Published: May 2026 | Data current as of: 30 April 2026 (DBS FY2025 Annual Report and Q1 2026 results released April 2026; OCBC FY2025 Annual Report 2025 and full-year results release February 2026; UOB FY2025 Annual Report 2025 and full-year results release February 2026. Where the annual report figures differ from the preliminary results releases, the annual report figure is used.)
The three numbers that matter
If you can read three numbers on a Singapore bank’s earnings release and tell me what they say, you understand more about banks than most retail investors. Those three numbers are NIM, ROE, and CET1.
Most retail investors evaluate banks the same way they evaluate a retailer or a manufacturer: dividend yield, P/E multiple, headline net profit growth. For companies with ordinary balance sheets, those metrics work reasonably well. For banks, they are actively misleading. Dividend yield tells you about payout policy, not value creation. P/E is noisy because bank earnings move hard with the rate cycle and provisioning decisions. Headline net profit can grow while the bank quietly becomes more fragile.
What actually drives a bank’s intrinsic value over the long run is the productivity of its capital (ROE), the efficiency of its core lending engine (NIM), and its capacity to absorb shocks without raising emergency equity (CET1). Everything else — dividends, loan growth, fee income momentum — flows from these three.
This piece is a framework reference for reading SGX bank disclosures. The primary worked example is DBS Group Holdings (SGX: D05); OCBC (SGX: O39) and United Overseas Bank (SGX: U11) appear at peer-comparison points throughout. All three are Singapore’s locally-incorporated “Big Three” banks — the universe most Singapore retail investors work with. Data is from FY2025 full-year results, with Q1 2026 used only for DBS where it adds to the worked example.
New to reading SGX annual reports? This guide walks through what to read, in what order, and what disclosures actually matter.
Why bank metrics are different
Banks are not industrial companies with leverage. They are leverage businesses with a layer of equity on top.
The balance sheet is inverted from what you are used to. Deposits — money the bank owes its customers — are the primary liability. Loans — money borrowers owe the bank — are the primary asset. Interest paid on deposits flows out; interest received on loans flows in. The spread between the two is the core of the income statement. When you model a manufacturing company, you build to free cash flow and discount it at the weighted average cost of capital. You cannot do this with a bank. For an institution whose business is intermediating money, drawing down a loan book generates a “cash inflow” in that framework — which would make banks look fantastically cash-generative while doing the opposite. The framework breaks.
The correct toolkit for bank valuation is different: dividend discount models (DDM), price-to-book (P/B) analysis using the Justified P/B formula, and reverse P/B to back out the ROE the market is pricing in. The full treatment of why this is and how it applies to DBS is in the published DBS article.
The metrics follow from the structure. NIM measures what the bank earns on its lending spread. ROE measures what the whole operation delivers to shareholders. CET1 measures how much loss-absorbing capital sits above the regulatory floor. Industrial metrics — EBITDA margin, free cash flow conversion, gross margin — applied to banks point you toward the wrong conclusions.
Net Interest Margin (NIM)
What it measures
NIM — net interest margin — is the ratio of net interest income to average interest-earning assets.
NIM = Net Interest Income (NII) / Average Interest-Earning Assets
Net Interest Income is gross interest earned on the loan and bond portfolio minus interest paid on deposits and wholesale funding. NIM expresses that net spread as a yield relative to the balance sheet that produced it — the gross margin of the lending business.
What drives it
Three variables move NIM meaningfully. The first is the spread between asset yields and funding costs — the most rate-sensitive component. When benchmark rates fall, floating-rate loan yields reprice down quickly; deposit costs fall more slowly. The resulting squeeze on NIM is the core mechanism behind the compression the Big Three have been navigating since late 2024.
The second is asset mix: the share of the loan book in floating-rate versus fixed-rate loans, and the proportion of interest-earning assets held in higher-yielding loans versus lower-yielding liquidity assets. DBS’s deliberate deployment of excess liquidity into high-quality liquid assets (HQLA — broadly, government bonds and central bank reserves counting toward regulatory liquidity requirements) in FY2025 lowered NIM while being accretive to NII and ROE overall.
