## STAMP€ (Stress Test Analytics for Macroprudential Purposes in the euro area)

## Source details

**Canonical URL:** [STAMP€ (Stress Test Analytics for Macroprudential Purposes in the euro area)](https://www.imf.org/-/media/files/news/seminars/mcm-lse-henry_session2.pdf)

## Other formats

- [Markdown version](/-/media/files/news/seminars/mcm-lse-henry_session2.pdf.md)
- [Structured JSON version](/-/media/files/news/seminars/mcm-lse-henry_session2.pdf.json)

---

### Context and purpose
- Presented by Jérôme HENRY, European Central Bank, at the IMF – LSE conference panel “Objectives of macroprudential stress-testing”, 15th December 2016, Washington DC.  
- Objective: provide a flexible top-down tool and running infrastructure for macroprudential analyses beyond microprudential stress-test quality assurance, enabling stand‑alone projections conditional on any scenario and embedding externalities, spillovers, and agents’ behaviour.

### The new territory for macroprudential stress testing
- Macroprudential function requires stress-testing frameworks that embed spillovers:
  - within the banking sector;
  - to other sectors, including the real economy;
  - allowing for banks’ own reactions that can spill over to other segments of the economy.  
- Quoted source: Vítor Constâncio, “The role of stress testing in supervision and macroprudential policy”, London 29 October 2015.

### Top-down, bottom-up, and inputs to supervision
- ECB staff conduct a variety of Top-Down stress tests for macroprudential purposes:
  - Quarterly risk impact assessment for the ESRB (EU‑wide).
  - Bi-annual exercise for the ECB FSR.
  - Macroprudential impact assessment for the Eurosystem.
  - Macroprudential extension of micro system-wide stress tests (see MPB).
- Top-down exercises also feed supervisory system-wide exercises (country-specific and EBA/SSM-wide; e.g. SSM CA2014, GR CA2015, EBA‑SSM2016).
- Bottom-up (banks’ results) are used for microprudential purposes and feed SSM-wide publication exercises (CA2014, EBA2016) and bank-specific supervision (SREP, ICAAP).

### ECB Top-Down stress-testing framework (workhorse for STAMP€)
- Core components:
  - Scenarios → Balance sheet → Satellite models → Macro models → Credit risk models → Profit models (incl. NII) → Market risk and loan loss models → Balance sheet and P&L tool → Solvency.
  - Contagion models, macro feedback models, financial shock modules, dynamic adjustment models, funding shock, RWA adjustments.
- Coverage indications: insurance + shadow banks, fire sales, micro households and NFC data (as extensions).

### Financial‑real feedback loops and second‑round effects
- Stress test results are fed into macro models incorporating “dynamic” balance sheet reactions from banks (example simulation based on Darracq Pariès et al. (2011)).
- Distinction emphasized between first‑round losses under the adverse scenario and second‑round losses that account for macroeconomic feedback of deleveraging.

### Spillovers, interconnectedness, and amplifications
- Systemic risks from interconnectedness often appear contained, but further analysis needed on:
  - price contagion and funding stresses;
  - cross-sector spillovers (investment funds and pension funds strongly affected by revaluation of bank equities).
- Interbank contagion related to direct bilateral exposures reported as immaterial in many simulations: below 10 basis points for most “simulated” interbank networks.
- Example metric categories shown:
  - Interbank contagion: distribution of second‑round loss on interbank exposures (pp of CET1 ratio).
  - Cross-sector spillover: losses triggered by revaluation of bank equities (% of total financial assets).

### Long‑term low interest rate (LIR) scenario implications
- Scenario characteristics and implementation issues:
  - Low growth and low interest rates for long—affecting profitability, viability, credit supply, risk‑taking, market structure.
  - Recommendation to elaborate a long-term scenario, i.e., well beyond 3 years.
  - Compare top-down stress test outcome with a back‑to‑normal scenario (higher interest rate levels, steeper yield curve, higher GDP).
  - Need to specify dynamic balance sheet (DBS) responses and consider other credit sources if not modelled.
  - Banks may search for duration (longer maturity portfolios) and tilt assets toward risk-taking (credit or market assets).

