🇨🇳Lending in China: Interest Rates and Borrower Eligibility

Concept

Lending in China is access to money that costs roughly 3% a year in renminbi as of September 2026, at a time when the opportunity cost of capital for most borrowers in the CIS, Türkiye, the Gulf and other emerging markets is higher, at the extreme sixteen times higher: from about 4–6% for a dollar, euro or dirham loan to the 49.3% average rate on Turkish lira commercial loans.

The one-year Loan Prime Rate is 3.00% and the five-year-plus LPR 3.50%, confirmed at the 20 September 2026 fixing, the sixteenth consecutive month without a change (PBoC/NIFC announcement of 20 September 2026). The weighted average rate on newly issued corporate loans in June 2026 was 3.04%, and the bank acceptance discount rate — the cheapest working capital in the system — averaged 0.94% in the same month (PBoC quarterly monetary policy report, Q2 2026, published 12 August 2026). On newly issued loans, manufacturing borrowed at 2.74%, below the LPR itself, and wholesale and retail trade at 3.24% (PBoC press conference, 15 July 2026).

This is not cheap because China is in deflation. PPI ran at +3.5% year on year in July 2026 (NBS). It is cheap because credit demand is weak — renminbi loans outstanding grew 5.2% year on year at end-June 2026 (PBoC Q2 2026 report) — and banks compete for good borrowers by pricing below the LPR. Easing also occurs outside the LPR: relending and rediscount rates were cut by 25 basis points with effect from 19 January 2026 — the one-year targeted relending rate was taken to 1.25% and the three-month rate to 0.95% — and on 27 January 2026 a CNY 1 trillion facility for lending to private business was created inside that same relending line, carrying the same rate (PBoC Q1 2026 monetary policy chronicle).

The borrower does not have to be a Chinese company. The claim that a mainland bank will not lend without a PRC legal entity has been wrong since 1 March 2022, when PBoC and SAFE circular Yinfa 2022/27 expressly allowed mainland banks to lend to non-financial enterprises incorporated abroad, in renminbi and in foreign currency (text at SAFE). Three practical access routes exist.

1. With no PRC entity — supplier credit under a Sinosure limit, a usance letter of credit, a direct cross-border loan under Circular 27, or a Free Trade Non-resident (FTN) account in the Shanghai free trade zone.
2. Through a Hong Kong structure — a loan in offshore renminbi (CNH), which today prices below onshore, supported by a mainland cross-border guarantee for offshore loan (内保外贷).
3. Through a WFOE or another onshore foreign-invested enterprise — an ordinary renminbi loan on national treatment, including one secured by a standby letter of credit from a foreign bank under the offshore guarantee for onshore loan (外保内贷) regime.

This article covers the credit perimeter: the price of money, access to it, and the conditions under which a rate differential turns into real savings. The documentary side — letters of credit, collections, guarantees and their operating mechanics — sits in a separate piece on trade finance in China; here those instruments appear only to the extent that they function as credit, with a usance LC as deferred payment and Sinosure cover as supplier credit priced at the premium.

The key limitation is that a rate differential is not, by itself, a profit. Hedge the currency risk and the forward premium eats almost all of it. Real savings arise in two configurations — a borrower with a natural hedge, and an importer who borrows goods rather than money.

Cost of capital: where China sits globally as of September 2026

Cost of capital is the all-in price of money for a corporate borrower: the benchmark plus the bank's credit margin, before any currency hedge. Published averages below are the figures of the central bank concerned; where a cell shows a benchmark plus a margin, the margin is an illustrative assumption, not a bank quote.

How durable is the gap. The 20 September 2026 fixing left both LPR tenors unchanged (PBoC), but the pressure is downward: CPI ran at just +0.5% year on year in July 2026 against a target near 2% (NBS), and credit demand remains soft. From the Chinese side the gap is more likely to widen than to close. It narrows from the other end, as policy rates fall in the rouble and tenge zones: the Bank of Russia already cut to 14.00% by its decision of 24 July 2026 (effective 27 July 2026), and the NBK to 16.25% on 4 September 2026. A structure planned over 12 to 24 months should assume the arbitrage window closes because of the borrower’s own currency, not because of China's.

One further reference point: top-rated foreign issuers borrow in renminbi more cheaply than the average Chinese corporate loan. Kazakhstan priced a debut sovereign panda of CNY 3.4 billion at 1.90% on 29 May 2026 (AIFC), while the domestic base rate stood at 16.25% by September 2026. Panda issuance exceeded CNY 160 billion in the first half of 2026, up 69% year on year (PBoC briefing, 15 July 2026). That is pricing for an issuer with capital-markets access — a floor no trading company will reach, but a measure of the market's depth.

The arbitrage mechanism: why a hedged renminbi loan is no cheaper than a dollar one

Credit arbitrage is borrowing in a cheap currency against an obligation denominated in an expensive one; its economics are set not by the rate differential but by whether the currency risk is covered, and with what.

The arithmetic of covered parity. Covered interest parity sets the forward price of the renminbi against the dollar: the renminbi trades at a forward premium roughly equal to the gap between dollar and offshore-renminbi money-market rates. With SOFR at 3.65% on 2 September 2026 (Federal Reserve Bank of New York) and three-month CNH HIBOR at 1.64% on 3 September 2026 (TMA), that gap — and so the computed forward premium — is about 2.0% a year. This is a calculation, not a dealer quote: an executable forward also carries the cross-currency basis and the bank's spread, and is priced on the day of the trade.

The consequence is direct. A borrower takes a renminbi loan at 3.0% and buys renminbi forward to close the position: 3.0% + 2.0% = 5.0% in dollar terms. That is SOFR plus about 135 basis points — inside the SOFR + 100–220 range taken above for an ordinary dollar loan. Covered interest parity closes the window almost entirely.

