The Visibility Standard: What BGN Must Publish Before Any MBG Canteen Pivot Scales

MBG Watch · 2026-08-11

The pivot is a data-design decision

Indonesia's National Nutrition Agency (BGN) is studying whether school canteens could become an alternative channel for the Free Nutritious Meals program. ANTARA reported on July 15 that BGN Deputy Head Agustina Arumsari said President Prabowo Subianto had opened the door to alternatives beyond Nutrition Fulfillment Service Units (SPPG), while asking that every option be backed by research and assessment before a final decision. The same report said the program targets 82.9 million beneficiaries and had resumed after a school-break suspension for evaluation. Source: https://en.antaranews.com/news/422828/indonesia-weighs-school-canteens-to-deliver-free-meal-program

That makes the canteen question larger than a choice between one kitchen model and another. It is a data-design decision. If procurement moves from centralized SPPG kitchens toward school canteens, the program may become more local and possibly more responsive. It may also become more fragmented, harder to audit, and easier for failures to hide at the school or vendor level.

MBG Watch does not take a position here on whether BGN should choose canteens, SPPGs, or a mixed model. The narrower accountability question is this: before a canteen pilot is judged safe, equitable, or ready to scale, what must the public be able to see?

Rupiah Stability Watch's July 28 crossing, "The MBG Canteen Pivot and the Rupiah," sharpened this point from the macro side. Its core finding was that opacity in a program this large does not stay inside the education or nutrition file. It can transmit through food demand, procurement credibility, fiscal expectations, and risk premium. MBG Watch's narrower child-meal standard is therefore simple: the model that feeds children must also make the meal, the money, and the incident trail visible enough to verify.

What is visible now, and what could disappear

The current SPPG model has not been sufficiently transparent. MBG Watch's prior pieces have documented the consequences: procurement chains that are hard to follow, kitchen approvals allegedly traded or rigged, forced tray purchases, hygiene and sanitation failures, weak outcome measurement, and evidence channels that can close just when scrutiny matters most.

But a centralized model at least has visible control points. There is a named SPPG. There is a kitchen location. There is a head of unit. There is an approval process. There are operational standards that BGN can suspend against. On July 27, BGN itself said it had suspended 833 SPPG for food-safety, hygiene, food-quality, and wastewater-treatment violations; ANTARA's English report and BGN's own release both cite the same number and describe stopped operations and payment consequences for noncompliance. Sources: https://en.antaranews.com/news/424181/indonesian-govt-freezes-833-mbg-kitchens-over-violations and https://www.bgn.go.id/news/siaran-pers/bgn-tindak-tegas-pelanggaran-internal-dan-sppg-demi-menjaga-integritas-program-makan-bergizi-gratis

A canteen model changes the audit map. Some risks could improve if local schools and parents can see meals before children eat them. But other risks become harder: more vendors, more price points, more local conflicts of interest, more small procurement decisions, and more difficulty knowing whether a child actually received a meal on the promised frequency.

BGN has already announced a useful first move. Its July 27 digital-transparency release says it is building a portal where parents can see which schools receive service, daily menus, the SPPG kitchen preparing the food, the responsible SPPG head, and service coverage. It also says a public dashboard will show the number of operating SPPG, their distribution, schools served, and national implementation progress. Source: https://www.bgn.go.id/news/siaran-pers/bgn-bangun-sistem-transparansi-digital-orang-tua-dapat-pantau-langsung-menu-mbg

That portal is a necessary start. It is not yet a visibility standard for a canteen pivot. Menus and kitchen names show what is served and who is nominally responsible. They do not, by themselves, show who owns the vendor, what was paid, what ingredients were purchased, whether the cold chain was maintained, whether the meal met nutrition composition, or whether incidents were corrected before expansion.

The minimum visibility standard

The table below is not a request for perfect data. It is the minimum public control set that should exist before any canteen pilot is scaled. Personally identifiable child data should remain private. Aggregated school, canteen, vendor, price, and incident data should not.

This standard is more demanding than a parent-facing menu portal, but it is also more useful. It treats transparency as an operating control. If a school canteen is safe and well-run, the data should help prove it. If it is unsafe, captured by a local vendor network, or delivering fewer meals than claimed, the data should make that visible before children bear the cost.

Four gates before scale-up

The canteen pivot should not scale as one political decision. It should pass four evidence gates.

Before a pilot starts, BGN should publish the participating schools, the delivery model at each school, vendor identities and beneficial ownership, food-safety certification status, inspection readiness, planned menus, expected beneficiary counts, and the price formula. Without these, a pilot begins as an opaque procurement transfer.

