Limitation

Maintaining the confidentiality of individual firm responses was prioritised throughout the data gathering and analysis period. The bulk annuity market consists of a small number of insurers. This limited the working party’s ability to discuss individual responses with firms directly. To mitigate this, the working party took particular care to ensure that the questions were drafted as clearly as possible and performed a high-level sense check of the post-submission responses.

5.1. Board & executive committee

An insurer’s Board has overall responsibility for the management and oversight of the insurer and its activities. Boards may establish committees to assist in fulfilling their oversight responsibilities.

Figure 3 shows that all respondents Boards delegate responsibility for reinsurance counterparty exposures to one or more committees. Delegating technical topics leaves more time for the Board to focus on, and provide leadership over, the insurer’s strategic direction, culture and setting the general tone from the top.

Figure 3. Key committees responsible for Reinsurance counterparty exposure.

The Board Risk Committee was the most frequently cited oversight forum (4 of the 6 respondents) followed by a Credit Risk Committee (3 of the 6 respondents). The Other category included an Insurance Risk Committee and Capital Management Committee.

What information is communicated to Board and Executive Committees?

Respondents were asked to indicate the information communicated to their Board and Executive Committees.

Figure 4 shows that 4 of the 6 respondents provide the current level of exposure relative to a Red, Amber and Green (“RAG”) status to their Board and Executive Committees. This metric is likely to be chosen for its visual nature, simplicity, ability to be used as a signal to focus management’s attention or as an indicator to take actions.

Figure 4. Information provided to Board & Executive Committees.

The “remaining headroom” and “impact on solvency ratio” metrics are quantitative in nature and require a deeper understanding of the technical detail including assumptions and management actions that sit behind the metrics. This is more challenging for individuals one or more steps removed from the process to acquire; and could explain why the majority of respondents do not provide this information to their Board (5 of the 6 respondents) or Executive (4 of the 6 respondents) Committees. Twice as many respondents provide information to Executive Committees relative to Board Committees. The Executive is closer to their firms’ reinsurance counterparty exposure and is likely able to more meaningfully contribute and engage on the topic than their Board Committee counterparts.

How frequently is information communicated to Board and Executive committees?

Respondents were asked how frequently information is communicated to their Board and Executive Committees.

Figure 5 shows that reinsurance counterparty risk exposures were most frequently communicated to Board Committees and Executive Committees on a quarterly basis (3 of the 6; and 4 of the 6 respondents respectively). This frequency aligns with the typical number of Board Committee meetings held annually. A reinsurance counterparty risk limit breach is one reason cited that could trigger 3 of the 6 respondents reporting to a Board Committee more frequently than once a quarter on an exceptions basis.

Figure 5. Frequency of reporting to Board & Executive Committees.

5.2. Capital modelling – methodology & approach

The SII standard formula includes neither an explicit reinsurance recapture risk (sub) module nor parameters within one or more risk (sub) modules to fully capture the balance sheet effects, such as the increase in RM that the ceding entity would need to recognise were previously reinsured risk returned to their regulatory balance sheet in practice. The discussion below focusses on the methodology and approach adopted by internal model firms.

How do insurers capture the balance sheet effects following a reinsurance counterparty default within their SCR calculation?

All respondents consider reinsurance recapture scenarios and associated financial impacts within the reinsurance counterparty risk model.Footnote ⁶ The reinsurance counterparty risk SCR will be sensitive to the relationship between the reinsurance counterparty’s financial soundness, e.g. expressed as a solvency ratio or credit rating; and the resulting PD. The effectiveness of capital as a mitigation tool is low when the PD is small relative to the number of exposures, which would be the case if a ceding entity used a small number of reinsurance counterparties. Actual reinsurer defaults are binary events.

Do insurers assume management actions following reinsurance recapture within their SCR calculation?

Figure 6 summarises respondents assumed management actions taken in recapture scenarios. Management actions reduce the extent of the loss to own funds that would otherwise have occurred.

Figure 6. Assumed management actions.