The third is funding mix: the share of CASA (current and savings accounts — cheap, often near-zero-cost deposits) versus term deposits versus wholesale funding. A higher CASA share means lower funding costs and a wider spread at any given level of asset yields.
How to read it for SGX-listed banks
DBS posted NIM of 2.01% in FY2025, down 12 basis points from FY2024 as benchmark rates declined. By Q1 2026, NIM had fallen a further 12bps to 1.89%. OCBC came in at 1.91% for FY2025 — a 29bps year-on-year compression, the steepest of the three. UOB posted 1.89%, down 14bps year-on-year.
| Bank | NIM FY2024 | NIM FY2025 | Change (bps) |
|---|---|---|---|
| DBS | 2.13% | 2.01% | −12 |
| OCBC | 2.20% | 1.91% | −29 |
| UOB | 2.03% | 1.89% | −14 |
Source: FY2025 annual reports.
My reading: an SGX-listed local bank sustaining NIM at or above 2.0% through the current normalising-rate cycle is operating at the top of the SGX peer set. None of the three is there as of Q1 2026. Historically, the Big Three’s steady-state has been 2.0–2.2%; current compression reflects the rate environment, not structural deterioration. If NIM were to fall persistently below 1.7% for one bank while peers held above it, that would warrant a harder look at whether funding costs or asset mix are the cause.
One important shift: NIM sensitivity to further rate moves has fallen materially versus the 2022–23 cycle as the banks have built substantial hedge books. The next rate move will produce less NIM swing than investors accustomed to that cycle might expect.
Common traps
Reading the absolute level rather than trajectory and trend. A NIM of 1.95% stabilising quarter-on-quarter is a different story from one compressing 8bps every quarter.
Confusing NIM compression with NII decline. NIM is a rate effect; net interest income is rate plus volume. If the loan book grows 8% in a year when NIM falls 15bps, NII may still rise. Always check whether total NII is growing or shrinking alongside NIM.
Not asking why NIM is changing. A NIM decline from funding costs rising faster than asset yields is a different credit story from a NIM decline driven by deliberate deployment of liquidity into lower-yielding assets.
Return on Equity (ROE)
What it measures
ROE — return on equity — is net profit attributable to ordinary shareholders divided by average shareholders’ equity.
ROE = Net Profit / Average Shareholders’ Equity
ROE is the single best summary metric of capital productivity for a bank. It tells you how much profit the bank generates per dollar of equity entrusted to it.
What drives it
The clearest decomposition uses a version of the DuPont framework adapted for banks. Return on Assets (RoA) measures profit per dollar of balance sheet. For SGX’s Big Three, RoA runs well below 1.5% in FY2025 — low in absolute terms because the business model involves deploying large quantities of other people’s money, so the balance sheet is large relative to earnings.
The equity multiplier — total assets divided by shareholders’ equity — is the leverage component. SGX banks typically run 8–12×. That is the mechanism converting a sub-1.4% RoA into double-digit ROE: a 1.2% RoA on 12× leverage produces roughly 14% ROE. This is also why CET1 and ROE pull in opposite directions — more capital reduces leverage, which, all else equal, reduces ROE.
How to read it for SGX-listed banks
DBS posted ROE of 16.2% in FY2025, compressing from FY2024’s 18.0% as NIM fell. By Q1 2026, ROE had recovered to 17.0% as capital return began and non-interest income remained robust. OCBC delivered 12.6%. UOB posted 9.6% for FY2025 — meaningfully depressed by a S$615m pre-emptive general provision booked in 3Q FY2025 to strengthen coverage against anticipated macroeconomic headwinds; the underlying ROE, adjusting for that provision, is closer to 12%.
Among the three SGX-listed local banks, only DBS has sustained mid-teens ROE through the full FY2021–FY2025 rate cycle. OCBC has hovered in the low-to-mid teens; UOB has generally run in low-double-digit territory — FY2025’s 9.6% being an exception depressed by the pre-emptive provision noted above. That is a factual description of what has occurred — not a forward threshold. If DBS’s ROE were to drop to high single-digits through the next cycle, this framing would need to be revisited. For now, the through-cycle record places DBS at a clear premium on capital productivity relative to its local peers.
