Bank Stress Testing Programme Template (EBA/GL/2018/04)

Stress Testing Programme: [Institution name] [Programme year, e.g. 2026 to 2027]

Stress testing programme for a credit institution: objectives and scope, governance, the inventory of stress tests, scenario design and severity calibration, the macro-financial variables, methodology by risk type, data and balance sheet assumptions, results, management actions, reverse stress testing, challenge and validation, and the use of results in the risk appetite framework, the ICAAP, the ILAAP and the recovery plan. Based on EBA Guidelines on institutions' stress testing (EBA/GL/2018/04), Directive 2013/36/EU (CRD) Articles 73, 86 and 100, and the revised EBA SREP and supervisory stress testing Guidelines (EBA/GL/2026/06, applying from 2027-01-01); structure also aligned with the Basel Committee "Stress testing principles" (October 2018), cited by name only. Requirements are referenced by number; obtain the guideline, directive or principles for the normative text and check the current version. Replace every [bracketed] value and every row marked Example. This is a document format, not financial, investment, legal, regulatory or accounting advice.

Document control

Approvals

Revision history

Contents

1. Objectives and scope
2. Governance and roles
3. Stress test inventory
4. Scenario design
5. Methodology by risk type
6. Data and balance sheet assumptions
7. Results
8. Management actions
9. Reverse stress testing
10. Challenge, validation and audit
11. Use of results
12. Annual calendar
13. Regulatory map and jurisdiction notes
14. Glossary and references
15. Standard and licence note

1. Objectives and scope

State the objectives in terms of decisions: setting and testing the risk appetite, assessing capital and liquidity adequacy in the ICAAP and ILAAP, calibrating the recovery plan indicators, informing the strategic plan and the budget, setting limits and identifying concentrations and hidden vulnerabilities. Then state the scope: risk types, entities, portfolios, horizons and the frequency of each test.

Example: Fernhollow Bank's programme has five objectives: (1) confirm that capital stays above the risk appetite floor of 12.0% CET1 in a severe but plausible adverse scenario over three years, or define the actions that restore it; (2) confirm a liquidity survival horizon of at least 60 days in a combined liquidity stress; (3) test the commercial real estate concentration and the deposit franchise, the bank's two main vulnerabilities; (4) calibrate recovery plan indicators; and (5) support the annual strategic plan with a downside case. The scope covers all material risks in the ICAAP risk inventory and the consolidated perimeter. The small client-hedging trading book (internal model for general market risk) is stressed through the trading book scenarios and the stress loss limit MRL-003 (EUR 25 million) of the Market Risk Limit Framework (finance/market-risk-limit-framework); the result is immaterial at group level and is not modelled separately in SCN-002.

2. Governance and roles

Assign clear responsibilities across the three lines. The management body approves the programme and the main scenarios and discusses the results and actions; senior management runs the programme and ensures resources; the risk function owns methodology and coordination; finance owns the baseline projections; business lines provide expert input; model validation and internal audit provide independent challenge.

3. Stress test inventory

Cover the main types: sensitivity analysis (one risk factor moved at a time), scenario analysis (several factors moved together under a coherent narrative), reverse stress testing (starting from a defined failure point), and ad hoc tests requested by management or the supervisor. Include supervisory exercises in which the bank participates.

4. Scenario design

Start from the bank's vulnerabilities, write a narrative that explains how a shock unfolds and spreads, and then derive consistent paths for macro and market variables. Calibrate severity against history (for example the worst observed outcomes in comparable downturns), against supervisory scenarios and against expert views. Document why the scenario is plausible and why it is severe enough. Keep a baseline aligned with the budget so the stress effect can be read as a difference.

4.1 Scenario narratives

4.2 Macro-financial variables

Cumulative property price changes in SCN-002: residential 0.93 x 0.95 x 1.00 = 0.8835, a fall of 11.7%; commercial 0.85 x 0.90 x 0.98 = 0.7497, a fall of 25.0%. The 10-year swap rate in SCN-002 is 150 bp above the baseline at the end of 2027 (4.1% minus 2.6%).

