When Climate Stops Being an Exception: The Baseline Stress Test MBG Now Needs

MBG Watch · 2026-09-03

The premise

Indonesia's Makan Bergizi Gratis program is being scaled as a daily public-health operation, not as an occasional relief activity. That matters because climate stress is also becoming daily and operational. It does not arrive only as a named disaster. It arrives as a hotter kitchen, a school day shortened by smoke, a road that cannot be crossed after rain, a water source that is unsafe, a cold chain that lost power, or a family that cannot be reached before a substitute meal is delivered.

The global signal is no longer abstract. UN News, summarizing the UN's September 2026 climate warning, reported that the world is set to breach the 1.5°C limit within the next few years, with higher temperatures intensifying extreme weather, health risks, water stress, food-security pressure, and infrastructure disruption (UN News, 2 September 2026). That does not prove that any one MBG kitchen will fail. It does support a narrower accountability question: has BGN treated heat, haze, floodwater, water disruption, power loss, and disaster continuity as baseline operating conditions rather than exceptional footnotes?

MBG Watch has already examined many of these ledgers one by one: floodwater and the food-safety clock; haze and the school-day status record; smoke and outdoor exposure; heat at the kitchen door; operating-window shifts; water and wastewater at the kitchen tap; delivery routes as part of the kitchen; last-mile communications; compounding logistics shocks; and the measurement boundary for mothers, toddlers, and 3T areas in “Earlier Care Is Not Earlier Proof.” This synthesis does not repeat those pieces. It joins them into one test: before MBG delivery assumptions are treated as reliable through the 2027 budget cycle, the public should be able to see how climate-normal stress changes the meal clock, the worker-safety record, the route record, and the correction trail.

What the evidence supports

First, Indonesia has credible national and international evidence for rising heat-health concern. WHO's July 2026 account of Indonesia's KATALIS-HEAT launch says extreme heat is one of the deadliest climate-related hazards globally and that Indonesia's National Adaptation Plan 2026–2030 identifies heat exposure as a key climate risk. WHO reports that Indonesia's temperatures have already increased by about 0.81°C relative to the 1991–2020 baseline, with national mean temperatures projected to reach +1.0°C to +1.5°C by 2050; hot-day events are projected to double by 2050, and warm-night events may triple by 2100 (WHO Western Pacific, 23 July 2026). For MBG, the operating implication is not only child exposure. It is also kitchen heat, worker fatigue, cooling time, delivery timing, and the practical ability to keep freshly cooked food inside its safe window.

Second, worker heat is an operational safety issue, not a welfare add-on. WHO and WMO's 2026 guidance says the frequency and intensity of extreme heat events have risen sharply, increasing risks for outdoor and indoor workers, and that health risks include heatstroke, dehydration, kidney dysfunction, and neurological disorders. The same release says worker productivity drops by 2–3 percent for every degree above 20°C and recommends occupational heat-health policies tailored to local weather, jobs, and worker vulnerabilities (WMO, WHO/WMO worker heat-stress guidance). An MBG kitchen is a workplace. A baseline stress test therefore needs heat-hour records, rest and hydration controls, and clear restrictions on work that becomes unsafe.

Third, Indonesia's own meteorological agency is already producing the kind of operating signal MBG should be able to use. BMKG's Climate Outlook 2026 says most Indonesian regions are predicted to have average 2026 air temperatures in the 25–29°C range and, compared with the 1991–2020 normal, most regions are expected to be 0.2–0.6°C warmer. It also says annual rainfall is generally expected to be normal, while many areas may exceed 2,500 mm/year, and its sector recommendations cover agriculture, air quality, health, water resources, energy, and disaster management (BMKG Climate Outlook 2026 PDF). BMKG's rainy-season outlook also shows why national averages are not enough: it says the 2025/2026 rainy season began at different times across zones, was predicted to come earlier than usual in many areas, and had different peak timing between western Indonesia and southern/eastern Indonesia (BMKG rainy-season outlook).