The ROE gap drives the valuation gap. DBS trades at roughly 2.4× book value; OCBC and UOB at roughly 1.2–1.5×. This is the Justified P/B formula at work: a bank generating ROE above its cost of equity creates value per share, and the market prices that. The full derivation applied to DBS is in the published DBS article; the OCBC article works through OCBC’s profile in the same framework.
Common traps
Reading current-year ROE in isolation. Bank ROE moves materially with the rate cycle. A through-cycle average is more informative than any single year.
Not adjusting for one-offs. UOB’s 9.6% is depressed by the pre-emptive provision discussed above; DBS’s FY2025 earnings reflect the first-year impact of the OECD global minimum tax framework. Normalise before drawing conclusions.
Confusing high ROE with high quality. High leverage produces high ROE but also higher tail risk. Always read ROE alongside the equity multiplier and CET1.
Ignoring the cost of equity (Ke). The value creation embedded in ROE depends on the spread between ROE and Ke — not ROE alone. If Ke is 12% and ROE is 14%, the bank is creating S$0.02 per dollar of equity per year. That spread is the basis of P/B valuation.
CET1 Capital Ratio
What it measures
CET1 — Common Equity Tier 1 — is the highest-quality, loss-absorbing layer of bank capital. It is broadly equivalent to common shareholders’ equity, adjusted downward for items the regulator considers insufficiently loss-absorbing: principally goodwill and other intangibles, deferred tax assets, and certain investments in other financial institutions. Divided by RWA — risk-weighted assets, where each asset class is scaled by a regulatory risk weight — it gives the CET1 ratio.
CET1 Ratio = CET1 Capital / Risk-Weighted Assets
CET1 is the regulator’s primary solvency metric: if a bank takes large unexpected losses, does enough CET1 remain to stay solvent without drawing on depositor or senior creditor funds?
What drives it
CET1 rises when the bank retains earnings or reduces RWA; it falls when the bank distributes capital through dividends or buybacks, takes large provisions, or grows the loan book. Capital management decisions — DPS policy, buyback programmes — are directly observable in the CET1 trajectory.
The regulatory framework for Singapore-incorporated banks sits in MAS Notice 637. For D-SIBs — Domestic Systemically Important Banks, a designation held by DBS, OCBC, and UOB — the effective CET1 minimum is 9.0%: a base minimum of 6.5% plus a mandatory capital conservation buffer of 2.5%. Non-D-SIB Singapore-incorporated banks face a 7.0% effective minimum. The D-SIB designation adds 2 percentage points above base Basel III minimums. On top of this, MAS can impose a countercyclical buffer of up to 2.5% in periods of elevated credit growth; no countercyclical buffer was in effect as of FY2025.
How to read it for SGX-listed banks
DBS reported fully phased-in Basel IV CET1 of 15.0% at FY2025 year-end, edging down to 14.8% by Q1 2026 as capital returns began. OCBC reported fully phased-in CET1 of 15.1%, while UOB posted 14.9%.
All three operate with CET1 well above the 9.0% effective D-SIB minimum — by at least 5.9 percentage points on a fully phased-in basis. This is not the regulator forcing them to hold more capital than they want. The banks themselves target ratios in the 13–14% range; the surplus above those internal targets has been running higher since the post-2022 capital build. “Fortress capital” is the SGX bank sector norm, not the exception.
The more interesting question is not whether the banks are well-capitalised — they are — but what they plan to do with the excess. DBS has been most explicit: the S$8 billion capital return programme targets the 13–14% range, with the Q1 2026 payout of S$0.81 per share — a S$0.66 regular interim dividend plus a S$0.15 capital return dividend — as its first quarterly instalment. That mechanism is visible in the CET1 move from 15.0% at FY2025 to 14.8% by Q1 2026.
