What makes a stress test report MAS / Central Bank-ready — and what makes it insufficient
A MAS / Central Bank stress test report that is insufficient is not one that failed the stress — it is one where the regulator cannot assess whether the stress was run correctly. The most common insufficiencies across SEA finance company stress test submissions are: scenarios that are not severe enough (testing 50bps when the guidance specifies 200bps), methodology descriptions that are vague (no indication of whether the bank's own historical data or standardised parameters were used), results that do not disaggregate by segment (a portfolio-level NPL figure provides no information about where the vulnerability sits), and management actions that are generic ("we will monitor the situation") rather than specific ("we will reduce the floating-rate funding proportion from 68% to 55% by Q2 2026").
The Stress Testing Agent AI generates a report structured to pre-empt each of these insufficiency types: scenarios specified at or above the MAS / Central Bank minimum; methodology stated with the data sources and assumptions used; results disaggregated by product, geography, and risk tier; and management actions derived from the scenario output with specific targets, timelines, and responsible owners.
A sample MAS / Central Bank stress test report: section by section
The institution has conducted stress testing of its loan portfolio (outstanding book: SGD2,840 million as of November 14, 2025) across three interest rate shock scenarios and two credit quality deterioration scenarios, in compliance with the MAS / Central Bank's Circular DNBR.PD.007/03.10.119/2015-16 and subsequent guidelines on stress testing for finance companies.
Key finding: Under the MAS / Central Bank benchmark 200bps instantaneous rate shock scenario, the institution's Capital to Risk-weighted Assets Ratio (CAR) falls to 15.6% — above the regulatory minimum of 15.0% but with a buffer of only 60 basis points. The institution's management has identified capital adequacy as the primary vulnerability requiring proactive management action, detailed in Section 5.
Portfolio covered: 100% of the outstanding loan book as of the test date. No sub-portfolios excluded.
Data source: Institution's Loan Origination System (as at November 14, 2025), core banking system balance data, and bureau consortium portfolio data. All figures auditor-reviewed as of the last quarterly close (September 30, 2025) and updated with October and November transactions.
Methodology: Top-down stress testing applied to segment-level data. Interest rate shock scenarios applied instantaneously to the rate curve per MAS / Central Bank guidance (not as a gradual shift). Asset quality deterioration estimated using the institution's own historical migration matrices for each product segment, calibrated to a 24-month stress horizon. Expected credit loss (ECL) provisions computed under Ind AS 109 Stage 1/2/3 migration assumptions consistent with the shock magnitude.
Key assumptions: (1) No management action is assumed to be taken during the shock period — this is a static balance sheet test. (2) Floating-rate borrowings reprice fully at the first reset date after the shock. (3) Fixed-rate lending assets do not reprice. (4) Deposit franchise (where applicable) assumed stable. (5) New origination during the stress period excluded.
| Scenario | Type | Shock Magnitude | MAS / Central Bank Status |
|---|---|---|---|
| Scenario 1 — Mild rate shock | Interest rate | +100 bps instantaneous | Below benchmark (reference) |
| Scenario 2 — Severe rate shock | Interest rate | +200 bps instantaneous | MAS / Central Bank benchmark minimum |
| Scenario 3 — Extreme rate shock | Interest rate | +300 bps instantaneous | Above benchmark (severe) |
| Scenario 4 — Mild credit deterioration | Credit quality | NPL +150bps from base | Standard stress |
| Scenario 5 — Severe credit deterioration | Credit quality | NPL +350bps from base | Severe stress |
| Metric | Base (no shock) | Scenario 1 (+100bps) | Scenario 2 (+200bps) | Scenario 3 (+300bps) |
|---|---|---|---|---|
| Net interest margin | 2.60% | 2.12% | 1.48% | 0.68% |
| Portfolio NPL (stressed) | 2.90% | 3.80% | 5.40% | 8.10% |
| Additional provisions (SGD million) | — | SGD34 Cr | SGD96 Cr | SGD188 Cr |
| CAR | 19.0% | 17.8% | 15.6% | 12.2% |
| CAR vs 15% floor | +4.0pp | +2.8pp | +0.6pp | −2.8pp (BREACH) |
| Borrowers TDSR (Total Debt Servicing Ratio) >45% | Base count | +2,840 | +7,240 | +14,600 |
Action 1 — Capital buffer strengthening (Primary): The Board has noted that the 200bps scenario produces a CAR buffer of only 60 basis points above the regulatory floor. Management has been directed to present a capital plan by January 31, 2026 that restores the CAR buffer to a minimum of 1.5 percentage points above the regulatory floor under the 200bps scenario. Options to be explored include retained earnings conservation, subordinated debt issuance, and selective portfolio rundown in high-risk-weight segments.
Action 2 — Funding duration extension: The current floating-rate liability proportion is 68% of total borrowings. To reduce cost-of-funds sensitivity to rate shocks, the treasury function has been directed to increase the fixed-rate proportion to 50% by June 30, 2026, targeting long-dated NCD issuance in H1 FY2026–27.
Action 3 — SE personal loan concentration limit: The self-employed personal loan segment shows the steepest NPL migration at 200bps (+4.6pp). Effective January 1, 2026, the maximum portfolio concentration in SE personal loans is capped at 12% of total AUM (currently 18%). New origination to be redirected to secured products (LAP, asset-backed loan) which show significantly lower rate sensitivity.
The regulatory report that requires follow-up questions has failed its purpose
A stress test report submitted to MAS / Central Bank that prompts a follow-up query letter costs more than the report cost to produce — in management time, in regulatory relationship quality, and in the signals it sends about the institution's governance. The Stress Testing Agent AI generates a report that is pre-emptively complete: scenarios at or above the MAS / Central Bank minimum, methodology stated with data sources and assumptions, results disaggregated to the level that allows the regulator to assess segment-level vulnerability, and management actions with specific targets and timelines. The institution that can say "we submitted a 5-section report with 5 scenarios tested, 3 management actions with specific timelines, and CAR analysis at 200bps — and received no follow-up queries" has satisfied regulatory intent, not just regulatory form.