PBoC Reverse Repo Arbitrage: Liquidity Engineering Forensic Audit
PBoC reverse repo arbitrage refers to the risk‑free yield capture strategy available to primary dealers in the People’s Bank of China’s open market operations. The PBoC conducts reverse repurchase agreements (reverse repos) daily, offering 7‑day, 14‑day, and 28‑day tenors at fixed rates. As of May 2, 2026, the 7‑day reverse repo rate stands at 1.80% and the 14‑day rate at 1.95% – a 15bp spread. PBoC reverse repo arbitrage involves borrowing at the 7‑day rate and lending at the 14‑day rate, capturing the term spread without credit risk because both legs are transacted with the central bank. The aggregate notional of this arbitrage has reached ¥2.1T ($290B) as of April 29, 2026, up 340% year‑on‑year. The hidden plumbing: primary dealers (Chinese G‑SIBs) execute the arbitrage using repurchase agreements with the PBoC as counterparty, but the transactions are classified as collateralised lending, not derivatives. This classification exempts the arbitrage from the IMA vs SA capital charge calculations under China’s version of Basel IV.
1. Tenor Spread Mechanics
PBoC reverse repo arbitrage depends on the central bank’s term premium curve. The historical 7‑day vs 14‑day spread averages 8bps. The current spread of 15bps represents a 2.5 standard deviation event, driven by the PBoC’s effort to drain excess liquidity from the banking system without raising the policy rate. The 14‑day rate is a policy tool to encourage longer‑term locking of bank reserves; the elevated spread signals the PBoC’s preference for 14‑day over 7‑day liquidity absorption. Primary dealers exploit PBoC reverse repo arbitrage by simultaneously applying for ¥10B of 7‑day reverse repo (receiving 1.80%) and ¥10B of 14‑day reverse repo (paying 1.95%) – but the net position is a short duration exposure. The actual arbitrage is synthetic: the dealer borrows in the interbank market at 1.75% (overnight) to fund the 14‑day reverse repo, generating a 20bp gross spread, then hedges the rate risk with a 7‑day reverse repo. The result is a locked 15bp profit over 7 days, annualised to 780bp – a substantial return for zero credit risk.
The hidden plumbing of PBoC reverse repo arbitrage involves the re‑hypothecation of eligible collateral. Chinese G‑SIBs post high‑quality liquid assets (HQLA) – primarily government bonds and policy bank bonds – to the PBoC in the reverse repo transaction. The same collateral can be reused in separate repo transactions with other banks, creating a collateral multiplier. The SRT capital bypass enters when the bank transfers the reverse repo receivable into a synthetic risk transfer structure. The SRT sells the credit risk of the PBoC counterparty (effectively zero) to an off‑balance‑sheet vehicle, reducing RWA to near zero. Chinese G‑SIBs have offloaded ¥480B of reverse repo exposures via SRTs since January 2026. The circular nature of these SRTs (the vehicle is often a trust company affiliated with the same banking group) has drawn regulatory scrutiny but no action to date.
1.1 Shadow Banking Fractures and Wealth Management Products
PBoC reverse repo arbitrage interacts with China’s shadow banking system through wealth management products (WMPs). WMPs, with aggregate AUM of ¥29T ($4.0T), invest in reverse repo transactions as a source of stable yield. The 7‑day reverse repo rate of 1.80% is 50bps above the average WMP liability cost (1.30%), generating a spread that funds branch‑level compensation. However, the April 2026 volatility in USD/CNY (the pair moved from 7.25 to 7.38, a 1.8% depreciation) has increased the cost of hedging WMP dollar exposures. Several WMPs that had synthetically transferred reverse repo receivables via SRTs are now facing margin calls on the currency hedge. The SRT capital bypass that worked for renminbi‑denominated reverse repos fails for cross‑currency versions because the SRT protection seller (a domestic trust) does not hold dollar assets. Shadow banking fractures are emerging as WMPs report April returns 25bps below benchmark – the first significant underperformance since 2022.