Replacement reinsurance was the most frequently assumed management action for 4 of the 6 respondents. This assumption is likely to materially simplify the capital modelling for MA portfolios and its interaction with the MA eligibility criteria and PRA matching tests in stress at the cost of potentially being MA dilutive (PRA supervisory statement 7/18: Solvency II Matching Adjustment confirmed that the reinsurance recoverable should not be mapped to a fundamental spread, i.e. and earn the risk free rate). Insurers will need to assume they have obtained replacement reinsurance cover within the one-year stress period in order to be able to claim credit for it within the SCR. An instantaneous reinsurance counterparty default (or other recapture event) will have the biggest financial impact but gives insurers the longest time to obtain replacement reinsurance cover. A reinsurance counterparty default (or other recapture event) assumed to occur later in the year would leave less time for replacement reinsurance cover to be found in order to qualify as risk mitigation within the SCR.

2 of the 6 respondents assume regulatory approval to recalculate transitional measures on technical provisions (TMTP). This would alleviate some of the impact from the recaptured RM for business written before 1 January 2016. The PRA’s supervisory statement 6/16: Maintenance of the “transitional measure on technical provisions” under Solvency II Footnote ⁷ sets out, amongst other things, the circumstances that might support insurers’ requests to recalculate their TMTP for regulatory reporting purposes. The decision to approve a TMTP recalculation more frequently than 24 months is at the discretion of the PRA. However, we think that the materiality of the loss to respondents’ solvency coverage following a reinsurance counterparty default (or other recapture event) is likely to meet the “material and sustained change in risk profile” test to support a TMTP recalculation application.

2 of the 6 respondents indicated that they do not assume any management actions in scenarios that result in the recapture of previously reinsured risk. This approach would be conservative if respondents had identifiable actions that they could implement within the SCR’s one-year time horizon but chose not to capitalise it upfront.

One respondent assumes regulatory approval to apply a MA to eligible liabilities. This suggests that the collateral received or assets injected into the MA portfolio from elsewhere in the business following recapture would contain new features outside of their existing MA approval. The PRA has up to six months to approve or reject a MA application from the date that the complete application is received. The time taken by the PRA to complete its review is outside of insurers’ control. Engaging the PRA early; being clear on the timelines and the activities on the critical path; and being responsive to any follow-up questions would help to minimise the time taken by the PRA to reach a decision on the outcome of the MA application.

Who is responsible for ensuring that the reinsurance counterparty risk SCR remains appropriate for regulatory reporting purposes?

The questions in this area were divided into three parts.

• The forum that provides oversight: 3 of the 6 respondents named the Board Risk Committee as the forum that makes the final decision as to whether the capital model continues to remain appropriate for calculating the reinsurance counterparty risk SCR. A Model Governance Committee (2 of the 6 respondents) and an Internal Model Change Board (one respondent) completed the list. The choice of forum is likely to be impacted by insurers’ governance processes and the expertise demonstrated by specific decision-making forums.

• The function responsible for reporting to the decision-making forum: 4 of the 6 respondents named the actuarial function as being responsible for reporting on whether the SCR continues to adequately capture reinsurance counterparty risks for regulatory reporting purposes. The SII Directive 2009/138/EC (Article 48: Actuarial function) sets out a formal role for the actuarial function to express an opinion on insurers’ reinsurance arrangements and to contribute with respect to the risk modelling underlying the SCR. Giving the actuarial function a formal role for reporting whether or not the SCR continues to appropriately capture reinsurance counterparty risks would appear consistent with its remit. The first line (one respondent) and risk function (one respondent) completed the list.

• The frequency of model signoff: 5 of the 6 respondents stated that the capital model used to calculate the internal model SCR was signed off annually. One respondent stated that its model was signed off less frequently than annually. This is likely justified on proportionality grounds.

How do insurers internally modelled reinsurance counterparty risk SCRs compare with their ORSA capital?

Table 2 summarises the similarities and differences between the internally modelled SCR and Own Risk and Solvency Assessment (ORSA) capital across 7 dimensions.

Table 2. Internal Model SCR versus ORSA – Is There a Difference?