Fourth, longer-term climate-risk assessments support planning for flood, drought, fire, haze, and water stress without pretending local certainty is higher than it is. The World Bank and Asian Development Bank's Indonesia Climate Risk Country Profile says Indonesia is highly vulnerable to floods, droughts, sea-level rise, rainfall shifts, and increasing temperature. It also states that precipitation projections are more uncertain than temperature, but that the intensity of sub-daily extreme rainfall appears to be increasing with temperature, while impacts remain highly dependent on local geography. The same profile reports high exposure to flooding, rising coastal-flood concern, water deficits in Java, Bali, East Nusa Tenggara, and parts of Sulawesi, and climate-linked fire and haze risk associated with drought and warmer dry-season conditions (World Bank/ADB Indonesia climate-risk profile).

Fifth, Indonesia already has public disaster and air-quality information systems that can be part of a monthly watchlist. BNPB describes InaRISK as a portal of disaster-risk assessments showing hazard coverage, exposed population, physical and economic loss potential, and environmental damage, intended for government, local government, stakeholders, and the public (BNPB InaRISK). BNPB's data portal offers integrated disaster datasets and dashboards, including disaster-event and impact datasets by type and region (BNPB Satu Data Bencana). BMKG's air-quality page provides three-day early warning for extreme air quality based on PM2.5 levels, smoke-particulate forecasts from forest and land fires, numerical pollutant dispersion modelling, and near-real-time PM2.5 monitoring from BMKG's national network (BMKG air-quality information). These systems are imperfect, but they show that MBG does not need to invent every signal from scratch.

Sixth, BGN has already acknowledged time and accountability as safety issues. ANTARA reported that BGN introduced a policy requiring MBG kitchens to state safe consumption time limits on every meal serving, with BGN's head saying the meals generally have a four-hour safe consumption window because they are freshly prepared without preservatives (ANTARA, 10 August 2026). ANTARA also reported BGN's statement that 93 percent of the 2026 national nutrition-program budget is distributed directly to regions through service units or MBG kitchens, and that budget distribution goes from KPPN to SPPG accounts (ANTARA, 26 July 2026). For 2027, ANTARA reported that the government proposed Rp240 trillion for MBG to serve 60.6 million residents nationwide, and that Treasury regional offices had begun surveying MBG kitchens for oversight (ANTARA, 1 September 2026). A climate baseline stress test belongs beside those financial and food-safety records, not outside them.

What the evidence does not show

The evidence does not show that every MBG kitchen faces the same climate hazard. It does not show that a national warming signal can be translated into a single stop/go rule for every district. It does not show that a normal annual rainfall forecast means daily road access, drainage, water quality, or wastewater handling will be normal at the hour a meal is cooked and delivered.

It also does not show that central punishment is the best answer to local ambiguity. A kitchen may be right to pause dispatch when floodwater enters a route. Another may be right to continue with a shorter route, a substitute menu, or a changed operating window if water, power, worker safety, and school status remain adequate. The public record should make those decisions visible and reviewable. It should not force local staff to hide uncertainty because they fear a punitive dashboard.

Nor does the evidence prove MBG outcomes under climate stress. Serving a tray during a hot, smoky, wet, or disrupted week is not the same as proving nutrition benefit, safe consumption, worker protection, and fiscal value under those conditions. That is why the stress test should be an operating-record standard, not a claim of success.

What a proportionate baseline stress test would contain

A useful MBG climate stress test would be local, monthly, auditable, and reversible. It would not require a perfect model. It would require every kitchen and route to keep the small set of records that make judgment possible.

1. Kitchen heat-hour threshold: indoor kitchen temperature, heat index where available, hours above the local threshold, mitigation used, rest/hydration actions, and any task restrictions.

2. Haze and school-status link: PM2.5 or official air-quality status, school opening or schedule status, outdoor queue and delivery exposure controls, and the decision made for that day's meals.

3. Water and wastewater status: potable-water source, interruption record, boil/treatment/substitution step if used, handwashing capacity, wastewater overflow or drainage problem, and clearance before restart.

4. Route passability: roads, bridges, docks, footpaths, and last-mile handoff points marked as open, delayed, blocked, or unsafe, with time stamps from dispatch to delivery.

5. Power and cold-chain continuity: outage start and end time, generator or backup status, refrigeration/cooling log, disposal decision if time-temperature control is breached, and restart sign-off.

6. Communications confirmation: which school, local government unit, driver, kitchen, health post, and parent-facing channel confirmed the changed status; which did not; and when the record was corrected.

7. Substitute-menu trigger: when the normal menu cannot be delivered safely, what shelf-stable, locally acceptable, nutritionally defensible substitute was used; who authorized it; and whether children with allergies or special needs were protected.