The same holds for soft currencies, where the effect is starkest. Market quotes for CNY/TRY forwards are not published — the market is over the counter — and an interest rate parity estimate puts the premium at around 35 percentage points against the lira at a TCMB policy rate of 37.00%; that is a computed figure, not a quote. The rouble is different, and it is usually described wrongly: CNY/RUB futures (Moscow Exchange futures list) and swaps trade on the Moscow Exchange with public quotes, and since 13 June 2024 exchange trading in dollar and euro pairs has been suspended while the other pairs, the yuan among them, trade as normal (Moscow Exchange notice of 12 June 2024); the rate differential at a key rate of 14.00% (effective 27 July 2026) implies a premium of around 11–15 points a year. The practical consequence matters more than the figure: a Russian borrower's yuan liability has an observable exchange rate and exchange liquidity for hedging — an argument for denominating the debt in yuan, and an infrastructural argument rather than the political one it is usually presented as.

For a Turkish importer, the lira alternative in this example is the TCMB's 49.3% average commercial-loan rate. He borrows renminbi at 3.0% and sees a 46-point gap. To turn that gap into profit rather than a bet on the exchange rate, he must buy renminbi forward to the repayment date — and the CNY/TRY forward already prices in the money-market differential, roughly 35 points a year at a 37.00% policy rate. The result is 3.0% + 35% ≈ 38%. About 35 of the 46 points were never profit; they were compensation for the risk of lira depreciation, and the forward market priced them at almost exactly that. The remaining 11 or so points are the difference between the credit margin in the average lira loan and the margin a renminbi lender charges him — and they exist only if a renminbi lender actually extends him credit on those terms.

A hedged renminbi loan costs a Turkish or Russian importer roughly what local money costs at money-market rates plus his own credit margin, give or take basis and fees. An unhedged renminbi loan is not arbitrage; it is an open currency position — and over the year to September 2026 it ran against the borrower: the CFETS central parity moved from 7.1089 yuan per dollar on 2 September 2025 to 6.7829 on 2 September 2026, a renminbi appreciation of about 4.8% (CFETS), and the forward premium prices further appreciation.

Two configurations remain, in which the rate is not the main consideration.

Configuration one: a natural hedge on the asset side. The familiar formulation — "the importer repays a renminbi loan out of a renminbi trade obligation" — is only half right. The payment to the supplier happens when the loan is drawn; the importer repays out of local revenue. The hedge is created not by the payment but by the goods. A distributor who reprices in line with the renminbi replacement cost — recalculating the price of a consignment at the current exchange rate and the current supplier price — carries an asset that is renminbi-denominated in economic substance: renminbi appreciation lifts both the liability and the local price of inventory. To the extent that sale prices actually follow the renminbi replacement cost within the loan's term, the position is largely closed and most of the rate differential stays with the borrower; the offset is only as good as the repricing — its timing, demand and inventory turnover.

The limiting condition is as follows: if the goods have to be sold at a fixed local price — a long-term contract, a won tender, a regulated price — there is no natural hedge, and the renminbi loan becomes a currency bet the size of the principal. On a 180-day cycle, renminbi appreciation of 8.5% consumes the entire gain against a 20% rouble alternative, and the pace observed over the year to September 2026 (about 4.8% on the central parity) is enough to cut more than a quarter of it over one cycle.

Panda bond issuance illustrates who can retain the difference in practice. Foreign panda bond issuers that print at around 1.9% in renminbi and swap the proceeds into dollars through a cross-currency swap end up funding at roughly 1.9% + 2.0% ≈ 3.9%, close to SOFR. That can undercut the issuer's ordinary dollar borrowing, but the saving comes from Chinese institutional demand for renminbi risk and from access to the swap market, not from the rate differential. For a trading company the channel is closed by ticket size: the panda market starts in the hundreds of millions of renminbi and requires a rating.

Configuration two, the strongest: arbitrage that is credit rather than currency. When an importer obtains payment terms from a Chinese supplier under Sinosure cover, it is borrowing on the supplier's credit market rather than its own. The cost of those terms is set by the insurance premium and by the price of money to the Chinese exporter — who funds himself at the 0.94% average bill-finance rate, or around 3% against receivables. The rate is anchored to the Chinese curve, and where the invoice is in dollars the deferral adds no renminbi position: it moves the date of the same dollar payment, and the importer's exposure stays what it already was on a dollar invoice — the dollar against his own revenue currency. This is credit arbitrage without a renminbi leg — the only construction on the list that adds neither a hedge nor a new directional view.

The numbers: a $1 million contract over 180 days

The calculation shows what a single trade cycle costs in cash under each instrument — a direct comparison on the same amount and the same tenor.

Assumptions: a $1 million contract, shipment from China, the full 180-day cycle financed, rates from the table above, computed as rate × amount × 180/360. The amounts illustrate the arithmetic and are not quotes: the rouble and tenge rows assume working-capital rates of 20% and 20.5% for a mid-market importer (the Bank of Russia's June 2026 averages for loans of up to one year were 17.07% overall and 18.02% for SMEs), the Sinosure row an annualised supplier uplift of 2–6%, and the hedged row the computed 2.0% forward premium.

The borrower’s model must also include the following costs omitted from the table: arrangement and issuance fees (for illustration, 0.1–1% of the amount, one-off); the conversion spread, which at an assumed 0.2–0.5% each way on a renminbi pair is equivalent over a 180-day cycle to 0.8–2 percentage points a year and can outweigh the difference between two adjacent rows; cargo and transaction insurance; warehousing and consignment control where goods serve as collateral; and the guarantor's fee where an SBLC or a surety is used. Instruments compare correctly only on the all-in cost of a cycle, never on the headline rate — most sales-driven calculations substitute the headline rate for that total cost.

The spread between the extreme rows is $236,500 on a single six-month turn of a single contract. Against the rouble alternative, a renminbi loan saves about $85,000 per cycle, Sinosure-backed terms $70,000–90,000, and dollar trade finance about $70,000.

Checking the rule against a rouble importer: (20% − 3.04%) × $1 million × 0.5 = $84,800, which matches the line-by-line figure. The same importer without a natural hedge: (20% − 5.04%) × $1 million × 0.5 = $74,800, also a match. The difference between the two numbers, $10,000, is the price of currency risk in that transaction. That is precisely the amount worth negotiating over when arguing with the supplier about whether the invoice is denominated in renminbi or dollars.