The internal model is designed to reflect insurers’ own risk profiles and so we would expect the methodology and approach to calculating the SCR and ORSA capital to be broadly the same. This is confirmed by the “no” responses in Table 2. Differences in the methodology and approach between the internally modelled SCR and ORSA capital likely indicates the areas that would have been most debated between an insurer and the PRA at the point of internal model approval or when subsequent changes were made to it.

5.3. ORSA (including stress & scenario testing)

The ORSA is a key risk management tool for all insurers. The Stress and Scenario Testing (SST) carried out as part of or alongside the ORSA process and the wider risk monitoring within an insurer provides a useful tool in analysing “what if” reinsurance counterparty risk stress or scenario impacts. Insurers have limited real world experience of a reinsurance counterparty default (or other recapture event) in a stressed environment and so rely on expert judgement when defining the SST process.

What approach do insurers take to model reinsurance counterparty default (or other recapture event) within the ORSA (including SST)?

Figure 7 shows that respondents take a wide variety of approaches to model the impact of reinsurance counterparty default (or other recapture event) on their business. This is likely driven by the reinsurance exposures and reinsurance strategies adopted by the different insurers.

Figure 7. Approach to modelling reinsurance counterparty risk (or other recapture event) within the ORSA (including SST).

There are several possible explanations for why the relatively simple single factor stress approach is the most common response (3 of the 6 respondents).

• It can be hard to model how a reinsurance counterparty default (or other recapture event) would interact with other stresses. There could also be the assumption that a reinsurance counterparty default (or other recapture event) alongside other events, e.g. a longevity stress, is too remote a possibility to include it as a severe but plausible scenario in the ORSA.

• Insurers may make the assumption that the most likely driver of reinsurance counterparty default (or other recapture event) would be idiosyncratic in nature, e.g. a major operational risk event or fraud that only impacts a single reinsurer, and not the result of an industry wide or more systemic stress.

• It will be easier to understand the severity of the stress by focussing only on a reinsurance counterparty default (or other recapture event). Incorporating other stress factors may make the story less clear and divert attention away from the impacts of a reinsurance counterparty default (or other recapture event).

On the other hand, a single factor stress approach could oversimplify the real world dynamics and dependencies with other risk factors, which could lead to the modelled financial impacts understating the true risk exposure. This could frustrate the attempts of decision makers to ask the right questions and leave the insurer less well positioned to deal with a reinsurance counterparty default (or other recapture event) were it to occur.

How many stresses and scenarios related to reinsurance counterparty default (or other recapture event) are considered in the ORSA (including SST)?

Figure 8 shows that 2 of the 5 respondents may be very concerned about a reinsurance counterparty default (or other recapture event) and run a large number of reinsurance counterparty related stresses and scenarios to gain insight into the potential solvency impact. Interestingly, the level of concern is not correlated with the size of the loss to solvency coverage that would be incurred if respondents’ most material reinsurance counterparty was recaptured before management actions. Respondents with the lowest losses to solvency coverage carried out the largest number of stresses and vice versa. This could indicate the reliance being placed on management actions to protect solvency.

Figure 8. Number of stresses and scenarios considered within the ORSA (including SST).

What does an insurer’s approach to reinsurance counterparty default (or other recapture event) say about the strength of the ORSA (including SST) scenarios?

3 of the 6 respondents contemplate the recapture of multiple treaties relating to two or more reinsurance counterparties in the ORSA (including SST). This scenario suggests that respondents had a reinsurance counterparty default or other “common to all” recapture provision(s) in mind when designing the strength of this scenario. 2 of the 6 respondents considered the recapture of all treaties relating to a single reinsurance counterparty. One respondent indicated that they consider the recapture of all treaties relating to a single reinsurance counterparty in the ORSA (including SST) but had done so only once in the last five years. An event leading to more than one reinsurance counterparty default (or other recapture event) is likely to represent a stronger scenario than an idiosyncratic stress that led to a single reinsurance counterparty default (or other recapture event).

What terms and timing do insurers assume for replacement reinsurance within the ORSA (including SST) projections?

Respondents were asked to indicate the terms they assumed for replacement reinsurance within their ORSA (including SST).

