On July 22, 2019, the European Insurance and Occupational Pensions Authority (EIOPA) published a discussion paper on methodological principles of insurance stress testing (the Discussion Paper)¹, which sets out proposed principles and guidelines for its next EU-wide stress test. Relevant stakeholders are invited to provide feedback by October 18, 2019. The Discussion Paper, which applies to reinsurers and insurers, was published against the backdrop of criticism of EIOPA’s previous techniques and its own admission that its testing methodologies are overly complex and disjointed, and that there is a need for a more consolidated approach. This Client Alert, as part of our Insurance Insight series, looks at this new proposed methodology and how affected parties may want to take preparatory action given recent discussion amongst policymakers and market participants on stress testing as an appropriate supervisory tool.

EIOPA’s current approach

At present, EIOPA uses a bottom-up insurance stress test designed to assess insurers’ individual resilience as well as the systemic risk posed by insurers as a group. It is conducted in collaboration with the European Systemic Risk Board (ESRB) and finds its legal basis in Regulation (EU) No. 1094/2010.² Whilst the stress test’s scope and the risks it identifies are generally held to be pertinent, it does have other “shortcomings”.³

Concerns raised by the European Banking Authority (EBA)

EIOPA is not the only organization to have criticized the current approach to stress testing: other EU institutions such as the EBA have also voiced their disapproval.⁴ The EBA is worried that bottom-up stress tests, especially those linked to supervisory actions and publicly disclosed, frequently equate to beauty contests, where participants are more concerned with outperforming others than with identifying their own weaknesses. The EBA feels that bottom-up stress testing cannot be made compatible with incentives, and it therefore suggests a move towards a top-down approach.⁵

Concerns raised by the European Court of Auditors (ECA)

The ECA’s negative feedback came in the form of a report⁶ primarily covering EIOPA’s supervisory activities between 2015 and 2017, which stated that the two scenarios used in the 2016 stress test presented certain “shortcomings”.⁷ One of these scenarios was the “Low Yield” scenario, which tests a key risk for the insurance sector.⁸ However, the scenario used in the 2016 stress test was insufficiently challenging, given that yields are now even lower than in the stressed scenario.⁹ The other issue identified was the “Double Hit”¹⁰ scenario, which dealt with falling interest rates as well as market shocks. The ECA also found fault with this, stating that the period selected for the financial shock simulator was too short and that there was insufficient stress on the stock of insurance companies outside the EU – a big issue, as dependence between stock markets goes up by 90% during a crisis, according to the ECA.

Concerns raised by stakeholders

EU institutions are not the only ones with concerns. Various stakeholders question whether asset class calibrations under the Solvency II Directive¹¹ (SII) (such as models for interest rates, equity, property, spreads, currencies and market risk concentrations) are “up-to-date” enough to determine capital requirements. The greatest unease relates to interest rate risk: the current methodology underestimates the amount of interest rate risk in low interest rate periods and cannot accommodate negative interest rates – it is simply behind the times. Despite these concerns, the European Commission has decided not to review interest rate risk until it reviews SII in 2020.¹² Until then, stakeholder concerns could only be appeased by EIOPA moving away from stringent SII adherence under the insurance stress test.

So what’s in the Discussion Paper?

In light of all this criticism, EIOPA’s recent Discussion Paper may be of great interest to various stakeholders, not just for the principles it outlines, but also for the wind of change it symbolizes. This is the case even if the Discussion Paper hints at EIOPA considering to slow down the frequency that it carries out stress testing and use the period in-between to perform better sensitivity analysis and follow-up on the recommendations issued to relevant institutions. 

The Discussion Paper sets out methodological principles necessary for the conduct and assessment of an EU-wide insurance stress test exercise, which will form the “toolbox to facilitate both the design and execution phase of future EIOPA stress test exercises”.¹³ It states that these future stress test exercises can be used by a variety of stakeholders such as supervisors, insurers, academics and rating agencies.¹⁴ The Discussion Paper specifically focuses on the bottom-up supervisory stress test (the method used by EIOPA so far), which it wants to improve and deepen in order to enhance the overall stress testing methodology.¹⁵

The Discussion Paper covers the following topics:

Stress test process and objectives¹⁶

The Discussion Paper states that the stress tests’ objectives must be articulated at the outset as this will influence the design, modelling and process for each stress test in addition to determining the type of stress test (e.g. top-down vs. bottom-up).¹⁷ The Discussion Paper further recommends that the micro-prudential approach should now include macro-prudential elements (such as a quantitative assessment of post-stress reactions by insurers) in order to address systemic risk, in line with achieving stability in European markets more generally.¹⁸
The Discussion Paper outlines three different approaches (recalculation of the baseline, time horizon and management actions) and their various uses in light of certain stress test objectives, such as assessing insurers’ sensitivity to specific shocks, assessing industry vulnerability and assessing the spill-over effects of a specific approach. It includes a table linking these approaches to the specific stress test objectives and requests stakeholder feedback on these approaches.¹⁹

Scope²⁰

The scope of a stress test can include the targeting of either solo, group or synthetic group undertakings, with the latter only including specific types of solos (e.g. based on geographical location). The Discussion Paper states that whilst the scope of a stress test generally depends on its objectives, the target (in the case of micro-oriented stress tests) should always be solo undertakings, as stress tests at group level (either real or synthetic) are harder to validate and less useful overall.²¹

Scenario design²²

Scenarios are severe yet plausible hypothetical situations that may “adversely affect the balance sheet and solvency position of insurance undertakings” and which can be made up of a single shock or a variety of market, financial and insurance specific shocks that impact the insurance sector on both a micro and macro level.²³

There are several factors to take into account regarding scenario design, one of which is data selection. The Discussion Paper recommends a hybrid approach to data selection over a purely historical or forward-looking one, in line with the scenarios used by EIOPA so far.²⁴ With regard to the other factors that shape the stress scenario, it recommends the following:

• Whether a single-shock, single scenario or combined scenario is most appropriate shall be determined by the objective of the exercise.²⁵
• Whereas consistency with the SII framework, a common ground for assessing the insurance industry’s resilience, is generally desirable, it is more important to design market-compatible scenarios. The paper therefore suggests departing from SII in some situations, most notably regarding the manner in which the risk-free rate curve is derived.²⁶
• The bucketing²⁷ (i.e. general) approach, as opposed to the granular approach used in previous stress test scenarios, is preferred for EU-wide assessments. As opposed to the granular approach, which allows for country-based analysis, the bucketing approach is more efficient. However, the bucketing approach requires homogeneity criteria to be established in order to avoid unfair or unreasonable results. This can be achieved by using objective criteria, such as ratings or volatilities.²⁸

Shocks and their application²⁹

The Discussion Paper distinguishes between market shocks, insurance-specific shocks and other impacts on the balance sheet.

Market shocks are calibrated in cooperation with the ESRB and are based on the financial shock simulator developed by the European Central Bank for the design of various stress tests as well as policy analyses. A market shock scenario focuses on one or more trigger events and the joint distribution of the resulting event with other financial variables, whose reaction is captured via their conditional distributions (i.e. the probability distribution for those financial variables).³⁰ Market shocks include:

• Shocks to bonds
• Shocks to equity
• Shocks to SWAP rates
• Shocks to real estate
• Shocks to loans and mortgages
• Shocks to collective investment undertakings
• Shocks to type 1 exposures (reinsurance recoverables, insurance intermediate receivables and reinsurance receivables)

Insurance-specific shocks are mainly concerned with the risk exposure of the European insurance industry. These shocks generally cover underwriting risks, catastrophe risks or the risk of deterioration of technical provisions, and are generally short-term or medium-term shocks.³¹ The Discussion Paper recommends that one or more of these shocks should be included in any stress test as long as they fit into the narrative, and any overlap with market shocks or other insurance-specific shocks are kept in mind (though overlap is not necessarily an issue). Insurance-specific shocks will generally be applied to the entire business, save for where these shocks only affect a specific line of business.³² Insurance-specific shocks include: 

• Life insurance shocks
• Non-life insurance shocks

Lastly, the Discussion Paper covers other impacts on the balance sheet such as:

• Deferred tax assets or liabilities
• Derivatives 

In line with EIOPA’s view that the Discussion Paper is overly complex, the section on shocks concludes by proposing several simplifications:

• Stress test participants may exclude part of a business from a scenario where it is insensitive to shocks due to its nature or geographical location.³³
• The loss-absorbing capacity of deferred taxes (LACDT) should only need to be substantiated by the deferred tax liability on the balance sheet, as future profitability as substantiation would require deeper analysis by national competent authorities. Further reducing the subjectivity involved in calculating the LACDT could be achieved by capping the LACDT to the amount of net DTL on the base case balance sheet.³⁴
• Several approaches (Curve-Fitting, Replicating Portfolios and Least Square Monte Carlo) have been developed by the insurance industry in response to issues experienced by traditional life and health businesses due to Article 122 of SII. However, the implementation of these approaches are difficult, as they require expert judgment and validation. In order to continue to use these techniques, there would need to be a full recalibration post stress test, which is operationally complex and lacks a technical framework. Therefore, the Discussion Paper suggests that companies apply approximations or simplifications rather than a full recalibration. The assessment of these approximations should be a central point of the validation process (see below), with companies providing quantitative and qualitative evidence that the approximations are appropriate. A dialogue between supervisors and companies concerning this topic is recommended.³⁵
• The calibration of long-term guarantees (LTG) shall remain unchanged, unless the shocks prescribed trigger a material change in the LTG measure, which would lead to the values being calibrated in line with EIOPA methodology. Further details are provided.³⁶
• To calculate the post-stress risk margin, companies may use a simpler calculation method than the one used for the calculation of end of year balance sheets, provided this does not negatively affect the assessment of technical provisions.³⁷
• Potential simplifications may be applied to the calculation of the post-stress position of solos according to the principle of materiality. The 2018 stress test experimented with a group-consolidated approach but the Discussion Paper holds that this is not a suitable way forward, unless more homogenous definitions of certain model points are reached.³⁸
• All simplifications and approximations used for post-stress results calculation should be discussed with the supervisor prior to implementation.³⁹

Data collection and validation⁴⁰

The Discussion Paper states that templates used to convey results (in the form of identifying vulnerabilities and risks) under baseline and stress scenarios should be as similar to SII QRT reporting templates as possible. The templates should also match the specifications, approaches and scope of the stress test. Where an ad hoc template or new data points are required, the Discussion Paper states that the company must justify this and that it may be subject to discussion. 

Additional forms required for validation purposes shall supplement these templates, as any reported data must be validated in order to ensure the consistent application of the stress test among participants, as well as the comparability of results. There are four levels of validation: level 0 (consistency and completeness check), level 1 (consistent application of shocks (closed-form formulas validation)), level 2 (benchmark analysis against peer-levels) and level 3 (proprietary in-house model of analysis). They are used in different scenarios and are of different complexities, with 0 being the simplest and 3 being the most complex.

Request for feedback

The Discussion Paper asks stakeholders to submit feedback, specifically with regards to the (69) questions asked throughout the Discussion Paper, for which the comment due date is October 18, 2019.

The bigger picture

The Discussion Paper is part of a broader movement to upgrade the stress-testing framework. EIOPA is increasingly prioritizing its focus on broader but connected issues such as the assessment of liquidity positions under detrimental scenarios, assessment of climate-related risks, and the matter of multi-period stress tests. Additionally, as mooted, a reduction in the frequency of the EU-wide stress test exercises by EIOPA, in order to allow for more in-depth analysis of stress-test results and to implement issued recommendations more effectively might be one of the easiest changes to implement to enhance the effectiveness of its stress testing exercises. 

A reduction in frequency would shadow the EBA’s approach to comprehensive bank stress tests. It will be interesting to see how the overall development of bank stress tests will influence insurance sector stress tests, and whether changes made in one sector will seem appropriate to the other. Currently, however, the approaches seem to be diverging: regulatory changes are leading to bank stress tests’ calculation methods being adapted and new components being added, all of which is increasing their complexity. This is something the insurance stress test is trying to move away from.

