Changes to the LSI Stress Test 2026

Significance of the 2026 LSI Stress Test and its relevance for institutions

In line with the regular two-year cycle, the 2026 LSI Stress Test is being conducted again this year for German credit institutions. The aim of the survey is to analyse the earnings situation and capital development of Less Significant Institutions (LSIs) under baseline and stress conditions over a multi-year horizon.

The 2026 LSI stress test is divided into two closely interlinked components: a survey section based on the institution’s own planning, and a standardised stress test section to simulate adverse developments. The 2025 annual financial statements serve as the data basis.

The 2026 LSI stress test is a central component of the supervisory review of LSIs and serves to assess their sustainable profitability and resilience to adverse macroeconomic and financial conditions. As a key instrument in the Supervisory Review and Evaluation Process (SREP), the results of the stress test have a direct impact on institutions’ capital requirements, as they play a decisive role in the derivation of the Pillar 2 Guidance (P2G) by the Federal Financial Supervisory Authority (BaFin).

Against a backdrop of persistently high macroeconomic uncertainty, structural changes in the interest rate environment and increasing volatility in the capital markets, the LSI stress test is gaining further significance this year.

Whilst previous surveys also relied on qualitative assessments and long-term evaluations, the 2026 LSI stress test shows a clear focus on quantitatively robust, consistent and cross-institutional comparable data.

At the same time, the trend in regulatory law towards greater consideration of the principle of proportionality is becoming apparent, particularly with regard to small and non-complex institutions (SNCI), for which targeted simplifications are envisaged. By pre-populating supervisory data from the regulatory reporting system, the operational burden of data entry is reduced for all institutions; at the same time, however, expectations regarding the careful validation of these values are rising.

General approach: Survey and stress test

The survey section focuses on the dynamic, institution-specific scenario derived from capital planning. This serves as a basis for comparison for the institutions’ subsequent scenario analyses. For the first time, the analysis covers a three-year period from 2026 to 2028, which promises to further ease the operational burden on the institutions.

In addition, three standardised interest rate scenarios are applied to a static balance sheet:

Particular emphasis is placed on net interest income as the key driver of earnings for most LSIs. Comparing dynamic planning with the results of static interest rate shocks provides important insights into the stability of the interest margin, maturity transformation and the institutions’ ability to absorb interest rate risks even during periods of stress.

In parallel, the stress test component simulates a baseline and a stress scenario over the same three-year horizon, assuming a static balance sheet. It covers the key risk categories of net interest income, counterparty risk and market risk, and is supplemented for the first time in the 2026 LSI stress test by the loss-free valuation of the banking book in accordance with BFA 3. The key control variable here is the development of the Common Equity Tier 1 (CET1) ratio, the strain on which in a stress scenario is used as a key factor in deriving the P2G within the framework of the capital adequacy recommendation.

The methodology is largely standardised and follows, in its basic logic, the approach of previous LSI stress tests. The stress results are determined on the basis of regulatory, uniformly specified parameters. Items not relevant to the stress test – such as administrative and operating expenses – are initially carried forward in the baseline scenario using historical values; in the stress scenario, the stress impact is reflected through predefined haircuts. The use of uniform scenarios enhances comparability between institutions and ensures a consistent, supervisory-coordinated assessment framework.

Changes compared to the 2024 LSI Stress Test

Compared to the 2024 survey, the 2026 LSI Stress Test introduces a number of operational changes aimed at simplifying processes, improving data quality and enhancing proportionality.

In the survey section, the number of interest rate scenarios has been reduced from five to three. The ‘inversion of the yield curve’ and ‘gradual rise in interest rates’ scenarios have been omitted, significantly reducing the calculation workload.

At the same time, the planning horizon has been shortened from five to three years. This adjustment takes into account the fact that long-term forecasts are inherently subject to greater uncertainty and improves the informative value of the reported data. The intended removal of qualitative questions underscores the clear focus on a data-driven stress test. Additional detailed information, such as the breakdown of the structural contribution, has been removed.