Figure 9 shows that 2 of the 4 respondents assumed that the terms agreed to replace reinsurance following the recapture of previously reinsured risks would be on less favourable terms. This suggests that respondents are assuming that the wider reinsurance market remains in stress from the point of recapture until replacement reinsurance is obtained. Whilst a worsening of the cost of replacement reinsurance is consistent with this backdrop, a more conservative approach (adopted by one respondent) would be to assume that the recaptured business cannot be reinsured. The Other category include assumptions that the affected business could not be reinsured, could be reinsured on the same terms, or could be reinsured on less favourable terms depending on the scenario.

Figure 9. Assumed terms for replacement reinsurance within the ORSA (including SST).

Respondents were also asked when they assumed replacement reinsurance could be obtained.

Figure 10 shows that 3 of the 6 respondents assumed that it would take longer than six months to obtain replacement reinsurance; 2 of the 6 respondents did not assume any replacement reinsurance at all within the ORSA; and one respondent assumed they could obtain replacement reinsurance within three months.

Figure 10. When replacement reinsurance is assumed to be obtained within the ORSA (including SST).

What are insurers’ views on the likelihood of a reinsurance counterparty default (or other recapture event) and the financial impact were it to occur?

5 of the 6 respondents consider the likelihood of a potential reinsurance counterparty default (or other recapture event) as “not very likely.” Views amongst respondents on the materiality of the financial impact of a potential reinsurance counterparty default (or other recapture event) were more mixed with 3 of the 5 respondents considering the impact as “very material” and 2 of the 5 respondents considering the impact as “material.” Table 3 compares respondents’ self-assessments of the materiality of the financial impact following recapture and the consequent loss of solvency coverage before management actions.

Table 3. Relationship between respondents’ loss of solvency coverage (before management actions) and their self-assessed view of the financial impact following recapture

The lack of a clear correlation between the financial impact following recapture before management actions and respondents self-assessed view of the financial impact of a recapture could reflect insurers:

• Confidence in the management actions available to mitigate the financial effects of a recapture within “in appetite” timescales

• View that the risk is contingent in nature and not very likely to occur

• Different interpretation of the meaning of “material” and “very material.”

5.4. Reverse stress & scenario testing

Reverse stress and scenario testing is a tool to help insurers identify one or more events that could cause the entity to fail or breach key risk indicators. However the stresses, scenarios and actions to mitigate risks to an insurer’s ongoing viability as a going concern is not formally defined. Instead, it is left to individual insurers to define; considering their business model, risk profile and attitude to risk.

Do insurers consider reinsurance counterparty default (or other recapture event) within recovery plans?

All 6 respondents have explicitly considered reinsurance counterparty default (or other recapture event) as part of a reverse stress or scenario test for recovery planning purposes since the introduction of SII.

The large loss of solvency coverage respondents reported following the reinsurance counterparty default (or other recapture event) of their most material reinsurance counterparties would appear to make this risk event a credible candidate to consider within a reverse stress and scenario testing setting.

Have insurers identified management actions to rectify SCR breaches within recovery plans?

5 of the 6 respondents have identified management actions to rectify a SCR breach within their recovery planning. The Undertakings in Difficulty Part of the PRA Rulebook transposes the SII Directive 2009/138/EC text that relates to the timescales that an insurer has to rectify a breach of its SCR. The timescales are prescribed in the Rulebook and an insurer in breach of the SCR must restore compliance within 6 months from the date that the non-compliance is observed. This could be extended, at the discretion of the PRA, by an extra 3 months. There is no supervisory discretion to extend beyond this period.

• 4 of the 5 respondents estimated that their longest identified management action would take more than 9 months to implement following an SCR breach. Insurers would need to rely on their quicker to implement management actions in order to continue as a going concern. The longest to execute management actions could provide additional capital buffers once the SCR breach had been rectified. This may be an area to (re)visit to ensure that all identified management actions are credible and robust.

• One respondent estimated that their longest identified management action to rectify a SCR breach would take between 6 – 9 months to execute. However, the extension from 6 to 9 months is not automatic and is likely to be influenced by a range of factors: e.g. an insurer’s progress against its recovery planFootnote ⁸ and the likelihood that SCR coverage could be restored within the extended time period.

