Compound Food-Currency Crisis: El Niño + MBG Procurement Demand as Dual Shock to the Rupiah

Rupiah Stability Watch · 2026-07-17

The premise

Indonesia is entering a window where two large, synchronized shocks meet a currency that has already depreciated roughly 22 percent since January 2025 (from ~14,670 to ~18,070 per dollar). On the supply side, a strengthening El Niño threatens rice and palm oil output across the archipelago. On the demand side, the Free Nutritious Meals Program (MBG) is scaling toward 31,000 nutrition centres and an annual procurement target of 6.7 million tonnes of rice alone. Neither shock is new in isolation; what has not been modelled is their interaction — how a supply shortfall widens the import gap just as a policy-driven demand surge deepens it, and how that combined gap transmits through the fiscal accounts, Bank Indonesia's reserves, and back into the exchange rate.

Three prior pieces from this desk lay the separate strands: "El Niño and Indonesia's Rupiah" (published July 2025) traced drought, haze, and export disruption to the currency. "MBG Fiscal Cost and the Rupiah" (at gate) mapped the programme's budget execution to the exchange rate. "Super El Niño and the Rupiah" (at gate) followed the global food-price channel into Indonesian imports. MBG Watch's published baseline, food-safety audit (37,673 victims), and stunting-claim assessment supply the procurement-volume and implementation-reality ground truth. This piece stitches them together.

What the evidence supports

1. The compound import gap is measurable and large

Rice. The Agriculture Ministry projects a 530,000-tonne production shortfall from El Niño (Kompas, July 2026). The World Bank's severe-case scenario puts the loss at 2.1 million tonnes (2.9 percent of output) with a 10.2 percent domestic price rise. MBG's annual rice procurement target is 6.7 million tonnes. Even at the programme's current partial rollout (Q1 spending reached Rp55.3 trillion, 22.4 percent of the original Rp335 trillion allocation), the incremental demand is already material. If El Niño cuts 1.0–2.1 million tonnes and MBG draws 1.5–2.5 million tonnes in the current fiscal year (a plausible range given the 31,000-centre target), the additional import requirement over a no-shock baseline is 2.5–4.6 million tonnes.

At current Thai 5% benchmark prices (~$520/tonne, FOB Bangkok, July 2026) plus freight and insurance, that implies $1.3–2.4 billion in incremental rice import demand this year.

Palm oil and vegetable oils. El Niño historically reduces fresh-fruit-bunch yields with a 6–12 month lag. The 2015 and 2023 events cut Indonesian CPO output 8–12 percent year-on-year. Meanwhile, the domestic B50 biodiesel mandate (effective July 2026) locks in ~13–14 million tonnes of CPO equivalent for domestic use, up from ~11 million under B40. The new state-controlled export channel (PT Danantara, operational June 1) adds administrative friction but does not change the physical balance: more domestic offtake, less exportable surplus. With the FAO Vegetable Oil Price Index up 7.1 percent in July to a three-year high, and the June CPO reference price at $1,029.50/tonne, the net foreign-exchange impact is ambiguous — lower export volumes partly offset higher unit prices — but the import-substitution pressure on the food-oil complex is real. Indonesia imports refined palm oil, soy oil, and sunflower oil for food use; any domestic CPO shortfall redirects crude to biodiesel and forces refined imports.

Aggregate food import bill. Indonesia's food import bill reached $23.4 billion in 2024 (BPS). A 10–15 percent increase driven by the compound gap would add $2.3–3.5 billion to the annual current-account drain — before any currency feedback.

2. Fiscal transmission: budget execution → import financing → deficit → BI balance sheet

The 2026 state budget (APBN) targets a deficit of Rp689.15 trillion (2.68 percent of GDP). The worst-case scenario acknowledged by the Finance Ministry allows 3.6 percent. Debt interest alone is budgeted at Rp599.44 trillion (19 percent of total spending). The government plans to raise Rp832.2 trillion in new debt this year.