8. Worker-safety restriction: which tasks stop under heat, smoke, floodwater, electrical risk, gas risk, or unsafe travel; how workers report unsafe conditions without retaliation; and how staffing gaps are covered.

9. Beneficiary privacy boundary: in disaster or displaced settings, how eligibility, delivery, and complaints are recorded without exposing children, pregnant women, breastfeeding mothers, toddlers, or 3T households to unnecessary public identification.

10. Correction trail: every changed decision should show the first entry, the correction, the reason, the person or role authorizing it, and the time. A climate ledger without corrections will become a performance display, not an accountability record.

What should stay local and reversible

High-risk gates should be clear. If water is unsafe, wastewater has backed into the preparation area, floodwater contaminates the route or kitchen, power loss breaks time-temperature control, PM2.5 reaches an official dangerous category that makes the school day unsafe, or worker heat exposure crosses a defined danger threshold, the record should support pause, disposal, substitution, or rerouting without waiting for a national office to approve the obvious.

Ambiguous conditions should remain guided by human judgment. A forecast of heavy rain is not the same as a flooded route. A warm day is not automatically an unsafe kitchen. Moderate smoke may require shorter outdoor exposure, changed pickup timing, or school-specific judgment rather than a blanket cancellation. The better system is not the most centralized one. It is the one where local decisions are bounded by clear gates, backed by official signals, and reviewed through records that families, workers, local governments, auditors, and BGN can understand.

That is the least-harm path: stop when the hazard is clear; guide and record when the hazard is uncertain; correct quickly when the first decision was wrong.

What BGN should publish before 2027 assumptions are treated as reliable

Before Rp240 trillion and 60.6 million-beneficiary assumptions are treated as operationally reliable for 2027, BGN should publish evidence that the climate baseline has been stress-tested against actual kitchens and routes. At minimum, the public record should show:

• the number and share of SPPG kitchens with heat, water, wastewater, power, route, communications, and substitute-menu protocols tested;
• district-level monthly counts of climate-related delays, pauses, reroutes, substitutions, disposal decisions, and school-status changes;
• how often four-hour safe-consumption windows were missed or compressed by heat, route delay, school schedule change, or power interruption;
• worker-safety incidents or near misses linked to heat, smoke, floodwater, electrical interruption, or unsafe travel;
• cold-chain and refrigeration breach counts, with disposal and restart decisions;
• complaint and correction timelines, including whether families were notified before changed delivery or substitute menus;
• how climate disruption affects non-school beneficiaries, including pregnant women, breastfeeding mothers, toddlers, and 3T households, without using them as proof of nutrition outcomes before outcome data exists; and
• whether Treasury, BGN, local government, school, health, and disaster-management records can be reconciled when they describe the same disrupted meal day.

Monthly watchlist through the wet season and 2027 budget cycle

MBG Watch should monitor these signals monthly, and more often during named incidents:

• BMKG: seasonal outlook updates; heavy-rain warnings; extreme-weather warnings; temperature and heat-risk products; PM2.5 early warnings; smoke-particulate forecasts; and near-real-time PM2.5 monitoring.
• BNPB: disaster-event datasets; flood, landslide, drought, fire, and evacuation dashboards; InaRISK hazard updates; emergency-response dashboards; and district-level incident reports affecting schools, kitchens, roads, water, or power.
• Ministry of Health and local health offices: heat-health planning outputs, diarrhoeal disease and food-poisoning alerts, dengue signals during warm wet periods, and any guidance affecting school feeding or vulnerable groups.
• BGN: SPPG activation counts, suspended kitchens, hygiene certificate status, safe-consumption time compliance, kitchen-accounting records, food-safety investigations, and budget absorption by region.
• Ministry of Finance and Treasury regional offices: kitchen survey findings, KPPN disbursement timing, variance between planned and actual operating costs, and whether climate-related substitutions or pauses are visible in spending records.
• Education and local-government signals: school closure, shortened school day, remote learning, evacuation use of school buildings, and local route access decisions.
• Utilities and logistics records: power outages, water-service interruptions, road and bridge closure notices, fuel availability, and last-mile communications failures.

The accountability question is not whether MBG can avoid climate stress. It cannot. The question is whether BGN can show, calmly and concretely, how the program knows when to continue, when to change, when to pause, and how to correct the record after the day has moved.