### Non‑performing exposures (NPE/NPL) surge and modelling implications
- Empirical patterns shown for Non-performing Loans to Total Gross Loans across countries (chart labels: CY FR GR IE IT ES) indicating significant variation across jurisdictions.
- Process and implementation issues for NPL modelling:
  - Necessity for conservative yet realistic NPL modelling (baseline vs. added stress).
  - Use of NPE buckets and income recognition rules for some classes.
  - Cure rates non‑zero and linked to macrofinancial assumptions.
  - New loans may have improved risk parameters—aggregate “PIE”.
  - Differing behaviour of NPE after write‑offs/restructuring.
  - Consistent and prudent use of provisions.
  - Data needs: NPE composition, cure rates, coverage ratios.
  - Secondary market sales as an option—need for pricing assumptions.
  - Emphasis on cure and provisioning modelling/calibration.

### Input to CCyB calibration: cost–benefit approach using top‑down STs
- Purpose: develop a stress‑test approach to inform CCyB calibration alongside other indicators/models.
- Costs and benefits:
  - Cost: higher capital in the short term with deleveraging.
  - Benefit: more resilient credit supply in a severe downturn.
- Procedure elements:
  - Country-specific, procyclical severity scenarios.
  - Two-step process: 1st and 2nd round effects from simulations.
    - 1st round: capital shortfall to a given target (required/market).
    - 2nd round: triggered banks’ reactions (deleveraging vs issuance), consider range of options (e.g., deleveraging good loans, all loans; pecking order non-core first; portfolio behaviour).
  - Metric: GDP under various assumptions (also other models).
- Implementation issues:
  - Policy process sequencing and interaction with P1R, P2R, or P2G not yet clear.
  - Data considerations: consolidated vs solo; macroprudential is a country-level step.
  - Defining cyclical scenario and coverage of risks.
  - Trade-off between country‑specific scenarios and euro‑area relevance.
  - Scenario horizon should be preferably a full cycle.
  - Further medium-term benefits potentially on funding costs or interest rates.
  - Defining “hurdle” rate for capital shortfall (aggregates).
  - Specification of banks’ reactions (DBS) remains critical.
  - Consideration of “pre-emptive” add‑on‑like CCyB if build‑up is detected.

### Liquidity stress testing (TD LST) and interaction with solvency
- TD LST framework combines quasi‑accounting and agent‑based modelling (ABM), covering:
  - Scenarios → Liquidity shocks → Balance sheet response → Impact measures → Feedback loops to solvency.
- Typical liquidity shock channels:
  - Deposit outflows, wholesale market seizure, derivative market seizure, use of committed credit facilities, margin calls, use of inflows from maturing assets, repo-eligibility with central bank, fire sale of assets.
- Impact measures and indicators include impaired liquidity position, lower flow of new credit, P&L loss (markdown of AFS and HFT assets), open market risk positions, macro deterioration, capital ratio / P&L, LCR / NSFR, Eurosystem recourse, ELA recourse.
- Historical scenario anchors: past liquidity shocks (1994, 2008, Greece 2012, EM sudden stops).
- TD LST highlights that funding, herding and solvency interactions are needed to observe market action; specific channels enumerated (A) Collateral limits; (B) Fire‑sale impact; (C) Interbank losses due to cash hoarding; (D) Funding cost shock following ∆CAR; (E) Peers’ funding cost impacted; (F) Insolvency spread via cross holding of debt.

### Coverage details and illustrative modelling features
- Asset/liability granularity: 10 asset / liability classes.
- Geographic coverage example: 28 EU countries, 20 RoW countries/areas.
- Network amplification: simulated interbank networks and system “topography” used to assess amplification via interconnectedness (source: Halaj and Kok (2013)).
- Satellite models translate scenario drivers (GDP, inflation, unemployment, reference interest rates) into risk parameters (PD, LGD, NII estimation output illustrative).

### Key conclusions and development priorities (STAMP€ outlook)
- STAMP€ e-book to be published shortly (as of presentation).
- STAMP€ provides a running infrastructure for macroprudential analyses and a stand‑alone projection tool conditional on scenarios.
- Externalities, spillovers, and agents’ behaviour already introduced, but further work needed:
  - Refine the DBS approach; shift to portfolio‑factor (PF)-based ALM type of bank behaviour (e.g., deleveraging).
  - Specify deleveraging implications (for example for NPL).
  - Go beyond bank solvency only: integrate liquidity stress‑test set‑up (accounting/contagion), connect insurers and extend to shadow banking in a broad sense.

---


_Source: https://www.imf.org/-/media/files/news/seminars/mcm-lse-henry_session2.pdf_