During the pilot, BGN should publish daily menus and meal counts, monthly price and procurement summaries, food-safety compliance rates, inspection outcomes, cold-chain compliance summaries, complaints, and illness reports. This is the period when the model is being tested. Waiting until the end converts monitoring into storytelling.

Before scale-up, BGN should publish a pilot evaluation that compares canteen, SPPG, and hybrid sites on at least six dimensions: reach, meal frequency, nutrition composition, food-safety incidents, per-meal cost, and vendor concentration. The evaluation should name negative results as well as positive ones. ANTARA reported on July 29 that the government is restructuring more than 13,000 MBG service centers and updating beneficiary data for 63 million recipients, with high-stunting regions prioritized. That is a large operating reset. It raises the cost of scaling before the measurement system is ready. Source: https://en.antaranews.com/news/424423/ri-accelerates-overhaul-of-13000-mbg-service-centers

After any incident, the public record should show the date, site, suspected failure mode, number of children affected if known, immediate suspension or substitution decision, laboratory testing status where relevant, corrective action, reopening criteria, and whether payments were stopped. A food-safety incident is not only a health event. It is a test of whether the program can learn.

How prior failures translate into canteen-pivot requirements

The forced ompreng-tray episode translates into equipment-procurement visibility. If canteens must buy trays, utensils, storage boxes, refrigerators, or stoves, the supplier, unit price, financing terms, and whether purchase is mandatory should be public.

The alleged rigging of SPPG approvals translates into approval-trail visibility. If a canteen is cleared to serve MBG, the evaluator, criteria, inspection date, failed items, and remediation should be visible in a machine-readable record.

The hygiene-certificate and suspension record translates into certification visibility. No school or canteen should be listed as active unless its food-safety status is current, and any exception should be explicit.

Cold-chain and heat-risk failures translate into time-temperature visibility. Perishable proteins, dairy, and cooked meals need logs that can be audited after a complaint. District-level compliance rates should be public even if raw logs remain inspection records.

Halted or weakened evidence collection translates into incident-record resilience. Reports, corrections, lab tests, and reopening decisions should be retained in a public archive so the accountability trail does not depend on one office's willingness to keep asking.

Unverified beneficiary counts translate into attendance-linked but privacy-protected service data. The public does not need child names. It does need the school denominator, meals prepared, meals served, opt-outs, missed deliveries, and disposal or leftover counts.

KPK's April 2026 study already pointed toward the same control logic. VOI's English report of the KPK findings said the program lacked short- and long-term success indicators and baseline measurements; flagged Banper-chain opacity, centralized partner and kitchen decisions, conflicts of interest, weak verification of foundations and kitchen locations, many kitchens failing technical standards, and limited Health Office and BPOM involvement in food-safety supervision. Source: https://voi.id/en/news/570797

What MBG Watch will monitor monthly

If BGN proceeds with a canteen pilot, MBG Watch will track whether the public record includes these fields, not only whether officials announce them.

The monthly watchlist should include: number of pilot schools by district; delivery model by school; beneficiary denominator and daily meals served; menu-nutrition completeness; average and range of per-meal costs; staple-input prices paid against district market references; vendor concentration and repeat ownership links; food-safety certification coverage; inspection pass and fail rates; cold-chain compliance summaries; complaint and illness counts; suspension, payment-stop, correction, and reopening records; and opt-out or missed-meal reasons.

The red flags are also concrete: expansion before baseline data is published; menu visibility without price or vendor visibility; active canteens without current food-safety status; repeated vendor names across many schools without ownership disclosure; per-meal prices that rise faster than local market references; high reported coverage with no opt-out or missed-meal accounting; incident reports that name illness but not correction; and reopened sites without stated criteria.

What this standard does not decide

This standard does not decide whether canteens are better than SPPG kitchens. Some schools may already have competent canteens, local trust, and shorter serving chains. Some may not. Some SPPG kitchens may be unsafe or captured; some may be disciplined and auditable. The evidence should sort those cases rather than assume them.

The standard also does not require BGN to expose private student identities, confidential whistleblowers, or raw medical records. The public-interest data are operational: school, vendor, price, menu, safety, incident, and aggregate beneficiary fields.

What the record supports is narrower and firmer: a canteen pivot that makes procurement and safety less observable is not yet a reform. It is a redistribution of opacity. A canteen pivot that publishes the minimum control record could become a genuine test of whether localization improves meals, safety, equity, and value for public money.

The policy choice remains with BGN and elected officials. The evidence condition should not. Before any scale-up, the public should be able to see the meal, the money, the vendor, the safety status, and the correction trail clearly enough to know whether children are being served well.