What to look for in disclosure
The bank’s stated CET1 target range alongside the actual ratio. The gap between the two is the clearest signal of available return capacity — DBS’s transitional CET1 of 16.9% at Q1 2026 sits roughly 3 percentage points above the upper end of its 13–14% target range, which is the explicit basis for its current capital return programme.
The basis of reporting — all figures in this article are on a fully phased-in Basel IV basis. The banks also disclose a higher transitional CET1, which applies temporary regulatory relief during the Basel IV phase-in period. For the Big Three, transitional CET1 runs 1.8–2.2 percentage points above the fully phased-in figure; DBS’s transitional ratio at Q1 2026 was 16.9% versus 14.8% fully phased-in. The fully phased-in figure is the more conservative and forward-looking anchor for valuation analysis.
The countercyclical buffer in the Pillar 3 report. No buffer is currently active in Singapore; if MAS activates one, the effective 9.0% minimum rises accordingly.
Common traps
Confusing CET1 with other capital measures. Tier 1 capital includes CET1 plus AT1 — Additional Tier 1, principally perpetual capital securities — which is lower-quality. Total CAR (Total Capital Adequacy Ratio) adds T2 (Tier 2, mainly subordinated notes) on top. For solvency analysis, always start with CET1.
Treating high CET1 as uniformly positive. A ratio of 20% means the bank is sitting on capital well above its target, earning the risk-free rate on the surplus. That is value-destructive if the excess sits idle for years. The right level depends on the capital return commitment and the loan book’s credit profile.
Reading CET1 in isolation from the leverage ratio (Tier 1 capital as a percentage of total unweighted exposure). CET1 is risk-weighted; a bank heavily weighted toward low-risk assets can show a high CET1 ratio while still being highly leveraged in absolute terms. The leverage ratio is the backstop metric.
Secondary metrics
Non-Performing Loan ratio
The NPL ratio — gross non-performing loans divided by gross loans — is the headline credit quality measure.
For the Big Three in FY2025: DBS 1.0%, OCBC 0.9%, UOB 1.5%. All three are benign in absolute terms. OCBC has sustained sub-1.0% NPL for seven consecutive quarters. UOB’s 1.5% — the highest in the peer group — reflects its ASEAN-retail loan mix, which carries more exposure to emerging-market credit cycles than the more Singapore- and Hong Kong-concentrated DBS and OCBC books.
The stock ratio is a lagging indicator — NPLs accumulate over multiple periods. The figure to watch is NPL formation: new NPLs recognised in the most recent period. A rising formation rate with a stable stock ratio means provisioning is absorbing the flow; formation spiking while the stock ratio also rises tells you the bank is running behind. The NPL coverage ratio (allowances divided by NPL balance) above 100% signals conservative provisioning.
Non-interest income share
Non-interest income (non-II) covers fees, commissions, trading income, insurance contributions, and investment gains — everything except the lending spread. As a share of total operating income, it measures how much of the business is insulated from interest-rate cycles.
For FY2025: DBS approximately 37%, OCBC approximately 37%, UOB approximately 32%. The DBS and OCBC wealth management and bancassurance franchises have structurally grown their non-II share over the last decade; UOB’s lower share reflects its more loan-intensive ASEAN-retail positioning. A higher non-II share supports a higher P/B multiple because that income is more recurring and less rate-sensitive — one component of the structural premium at which DBS and OCBC trade relative to UOB.
Cost-to-Income Ratio (CIR)
CIR — operating expenses divided by total operating income — is the headline efficiency measure. Lower is better.
SGX banks have run CIR in the 38–45% range in recent years. DBS came in at 40.4% and OCBC at 40.2% for FY2025. UOB ran at 44.6%, reflecting its ongoing investment cycle in ASEAN post-integration. The direction of CIR matters as much as the level. Positive operating leverage — income growing faster than costs — is what creates compounding quality in earnings: a bank growing total income at 7% while holding cost growth at 3% is producing a progressively cheaper cost base. The reverse compounds equally unfavourably.
