Who Is Buying the Rupiah? Capital-Account Composition Through the Hormuz Escalation

Rupiah Stability Watch · 2026-07-23

The Premise

Three publications at the gate document the same paradox: the rupiah appreciated through seven consecutive nights of US strikes on Iranian targets around the Strait of Hormuz, reaching ~17,940–17,950 per dollar while oil held above $85 and regional currencies softened. The observation is solid. The mechanism is not.

This piece traces the capital-account anatomy of that resilience. The question is not whether the rupiah held—it did—but who was buying, what they bought, and whether the buying is structural or reversible. The answer determines whether the current level is a new equilibrium or a carry-trade crowded trade waiting to unwind.

The short verdict: the Hormuz resilience was funded primarily by a USD 8.48 billion surge into Bank Indonesia Rupiah Securities (SRBI) in Q2 2026—a high-yield, short-tenor carry trade attracted by a 200+ basis-point premium over the Fed. Equity flows remained negative (USD 2.3 billion outflow in Q2). Government bonds (SBN) saw a modest USD 1.78 billion inflow, recovering from a 19-year low in foreign ownership documented in our Pacific Security piece. BI's reserves rose to USD 145.6 billion in June only because inflows exceeded the central bank's intervention sales. GR 24/2026's export-centralization mechanism is generating structural USD-IDR conversion, but the volume is not yet large enough to explain the exchange-rate level. Global EM funds are underweight Asia but overweight Indonesia by 0.9 percentage points—a positional anomaly that could reverse.

The rupiah's Hormuz strength is real but compositionally fragile. It rests on hot money, not deep money.

What the Evidence Supports

1. Q2 2026 Capital Flows: The Carry Trade Dominated

Bank Indonesia Governor Perry Warjiyo's July 7 testimony to the DPR Budget Committee provides the most granular official breakdown to date. The numbers are stark:

Source: BI Governor Warjiyo, DPR Working Meeting, July 7, 2026. Exchange rate used: IDR 17,960/USD.

SRBI alone accounted for 106% of total Q2 net inflows. The 12-month SRBI yield reached 6.45% in July, against a BI rate of 5.75% and a Fed funds range of 3.50–3.75%. That is a 200–250 bp carry on a central-bank-issued, rupiah-denominated instrument with tenors as short as 6–12 months. For a global carry trader, it is a near-ideal vehicle: high yield, high liquidity, explicit central-bank backing, and no duration risk.

The equity outflow is equally telling. Our "Governance Risk Premium Deepens" piece documented USD 3.9 billion in stock outflows through late June—the largest since 1997. The Q2 data shows the bleed continued at USD 2.3 billion. Foreign investors are not buying Indonesian equities; they are selling them even as the currency strengthens. That divergence is a warning signal, not a contradiction.

SBN inflows of USD 1.78 billion represent a genuine recovery from the 19-year low in foreign ownership we documented in "Pacific Security Realignment." But at USD 1.78 billion, they are one-fifth the SRBI volume and barely offset the equity outflow.

2. BI Reserves Rose Because Inflows Exceeded Intervention, Not Because Intervention Stopped

BI's international reserves reached USD 145.6 billion at end-June 2026, up USD 700 million from May's USD 144.9 billion—the first monthly increase in 2026 after five consecutive declines. The Tempo report of July 7 quotes BI Executive Director Ramdan Denny Prakoso: the increase was "mainly influenced by tax and service receipts amid the government's foreign debt payments and Bank Indonesia's exchange rate stabilization policy."

The phrasing matters. BI acknowledges it intervened. The reserve increase means gross inflows (SRBI + SBN + tax receipts) exceeded gross outflows (debt service + intervention sales). BCA Research's July 8 note "FX Reserves: The Cost of Calm" confirms the reserve recovery was "primarily driven by debt-based inflows, specifically higher foreign investment in SRBI and SBN."

We do not have a daily intervention ledger. But we can infer: if reserves rose USD 700 million while Q2 portfolio inflows totaled USD 7.98 billion (mostly in April–June), and government external debt payments typically run USD 1–2 billion monthly, then BI likely sold USD 3–5 billion in the spot/NDF/DNDF markets during Q2 to prevent the rupiah from strengthening further. The rupiah's stability at ~17,950 was not a free-market outcome; it was a managed outcome funded by carry-trade inflows.

3. GR 24/2026 Export Centralization Is Generating Structural Demand—But the Volume Is Still Small

Government Regulation 24/2026, effective June 1, mandates that exports of coal, palm oil, and ferro-alloys flow through state-owned enterprises (SOEs) under the Danantara Sumber Daya Indonesia (DSI) umbrella, with 100% of export proceeds (DHE) channeled through state-owned banks and converted to rupiah.

This creates a mechanical, non-discretionary source of USD-IDR conversion. Every tonne of coal or palm oil exported now generates rupiah demand that did not exist under the old fragmented system. Our July 3 piece "Indonesia's New State-Controlled Export Chain" estimated the annualized export value of covered commodities at ~USD 60–70 billion. If even half converts through the new channel, that is USD 30–35 billion of structural annual demand—roughly USD 2.5–3 billion monthly.

But implementation is phased. Reporting from HSF Kramer (July 9) and ABNR (June 9) indicates the DSI licensing and SOE onboarding process is still ramping. The actual forced conversion in June–July 2026 is likely well below the steady-state potential. We estimate USD 0.5–1.0 billion monthly in the current quarter—real, structural, and growing, but not the primary driver of the ~17,950 level.

4. Real-Money Positioning: Underweight Asia, Overweight Indonesia

Nomura's July 2026 EM fund survey (via Investing.com) finds emerging-market funds underweight Asia broadly—reducing India, China, and Taiwan exposure. But Indonesia is an exception: 0.9 percentage points overweight relative to the MSCI EM benchmark. Korea is the only other Asian overweight (+0.8 pp).