Translate the narrative into variables with a documented chain of reasoning: which variable moves first, how the others respond and why. Use the bank's own macroeconomic model or a recognised external forecast for the baseline, and apply shocks that are consistent with each other (for example, higher inflation with higher rates and weaker real income). Add bank-specific variables that the macro paths do not capture, such as deposit pricing pressure from competitors, the loss of a major client or a rating downgrade, and state how they interact with the macro shock. Keep every variable path in a scenario file with a version number so each model run can be traced to the scenario it used.

4.3 Severity calibration

Example: The scenario committee compared the GDP fall of SCN-002 (a cumulative decline of about 3.5% over two years, since 0.975 x 0.990 = 0.9653) with the deepest two-year contractions in the bank's markets since 2000 and with the most recent EU-wide adverse scenario, and judged it between the two. The commercial property fall of 25% is deliberately more severe than the residential fall because the bank's vulnerability lies there. The scenario is plausible because each element has occurred in the last 25 years, though not all together.

5. Methodology by risk type

For each risk type, describe the transmission from scenario variables to financial outcomes, the models or expert methods used, key parameters and limitations. Make the links between risks explicit, for example how higher rates raise both net interest income and credit losses, or how a rating downgrade affects both funding costs and deposit outflows.

Explain how the ECL under IFRS 9 is projected in stress (see ECL Methodology, finance/ecl-methodology) and how the results are kept consistent with accounting rules. Describe where expert judgement overrides model outputs and who approved it.

6. Data and balance sheet assumptions

Choose and justify the balance sheet assumption. A static balance sheet (maturing exposures replaced like for like) isolates the stress effect and supports comparison across years; a dynamic balance sheet reflects business plans and management reactions but must be credible under the scenario. Many banks run both. Record data sources, reconciliations with accounting and supervisory reports, and data limitations.

Example: Fernhollow Bank runs SCN-002 on a dynamic balance sheet for the ICAAP, with loan growth cut to zero in 2027 and to half the baseline rate in 2028 and 2029, and on a static balance sheet for the comparison with the supervisory exercise. Dividends are set to zero in the adverse scenario because the board's capital policy suspends distributions when the projected CET1 ratio falls below 13.5%.

Reconcile the starting point with the audited financial statements and the supervisory reports at the reference date, and record every manual adjustment in a data log with its approver. Where data are missing (for example historical loss data for a new product), use proxies from comparable portfolios or external data and add a conservative overlay. State the treatment of new business, prepayments, rating migration and collateral values, and keep the same assumptions across the capital and liquidity runs unless there is a documented reason to differ.

7. Results

7.1 Capital results (ST-001, SCN-002, before management actions)

Main drivers of the 3.1 percentage point CET1 depletion to the 2028 trough: impairments of EUR 1,000 million over two years, of which commercial real estate EUR 430 million and SME lending EUR 390 million (the rest mortgages and consumer), valuation losses of EUR 130 million on the liquidity portfolio, operational losses of EUR 40 million, and RWEA inflation of EUR 2,000 million, mainly from rating migration and partly from the reduced loan growth in 2028, partly offset by pre-provision profit of EUR 680 million and tax effects of EUR 110 million. Check: 3,000 minus 1,000 minus 130 minus 40 + 680 + 110 = 2,620. Full details are in the ICAAP Report, section 8.

7.2 Liquidity results (ST-002)

The liquidity results stay within appetite, but two of them fall below early warning levels in the adverse macro scenario. The programme therefore treats the 2028 senior bond maturity as the key liquidity vulnerability and links it to the pre-funding decision recorded in section 11. Read these results together with the capital results: the same scenario that lowers the CET1 ratio also slows deposit growth and raises funding costs, so capital and liquidity actions must be planned together.

7.3 Sensitivity results

Recompute: 300 / 3,400 = 8.8% of Tier 1. A CET1 effect of 1.6 percentage points on TREA of 20,000 is EUR 320 million, which corresponds to an ECL increase of EUR 360 million less a tax effect of EUR 40 million (only part of the provision is tax deductible in the example).

8. Management actions

Define management actions in advance, linked to triggers in the risk appetite framework and recovery plan, and show results before and after them. Each action needs an owner, an estimated effect, a time to execute, preconditions and an assessment of feasibility under the specific scenario. Do not include actions that depend on market conditions the scenario has removed.