P/B versus ROE — the valuation linkage
The bridge between operating metrics and valuation sits in the Justified P/B formula:
Justified P/B = (ROE − g) / (Ke − g)
where Ke is the cost of equity and g is the long-run sustainable growth rate in book value per share. A bank earning ROE above Ke creates value per share, and the market prices the stock above book. A bank earning ROE below Ke destroys value, and the stock trades below book.
This is why DBS trades at roughly 2.4× book while OCBC and UOB trade at roughly 1.2–1.5×: the ROE differential maps directly through the formula. The gap is not the market misevaluating similar businesses; it reflects materially different capital productivity records. The full derivation and application to DBS is in the published DBS article. The companion OCBC article applies the same framework to OCBC’s different capital structure and ROE profile.
How to read a bank earnings release
Framework in hand, five habits worth building.
Read the three headline metrics first — NIM, ROE, CET1 — before reading the press release narrative. The narrative is management’s attempt to frame the story; the metrics are what actually happened. If NIM fell 15bps but the press release leads with “record non-interest income,” that is useful context. If NIM fell 15bps and the press release says “resilient performance,” draw the conclusion from the metrics independently.
Compare to the bank’s own recent history first, then to peers. A NIM of 1.95% means different things at different points in the rate cycle. The signal is in trajectory — is this bank’s NIM compressing faster or slower than peers? Is the ROE gap to DBS narrowing or widening? Peer-relative movement often reveals more than any absolute level.
Use the secondary metrics to direct attention to the right question. If ROE fell but NIM held steady, the deterioration is in non-II, CIR, or provisions — look there. The metrics should generate questions, not substitute for thinking.
Adjust for one-offs before drawing conclusions. UOB’s FY2025 ROE of 9.6% is depressed by the S$615m pre-emptive provision. DBS’s FY2025 earnings reflect the initial impact of the OECD global minimum tax framework. Both are disclosed and explained; the adjustment is the analyst’s job. One-off-adjusted figures are what is comparable across periods.
Read the Pillar 3 disclosure once a year per bank. The Pillar 3 report — separately filed alongside the annual report — contains the RWA breakdown by exposure class, the leverage ratio, the LCR (Liquidity Coverage Ratio, a 30-day liquidity stress metric), the NSFR (Net Stable Funding Ratio, a structural one-year funding metric), and the countercyclical buffer disclosure from MAS. Most retail investors never open it. For the amount of information per page about the bank’s actual risk profile, it is the second-most useful document a bank produces.
The framework underneath
This piece is the reference anchor. When valuation work on a specific bank lives elsewhere on this site — DBS and OCBC are published; UOB and others to follow — the three metrics above are the diagnostics running underneath every model. The DDM inputs, the Justified P/B implied by the ROE and cost-of-equity assumptions, the stress test on what an NPL cycle would do to CET1: it all flows from understanding what these metrics actually measure and what moves them.
For readers newer to bank analysis, the best next step is to pull the last four quarterly results releases for any of the three local banks and watch the metrics move. The rate cycle of 2022–2025 — from near-zero rates through tightening and into normalisation — compressed and expanded NIM, ROE, and CET1 in ways that are instructive precisely because they actually happened.
Disclosure: At time of writing, the author holds 1,100 units of Oversea-Chinese Banking Corporation Ltd (SGX: O39) and 64 units of United Overseas Bank (SGX: U11). The author holds no position in DBS Group Holdings (SGX: D05), and has no plans to initiate any such position within the next 72 hours. All three banks are referenced in this article solely as worked examples of the metrics discussed — not as recommendations. No view is expressed on whether any of the three is a buy, hold, or sell at current prices.
Disclaimer: This article reflects the author’s personal analysis and opinions, written in a strictly personal capacity. It is not financial advice and does not take into account any individual reader’s financial situation, investment objectives, or risk tolerance. Information is sourced from publicly available filings as of the date noted but accuracy cannot be guaranteed. The author may hold positions in securities discussed (see disclosure above). Readers should conduct their own research and consider consulting a licensed financial adviser before making any investment decisions. The author is not a licensed financial adviser under the Financial Advisers Act of Singapore.
An ordinary man with extraordinary inspiration