The usually omitted part of the calculation concerns tied-up capital. The rate is not the only price, and often not the main one. An importer working on 100% prepayment finances not 180 days but the whole cycle from payment to receipts: 30–60 days of production, 30–45 days of transit, 30–60 days of sale and inventory turnover. Those are the same 90 to 165 days, but funded with own capital taken out of circulation. Moving to payment terms pushes the payment to the end of the cycle and releases that money: on a $1 million contract with two turns a year, prepayment keeps $1 million tied up permanently, whereas 120 days of terms frees roughly two thirds of it for the next purchase. The effect here is not a rate but a volume effect — the same capital turns more goods — and it is frequently larger than the interest saving.

Hence a practical rule for negotiations: bargain first over the length of the terms, then over their price. Each additional 30 days on a $1 million contract removes the need to fund $1 million for a month: at a 20% alternative rate that is $16,670 per shipment, at 25% it is $20,830. That gives the break-even point of the negotiation: a supplier uplift below 1.67% of the contract for each additional month of terms is arithmetically worthwhile for a borrower whose alternative is a 20% rouble loan, and not worthwhile for one whose alternative costs 8%. The threshold moves linearly with the borrower’s own rate — and it is that threshold, not the Sinosure tariff, that should be considered in the negotiation.

Stretching the cycle to 180 days also cuts average tied-up capital: from $1 million under full prepayment to roughly $833,000 with one month of terms and $667,000 with two. The released amount goes straight into the next purchase, so for a growing importer the effect of payment terms is better counted as incremental turnover on the same capital than as a saving.

The Russian borrower: what the yuan costs at home and what the law says

A 20% rouble alternative is not the only benchmark for a Russian company: Russian banks lend in yuan too, and the price of that money is known from official statistics rather than from consultants' estimates. The Bank of Russia has published weighted average rates on yuan loans to non-financial organisations since 16 February 2024, with the series disclosed back to January 2023 (Bank of Russia press release; data in loans_nonfin.xlsx, sheets «ставки_юань» and «структура_юань»). The series is short, but it settles the main question: what the yuan costs in Moscow rather than in Shanghai.

The price of the yuan in a Russian bank

The rate cycle. In 2023 a yuan loan of up to one year cost 4.8–7.2% a year. The break came in the second half of 2024: in September and November 2024 the 91–180 day segment cost 12.77% and 14.16% against an up-to-one-year aggregate of 11.90% and 11.74%, and in December 2024 the 31–90 day segment cost 14.77% against an aggregate of 12.25%. The peak came in March 2025, at the long end: 13.93% on loans over one year against 10.58% on loans of up to one year in the same month. By June 2026 the rate was back at 7.54% up to one year and 8.81% over one year, a premium of 4.5–5.8 percentage points over the LPR, which has held at 3.00% since May 2025. The series is published by maturity segment and some sub-segments are not disclosed in certain months, so the tables below show only the up-to-one-year and over-one-year aggregates, set against the Chinese anchor (PBoC LPR announcements) and the key rate (Bank of Russia) on the same six dates.

Over two and a half years the Chinese anchor fell by 45 basis points and has not moved since 20 May 2025, while the key rate on the same dates went from 16.00% (December 2023) to 21.00% (December 2024) and 14.25% (June 2026), and the up-to-one-year yuan rate in a Russian bank from 6.72% to 12.25% and 7.54%. Only the Russian side moved. Everything that happened to the cost of the yuan for a Russian borrower was set by the domestic funding market rather than by PBoC decisions — which is also why the gap between the LPR and the Russian bank rate changed threefold in two years.

Tenor matters more than rate. In April 2026 97.77% of yuan lending by volume was for up to one year, and the 31–90 day segment alone took 73.23% of the month's lending — 74.9% within the up-to-one-year segment; in January 2023, by contrast, almost 90% was for more than one year. The instrument has changed purpose: in 2026 a yuan loan from a Russian bank finances a specific delivery for one to three months and is not a source of long money. An investment cycle cannot be planned on it — not because this is prohibited, but because such loans barely appear in the statistics.

Volatility and spread

Where the volatility comes from. The cost of overnight yuan inside Russia is shown by the Moscow Exchange's RUSFAR CNY indicator, and its behaviour explains both the premium and the short tenors. It averaged 40.14% in September 2024, peaking at 212.08% on 4 September 2024; in March 2026 the average was 11.11% with a peak of 43.93% (RUSFAR CNY archive, Moscow Exchange). Between shortage episodes the rate falls to almost zero and below it. The Bank of Russia raised its daily yuan swap limit to CNY 30 billion in August 2024, and since 5 February 2025 it has stood at CNY 5 billion (Bank of Russia, FX swap terms); since 1 April 2026 the Bank of Russia has published a separate indicator, the implied overnight yuan swap rate (Bank of Russia press release). That amplitude is what explains the tenor of lending: yuan funding inside Russia is short, and a bank lends yuan against a delivery rather than for years.

What is left of the spread. The gap between the Chinese benchmark rate and the rate at which a Russian company actually obtains yuan was 1.2–3.6 percentage points on loans of up to one year in 2023, widened to 7.5–10.8 points in March 2025, and stood at 4.5–5.8 points in June 2026. The "LPR at 3.00% (unchanged since May 2025) against a key rate of 14.00% (effective 27 July 2026)" arbitrage is not available to a Russian company: between those two numbers stands a Russian bank that obtains yuan funding at high rates and short maturities, and the difference goes to it.

The spread a Russian company really has is "yuan in a Russian bank against roubles in a Russian bank": about 5.4–6.7 points below the key rate in June 2026. Three things come off it. The first is tenor: in April 2026, 97.8% of lending was for less than a year, so the comparison is not with a three-year rouble loan but with rouble working capital for a quarter. The second is currency risk, and here it is created rather than hedged: in June 2026, 51.2% of payments for imports from Asia were made in roubles (Bank of Russia, external sector statistics), and if the supplier is paid in roubles, a yuan loan does not close a currency position but opens one. The third is the unpredictable price of a rollover: a three-month loan taken in a calm month is rolled over at the rate of whatever month the rollover falls in, and September 2024 showed what that rate can be.