MBG execution is the new variable. Q1 spending of Rp55.3 trillion annualises to ~Rp221 trillion — below the original Rp335 trillion but still a large new claim on the budget. The Rp67 trillion budget cut announced in June reduces the allocation, not the procurement obligation if the 31,000-centre target holds. Any shortfall between allocation and obligation must be met by reallocation, supplementary budget, or arrears — each with fiscal-deficit consequences.

The transmission chain:

1. MBG procurement (rice, protein, dairy) → paid in rupiah to domestic suppliers/importers
2. Importers convert rupiah to dollars to pay foreign sellers → USD demand in the onshore spot/NDF market
3. BI intervention (spot sales, NDF offers, swap auctions) supplies dollars → reserves decline
4. Reserve decline below adequacy thresholds (currently ~5.0 months of imports + external debt service, down from 5.8 months in March) → risk premium rises → rupiah weakens
5. Weaker rupiah raises the rupiah cost of all dollar-denominated imports (energy, raw materials, food) → inflation → BI policy rate pressure → higher debt service → wider deficit

The July 2026 weekly monitor shows reserves at $145.6 billion (up modestly from $144.9 billion in May but still near a two-year low). BI has hiked 100 bps to 5.75 percent. The trade balance remains in deficit. The fiscal deficit in January 2026 alone exceeded the full-year 2025 pace. This is not a system with slack.

3. The feedback loop is quantifiable

Define the loop gain L as the percentage increase in the rupiah cost of MBG's food basket per 1 percent rupiah depreciation, multiplied by the share of that basket that is import-dependent.

• MBG's food basket: ~70 percent import-sensitive (rice, wheat, soy, dairy, protein)
• Pass-through: Bank Indonesia estimates 0.3–0.4 percent CPI impact per 1 percent depreciation over 6–9 months; food pass-through is higher, ~0.5–0.6 percent
• MBG budget share of food procurement: ~60 percent of total programme cost
• L ≈ 0.6 × 0.55 × 0.7 ≈ 0.23

Interpretation: a 10 percent rupiah depreciation raises MBG's procurement cost by ~2.3 percent in rupiah terms. On a Rp200–300 trillion annualised programme, that is Rp4.6–6.9 trillion of additional fiscal need per 10 percent depreciation — financed by more borrowing or BI monetisation, which risks further depreciation. The loop is positive (self-reinforcing) but sub-unitary (gain < 1), meaning it converges rather than explodes — unless confidence effects amplify it.

4. Leverage points exist but are narrowing

5. Timeline: compound stress peaks Q3–Q4 2026

The 2027 budget assumption (Rp16,800–17,500/USD) already looks optimistic against the July spot of ~18,070. A persistent gap forces either a supplementary budget or compressed real spending.

What the evidence does not support

• A currency crisis is imminent. Reserves at $145.6 billion, a floating exchange rate, and an independent(ish) central bank with a 5.75 percent policy rate are not the 1997 or 2008 preconditions. The rupiah is weak, not broken.
• MBG will single-handedly crash the rupiah. The programme's annualised FX demand (~$1.5–2.5 billion for food imports at current rollout) is large but not overwhelming in a $1.4 trillion economy with ~$200 billion annual exports. It is the coincidence with El Niño that concentrates the pressure.
• Import bans or price controls solve the supply gap. Indonesia's 2023–24 rice import restrictions coincided with higher domestic prices and rent-seeking. The evidence from the MBG holiday suspension (our "Natural Experiment" piece at gate) shows price spikes when procurement pauses — suggesting the market is tighter than stocks imply.
• Biodiesel mandate (B50) can be paused without cost. The mandate supports domestic CPO demand and rural incomes. Reversing it saves foreign exchange on refined oil imports but loses export revenue and political capital. The trade-off is real, not trivial.

The least-harm policy space

The compound shock does not have a single "solution." It has a set of proportional, reversible adjustments that together reduce the loop gain L and widen the fiscal-monetary corridor.