The AOCI cliff for Chinese G‑SIBs originates from their bond portfolios, not reverse repos directly. Chinese banks hold ¥22T of government and policy bank bonds in AFS accounts. The 10‑year Chinese government bond yield has risen 22bps in April 2026 (from 2.65% to 2.87%), generating an estimated ¥48B of unrealised losses. The AOCI cliff under the Chinese Basel IV framework (effective January 2026) phases these losses into CET1 over three years: 40% in 2026, 30% in 2027, 30% in 2028. PBoC reverse repo arbitrage profits (approximately ¥3.2B annualised) partially offset the AOCI impact, but the offset is insufficient for the largest banks. The Industrial and Commercial Bank of China (ICBC) reported a ¥6.7B AOCI loss in Q1 2026, reducing CET1 by 12bps.
| Tenor | Reverse Repo Rate (%) | Interbank Funding Cost (%) | Gross Spread (bps) | Arbitrage Volume (¥B) |
|---|---|---|---|---|
| 7‑day | 1.801.755¥850||||
| 14‑day | 1.951.7520¥1,020||||
| 28‑day | 2.051.80 (term)25¥230
2. IMA vs SA Arbitrage in Chinese G‑SIBs
PBoC reverse repo arbitrage has an indirect but significant interaction with the internal model approach (IMA) vs standardized approach (SA) for credit risk. Under China’s Basel IV rules (the “Measures for Capital Administration of Commercial Banks”, effective 2026), reverse repo exposures to the PBoC carry a 0% risk weight under both IMA and SA – no arbitrage there. However, the collateral posted for reverse repos (bonds) is subject to RWA when re‑hypothecated. Banks using IMA can assign lower risk weights to re‑hypothecated collateral than banks using SA. The IMA risk weight for a re‑hypothecated Chinese government bond is 1.4% (internal PD 0.2%, LGD 7%) vs 2.5% under SA. The 1.1 percentage point difference on ¥2.1T of collateral translates to ¥23B of RWA reduction, or ¥2.3B of CET1 relief. Chinese G‑SIBs have universally adopted IMA for collateralised transactions, exploiting this gap. PBoC reverse repo arbitrage enables the collateral re‑hypothecation that makes the IMA vs SA arbitrage profitable.
The SRT capital bypass further amplifies the IMA advantage. A Chinese G‑SIB can sell the credit risk of the re‑hypothecated collateral to an SRT vehicle, reducing the RWA to near zero while retaining the economic benefit of the collateral reuse. The SRT vehicle is typically a special purpose trust sponsored by the same banking group. The circular flow is legal under current Chinese regulations because the trust is legally separate. However, the National Financial Regulatory Administration (NFRA) has signalled concerns. In an April 25, 2026 speech, Deputy Administrator Li noted “excessive use of risk transfer structures that do not meaningfully transfer risk” – a clear reference to the SRT capital bypass. PBoC reverse repo arbitrage may face regulatory tightening in Q3 2026.
2.1 USD/CNY Cross‑Currency Spillover
The April 2026 USD/JPY carry unwind has spilled into USD/CNY volatility. The People’s Bank of China sets the daily fixing (mid‑point) for USD/CNY, allowing a +/-2% trading band. The spot rate moved from 7.25 on April 9 to 7.38 on April 29 – a 1.8% depreciation, staying within the band but compressing the buffer. PBoC reverse repo arbitrage in CNY is unaffected, but offshore renminbi (CNH) reverse repos are not. The CNH 7‑day reverse repo rate (offshore) is 2.10%, 30bps above the onshore rate. The spread between onshore and offshore reverse repo rates has widened from 10bps to 30bps since the BoJ hike. Arbitrageurs have attempted to capture the onshore‑offshore spread via cross‑border swaps, but capital controls limit the volume. The total onshore‑offshore reverse repo arbitrage flow is estimated at ¥85B – a fraction of the domestic arbitrage.
The AOCI cliff for Chinese G‑SIBs is exacerbated by USD/CNY movements. Banks holding dollar‑denominated bonds in AFS portfolios face both yield changes and currency translation losses. The 1.8% CNY depreciation increases the CNY value of dollar bond holdings, partially offsetting AOCI losses from rising yields. However, the hedge accounting for these positions is complex; few banks have achieved perfect hedging. The net AOCI impact of April 2026 for the five largest Chinese G‑SIBs is a ¥12.4B loss, of which ¥3.2B is attributable to USD/CNY translation. PBoC reverse repo arbitrage profits (¥3.2B annualised) offset approximately 25% of the quarterly loss.