Currency control and tax

Currency control: what the borrower must do. A loan agreement with a non-resident is registered with the authorised bank and given a unique contract number where the obligations reach RUB 3 million — the same "strict" bracket as an import contract rather than an export one, where the threshold is RUB 10 million (Bank of Russia Directive No. 6819-U). The deadline is no later than 15 working days after a tranche is credited, and on repayment before the transfer order is submitted (Bank of Russia Directive No. 6663-U). A rouble loan from a Chinese bank is a currency transaction too: the regime follows the non-resident status of the other party, not the currency of payment. The type-"C" account regime (Decree No. 95) does not apply to repayment to Bank of China or ICBC, because the PRC and Hong Kong are not on the list of unfriendly states (Government Order No. 430-r); it does apply to a lender controlled by persons from a listed state, wherever the lender is registered — for example, a Hong Kong lender owned through a British, Singaporean or EU company.

Withholding tax on interest: 0%. Article 11 of the Russia–China double tax treaty was replaced in full by the Protocol of 8 May 2015, and interest is taxable only in the recipient's state of residence (Protocol of 8 May 2015). Zero is a rate, not an absence of obligations. The lender must provide confirmation of tax residence and confirmation of beneficial ownership of the income, both before the payment date (Article 312 of the Tax Code; for the PRC no legalisation or apostille is needed under paragraph 4 of the Protocol of 13 October 2014), and the MLI principal purpose test has applied to withholding tax under this treaty since 1 January 2024 (Russia's notification to the MLI depositary). The cost of a mistake has risen: without the confirmations the base rate is 25% from 1 January 2025, up from 20% (Federal Law No. 176-FZ) — a quarter of the interest.

Thin capitalisation. An ordinary loan from an independent Chinese bank does not fall under subparagraph 1 of paragraph 2 of Article 269 of the Tax Code — the bank holds no stake in the borrower. The debt becomes controlled under subparagraph 3 if the loan is guaranteed by the Chinese shareholder or its affiliate, a construction typical of a deal with a Russian borrower. Paragraph 9 of Article 269 saves it provided the lender is a bank unrelated to both the borrower and the guarantor, and the guarantor has never performed the obligation for the borrower; the lender's written confirmation is requested for this under paragraph 10. The asymmetry that makes this worth checking before signing: recharacterisation turns interest exempt from withholding into dividends, which are taxed at source. For yuan debt arising from a controlled transaction, deductible interest lies in the range from 1% to SHIBOR plus 7 percentage points (subparagraph 3 of paragraph 1.2 of Article 269), with SHIBOR taken for the tenor closest to the term of the debt.

Borrowing routes at three levels of presence in China

The level of presence in China determines the instrument set available: the deeper the legal presence, the cheaper the money and the higher the barrier to entry.

Level A — no PRC entity

Supplier credit under a Sinosure limit. The most accessible instrument: deferred payment terms that the Chinese exporter insures with Sinosure, whose short-term export credit cover normally runs to a credit period within one year and at most two (Sinosure 2024 annual report), priced as the supplier's uplift, which carries the premium. The buyer does not apply for the limit itself: the insurer sets it on the buyer at the exporter's request, on its own assessment.

Usance LC and post-financing. The documentary perimeter is set out in the article on trade finance in China. A usance LC is a loan from the issuing bank to your supplier against your credit risk, and its price to you is the issuance fee (usually 0.5–1.5% a year of the amount), the cash cover your bank demands, and the discount your supplier pays. With 100% cash cover the instrument is not credit at all — it merely removes risk from the supplier.

Direct cross-border loan under Yinfa 2022/27. Since 1 March 2022 a PRC bank may lend to a foreign non-financial enterprise in renminbi and foreign currency; the bank's portfolio ceiling is calculated as its net tier-one capital (operating capital for branches of foreign banks) × the outbound lending leverage coefficient × the macroprudential parameter (SAFE). The rule sets no single universal multiplier, and the coefficients are revised: under Yinfa 2026/72 of 11 April 2026, announced on 15 April, the coefficient for foreign-invested banks and PRC branches of foreign banks rose from 0.5 to 1.5, and for China Exim Bank from 3 to 3.5 (SAFE). The restrictions are strict and define where the channel applies: proceeds may not go into securities, into fictitious trade transactions, into repayment of external debt arising under a 内保外贷 structure, or — critically — back into China as loans or capital contributions. The money has to work outside the mainland.

The NRA account — a settlement tool that works as collateral. A non-resident account at a PRC bank is not itself a credit product, but balances held in it are accepted as security for an onshore loan. For a group with renminbi receipts from Chinese counterparties this is a way to turn idle balances into a credit limit without incorporating anything. The opening mechanics and constraints of such accounts are covered in payouts to Chinese suppliers.

FTN account in the Shanghai free trade zone. The non-resident version of a segregated-accounting account: an offshore company opens an FTN and receives a cross-border renminbi loan from a Shanghai bank (product description, 25 June 2026). The same product description gives an example: the Hong Kong operating platform of a large central state-owned group obtained a renminbi loan through its FT account, drawing on onshore renminbi pricing. The binding constraint is not the law but whether a given bank is willing to run compliance on a foreign beneficial owner.

A separate branch exists for equipment and capital goods — the export buyer's credit (出口买方信贷): contracts from $1 million, an importer down payment of at least 15% (20% for ships), Chinese content generally above 70% (50% for ships and aircraft), tenors generally up to 10 years, and mandatory export credit insurance (Bank of China). China Development Bank finances 80–85% of contract value on comparable terms (note, 24 June 2026). The borrower can be a foreign bank, a finance ministry or the importer itself (China Exim Bank).

Level B — through a Hong Kong structure

By 2026, low-cost renminbi funding was no longer confined to the mainland, subject to the distinction between bank funding and borrower pricing. CNH HIBOR 3M stood at 1.64% on 3 September 2026 against a weighted average of 3.04% on newly issued onshore corporate loans, and that 140 basis point gap is a difference in the bank's cost of funding, not in the price of credit to the borrower: on the illustrative all-in rates in the table offshore is no cheaper, at CNH 2.9–4.0% against an onshore average of 3.04%. The borrower's gain arises where a bank passes cheap funding through into its margin, and therefore depends on the borrower's quality and collateral.

