The credit profiles of 35 leading reinsurance carriers show significant overall resilience to major reserve shocks, but with some very big individual variances.
23 May 2022
1. Executive summary
- Litmus’ research suggests a significant overall resilience to major reserve stresses across 351 leading non-life reinsurance carriers. However, the individual reinsurer results vary considerably: from a negligible effect to a very profound impact (see sections 2 and 6).
- Reinsurer resilience to stress scenarios matters because the reinsurance asset2 is a fundamental source of exposure for many cedants: especially since that exposure can increase very substantially in stressed situations such as industry-wide reserve shocks.
- The stresses we have deployed are the point in time recognition of whole account loss reserve deficiencies3 of 15%, 25% and 35%.
- We calculate the impact of those stresses on each reinsurer by first manipulating its published balance sheet data4 to reflect each stress. We then input that modified data into our non-life reinsurer credit scoring models (“LitmusQ” models5) and compare the results to those produced using unstressed data. This process creates the “before stress” and “after stress” outcomes displayed in Charts 1, 4 and 5 below.
- Most cedants use ratings as a fundamental plank of their reinsurer counterparty risk management process. Individual reinsurers will often carry a rating that is a function of a group credit profile6, not its own. But cedants and brokers can also be concerned with the counterparty risk they take with the individual carrier they use7. Our results here reflect that “carrier not group” specific consideration.
- The material variance in impact between individual carriers noted here is also seen when we stress test at the consolidated group level and for other stressed scenarios, such as catastrophe PMLs and severe investment market hits.
- We believe the results shown here suggest that top-down approaches to stress testing the reinsurance asset, which can tend to mimic approaches to the stress testing of bond portfolios8, may not effectively capture the range and degree of exposures a cedant actually has, as observed in a bottom-up case-specific analysis of the type we describe here.
- The relevance of a case-specific approach reflects key characteristics of the reinsurance asset that differ from typical bond portfolios, namely: inherently very high sector concentration risk, commonly very high counterparty concentration risk, and the lack of liquidity (limiting the ability of a cedant to actively manage their reinsurer counterparty exposure during a stress event).
- Finally, we also reference S&P’s9 research reflecting group specific applications of standard reserve stresses on the combined aggregate capital adequacy of its global reinsurance group cohort.
As with some of our case specific carrier level results the S&P aggregated outcomes show a very material impact, but we also note how the S&P research inherently highlights that such stress tests (theirs and ours) should be understood for their “point in time” nature. This differs from the much wider and prospective basis for a rating analysis that would be deployed during and after a severe stress.
We welcome questions and deeper discussions on what lies behind our analysis, the details of the headline results referenced here and how to interpret those results.
2. Individual reinsurer reserve stress-test results summary
Our research reflects the application of a set of whole account, gross and net, loss reserve stress tests on the most recent fully reported financial year (2020) across our reinsurer carrier cohort (a 2021 data update will be created as soon as every carrier’s 2021 numbers are available to us).
The “before” and “after” stress results are calculated using the reinsurer models from our credit scoring application LitmusQ. A brief summary of LitmusQ financial scores and the context for our mapping of those scores to a lower-case version of the international rating scale10 (as shown in Chart 1) is given in section 3 below.
Distribution of reinsurer cohort LitmusQ financial scores after reserve stresses of 15%, 25% and 35%.
The pre-stress financial score results cluster around the ‘a’ and ‘bbb’ category mappings (Note: 13 of the 18 ‘bbb’ category mappings are at the ‘bbb+’ level). The severe whole account reserve book 15% stress still shows 30 out of 35 cohort members’ financial scores at a ‘bbb’ category mapping or higher.
For the exceptionally severe 35% whole account stress, 19 out of the 35 cohort members financial scores map to the ‘bb’ category or below, with 6 of those having financial scores that map to the ‘ccc’ category. None, however, “fail”11.
Conversely, and highlighting the wide range of impact seen, even at the 35% stress levels, 9 of the 35 cohort members’ financial scores still map to at least the ‘a’ category, 2 of whose financial scores still map to the ‘aa’ category.
3. Using the LitmusQ models to quantify the impact of the stresses
The models within LitmusQ apply a set of calculations to the financials of non-life insurers and reinsurers, combining key capital, performance and other ratios with adjustments based on areas such as growth, volatility and size.
The financial scores produced range from 0 to 100. Whilst a LitmusQ financial score outcome is not the same as a rating12 we deploy a standardised mapping of the numerical results to a lower-case version of the international rating scale to give context for the apparent strength of the numerical score.
In the case of stress testing, we then calculate and show the change in that financial score strength derived from each of the stressed data sets compared to the results derived from the pre-stress data.
The individual carrier outcomes reflect their own stand-alone financial data profiles and are often different from what we would see when scoring the consolidated group profile.
The data manipulation and financial score creation process works like this:
4. Background to – and creation of – the individual reinsurer carrier cohort
Reinsurance carrier ratings are often a function of the group to which they belong. While use of those ratings is typically a central plank of most cedants’ counterparty risk management process, the fact that a cedant’s contractual exposure typically exists at the carrier level can also lead cedants and brokers to pay considerable attention to individual reinsurance carrier strength and resilience. Our research here reflects that “carrier level” consideration.