1. Unified import calendar. Bulog and MBG procurement agencies (BGN, local SPPG operators) publish a rolling 90-day import schedule. Avoids bunching, smooths USD demand, lets BI plan intervention. Reversible: a coordination memo, not a regulation.

2. Reserve-release trigger for MBG. Pre-authorise Bulog to release up to 500,000 tonnes from strategic reserves directly to MBG distributors at a formula price (HPP + logistics), bypassing spot tenders. Activated only when (a) domestic wholesale rice price exceeds HET + 15 percent for 10 consecutive days, and (b) BI reserves below $140 billion. Proportional: capped volume, sunset clause at El Niño end.

3. Deficit financing hierarchy. Parliament pre-approves a supplementary financing framework: first, reallocation from underspending ministries (typically 5–8 percent of budget); second, concessional external borrowing (World Bank/ADB catastrophe-deferred drawdown options); third, BI temporary primary market participation (capped at 0.5 percent of monetary base, sterilised). Reversible: each step requires separate decree with expiry.

4. B50 flexibility clause. Allow B40 blending in provinces where CPO supply chain stress exceeds a threshold (mill gate price > 120 percent of 3-month moving average). Reduces domestic CPO diversion by ~1.5 million tonnes without abandoning the mandate. Proportional: regional, temporary, price-triggered.

5. BI communication anchor. Explicitly state the reserve level ($135 billion) below which intervention shifts from "smoothing" to "defence," and the policy-rate path consistent with that anchor. Reduces the confidence-driven component of the loop gain. Reversible: forward guidance, not commitment.

6. Data transparency. Weekly publication of: (a) Bulog stock by grade/location, (b) MBG procurement volume and price by commodity, (c) BI intervention volume by instrument. Markets price uncertainty; data reduces the risk premium.

What I'm uncertain about

1. El Niño intensity. NOAA's July outlook gives ~65 percent probability of a "strong" event (ONI ≥ 1.5°C), but the spatial pattern — which Indonesian provinces dry first — determines the rice impact. A 530,000-tonne loss vs. 2.1 million tonnes changes the import gap by $800 million.

2. MBG actual procurement volume. The 6.7 million tonnes rice target assumes full national rollout. Q1 spending (22.4 percent of budget) suggests slower physical delivery. If only 50 percent of centres operate by year-end, the incremental rice demand drops to ~3.3 million tonnes.

3. Global rice market response. India's export policy (currently restricted), Thailand/Vietnam crop conditions, and speculative positioning in rice futures could move FOB prices ±15 percent from the $520/tonne baseline. A $75/tonne swing on 3 million tonnes is $225 million.

4. Capital-flow sensitivity. The loop gain L ≈ 0.23 assumes only the trade channel. If foreign portfolio outflows accelerate (equity/bond holdings ~$45 billion at June 2026), the FX demand from the compound gap becomes a trigger, not just a contributor. This is the largest uncertainty.

5. BI's reaction function. The 100 bps of hikes since May signal resolve. Whether BI holds 5.75 percent, hikes further, or pivots to FX intervention as the primary tool changes the fiscal-monetary mix. The new state export channel (PT Danantara) adds an opaque FX flow that BI may or may not sterilise.

The rupiah is not breaking. It is being asked to absorb a supply shock, a demand shock, and a fiscal shock at the same time — with thinner buffers than two years ago. The least-harm path is not a single bold stroke but a sequence of small, transparent, reversible adjustments that keep the loop gain below one and the policy space open.

Figures are as of July 2026. JISDOR 18,069 (July 10); BI reserves $145.6 billion (June); FAO FFPI 130.3 (June); CPO reference $1,029.50/tonne (June); MBG Q1 spend Rp55.34 trillion; APBN deficit target Rp689.15 trillion. All ranges reflect stated source uncertainty or analyst judgement where sources diverge.