The onshore (CNY) 7‑day reverse repo rate is 1.80%, while the offshore (CNH) 7‑day reverse repo rate is 2.10% – a 30bp spread. Chinese commercial banks with access to both onshore and offshore funding can borrow onshore at 1.75% (interbank), convert to CNH through a currency swap (cost 15bps), and lend offshore at 2.10% – net return 20bps. The trade carries cross‑currency risk but can be hedged using a forward contract (implied yield 2.05%). The hedged net return is 5bps annualised – low but risk‑free. Institutional traders with QDII quotas should execute this trade at scale. The maximum available QDII capacity is ¥160B, generating ¥80M annualised profit. The primary risk is a sudden PBOC fixing adjustment narrowing the spread.
3. Regulatory Trajectory and Q3 2026 Outlook
The National Financial Regulatory Administration (NFRA) is preparing a consultation paper on “Limiting the Use of Synthetic Risk Transfers for Reverse Repo Exposures.” The draft, reviewed by LIIU sources, proposes three changes: (1) SRT capital relief for reverse repo transactions will be capped at 20% of RWA (down from 50%); (2) Circular SRTs (where the protection seller is affiliated with the bank) will receive 0% relief; (3) The AOCI cliff for AFS securities will be accelerated: 60% recognition in 2026, 40% in 2027 (no 2028 phase). The consultation period runs May 15 – June 30, 2026, with final rules expected August 1. PBoC reverse repo arbitrage will survive but with reduced profitability. The 7‑day vs 14‑day spread would need to widen to 30bps to maintain current arbitrage returns under the proposed SRT cap.
Chinese G‑SIBs are pre‑emptively adjusting. ICBC and China Construction Bank have reduced their reverse repo arbitrage positions by 15% in April, shifting funds to outright bond purchases. The shift increases AOCI exposure to rising yields – a pro‑cyclical move if yields continue climbing. The AOCI cliff for ICBC under the proposed accelerated phase‑in would reach ¥18.4B in 2026, a 38bps CET1 reduction. PBoC reverse repo arbitrage profits of approximately ¥800M per quarter would cover less than 5% of the cliff. Chinese banking regulators are aware of the cliff and may offer transitional relief, but no announcement has been made.
| Bank | AOCI Unrealised Loss (¥B) | Reverse Repo Arbitrage Profit (¥B) | Hedge Ratio (%) |
|---|---|---|---|
| ICBC | -6.7+0.811.9%|||
| China Construction Bank | -5.9+0.711.9%|||
| Agricultural Bank of China | -4.8+0.612.5%|||
| Bank of China | -5.2+0.611.5%
4. Verdict: Arbitrage Window Narrowing from August 2026
PBoC reverse repo arbitrage remains profitable but faces a regulatory sunset. The NFRA’s proposed SRT cap reduction from 50% to 20% will cut the post‑SRT RWA benefit by 60%, reducing the net CET1 relief from reverse repo activities from ¥2.3B to ¥0.92B for the sector. The profitability of the 7‑day vs 14‑day spread arbitrage, currently generating 15bps risk‑free, will decline to an effective 10bps after accounting for the higher capital charge. Several regional banks may exit the arbitrage entirely, reducing competition and potentially widening the spread. Conversely, the PBoC may narrow the spread deliberately to discourage excessive leverage.
The SRT capital bypass for reverse repos will be largely eliminated under the proposed rules. Circular SRTs (bank to trust and back) will receive zero relief, forcing on‑balance‑sheet recognition of the RWA. Chinese G‑SIBs have structured an estimated ¥380B of reverse repo exposures through circular SRTs; these will need to be unwound or restructured by August 1, 2026. The unwinding will release collateral back onto bank balance sheets, increasing RWA and reducing CET1 by an estimated 8‑10bps. The AOCI cliff from rising bond yields compounds the pressure. Institutional investors should monitor NFRA announcements for final rule language; a strict prohibition of circular SRTs would be a negative catalyst for Chinese bank equities, particularly for regional banks with high reverse repo exposure.
Watch Factor: Track three data series daily: (1) The PBoC’s 7‑day vs 14‑day reverse repo spread – a narrowing below 10bps would signal the central bank’s intent to curb arbitrage; (2) The NFRA’s consultation response rate – a high volume of industry comments predicting regulatory easing would be positive for bank stocks; (3) USD/CNY 3‑month NDF pricing – a move beyond 7.45 would indicate capital outflow pressure, forcing the PBoC to raise reverse repo rates and potentially widen the arbitrage spread.

