Mr Speaker, half of this Bill deals with what happens if one of our big banks fails. A bank's first cushion against losses is its capital, mostly its shareholders' money. If losses ever used that up, the Government would have to choose between letting the bank collapse, taking deposits and payments with it, and rescuing it with public money.
This Bill lets the Monetary Authority of Singapore, MAS, require our banks to hold a second cushion, called total loss-absorbing capacity: special debt that can be written down or turned into shares if a bank fails, so that its own investors rescue it rather than the public. This is called a bail-in, which MAS already has the power to do. The second cushion is meant to ensure there is enough such debt to give a failed bank fresh capital and keep it serving customers while it is restructured. The Financial Stability Board wrote this standard in 2015 for the world's largest banks. MAS proposes that capital and this debt together come to at least 14 per cent of risk-weighted assets, and our three local banks already meet it. The Bill also writes proliferation financing, the funding of weapons of mass destruction, into the Act, alongside money laundering and terrorism financing.
Sir, I have five questions and one proposal.
I. When the rule-makers disagree
The first concerns the measure. The standard is the Financial Stability Board's, but the measure it uses is the Basel Committee's. That Committee is where bank supervisors from 28 jurisdictions, MAS among them, agree how much capital banks must hold. MAS's own consultation says the Board developed the standard "in consultation with" the Basel Committee. The cushion is 14 per cent of risk-weighted assets, a Basel number. So any disagreement over Basel bears on this Bill.
There is one. The US Federal Reserve's Vice Chair for Supervision describes America's approach as tailoring (quote) "requirements to the U.S. financial system rather than adopting international standards wholesale" (end quote). America's proposed version of the final Basel reforms would lower its largest banks' core capital requirements by about 5 per cent. Britain has delayed them to 2027, and the European Union has put a temporary multiplier on its trading-book rules to protect its banks.
Singapore implemented the final reforms in July 2024, ahead of Europe, Britain and America. By the Basel Committee's latest figures, they raise required capital by 3.6 per cent outside Europe and the Americas, against about 1 per cent in each of them. At DBS, the floor in those reforms would add about 50 billion dollars of risk-weighted assets by 2029, largely, the bank says, because of its lending to companies.
I am not arguing for leaving the club. A small financial centre lives on its reputation, and a clean Basel grade lets foreign supervisors rely on ours. But when the rule's authors step back, we should say which course we are on: the text on the Committee's timetable, or what the largest members actually practise.
My first question. What is Singapore's Basel strategy now that the largest members are departing from the text?
II. What the tape counts as zero
The second concerns what that measure leaves out. The cushion is 14 per cent of risk-weighted assets. Think of risk-weighted assets as a measuring tape that counts each asset by how risky the rules say it is. On that tape, government bonds rated AA minus or better count as zero, or close to it, however many a bank holds, United States Treasuries included.
Nearly all of Silicon Valley Bank's loss sat in bonds the rules counted at zero or one-fifth. MAS's consultation points to "the failure of several US banks in 2023", Silicon Valley Bank among them. It had put its deposits into long-dated government-backed bonds, booked at the price it paid. When rates rose, those bonds lost about 15 billion US dollars, and none of its ratios showed it: its capital ratios were above every minimum, and its leverage ratio, capital against everything it held, was 8.1 per cent against a floor of 4. When its depositors, nearly all uninsured, pulled out 42 billion US dollars in a single day, the loss could no longer wait. Nor was it alone. At the end of 2022, American banks carried about 620 billion US dollars of paper losses on their bonds, and in the end the Federal Reserve lent against those bonds at face value. The loss no ratio counted ended up as public risk.
Our banks are better placed than Silicon Valley Bank was, and MAS already makes them publish what a rate shock would do: at DBS, a rise of around two points costs 1.7 billion dollars, 2.8 per cent of its core capital. The banks' accounts say how much of their government bonds they carry at cost: about 86 billion dollars across the three, and at DBS it more than doubled last year, to 56 billion. They do not say what those bonds are worth today, which governments issued them, or what the system would lose on them together.
My second question. Will MAS require our systemically important banks to publish every half-year, on a common template, their government bond holdings country by country and the market value of the bonds they carry at cost? And will MAS publish the same for the whole system?
III. Would the cushion work when it is needed?
The third concerns whether the cushion would work when needed. The answer depends on things beyond this Bill. The Financial Stability Board's own report in January says challenges persist in "funding in resolution and effective bail-in execution, particularly in cross-border contexts". Credit Suisse shows what that means.
In March 2023 Switzerland arranged UBS's takeover of Credit Suisse, backed by public guarantees, and its regulator wrote off about 16.5 billion Swiss francs of the bank's AT1 bonds, designed to absorb losses, while shareholders were paid. The order rested partly on an emergency ordinance, not on Switzerland's resolution regime, its law for handling a failing bank in an orderly way, which the Financial Stability Board later said offered "an executable alternative". The same week, the Bank of England did use its resolution powers on Silicon Valley Bank's British arm. It sold the bank to HSBC for one pound, protected depositors, used no public money, and wrote its loss-absorbing bonds down to nothing.
The Swiss write-off came undone at home. In October 2025 a Swiss court revoked the order, finding the bonds' own trigger had not been met and no statute authorised it; the ruling is under appeal, and the bonds are still worthless. Abroad, bondholders lost in New York, where courts held Switzerland immune, and investors from Japan, Hong Kong and Singapore have turned to investment treaties. And the doubt spread to all such bonds: within three days, regulators in the European Union, Britain, Hong Kong and Singapore had to say publicly that, under their own rules, shareholders would take losses first.
My colleague Jamus Lim will ask whether bonds like these should still count towards the cushion.
