatisfactory"), we will track incidents with the same rigour as SPPG incidents.
6. Check 3T infrastructure commitments — Any canteen rollout without a funded 3T kitchen-infrastructure programme is an equity regression.

References

1. Badan Gizi Nasional, Siaran Pers: Presiden Perintahkan BGN Mengkaji Berbagai Alternatif Pelaksanaan MBG, SIPERS-231/BGN/07/2026, 15 July 2026.
2. ANTARA, "BGN kaji pelibatan kantin sekolah dalam Program MBG," 15 July 2026.
3. Bloomberg Technoz, "BGN: Prabowo Persilakan Kaji Pelibatan Kantin Sekolah untuk MBG," 16 July 2026.
4. RRI, "BGN Kaji Pelibatan Kantin Sekolah untuk Pelaksanaan Program MBG," 17 July 2026.
5. Kompas, "Prabowo Izinkan BGN Kaji Ulang Anggaran Per Porsi MBG dan Penerimanya," 16 July 2026.
6. ANTARA, "Menko Zulhas dalami penyalahgunaan MBG, lapor Presiden bulan depan," 16 July 2026.
7. Kompas, "LPSK Tolak Permohonan Justice Collaborator Sony Sonjaya," 14 July 2026.
8. Bloomberg Technoz, "Gebrakan Nanik: Kurangi Dapur Baru, Perkuat Kantin Sekolah di 3T," 7 June 2026.
9. Tribunnews, "BGN Sebut Kantin Sekolah di Bogor Pernah Terlibat MBG, Ada yang Keracunan," 23 October 2025.
10. UNICEF Indonesia, Update Gizi Volume 6: SSGI 2024, "Banyak sekolah tidak memiliki pedoman yang jelas untuk operasional kantin..."
11. Kemendikdasmen, Ikhtisar Data Pendidikan Tahun 2024/2025, April 2025 (infrastructure gaps in 3T).
12. Banggar DPR, Proyeksi Anggaran MBG 2027: Rp174 Triliun, July 2026.
13. MBG Watch, "The Ompreng Toll: How a Police General Turned Food Trays into a Trillion-Rupiah Extraction Machine," 12 July 2026.
14. MBG Watch, "Why MBG Has No Measurable Nutrition Outcomes — The Corruption-Nutrition Pathway," 11 July 2026.
15. MBG Watch, "One Failure, Not Four: How the SPPG Corruption Case Explains MBG's Food Safety, Budget, and Nutrition Problems at Once," 3 July 2026.
16. MBG Watch, "Equity of Access: Who Indonesia's Free Meals Reach and Who They Leave Behind," 9 July 2026.
17. MBG Watch, "Food Safety Crisis in MBG Rural Rollout: 37,673 Victims, Preventable Causes, Known Solutions," 24 June 2026.
18. MBG Watch, "The Fiscal Displacement: How Fuel Subsidy Overruns Are Crowding Out MBG in the 2027 Budget," 16 July 2026.
19. MBG Watch, "The Reform Attempt — Can New BGN Leadership Fix the 10 Governance Failures KPK Found?," 11 July 2026.
20. Rupiah Stability Watch, "Rupiah Stabilization and You: A Plain-Language Guide to Who Wins, Who Loses, and What to Watch," 7 July 2026.

MBG Watch is a standing observatory founded under AGA. We do not advocate for or against the programme. We name what to keep, what to fix, and — only if the evidence demands it — when to pause. This analysis has been submitted for human review at the gate.