15.4 Stress Testing

Stress Testing

Stress testing asks what happens to earnings, capital, liquidity, and franchise value under severe but plausible shocks—and under structural breaks that historical VaR windows may never have sampled. VRM–8 connects rationale, enterprise-wide design, stressed risk measures, reverse stress tests, governance, and Basel’s high-level principles.

Why Stress Test?

Day-to-day risk metrics (VaR, sensitivities, ratings-based credit capital) are calibrated to recent history and model assumptions. They can fail when:

• Correlations spike toward 1 in crises (diversification vanishes).
• Liquidity evaporates and exit assumptions break.
• Nonlinear products gap through hedges.
• Credit migrations and defaults cluster far beyond mean EL.
• Operational or conduct events coincide with market stress.

Stress testing is therefore a complement, not a replacement: it explores joint tails, management actions, and capital trajectories that a single 99% one-day VaR number cannot show. Supervisors also use stress tests as a macroprudential and microprudential tool (e.g., CCAR/DFAST-style exercises in the U.S., EBA/ECB EU-wide tests, Bank of England ACS)—FRM expects conceptual familiarity, not jurisdiction-specific form trivia.

Enterprise-Wide Stress Versus Other Measures

Enterprise-wide stress testing links macroeconomic scenarios to credit losses, market P&L, fee income, OpRisk overlays, risk-weighted assets, and capital ratios over a multi-quarter or multi-year horizon. Narrow shocks (parallel +100 bp on a rates desk) are useful but incomplete: they miss feedback loops such as rating downgrades raising funding costs, which force asset sales, which further depress prices.

Stressed VaR and Stressed Expected Shortfall

Stressed VaR (sVaR) estimates VaR using model inputs from a continuous stressed historical period (for example, a 12-month window spanning 2007–2009) rather than the recent calm window. Basel market-risk frameworks historically required sVaR alongside current VaR to keep capital from collapsing in benign markets.

Stressed expected shortfall (ES) applies the same idea to ES (average loss beyond VaR), aligning with FRTB’s shift toward ES while retaining the need for stressed calibration. Conceptually:

Current VaR/ES → uses recent volatility and correlation Stressed VaR/ES → uses stressed volatility/correlation (or stressed P&L history)

Worked comparison sketch

A equity desk’s current 10-day 99% VaR is USD 12 million using the last year of data (realized vol ≈ 12%). Revaluing the same positions on returns from a stressed year with realized vol ≈ 35% and higher pairwise correlations produces stressed VaR ≈ USD 40 million. Capital add-ons keyed to sVaR prevent the desk from looking “low risk” merely because the recent sample was quiet. Exact multipliers depend on regulatory formulas; the exam point is the stressed window / stressed parameters logic.

Scenario and Model Challenges

Designing scenarios is hard:

1. Severity versus plausibility — too mild fails the purpose; too extreme is dismissed as science fiction.
2. Consistency — GDP, unemployment, FX, spreads, and property prices must cohere (a full macro narrative).
3. Second-round effects — fire sales, rating triggers, contingent facilities drawing, and behavioral assumptions for deposits.
4. Model risk — PD/LGD models estimated in expansions misstate downturn losses; market models miss jumps; OpRisk overlays are judgmental.
5. Data and systems — enterprise stress needs reconciled exposures across legal entities, currencies, and products.
6. Management actions — which mitigations are realistic under stress (deleveraging, dividend cuts, new issuance) versus wishful thinking?

Good programs document assumptions, run sensitivity to key elasticities, and compare model output to historical crisis benchmarks.

Reverse Stress Testing and Regulatory Examples

Reverse stress testing starts from a defined adverse outcome—breach of capital ratio, failure of the business plan, or resolution trigger—and works backward to identify scenarios that could cause that outcome. It answers: “What would have to happen for us to fail?” rather than “What is the P&L in this pre-specified scenario?”

Uses:

• Reveal hidden concentrations and fragile assumptions.
• Challenge official scenarios that never threaten viability.
• Support recovery and resolution planning.

Regulatory examples (high level): supervisors prescribe baseline/adverse/severely adverse macro paths and require projected capital ratios, sometimes with qualitative review of governance and controls. Banks must explain results, capital actions, and shortcomings. Reverse stress tests are often expected as part of ICAAP/ILAAP-style internal assessments even when not the centerpiece of the public supervisory exercise.

Roles: Board, Senior Management, and Audit

If stress results never change limits, dividends, or business appetite, the program fails the use test regardless of model sophistication.

Policies, Validation, and Independent Review

A sound stress-testing policy defines: scope (legal entities, risk types), frequency, scenario library, model inventory, data lineage, approval authorities, and escalation thresholds. Validation covers conceptual soundness of satellite models (PD, LGD, PPNR, market shocks), benchmark comparisons, and outcome analysis where feasible. Independent review (validation team and/or audit) should not be the same people who built the models. Findings must feed remediation with timelines.

Basel Stress-Testing Principles (High Level)

Basel Committee guidance on stress testing (principles for banks and supervisors) emphasizes themes FRM candidates should recognize:

1. Stress testing should be a firm-wide risk-management tool, integrated into governance and decision-making—not a compliance silo.
2. Board and senior management responsibility for the program’s effectiveness.
3. Proportionality — intensity commensurate with size, complexity, and risk profile.
4. Meaningful, relevant scenarios covering material risks and multiple horizons; include reverse stress tests.
5. Robust infrastructure and data to produce timely, reliable results.
6. Regular review and update of scenarios, models, and assumptions; independent challenge.
7. Supervisory expectations that results inform capital and liquidity adequacy assessments.

(Exact principle numbering evolves across BCBS papers; learn the themes, not rote IDs.)

Putting Stress Testing on the Exam

When a stem contrasts VaR with stress tests, emphasize joint tails, model breaks, and multi-period capital paths. When it mentions sVaR/sES, point to stressed calibration windows. When it asks about reverse stress, work backward from failure. When it asks who does what, map board (appetite/challenge), management (execution/use), and audit/validation (assurance). Stress testing is where Valuation & Risk Models meets governance—and where “severe but plausible” must still be decision-useful.