Outlook

At first glance, the Discussion Paper does little to alleviate the EBA’s concerns. Based on the paper, it appears unlikely that EIOPA will introduce a top-down stress test anytime soon. On the other hand, the ECA might be slightly more pleased: while its “Low Yield” concerns were left largely unaddressed, the Discussion Paper recommends an increased focus on macro-prudential elements, i.e. “large internationally active groups”⁴¹, as well as asking for feedback on appropriate time horizons.⁴² Other stakeholders may be feeling ambivalent: on the one hand, they likely support EIOPA’s proposal not to adhere to SII where this is not in line with market compatible scenarios (specifically against the backdrop of the low interest rate environment). On the other hand, several further concerns may have been left unaddressed. Despite that, this may lead to dialogue among market participants and constructive feedback for EIOPA. Another area that however remains unaddressed is how EIOPA’s deliberations might be viewed in the UK following Brexit and what the UK regulators may consider with their own stress testing supervisory agenda.

Overall, it remains to be seen whether EIOPA’s paper has done enough to address concerns and whether the insurance stress test will ultimately be reformed in a suitable manner, i.e. be reduced in its complexity whilst still allowing for effective supervision of insurance companies. For now, we await EIOPA’s feedback publication, keeping an eye open for stakeholder and market body commentary published in the meantime that may determine whether EIOPA’s starting of this important discussion is as much of a game changer as EIOPA hopes it might be. 

If you would like to discuss strategic options or any of the items mentioned above, in particular how to plan ahead for any operational impact from meeting compliance requirements and/or documentation or how these priorities may affect your business or your clients more generally, please contact our Eurozone Hub key contacts.

1. EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
2. Specifically, clause 42 of the preamble of Regulation (EU) No 1094/2010 of the European Parliament and of the Council of November 24, 2010 establishing a European Supervisory Authority (European Insurance and Occupational Pensions Authority), amending Decision No 716/2009/EC and repealing Commission Decision 2009/79/EC.↩
3. Page 8 of the ECA’s “Special report No 29/2018: EIOPA made an important contribution to supervision and stability in the insurance sector, but significant challenges remain” dated November 2018, available here.↩
4. The EBA’s paper titled “Are Stress Tests Beauty Contests? (And What We Can Do About It)” by Mario Quagliariello, dated July 2019, available here.↩
5. Page 13 of the EBA’s paper titled “Are Stress Tests Beauty Contests? (And What We Can Do About It)” by Mario Quagliariello, dated July 2019, available here.↩
6. The ECA’s “Special report No 29/2018: EIOPA made an important contribution to supervision and stability in the insurance sector, but significant challenges remain” dated November 2018, available here.↩
7. Page 32 of the ECA’s “Special report No 29/2018: EIOPA made an important contribution to supervision and stability in the insurance sector, but significant challenges remain” dated November 2018, available here.↩
8. Page 34 of the ECA’s “Special report No 29/2018: EIOPA made an important contribution to supervision and stability in the insurance sector, but significant challenges remain” dated November 2018, available here.↩
9. Page 34 of the ECA’s “Special report No 29/2018: EIOPA made an important contribution to supervision and stability in the insurance sector, but significant challenges remain” dated November 2018, available here.↩
10. [1] Page 35 of the ECA’s “Special report No 29/2018: EIOPA made an important contribution to supervision and stability in the insurance sector, but significant challenges remain” dated November 2018, available here.↩
11. Directive 2009/138/EC of the European Parliament and of the Council of 25 November 2009 on the taking-up and pursuit of the business of Insurance and Reinsurance.↩
12. BaFin’s expert article on “Solvency II: Reviewing the standard formula” dated July 31, 2019, available here.↩
13. EIOPA’s news update titled “EIOPA publishes Discussion Paper on Methodological Principles of Insurance Stress Testing” dated July 22, 2019, available here.↩
14. Page 8 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
15. Page 8 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
16. Page 10 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
17. Page 13 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
18. Page 14 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
19. Page 24 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
20. Page 25 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
21. Page 29 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
22. Page 30 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
23. Page 30 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
24. Page 31-21 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
25. Page 40 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
26. Page 32-33 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
27. Page 37 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
28. Page 36-37 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
29. Page 41 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
30. Page 42 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
31. Page 49 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
32. Page 48-50 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
33. Page 42 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
34. Page 73-74 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
35. Page 74-76 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
36. Page 76 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
37. Page 77 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
38. Page 77-78 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
39. Page 72 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
40. Page 79 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
41. Page 14 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩
42. Page 19 of EIOPA’s “Discussion Paper on Methodological principles of insurance stress testing” dated July 22, 2019, available here.↩