From 2 February 2026 the HKMA doubled its RMB Business Facility from CNY 100 billion to CNY 200 billion. The facility, launched in October 2025 to replace the RMB Trade Financing Liquidity Facility of February 2025, funds participating banks for trade finance and for capital-expenditure and working-capital term loans to their corporate clients (HKMA). For a group with a Hong Kong leg, moving the borrowing centre into CNH remains the rational move — but what grounds it is that HKMA facility and the depth of the offshore market, not the arithmetic of "3.04 − 1.64". The HKMA's own figures measure that depth: offshore renminbi bond issuance in Hong Kong reached RMB 1,100.9 billion in 2025 (+3%), and renminbi loans stood at RMB 935.0 billion (+29%) (HKMA Annual Report 2025).

An additional financing mechanism is the cross-border guarantee for offshore loan (内保外贷): a mainland bank issues a guarantee or SBLC and a Hong Kong bank lends to the offshore company. Prior approval was abolished in 2014 and the regime is registration-based: a bank guarantor reports the guarantee to SAFE through data submission, while a non-bank guarantor registers it with its local SAFE branch within 15 business days of signing (art. 9 of Huifa 2014/29). In practice CCB Asia accepts guarantees and SBLCs from mainland banks in HKD, RMB and EUR and lends to the offshore structure for one to five years for purchases, working capital, investment and dividends (asia.ccb.com).

Separate regimes apply in the pilot zones — Qianhai, Nansha and Hengqin in the Greater Bay Area, and the Hainan Free Trade Port — whose local regimes let an enterprise registered in the zone borrow renminbi from Hong Kong banks directly; eligibility and quotas follow each zone's own rules and are confirmed with the local PBoC branch before a structure relies on them. For a group with a manufacturing or procurement leg in Guangdong, this is a way to obtain offshore pricing on a mainland entity. The cross-border cash pooling rules, Yinfa 2025/251 of 24 December 2025, add a single multi-currency master account and self-allocation of quotas inside the group: a cross-border financing leverage ratio of 2 and a macroprudential parameter of 1.75, applied to the parent's owner's equity plus the weighted equity of the participants, so the concentrated external debt limit for the pool as a whole runs to 3.5× that base (2 × 1.75); outbound lending carries leverage of 1 and a coefficient of 0.8 (SAFE).

A Hong Kong bank's requirements differ from mainland ones and are set by each bank: its credit file usually covers the operating history, audited accounts, recent bank statements and, for an owner-managed company, personal guarantees from directors, and the timetable depends on the facility and the security. In exchange there is no mainland-style regulatory working capital needs calculation and no mandatory entrusted payment. The Hong Kong banking landscape and account-opening requirements are covered in the Hong Kong hub; for a group with mainland flow, Bank of China (Hong Kong) and CMB Wing Lung are the usual answers.

Level C — through a WFOE or another onshore FIE

A wholly foreign-owned enterprise borrows in China on national treatment: the Foreign Investment Law has applied since 1 January 2020, and article 17 expressly confirms an FIE's right to raise finance (text). No separate approval is required for a renminbi loan.

What the bank looks at. The framework is the NFRA working capital loan measures in force since 1 July 2024: a working capital loan runs up to three years (five where the cycle is long) against a calculation of working capital needs, proceeds may not fund dividends or financial investments, and any single disbursement above CNY 10 million must be made by entrusted payment directly to the borrower's counterparty (text of Order No. 2). Investment loans run up to ten years. The practical scoring shows through in mass-market products: a standard tax-record-based loan line requires at least two years of operations, 24 months of clean tax payment, a tax rating of A/B/M and no more than two lending banks. The bank must pull a corporate credit report from the PBoC Credit Reference Centre.

Collateral. Pledges over receivables, inventory and equipment are registered in the unified online movable-property security register (zhongdengwang); mortgages and guarantees from owners and affiliates are equally standard. A guarantee from a state-backed guarantee company costs no more than 1% a year under the small-business standard, and since 19 January 2026 a CNY 500 billion special guarantee programme for private investment operates at the same price cap (Ministry of Finance); commercial guarantors charge 2–3% in practice. All in, "loan plus state guarantee" for an SME comes to roughly 3.5% + 1% ≈ 4.5%.

The key structure for a foreign group is the offshore guarantee for onshore loan (外保内贷): a standby letter of credit or guarantee from a foreign bank secures a renminbi loan from a Chinese bank. CCB accepts SBLCs in USD, HKD, EUR, JPY, GBP and CHF, lends in renminbi for up to five years, and requires the SBLC to remain valid at least 15 days beyond the loan (ccb.com). This is the lawful route to "cheap renminbi credit against foreign credit risk" — and simultaneously the point at which the sanctions filter applies to the SBLC-issuing bank. The volume constraint: on a call of the guarantee, the debtor's external debt may not exceed audited net assets for the prior year.

What Level C does not solve. Cheap renminbi credit at a WFOE remains renminbi inside China. It serves purchasing, production and working capital in the Chinese perimeter, but it does not automatically become group liquidity outside the mainland. Profit repatriation is a separate regime with its own sequence — audit, withholding tax on dividends, statutory reserve allocations — and remitting borrowed funds abroad is directly constrained by the outbound lending rules. Building a WFOE for the sake of a 3% rate is therefore justified where the turnover itself happens in China — purchasing, finishing, consolidating consignments — and works poorly as a way to cheapen the funding of operations located entirely outside the country.

How long the process takes depends on the bank, the facility and the security; a pledge of real estate or goods adds a security registration. A benchmark bank's document pack: a written application, constitutional documents and a board resolution, the business licence, three years of audited accounts plus current figures, a credit rating and collateral documents (Bank of China).

