A quarterly stress test is a regulatory obligation. An on-demand stress test is a risk management tool. The difference matters because the events that create the most urgent need for stress test data rarely happen on a quarterly schedule. A 50bps RBI rate announcement, a sector NPA surge, a credit rating downgrade at a major counterparty — each of these events changes the stress picture materially and demands an updated test within hours, not at the end of the quarter.
The two distinct roles of stress testing — and why they require different cadences
The regulatory stress test is a periodic attestation: the institution demonstrates to the regulator that it has tested its resilience under specified scenarios, documented the results, and identified management actions for any vulnerabilities found. This test is backward-looking in the sense that it reports on the portfolio as it currently stands, run on a fixed schedule, and submitted as a formal document. The regulator wants consistency, comparability, and completeness.
The operational stress test is a decision support tool: it answers the question "given what just happened, how does our risk position change, and does it change our decisions about origination, pricing, or capital?" This test is forward-looking, triggered by events, and consumed by the ALCO and the risk team rather than submitted to a regulator. It needs to be fast, specific to the event that triggered it, and actionable rather than comprehensive.
The Stress Testing Agent AI runs both — the quarterly test on the regulatory schedule with the full five-scenario suite, and the on-demand test as a targeted single-scenario run within minutes of a trigger event being identified.
The quarterly test: what runs, when, and what it produces
| Quarter | Test Date | Scenarios Run | Data Snapshot | Output | Submission |
|---|---|---|---|---|---|
| Q1 FY2025–26 | Jul 14, 2025 | 5 scenarios (3 rate shock + 2 credit) | June 30 portfolio | Board report + RBI submission pack | By Jul 31 (within 30 days of quarter-end) |
| Q2 FY2025–26 | Oct 12, 2025 | 5 scenarios + liquidity stress added | Sept 30 portfolio | Board report + RBI submission pack | By Oct 31 |
| Q3 FY2025–26 (next) | Jan 12, 2026 | 5 scenarios | Dec 31 portfolio | Board report + RBI submission pack | By Jan 31 |
| Q4 FY2025–26 (next) | Apr 12, 2026 | 5 scenarios + annual reverse stress | Mar 31 portfolio | Board report + RBI submission pack + Annual disclosure | By Apr 30 |
The on-demand test: six trigger conditions
Any RBI repo rate change triggers an on-demand rate shock update
Within 2 hours of an RBI rate announcement, the Stress Testing AI re-runs the 200bps scenario with the new base rate and computes the updated NIM compression and CRAR impact. The ALCO receives the updated figure before their post-announcement discussion. A 50bps hike does not change the 200bps stress scenario magnitude — but it changes the starting point, which changes the absolute NIM and CRAR numbers.
→ Scope: 200bps scenario only · SLA: 2 hours from announcement · Output: ALCO brief, not Board reportActual NPA increases more than 50bps in any single month
If the portfolio's actual NPA rate increases by more than 50 basis points in a single month — either from a seasonal factor, a sector deterioration, or an origination quality issue — the Stress Testing AI re-runs the credit quality stress scenario at the new base NPA. If the actual NPA has already moved 50bps, the stress scenario applied to a higher base produces materially different CRAR outcomes than the quarterly test showed.
→ Scope: credit stress scenarios only · SLA: 4 hours from NPA data available · Output: risk committee briefA significant borrowing is raised or repaid — altering the funding mix materially
A large NCD issuance, bank line drawdown, or commercial paper rollover that changes the floating/fixed funding proportion by more than 5pp triggers an on-demand cost-of-funds update. The rate shock impact is highly sensitive to the floating rate proportion — a funding mix change can improve or worsen the 200bps scenario outcome significantly.
→ Scope: Channel 1 (cost of funds) only · SLA: same day · Output: treasury briefActual CRAR falls within 1.5pp of the regulatory floor
If the institution's actual (unstressed) CRAR falls within 1.5 percentage points of the 15% regulatory floor, an on-demand stress test is triggered immediately — regardless of the quarterly schedule. At this proximity, any moderate stress pushes CRAR below the floor. The on-demand test quantifies the exact headroom and the specific actions required to restore it.
→ Scope: full 5 scenarios · SLA: 4 hours · Output: Board + ALCO emergency briefA single sector exceeds 15% of portfolio AUM
The Stress Testing AI monitors sector concentration continuously. When a single sector (e.g., MSME textile, real estate LAP, SE personal) exceeds 15% of portfolio AUM, it triggers a sector-specific credit stress test — what is the CRAR impact if that sector's NPA rate doubles? The answer determines whether the concentration warrants a cap.
→ Scope: sector-specific credit stress · SLA: 6 hours · Output: portfolio management briefALCO, Board, or risk committee requests a scenario not in the standard suite
The ALCO may want to model a specific scenario that is not in the standard suite — a simultaneous rate shock and NPA spike, a liquidity crunch combined with a credit deterioration, or a specific geographic stress (what if Karnataka MSME NPA doubles due to a textile sector slowdown?). These bespoke scenarios are requested and run on-demand, typically within 2 hours of specification.
→ Scope: bespoke scenario · SLA: 2 hours from specification · Output: requesting committeeProduction time comparison: Stress Testing AI vs manual process
Why on-demand matters more than quarterly in a rising rate environment
In a stable interest rate environment, the quarterly stress test cadence is adequate. The portfolio's rate sensitivity does not change dramatically between quarters, and the ALCO has time to review results and act before the next test. In a rising rate environment — or at moments of macroeconomic uncertainty when RBI policy direction is unclear — the gap between quarterly tests becomes a period of managed blindness: the institution knows its stress picture from last quarter but not from today.
The Stress Testing Agent AI eliminates this blindness. Every significant event that changes the stress picture — an RBI announcement, a sector NPA spike, a CRAR proximity alert — triggers a targeted on-demand run within hours. The ALCO's view of the institution's stress position is never more than hours old, not months. For decisions made between quarterly cycles — a large NCD issuance, a portfolio acquisition, a pricing committee review — this current stress data is the difference between a decision made with accurate risk information and a decision made with stale data dressed as current intelligence.
Risk does not wait for the quarterly calendar — and neither should your stress test
The events that most change an institution's stress picture — an RBI rate announcement, a sector NPA deterioration, a CRAR that has drifted toward the regulatory floor — do not happen on a quarterly schedule. They happen on the day they happen. An institution that can update its stress position within 2 hours of a material event is an institution that can make decisions at the speed that risk actually moves. The Stress Testing Agent AI makes the on-demand test as routine as the quarterly one — because the ALCO that meets on the day of a 50bps RBI rate hike and does not have an updated stress number before the meeting is making its most consequential decision of the year with information that is already three months old.