Firmwide Scenario Models: Silo vs Integrated Stress Testing

Book: Financial Risk Management: Applications in Market, Credit, Asset and Liability Management and Firmwide Risk Authors: Jimmy Skoglund & Wei Chen ISBN: 978-1-119-13551-7

Chapter 9 is the capstone. All the silo techniques from earlier chapters feed into firmwide scenario analysis and stress testing. Regulators now require it. Banks that skipped it before 2008 paid the price.

Why firmwide stress testing grew

SCAP in 2009 started with credit losses under macro scenarios. Today CCAR, DFAST, and EBA tests want full balance sheet projections: assets, liabilities, earnings, losses, capital, and liquidity under multi-year stress.

CCAR covers US bank holding companies above $50 billion. DFAST starts at $10 billion. EBA runs parallel exercises in Europe. Emerging markets are adopting similar frameworks.

The shift from “how much credit loss in a recession” to “does the bank stay viable with adequate capital ratios for nine quarters” is a much harder question.

Bottom-up scenario approach

Generate joint scenarios for core risk factors (rates, FX, GDP, unemployment, credit spreads, etc.). Value each exposure under each scenario. Sum to firmwide P&L per scenario. Derive VaR, capital needs, or pass/fail thresholds from the scenario distribution.

This contrasts with Chapter 8’s top-down approach, which aggregates pre-computed silo risk numbers without joint scenarios.

Silo approach

Each risk system handles its own piece:

• Market risk: trading P&L under scenarios
• ALM: net interest income projections
• Credit risk: loan losses and provision needs

A firmwide scenario generator creates macro paths. Satellite models translate macro factors into portfolio-level risk factors (covered in Chapters 2-4). Results aggregate scenario by scenario.

Challenge: consistency. Different systems may use slightly different scenario definitions or timing. Aggregation is a separate step that can introduce gaps.

Firmwide risk model approach

One integrated model projects earnings, credit losses, and balance sheet items together per book of business. Avoids the inconsistency of running credit losses in one system and cash flows in another.

Often implemented per book (mortgages, cards, corporate, trading) with a second-stage firmwide aggregation.

Advantage: loan-level consistency on cash flows, accruals, and credit losses. Management action rules (cut dividends, raise capital, shrink balance sheet) can be modeled with feedback loops that are hard to coordinate across silo systems.

Approximate firmwide models trade precision for tractability when a full integration is too costly.

Risk capacity vs risk exposure

Firmwide capital analysis compares risk capacity (capital plus earnings buffer) to risk exposure (losses at a quantile or under a stress scenario). This extends the risk reserve logic used throughout the book to the entire firm.

ICAAP uses this framing internally. Regulatory stress tests use a related but stricter lens.

Data and organizational challenges

Stress testing needs clean data across risk and finance. Siloed systems, inconsistent definitions, and manual reconciliation were common pre-CCAR pain points.

Building the process takes years of iteration. The book is realistic about that.

Satellite models and macro mapping

High-level macro scenarios (GDP, unemployment, house prices, rate paths) must map to portfolio risk factors. Satellite models do that translation. They can be statistical regressions, factor models, or reduced-form links calibrated to historical stress episodes.

Weak satellites produce weak stress tests. A beautiful macro narrative with arbitrary PD shocks is not stress testing. It is storytelling.

Organizational reality

Firmwide stress testing forced risk and finance to share data definitions. Net interest income, provision expense, trading VAR losses, and operational loss scenarios must reconcile to one income statement per scenario path. Banks that treated CCAR as a regulatory filing exercise missed the internal management value.

ICAAP connection

Internal capital adequacy assessment uses the same bottom-up machinery with bank-chosen scenarios rather than only regulatory paths. Reverse stress testing (how bad must things get to break capital) complements prescribed scenarios.

My take

The silo approach is where most banks start because the systems already exist. The integrated model is where they want to end up because regulators and boards ask questions that span credit, rates, and balance sheet growth in one scenario.

Satellite models are the unsung hero. Without a credible link from “unemployment hits 10%” to “mortgage PD rises to X,” the whole exercise is theater.

The book is candid that neither approach is perfect. Silo aggregation is faster but fragile. Integrated models are coherent but expensive. Hybrid designs are the pragmatic end state for large banks.

Previous: Firmwide Risk Aggregation Next: Firmwide Stress Testing and the Future