Before using any of the three routes, an additional financing option is worth assessing. Where the beneficial owner holds a liquid securities portfolio at a European or Asian bank, a lombard loan against it is priced as a margin over SOFR or EURIBOR that the bank sets for the portfolio and the client, can be drawn quickly once the line is in place, and requires no Chinese entity and no new corporate credit profile; the bank still applies its own source-of-funds and purpose checks. The constraint is size: creditworthiness is set by the portfolio rather than by turnover, and the loan-to-value the bank grants depends on the portfolio's composition and liquidity. Building a Chinese perimeter starts to make sense when the funding requirement exceeds what your own balance sheet can carry, or when trading volume makes one or two percentage points material. The full map of Chinese instruments and jurisdictional forks is in the China hub.

Who lends: bank appetite

Bank appetite combines product set, compliance profile and willingness to work with a foreign beneficial owner — and in China those three diverge across segments far more than rates do.

The Big Four and joint-stock banks

The Big Four. Their published product lines cover this kind of trade credit — Bank of China's export buyer's credit and CCB's renminbi loans against a foreign bank's standby letter of credit are cited above — and the differences start in compliance. For flows connected to Russia, availability is decided bank by bank and transaction by transaction, including at Bank of China: a group's product list does not show whether it will take a particular Russia-linked payment, which is confirmed in writing with the bank for the actual parties and goods. The map of mainland banks by usability for foreign trade sits in a separate review of Chinese banks.

Joint-stock banks — the most active segment. Ping An Bank leads supply chain finance: CNY 911.28 billion of SCF financing in the first half of 2025 (+25.6% year on year) and CNY 155.72 billion of cross-border trade finance (+33.2%) (bank press release). Its platform covers prepayment financing, spot financing against goods pledged with a warehouse operator in a pledge–disbursement–consignment redemption–repayment cycle, reverse factoring and pledges of electronic warehouse receipts (bank.pingan.com). This is the main channel through which a trading company with no real estate obtains credit against turnover. Industrial Bank runs the benchmark mechanics for financing against a Sinosure policy, with a tripartite agreement assigning insurance proceeds (cib.com.cn).

Supply chain finance

One direction is developing faster than the rest: refinancing trade transactions through central bank instruments, above all PBoC rediscounting, whose rate was cut by 25 basis points with effect from 19 January 2026. The point of the construction is that the cost of money for the end trade is anchored to the central bank's rediscount rate rather than to the bank's own lending rate. For an importer this means the question "what rate will the bank give" is increasingly settled by whether the transaction falls into a policy-supported category, and not only by the borrower's credit quality. The practical consequence: prepare the transaction file so that it reads as financing of a real flow of goods with verifiable shipment — the most effective single way to lower the rate in the Chinese perimeter.

Supply chain regulation. Circular Yinfa 2025/77, issued by the PBoC with five other authorities, has applied since 15 June 2025, with a two-year transition for voucher business: the payment term on an electronic accounts receivable voucher should in principle be within six months and at most one year, and anchor enterprises may not impose unreasonable payment terms on small suppliers or stretch settlement through non-cash payment (text at gov.cn). The rule governs domestic supply-chain vouchers; it does not set the payment terms a Chinese exporter offers a foreign buyer, which are agreed contract by contract.

Instruments on the supplier's side that you end up paying for. Bank acceptance discounting (0.94% on average in June 2026), packing credit against an export contract, export bill negotiation (押汇) and forfaiting — the non-recourse purchase of export receivables under a letter of credit for one to five years, with a minimum ticket around $100,000 (Bank of China). Knowing them serves a practical purpose: when a supplier refuses payment terms because he "has no free cash", the conversation should move not to price but to which of these instruments he will use to close the gap. Where he funds at the Chinese rates above, his money is cheaper than a rouble or lira alternative, which is what leaves room to negotiate.

Settlement with Russia and the CIS

Regional banks. The real hub for cross-border settlement with Russia and the CIS is Harbin Bank: its cross-border clearing and settlement business exceeded RMB 400 billion in 2025 (Harbin Bank 2025 annual results, HKEX), and it is a direct CIPS participant (CIPS participant list). Alongside it, Longjiang Bank operates branches in Suifenhe and Dongning. Bank of Dalian, often credited in Russian-language surveys with a CIS specialisation, positions itself publicly on something else — the Shanghai free trade zone segregated-accounting platform; it has announced no CIS settlement programme of its own. For the credit question this segment yields exactly one conclusion: settlement access and credit access are different things, and a bank that processes a payment does not thereby open credit to a foreign borrower. The settlement side is covered in the articles on payments to China and Chinese bank restrictions.

A foreign beneficial owner

Foreign beneficial ownership changes the relevant product rather than preventing access to credit. Following the FX business reform (外汇展业), a Chinese bank assigns each client a due diligence rating before any transaction — by beneficial owner, sanctions profile and commodity codes — and processes transactions for top-tier clients under a general mandate without per-transaction control. The practical effect: foreign ownership does not close credit, it moves the borrower into a different product branch — 外保内贷 for an onshore loan against foreign security, and the mirror 内保外贷 for an offshore loan against mainland security. The logic of both is identical: the loan is booked where it is cheapest to fund, and the collateral stays where it physically sits. The question reduces to the quality of the guarantor and its sanctions profile — that is, which bank is prepared to issue the SBLC. A separate filter applies to beneficial owners with a Russian connection: refusals on sensitive commodity codes, week-long delays without explanation, queues for account opening.

Who can lend to a Russian company. Chinese banking groups are present in Russia through subsidiaries licensed by the Bank of Russia — for example AKB Bank of China (JSC), a Russian legal entity supervised by the regulator (Bank of Russia register card, OGRN 1027739857551). Such a bank funds itself and prices as a Russian bank rather than a mainland one: its yuan lending sits at the levels reported in the Bank of Russia's yuan loan series, not at the LPR.

Sinosure as quasi-credit

Sinosure is China's state export credit insurer, and for an importer its product functions as supplier credit: the insurer sets a credit limit on the buyer, within which a Chinese supplier ships on deferred payment terms. For its named specific-contract products the insurer states a maximum indemnity of 90%, agreed case by case (Sinosure); short-term export cover normally runs to a credit period within one year and at most two (Sinosure 2024 annual report). The institution itself — the policy line-up, how a limit is obtained, the tariff, subrogation and practice with Russian buyers — is covered in a separate article on Sinosure; for credit pricing, the instrument is relevant only as a financing cost.