I draw three lessons. First, the legal basis must be clear, and it is here: the bail-in power is in statute, and this Bill adds the cushion.
Second, recognition must be agreed in advance, and it runs both ways. Part of this cushion will probably be sold abroad. DBS raised two billion US dollars in March, and MAS allows the debt to be governed by foreign law, so a bail-in of it would rest on contract clauses and, if challenged, on a foreign court. Our own law is rightly careful here: section 103 lets MAS recognise another country's resolution only if that would not harm our financial system, treat our creditors unfairly, go against the public interest, or cost Singapore money. Every other country is entitled to the same care. We are a host as well as a home. MAS proposes that foreign banks' Singapore subsidiaries hold 75 to 90 per cent of their requirement, a figure the Financial Stability Board says should be set "in consultation with the home authority". Clear, published criteria on our side make it easier for others to agree theirs.
Third, who holds the new layer affects whether it will be used. DBS carries almost no AT1, about one million dollars, and OCBC and UOB about 1.7 and 3.4 billion. The new layer will be mostly new debt, and its holders are not yet known. MAS will keep bail-in debt sold to retail investors in Singapore out of the new layer, because concern over imposing losses on them "may delay decisions relating to bail-in". But such bonds can still be sold here to accredited investors, including individuals who meet an asset or income test, and to retail investors abroad.
My third question. Has MAS's resolution planning for DBS, OCBC and UOB been tested for legal recognition wherever it would need to be enforced? And has the Government assessed whether writing down foreign holders could be challenged under Singapore's investment treaties?
IV. What the FATF also asked for
The fourth concerns the Bill's other half, which answers the Financial Action Task Force. Its evaluation of Singapore in May rated us "Moderate" on proliferation-financing sanctions. It also found that information on who owns companies is largely not verified beyond the checks firms make on their own customers, and that Singapore "prosecutes few legal persons, local directors and professional intermediaries", which is "not in line with its risk and context". Its roadmap sets Singapore twelve key actions, none of them a new law. My colleague Sylvia Lim will speak to the three on proliferation financing. Two others bear on the companies our banks lend to: verifying ownership in ACRA's registries, and pursuing the directors and intermediaries who facilitate money laundering here.
My fourth question. How, and by when, will the Government meet those two?
V. What would actually help Asian banks
Finally, a proposal. This Bill prepares for a bank that fails. After 1997 the IMF found that our region's banks had "relied on collateral rather than credit assessment and cash flow analysis", and that connected lending made it worse. Cleaning up cost Indonesia's government more than half a year's output. There are those who would answer Asia's banking risks with an Asian Basel: softer capital ratios for Asian conditions. But I think Asian banking requires different advances.
It requires better lending. Asia's problem is often the allocation of credit, not its quantity: bank lending to the private sector is about 118 per cent of GDP in Malaysia, 112 per cent in Thailand and 125 per cent in Vietnam. Yet only about one firm in three in Malaysia and Vietnam, and one in seven in Thailand, uses a bank loan to finance investment. A bank in our region will readily lend against another building, and turn away an exporter who has orders in hand but little property to pledge. Our own small and medium enterprises face the same preference: in 2013, 55 per cent of their loans were secured on property; by June this year, about 70 per cent were. Cutting capital requirements across the board would simply finance more buildings. Better borrower information, enforceable claims and guarantees aimed at identifiable risks change where the credit goes. For the same reason, a regional approach should sometimes be tougher, on connected lending, hidden losses and implicit state guarantees.
Singapore chairs ASEAN next year. The Prime Minister has said of the region that "there are regulatory restrictions and there are ways in which we can make things more seamless". Credit is one of those ways. Here are three considerations to advance.
First, a firm's repayment record that crosses a border. Take a supplier in Johor that has served Singapore factories for ten years and never missed a repayment to its Malaysian bank. A Singapore bank cannot easily see that record, so it will tend to ask for property collateral. If credit bureaus in the region shared such records, with the firm's consent and a common way of identifying it, that supplier could show its history here instead of starting from nothing, and our firms could do the same there. In Hong Kong, banks that make more use of firms' shared records charge lower rates and ask for less collateral.
Second, a supervisory view of one corporate group across several countries, for supervisors only. Three borrowers in three countries can turn out to depend on one property group, and today no single supervisor sees the combined exposure. The euro area already does this: its central banks collect data on individual bank loans to companies above 25,000 euros, so supervisors can see a company's borrowing across banks and borders. Enforcement at home, against powerful interests, remains the hard part.
Third, verified trade records, so the same invoice or shipment cannot be financed twice. In 2020 Hin Leong left 23 banks owed about 3.5 billion US dollars, and some of its cargoes had been financed more than once. Our banks have already built such a registry. The Association of Banks in Singapore launched its Trade Finance Registry in June 2023; more than 40 institutions now use it, and the ASEAN Secretariat has expressed interest.
We could start with a two-year pilot involving Singapore, Malaysia, Indonesia and Thailand, covering one trade corridor and one cross-border banking group, and publish the results. It would have to be reciprocal: useful information flowing both ways, and credible assurances about how each other's banks would be treated in a crisis.
My fifth question. Will the Government put such a pilot on the agenda for its ASEAN chairmanship?
Close
Mr Speaker, this Bill makes our banks safer if they fail. The larger work is to make them more useful while they stand. The Basel rules were rebuilt after 2008 around the failures of wholesale-funded, trading-heavy banks, and Asian banks like ours, which lend out their customers' deposits, are not that kind of bank. I hope future legislative energy in banking will focus on rules better suited to an Asian context - rules aimed at how our banks actually lend, and how they would actually fail.
Thank you, Mr Speaker. I support the Bill.