Being clear on the actions and milestones needed to increase the PRA’s appetite to provide firms with limited extra time will be important.

6.1. Risk appetite and risk limit setting

Each insurer has its own set of key performance indicators that are used by management to assess the financial performance of the business. These heavily influence the metrics used by insurers to measure and calibrate their reinsurance counterparty risk appetite and limits.

What financial metrics are important to insurers when setting their reinsurance counterparty risk appetite and risk limits?

Figure 11 summarises the metrics used by respondents to calibrate their reinsurance recapture risk limits. Table 4 shows that 5 of the 6 respondents calibrate their risk limits with respect to a change in SCR or loss of surplus metric. The remaining respondent uses other SII metrics, e.g. based on a 1-in-X year longevity stress or proportion of SII Best Estimate Liability (BEL). Only one respondent considers the loss of earnings when calibrating their risk limit.

Figure 11. Metrics used for risk limit calibration.

Table 4. Figure 11 Breakdown by Respondent

How do insurers reflect their preferences between reinsurance counterparties when setting risk appetite and risk limits?

When assessing how much risk to take on, insurers not only need to worry about the quantum of the risk, but also who the counterparty is. Insurers consider a financially weaker counterparty to be more likely to default and therefore set risk limits that depend, in part, on the counterparty’s financial soundness. All respondents use the reinsurer’s credit rating as an approximation of its financial strength with a few also considering wider financial performance, risk profile, type of counterparty and geography.

How quickly do insurers assume a reinsurance counterparty defaults and are any post-stress management actions taken into account when setting risk limits?

All respondents assumed that their reinsurance counterparties default instantaneously when setting risk limits, which does not leave any time to take pre-emptive rectification actions. This measure is possibly chosen for its simplicity as a modelling approach. However, 3 of the 6 respondents took post-stress management actions into account when setting their risk limits. All 3 respondents assumed for example that they would be able to obtain replacement reinsurance 6–12 months following a reinsurance counterparty default (or other recapture event). The confidence in, and ability to take, management actions to protect the balance sheet may explain why insurers are prepared to accept a significant drop in solvency ratio when setting their risk limits.

What level do insurers set their risk appetite and risk limits?

All respondents set their reinsurance counterparty default risk appetite at counterparty level with insurers worrying about a single counterparty defaulting in isolation of other reinsurance counterparties. 2 of the 6 respondents also set a risk appetite at aggregate level across a number of counterparties to understand the risk of multiple counterparties defaulting at the same time. Simply adding the exposures together implies a perfect correlation between the fortunes of reinsurance counterparties, or an event severe enough to impact a large proportion of the re/insurance market. Such an approach prioritises simplicity, a desirable characteristic of a risk metric. By contrast, all respondents assumed a correlation in the 0%–25% range between reinsurance counterparties when modelling the reinsurance counterparty risk SCR. One interpretation of this observation is that a third of respondents take a more conservative approach when setting their risk appetite relative to the scenarios assumed for regulatory capital modelling purposes.

6.2. Use of collateral

The future payments made by a reinsurer under an annuity related reinsurance treaty change over time as longevity experience emerges and updates are made to future longevity assumptions, covering both base mortality and longevity improvement rates. Collateral posted reduces exposure to counterparty risk for either the reinsurer or insurer. The collateral received by the insurer provides partial protection from the balance sheet effects of a reinsurance counterparty default (or other recapture event).

All respondents have at least one collateralised annuity related reinsurance treaty and allow for collateral in their exposure calculations. How this is incorporated into the exposure calculation including the approach to haircutting assets and the frequency that collateral balances are settled may differ across the market. The importance of each of these is increased when considering quota share reinsurance where the pre-collateral counterparty exposure generally covers the full SII BEL and not only the net exposure (the difference, in present value terms, between the “floating” and “fixed” legs). Any growth in the market for quota share reinsurance to meet annuity writers’ demand is likely to see an increased focus on collateral.

Why does collateral only partially protect insurers’ solvency coverage following a reinsurance counterparty default (or other recapture event)?