The scale of the instrument explains why this is not exotic: the aggregate amount insured across all lines in 2024 — short-term and medium- and long-term export credit insurance, domestic trade insurance and overseas investment insurance — was $1,021.4 billion (+10% year on year), across 227,000 clients (+12.4%) (Sinosure). A 70-article NFRA supervisory instrument has applied since 1 January 2025, expressly requiring maximum use of policy insurance in support of foreign trade, and on 14 April 2026 MOFCOM and the Exim Bank issued a notice on strengthening credit support for exports and imports over the 15th Five-Year Plan (mofcom.gov.cn).

The credit logic. Terms under cover move the borrowing onto the supplier's credit market: he funds himself with a bill discount (0.94% on average in June 2026) or by assigning his policy rights to a bank, that is, on the Chinese curve rather than the importer’s. The premium is paid by the supplier and passed into the price, so the real interest rate of the transaction is the supplier's price uplift, not the Sinosure tariff. The conversion is always the same: effective rate = uplift × 360 / days of terms.

Where the break-even lies. An uplift below 1.67% of the contract for each additional month of terms is arithmetically worthwhile for a borrower whose alternative is a 20% rouble loan, and not worthwhile for one whose alternative costs 8%. The threshold moves linearly with the borrower's own rate, and the negotiation is about that threshold, not about the insurer's tariff.

The currency side. With a dollar invoice, terms add no renminbi position — they move the date of the same dollar payment, and an importer earning roubles, lira or tenge keeps the dollar exposure that any dollar invoice carries. That distinguishes them from a renminbi loan, where the hedge has to be priced separately.

What closed in 2026

The regulatory window narrowed in 2026 at one specific point: a mainland non-financial enterprise may no longer fund loans to offshore borrowers with its own borrowings.

The outbound lending rules (境外放款). PBoC and SAFE published the Administrative Rules on Outbound Lending by Domestic Enterprises on 20 March 2026, effective 20 April 2026 (primary text at SAFE). The outbound lending limit equals owner's equity × a macroprudential coefficient, and the new rules set that coefficient at 0.6, up from 0.5 (annex 1) — so the rule not only closed off debt-sourced funds but also widened the limit for lending out of equity; repayment must be in the currency disbursed; three earlier instruments were repealed (Huifa 2009/24, Yinfa 2016/306, Yinfa 2021/2). The key provision (art. 13): funds of debt origin are prohibited — the mainland lender must use its own funds and may not finance an outbound loan with its own borrowing. The rules govern domestic non-financial enterprises (art. 2); a bank's direct lending abroad under Circular 27 remains a separate channel. The "borrow in China at 3%, lend elsewhere at 15%" structure has been eliminated by rule, not by enforcement practice.

Use of proceeds under Circular 27. A direct cross-border loan to a foreign company may not go into securities, into fictitious trade transactions, into repayment of external debt under a 内保外贷 structure, or back into China. The channel therefore works for working capital financing of a foreign operation, and does not work as a way to bring cheap money back onto the mainland.

The 1.75 parameter as an exchange rate lever. The macroprudential parameter for cross-border financing was raised from 1.5 to 1.75 on 13 January 2025 (gov.cn) and holds through 2026; the December 2025 cash pooling rules set a concentrated external debt limit for the pool as a whole of up to 3.5× the base (leverage of 2 × a parameter of 1.75). The parameter is set by joint notice of the PBoC and SAFE and can be moved in either direction; the January 2025 increase was presented as widening cross-border funding sources, which makes it a countercyclical exchange rate tool rather than a fixed limit. Any structure built on the current value must budget for its revision.

What opened instead. The tightening touched only structures that transfer borrowed funds out of China; the trade perimeter was expanded through 2026. The HKMA doubled its RMB Business Facility to CNY 200 billion from 2 February 2026. In April 2026 MOFCOM issued notices — with the Exim Bank and separately with Sinosure — on strengthening credit and insurance support for foreign trade over the 15th Five-Year Plan. In its notice of 15 May 2026 on financial services for small and micro enterprises, the NFRA requires banks to apply loan rollover policy effectively, to strengthen support for first-time and unsecured lending and for medium- and long-term credit, and to set the rate on inclusive SME loans taking into account the LPR, the cost of funding and the cost of risk (text mirror). The policy distinction is explicit: financing a real flow of goods is encouraged, financial arbitrage is closed.

Lawful financing opportunities have shifted away from the intermediary company and towards structures where a bank is the lender — 外保内贷, panda bonds, dim sum, cross-border loans under Circular 27 — and towards instruments tied to a physical shipment.

Risks

The risks of renminbi funding fall into three groups — revaluation, regulatory and sanctions-related — and in 2026 each has distinct implications.

Appreciation, not devaluation. The consensus risk that "the renminbi will fall" is inverted in 2026: the CFETS central parity stood at 6.7829 yuan per dollar on 2 September 2026 against 7.1089 a year earlier, a renminbi appreciation of about 4.8% year on year (CFETS), and the forward premium prices further appreciation. Over that year an unhedged renminbi borrower with dollar revenue lost about 4.8% on the revaluation of the debt, more than the two to three points he saved on the coupon against a dollar loan. The risk is two-sided by nature, but as of September 2026 the sign runs against the borrower.

Exchange control and use of proceeds. Funds raised under 内保外贷 may not be freely brought into China, and misuse is punishable under arts. 43–44 of the Foreign Exchange Administration Regulations by a SAFE fine of up to 30% of the amount involved, and up to the full amount in a serious case of changed use of funds (State Council). A call on the guarantee creates registrable external debt, which runs into the net assets ceiling.

Use-of-proceeds control at the level of the individual payment. A renminbi loan in China is not money in a current account to be disposed of freely. Any single disbursement above CNY 10 million falls under the entrusted payment regime: the bank remits the funds directly to the borrower's counterparty, bypassing its free balance. Working capital loans may not fund dividends or financial investments; after disbursement the bank runs post-lending monitoring and may demand early repayment where proceeds are misapplied. For a group this means the loan has to be planned against a specific working capital needs calculation and specific suppliers, not against general liquidity, and that intragroup transfers need an economic rationale of their own.