In the stress test section, the basic methodology remains unchanged. However, it has been clarified in certain respects.

• In the P&L capital, the calculation of market value reserves has been corrected and supplemented with other, unallocated items, whilst individual items relating to market risk have been omitted to avoid inconsistencies. Consistency between survey and stress test data continues to be ensured through data quality checks.
• The net interest income remains largely unchanged. Only minor additions have been made, such as the option to choose between direct and indirect minimum reserve holdings, as well as the option to adjust the ‘Other net interest income’ item (yes/no).
• In counterparty risk, the number of data points to be collected has been reduced; this also includes the removal of individual items that previously served primarily for quality assurance purposes. Off-balance-sheet volumes will in future only need to be reported on an optional basis. Furthermore, methodological ambiguities have been resolved, particularly for institutions whose counterparty default risk RWA¹ (GPA-RWA) is below 1% of total risk-weighted assets (TREA²).
• Clarifications have also been made regarding market risk, for example concerning the proportional look-through of funds and the allocation of price reserves. The statistical method previously used has been replaced by a calibration based on the weighted average rating of the portfolio and the proportion of government bonds. Additionally, institutions may optionally report more detailed information on Portfolio A.
• A significant innovation is the standalone BFA-3 form. The impact on the profit and loss account is deliberately estimated conservatively here, by taking into account only negative effects arising from new or increased provisions for expected losses. Positive present value effects can only compensate for write-downs to a limited extent. The use of previously reported IRRBB data increases feasibility, whilst fallback solutions are provided for data gaps.

In addition, new colour coding has been introduced in the form to make it easier to navigate when completing it: cells highlighted in orange indicate values pre-filled by the supervisory authority, whilst pink fields are intended for voluntary disclosures.

A significant simplification of the 2026 LSI stress test is the introduction of a proportionality solution for SNCI institutions in accordance with the CRR. As a result, around 500 institutions may be exempted from having to complete the risk sheets in full. The assessment of eligibility criteria – in particular SNCI status and total assets – is carried out automatically in the ‘A2 ST-Exemption RK’ spreadsheet. In future, all institutions will continue to have the option of being partially exempted from selected risk assessments, thereby enabling the data collection burden to be reduced in a targeted manner.

Risk calibration for the proportionality solution and the exemption solution should be carried out as uniformly as possible and is generally more conservative than the complete completion of the survey forms.

Regulatory timetable and phases of the 2026 LSI stress test

The 2026 LSI stress test follows a tightly scheduled and binding timetable. The operational completion of the survey forms will begin in April 2026, whilst initial test runs and systemic checks may take place as early as the first quarter. The final documents must be submitted to the supervisory authority by 29 May 2026 at the latest. This means that the processing phase from April to May coincides with a period in which quarterly risk reports and other quarterly and annual activities are also due, placing particular demands on the institutions’ operational planning and resource allocation.

Between early June and late July 2026, the supervisory authority will carry out comprehensive data quality checks, during which validation errors must be corrected and warnings commented on. Building on this, the benchmarking phase will follow from late July to early September 2026, during which peer group comparisons will be made and notable deviations analysed.

Conclusion

The greater emphasis on quantitative data and the use of pre-filled fields mean that the quality and consistency of the reported figures are more critical than ever. Institutions are required to closely integrate their planning, risk management and reporting processes.

With the introduction of the proportionality solution, pre-population and clarifications in individual risk sheets, the supervisory authority aims to significantly reduce the operational implementation burden for institutions. The challenges therefore lie more in the technical modelling of the individual risk modules.

Given the importance of the stress test for deriving the P2G, early, structured and interdisciplinary preparation is therefore recommended. The LSI Stress Test 2026 is thus far more than a regulatory formality. It represents a key management tool that makes risks transparent and highlights areas requiring action.