Collateral is a key risk mitigation tool to manage insurers’ exposure to reinsurance counterparty default. However, it does not fully cover the combined increase in RM and the recaptured SCR net of the release in counterparty risk SCR over and above (for quota share arrangements) the SII BEL that insurers would need to hold upon recapturing previously reinsured risk. Table 5 shows that at least 30% of the loss of solvency coverage on recapture is attributed to the RM.

Table 5. The proportion of respondents’ loss of solvency coverage (before management actions) that is explained by the increase in RM following recapture

How frequently do insurers monitor and review their collateral positions?

All insurers monitor collateral positions with management information compiled quarterly for 3 of the 6 respondents. 2 of the 6 respondents produce management information on a monthly basis with the remaining respondent doing so on a weekly basis. The frequency of this monitoring may correlate to the size of the collateral amounts. Insurers with more quota share reinsurance in-place relative to peers may monitor their collateral positions more frequently due to the associated increased risk.

5 of the 6 respondents have also undertaken assurance that the collateral is being managed and monitored in-line with the appropriate policy. 4 of the 5 respondents undertake assurance reviews on an annual basis with the remaining respondent doing so on an ad hoc basis. One respondent has not undertaken any assurance that the collateral is being managed and monitored in-line with the appropriate policy; possibly on materiality grounds.

What protections do insurers employ to manage their exposure to a deterioration in a reinsurer’s financial standing?

5 of the 6 respondents include protections within their reinsurance treaties that trigger upon a breach of early warning signals or indicators that the reinsurance counterparty may be in difficulty. 4 of the 6 respondents introduce or increase collateral requirements; and 2 of the 6 respondents include stricter asset eligibility requirements to restrict the type of assets that can be posted as collateral by the reinsurer. One respondent includes a protection to restrict the level of new business that can be written within existing reinsurance flow treaties.

What risks do insurers take on upon recapturing the collateral following a reinsurance counterparty default (or other recapture event)?

The collateral received upon recapture is unlikely to fully match the insurance contract’s underlying features. Rebalancing costs, e.g. converting the collateral into MA eligible assets to support the recaptured liabilities, could therefore be significant. All respondents identify currency related basis risk between the assets in the collateral pool that would be received following recapture and the underlying insurance liabilities. Other sources of basis risk identified by respondents arose from:

• the indexation of the collateral assets (5 of the 6 respondents)

• the duration of collateral assets (5 of the 6 respondents)

• the collateral assets’ expected cash flows (5 of the 6 respondents).

6.3. Treaty recapture provisions

Reinsurance contracts are complex legal documents that are binding for both counterparties. Figure 12 below summarises the main recapture provisions included by respondents within their reinsurance treaties.

Figure 12. Main recapture provisions contained within two or more longevity and quota share reinsurance treaties.

The recapture provision following a breach of a reinsurer’s regulatory solvency position is noteworthy given its potential to cause a disorderly market disruption. The inability for insurers to recognise the full benefit of reinsurance within their SCR and balance sheet following a breach of a reinsurer’s regulatory solvency position could lead to a “run on the reinsurer.” This could have macroeconomic consequences if a single reinsurer was a counterparty to a number of insurers. It is unclear whether, and if so when, the reinsurance market would have the appetite and capacity to absorb all previously reinsured risk across the affected insurers in this scenario.

Do reinsurance treaties permit recapture if the cover ceases to qualify as a risk mitigation technique for regulatory purposes Footnote ⁹ ?

Risk mitigating techniques (RMT) that meet the SII qualifying criteria allow insurers to reduce the size of the SCR that would otherwise need to be held. Figure 12 showed that only 2 of the 6 respondents included recapture provisions that would enable the insurer to recapture reinsured business where the reinsurance arrangement no longer qualified as risk mitigation for regulatory reporting purposes. This means that 4 of the 6 respondents would not be able to take credit for the risk mitigation within the SII SCR and may need to recognise losses even if the reinsurer had not technically defaulted or breached their solvency coverage. An example could be a court ruling that favoured a reinsurer in dispute with an insurer that risked undermining the legal enforceability of similarly worded reinsurance arrangements and the effectiveness of the cover as risk mitigation.