Rule changes. March 2026 demonstrated the speed: rules published on 20 March, in force on 20 April — one month to restructure. Any structure with a 12 to 24 month horizon needs a plan for a change of regime.

Country and sanctions filters. Every institutional cheap-funding structure — SBLCs from large banks, 内保外贷 and 外保内贷, the panda market — runs through the compliance of a tier-one bank, so the beneficiary's country and profile determine eligibility, which cannot be changed through pricing. The same mechanism operates for any jurisdiction under comprehensive sanctions or carrying secondary-sanctions risk, and the best-documented case is the Russian perimeter, where the window is closed by bank compliance rather than by PRC exchange control.

For a Russian connection, that compliance has a named mechanism. Executive Order 14114 of 22 December 2023 allows the US Treasury to impose measures on a foreign financial institution for a significant transaction involving a blocked person — up to SDN designation or a ban on correspondent accounts; since 12 June 2024 every person blocked under EO 14024 counts as part of Russia's military-industrial base (OFAC FAQ 1181). OFAC sets no numeric threshold for a "significant transaction" and weighs the totality of the circumstances (OFAC FAQ 1151), so a bank cannot calculate a safe payment size and cuts off the whole category.

Two things change the assessment of this risk. First, as of 14 November 2025 the United States had not sanctioned any Chinese bank or financial institution for facilitating payments for Russia — as recorded in a staff research report of the U.S.-China Economic and Security Review Commission (USCC) — so the effect has been achieved by threat rather than use. Second, in its 18th package of 18 July 2025 the EU listed Chinese banks for the first time, and for Heihe Rural Commercial Bank the effect can be traced by date — the listing published on 19 July 2025 (Council Regulation (EU) 2025/1494), a ban for EU persons from 9 August 2025, and removal from Annex XLV, together with Heilongjiang Suifenhe Rural Commercial Bank, by the 20th package (Council Regulation (EU) 2026/506 of 23 April 2026), so the current position is checked against the consolidated annex.

Neither the PRC nor Hong Kong has issued a legal act prohibiting business with Russian persons: what is observed are the institutions' own de-risking decisions, which are therefore reversible and negotiable. The refusal attaches not to the Russian origin of the counterparty but to the sanctions status of a specific participant in the chain, above all the remitting bank: a Russian bank blocked under E.O. 14024 has, since June 2024, counted as part of Russia's military-industrial base for the foreign-financial-institution mechanism (OFAC FAQ 1181), so a payment it sends is the category a Chinese bank is least able to process. The full chronology of the narrowing is in the article on Chinese bank restrictions.

For the credit question it follows that a "Russian borrower" is not one position but four, and they must not be mixed.

The consequences for ownership structuring are set out in the articles on sanctions screening and source of funds.

CNH basis risk. Offshore renminbi is a separate market with its own liquidity, and its rate is more volatile than the onshore one: through 2026 CNH HIBOR overnight held around 1.3–1.5% — 1.34% on 31 March, 1.47% on 3 July and 1.38% on 3 September — and in periods of pressure on the currency the PBoC has historically drained offshore liquidity, spiking short-dated rates by tens of percentage points. A borrower on a floating CNH HIBOR rate with a short refinancing tenor takes that risk in full. The practical conclusion: in CNH, either fix the rate for the whole term or keep the loan tenor no shorter than the trade cycle, so that a rollover does not fall in a moment of stress.

Goods perimeter risk. Every structure secured by a consignment adds a layer of risk unrelated to the rate: the integrity and authenticity of warehouse records, the good faith of the warehouse operator, the possibility of the same consignment being pledged twice, the quality of the security registration. The Chinese market answers this institutionally, with a unified online register of movable-property security and a shift to electronic warehouse receipts inside bank SCF platforms, where the pledge–disbursement–consignment redemption–repayment cycle is run by the bank jointly with the warehouse operator. The practical conclusion for the borrower: in such a transaction the warehouse operator is a participant in the credit structure rather than a logistics contractor, its selection is agreed with the lender, and the security interest is registered before disbursement rather than after.

Global SWIFT payment share. The renminbi's share of global SWIFT payments was 2.74% in February 2026, sixth place, against 4.33% a year earlier (SWIFT RMB Tracker). Cheap renminbi does not mean the yuanisation of settlement: the arbitrage runs through the capital account, not through payment flows.

Choosing an instrument

The choice is determined by four parameters in a strict order — sanctions profile, currency of the financed asset, presence of a PRC entity, and deal size — and the first constraint that binds eliminates whole branches.

1. Sanctions and country profile. Checked first, because it cannot be solved with price. Where the beneficial ownership structure or the commodity list gives a tier-one bank grounds to refuse, the branches involving SBLCs, 内保外贷/外保内贷 and panda instruments are closed regardless of financial metrics — what the bank checks in the transaction itself (HS code, end use, documents, source of payment, country risk) is set out in the article on trade finance in China. What remains is the goods perimeter and intermediated financing.
2. Currency and repriceability of the asset. If the financed goods are sold at a price you reset in line with the exchange rate within the loan's term, the renminbi loan keeps most of the differential. If the sale price is fixed in local currency, calculate the hedged rate — and the differential will most likely disappear.
3. A PRC entity. With one, Level C opens at 2.7–4.5%; creating one makes sense where the turnover itself runs through China and the horizon is at least two years, otherwise maintenance and tax administration outweigh the rate advantage. Without one, Levels A and B.
4. Size and tenor. Below $1 million and under 180 days, only Sinosure-backed terms and a usance LC are economically sensible: bank structures do not repay their approval times and fees. From $1 million and from a year, export buyer's credit, SBLC-backed lending and the Hong Kong CNH perimeter come into range.

The fifth step is arithmetic, and it is the one most often skipped: calculate each eligible instrument’s all-in cost of one trade cycle in cash — fees, conversion, insurance and storage included — and only then compare. An instrument with a higher headline rate but no conversion and no goods perimeter regularly turns out cheaper than the "cheap" renminbi loan, particularly below $1 million, where the fixed costs of a structure are not spread across volume.