With MA-002 and MA-003 the 2028 CET1 ratio becomes 2,680 / 21,700 = 12.35%, above the floor of 12.0%.

9. Reverse stress testing

Define failure points for capital (for example breach of the MDA threshold, breach of TSCR as a proxy for non-viability) and for liquidity (exhaustion of counterbalancing capacity). Work backwards to the loss or outflow that would cause each, then explore several combinations of events that could produce it, including non-macro events such as fraud, cyber attacks, conduct fines and the loss of a key counterparty. Assess plausibility and feed the results into early warning indicators and recovery options.

Recompute: 3,000 minus (9.125% x 22,000 = 2,007.5) = 992.5; 3,000 minus (5.625% x 22,000 = 1,237.5) = 1,762.5. RST-003: 5,400 + 1,500 + 1,800 + 600 + 630 = 9,930 gross outflows; 9,930 minus 1,200 = 8,730, which exceeds the counterbalancing capacity of 8,500. The required outflow is about 1.8 times the 30-day net outflow of the combined scenario SCN-L03 (4,900), which is what makes the reverse scenario more extreme than the regular one.

10. Challenge, validation and audit

Document the challenge at each step: the scenario committee challenges narratives and variable paths; model validation reviews satellite models and overlays under the Model Risk Management Policy (finance/model-risk-management-policy) with findings recorded in Model Validation Reports (ai-data/model-validation-report); the risk committee challenges results and actions; internal audit reviews the whole programme periodically. Benchmark results against supervisory exercises and peer disclosures where available, and explain material differences.

11. Use of results

Record the decisions taken on the basis of stress test results during the year, with dates and owners, so the supervisor and internal audit can see that stress testing is part of management and not a separate exercise.

12. Annual calendar

Plan the cycle backwards from the date the ICAAP and ILAAP must be submitted. The management body should discuss and approve the scenario narratives before any numbers are produced, so that its challenge shapes the exercise rather than reacting to it. Leave time for at least one iteration between the first results and the final run, because the first results almost always reveal an assumption that needs to change. Run a lighter update at mid-year, or sooner if a trigger in the risk appetite framework fires or the economic outlook shifts materially, and record the reason for every off-cycle run in the stress test inventory.

Example: Fernhollow Bank approves scenarios at the end of August, produces first results in mid-November on 2026-09-30 data, challenges them in December, rolls them forward to the actual 2026-12-31 balance sheet in January and February, and signs the ICAAP and ILAAP in March. The roll-forward replaces the starting balance sheet and capital position with the year-end actuals and re-runs the projections; a material change from the first results goes back to the risk committee before sign-off.

13. Regulatory map and jurisdiction notes

14. Glossary and references

References (cited by number and title; obtain the official text):

• EBA, Guidelines on institutions' stress testing, EBA/GL/2018/04.
• EBA, Guidelines (revised) on common procedures and methodologies for SREP and supervisory stress testing, EBA/GL/2026/06, 26 June 2026, applying from 2027-01-01.
• Directive 2013/36/EU (CRD), Articles 73, 86 and 100, as amended. EUR-Lex.
• ECB, Guide to the ICAAP (July 2026) and Guide to the ILAAP (November 2018). ECB Banking Supervision.
• Basel Committee on Banking Supervision, Stress testing principles, October 2018 (name only).

15. Standard and licence note

Based on EBA Guidelines EBA/GL/2018/04 and EBA/GL/2026/06, Directive 2013/36/EU (CRD) Articles 73, 86 and 100, and the ECB Guides to the ICAAP and ILAAP. Structure aligned with the Basel Committee "Stress testing principles" (October 2018), cited by name only. EUR-Lex, ECB and EBA material is reused with acknowledgement; guideline and article numbers are cited and no normative text is reproduced. The structure, guidance, scenarios and the Fernhollow Bank example are original; Fernhollow Bank and every figure are fictional. Check the current official versions before use. This is a document format, not financial, investment, legal, regulatory or accounting advice.

Template licence: MIT.