This is a positional anomaly. It suggests global allocators are using Indonesia as a "clean" EM proxy—high carry, commodities, reform narrative—while reducing broader Asia risk. The overweight is modest in absolute terms (EM funds manage ~USD 320 billion per Jefferies' March 2026 survey), but it is directional: real money is adding to Indonesia while cutting peers. If risk sentiment deteriorates, this pocket of overweight becomes a source of selling pressure, not support.

5. Foreign Bond Ownership: Recovering from a 19-Year Low, But Still Historically Depressed

The "Pacific Security Realignment" piece documented that foreign SBN ownership fell to a 19-year low of ~13.5% in May 2026. The Q2 inflow of USD 1.78 billion likely pushed this to ~14.5–15% by end-June—still far below the 2017–2019 range of 35–40%. The recovery is real but embryonic. SBN duration risk (7–10 year tenors dominate foreign holdings) makes them sensitive to US rate expectations. A Fed hold or hike would stall this recovery.

What the Evidence Does Not Support

The rupiah's resilience is not driven by equity-market confidence. The USD 2.3 billion Q2 equity outflow—on top of the USD 3.9 billion documented through late June—means foreign portfolio managers are reducing Indonesia equity exposure while the currency strengthens. That is a bearish signal masked by a bullish price.

BI intervention has not ceased. The reserve increase reflects net inflows exceeding intervention sales, not an end to intervention. BI's July 7 statement explicitly cites "exchange rate stabilization policy" as a reserve use. The central bank is leaning against both depreciation and excessive appreciation—a two-way management that caps volatility but consumes reserves when flows reverse.

GR 24/2026 has not yet created a structural floor at 17,950. The forced-conversion volume is still ramping. It will become a more meaningful support in H2 2026 and 2027, but attributing the current level to it overstates the data.

Real-money inflows are not broad-based. The Nomura overweight is concentrated in a subset of EM funds. Pension funds, sovereign wealth funds, and index-trackers have not shown a coordinated rebalancing toward Indonesia. The overweight is tactical, not strategic.

The Least-Harm Path

The capital-account composition reveals a vulnerability profile that policymakers and market participants should price honestly:

The least-harm path for the weeks ahead:

1. BI should not raise rates further to defend the carry trade. The 200+ bp premium is already generous. Additional hikes increase the fiscal cost of the BI remuneration scheme (Warjiyo pledged to cover the government's added interest burden) and deepen the eventual unwind.

2. Let the rupiah appreciate modestly if carry inflows persist. A stronger rupiah lowers imported inflation, eases corporate USD debt service, and reduces BI's intervention need. The 17,950 level is not sacred; 17,800–18,000 is a tolerable range.

3. Accelerate GR 24/2026 implementation. The structural conversion channel is the only durable support. Fast-tracking DSI onboarding and SOE export licensing converts speculative support into structural support.

4. Prepare for SRBI outflow scenarios. Stress-test reserve adequacy against a USD 10–15 billion SRBI redemption shock (roughly 20% of outstanding). The current USD 145.6 billion covers ~5.5 months imports; a sharp outflow could breach the 5-month adequacy line Fitch watches.

5. Communicate the composition honestly. Market participants deserve to know that 85% of the Q2 inflow was SRBI carry. Transparency reduces the risk of a disorderly repricing when the carry trade reverses.

What I'm Uncertain About

1. The true scale of BI's June intervention. We know reserves rose USD 700 million. We know Q2 inflows were USD 7.98 billion. We don't know the monthly breakdown of inflows vs. intervention sales. If June inflows were USD 3 billion and BI sold USD 2.3 billion, the market is tighter than reserves suggest. If June inflows were USD 5 billion and BI sold USD 4.3 billion, the carry trade is the only thing keeping the rupiah at 17,950. BI's weekly intervention data is not public.

2. SRBI foreign ownership by tenor. The BI SRBI ownership page publishes aggregate foreign holdings monthly (latest: March 2026). We don't have the July 2026 weekly breakdown. The 6–12 month tenor concentration matters: 6-month paper rolls in October–December 2026; 12-month paper rolls in Q2 2027. The rollover calendar is a forward vulnerability map we cannot yet read.

3. GR 24/2026 actual conversion volumes. No public data tracks daily DSI FX conversion. Industry reports (HSF Kramer, ABNR) describe the framework, not the flow. The USD 0.5–1.0 billion monthly estimate is a reasoned bound, not a measurement.

4. EM fund positioning granularity. Nomura's 0.9 pp overweight is an aggregate. We don't know if it's driven by a few large active managers or broad index-tracker rebalancing. The former is sticky; the latter is mechanical and reversible.

5. The Hormuz-oil pass-through lag. Oil at $85+ should widen Indonesia's trade deficit (net oil importer). The trade data for June–July 2026 will show whether the current account is absorbing the shock or whether capital flows are fully offsetting it. The next BPS trade release (mid-August) is the next hard check.

This analysis builds on and distinguishes from three gate-pending pieces—"The Rupiah's Paradox: Appreciating Through Night 7," "Beyond the Hormuz Puzzle," and "The Rupiah's Silence in the Storm"—which establish the exchange-rate observation this piece explains. It cites the published "Governance Risk Premium Deepens" (equity outflows), "Pacific Security Realignment" (19-year low in foreign bond holdings), "S&P Family Divergence" (bond vs. equity signal split), and BI Governor Warjiyo's July 7 Q2 flow breakdown. Data gaps are noted inline; no channel is speculated beyond what the evidence supports.