The 35-carrier cohort is derived from the “reinsurers (carriers) by country” section of the 2021 S&P publication: “S&P Global Ratings Top 40 Reinsurers and Reinsurers by Country”. We identified the largest suitable carrier shown in the S&P list from any given group (defined by Net Written Premiums) and from these selected the largest 35.
5. S&P global reinsurance group cohort stress tests
S&P traditionally publishes an annual set of stress tests and their impact on the aggregate “capital adequacy” (as per S&P’s capital model) of the 20 or so groups that make up its global reinsurance cohort (the exact number of cohort members varies across the years due to M&A activity).
The S&P graph below (Chart 2) shows that a reserve shock of the level of the US 2002 reserve strengthening requirement would have reduced the S&P cohort’s aggregate risk absorbing capital (“Total Adjusted Capital” is S&P’s term) at the 2020 year-end from being redundant at the “AA” benchmark to close to the “BBB” level.
Standard & Poor’s Global Reinsurance Cohort Capital Stress Tests – 2020 Financial Year
In the 2019 edition of its “Global Reinsurance Highlights” publication, S&P also included a significantly more extreme reserve stress scenario – a 35% strengthening requirement – the impact of which would have dropped the cohort’s aggregate capital adequacy model outcome for the 2018 year-end to well below the “BBB” benchmark (see Chart 3., below). This is a degree of reserve strengthening of the same headline magnitude as we selected for our “Exceptionally Severe” reserve stress scenario.
Standard & Poor’s Global Reinsurance Cohort Capital Stress Tests – 2018 Financial Year
6. Litmus commentary on the Litmus individual carrier level results and the S&P group cohort aggregate results
Summaries of the scale of individual LitmusQ financial score changes by rating notch mapping
At first sight, the impact on some individual reinsurers in the Litmus research may seem dramatic, but the (also dramatic seeming) S&P global reinsurance group cohort reserve stress results, noted in section 5 above, indirectly highlight an important context for stress test outcome consideration.
This context is that stress tests of this type are point in time and, as long as insurance/reinsurance groups survive the stress with their ability to trade and underlying business model intact, their (forward looking) ratings would be likely to reflect key prospective issues such as expected ongoing profitability and the (associated) ability (and time frame) to rebuild capital adequacy if needed. As an example, S&P’s average rating level for its global reinsurance group cohort did not become “BBB” after the 2002 reserve shock.
An extra context for an individual carrier’s post-stress rating would be whether the agencies that rate it deem the parent group’s commitment to it has changed or not.
That said, our results show a very substantial variance across the carriers included in the research. This also aligns with the very wide range of impact the 2002 reserve shock had in practice on reinsurers at the time.
Distributions of the Litmus cohort financial score changes by rating notch mappings
A key observation (from our perspective at least) is that top-down stress testing of a cedant’s credit risk exposure to its reinsurance asset may miss important differences in how specific types and degrees of stress play out across their counterparties. We also see material degrees of impact variance when we run reserve stress tests at the group level, and across catastrophe and market risk stresses.
Unlike a typical portfolio of rated bonds, the (non-collateralised) reinsurance asset reflects very high industry concentration risk and, often, some very high counterparty-specific concentration risk, while the potential liquidity of the reinsurance asset is, at best, limited.
The significance of these important reinsurance asset characteristics can be increased further by a cedant’s own credit profile in a stressed context: the exposure to reinsurers may well grow substantially if the cedant itself is under a significant underwriting risk related stress. This would reflect the increase in the absolute scale of the reinsurance asset, the potential that at least some of its reinsurers may well be suffering from that same stress across their wider cedant portfolio, and the extent to which the cedant’s own risk absorbing capital is reduced by the stress.
There are, of course, many assumptions we have made in this research, and we would welcome the opportunity to discuss these, the results and the approach we have deployed with interested market participants.
1 The 35-carrier cohort is derived from the “reinsurers (carriers) by country” section of the 2021 S&P publication: “S&P Global Ratings Top 40 Reinsurers and Reinsurers by Country”. We identified the largest suitable carrier shown in the S&P list from any given group (defined by Net Written Premiums) and from these selected the largest 35. See note 12 below for the suitability criteria used.
2 The “reinsurance asset” represents the total amount caried in the balance sheet of a cedant for future payments due from reinsurers. This includes reinsurers’ share of IBNR.
3 The deployment of the stresses is conducted via increasing the total reported loss reserve by each noted percentage and adjusting the other impacted data items (total assets, reinsurers share of technical reserves and shareholders’ funds) accordingly. See also note 13 below.
4 All financial statement data used in this research was sourced with permission from A.M. Best (Best’s Financial Suite – Global).
5 “LitmusQ” is Litmus’ online non-life (re)insurer credit and financial profiling application. It contains four “scoring models” calibrated for different non-life (re)insurer types: Primary Insurers, Lower Risk Reinsurers, Higher Risk Reinsurers, Captive Insurers. The two reinsurance models were used for this research.
6 Rating agencies have differences in the detail of how they approach the concept of ratings that reflect the overall group credit profile (and which carriers are covered by that “group” rating level). But broadly speaking they cover the same ground. The credit profile of any given carrier within a group – including those covered by the group rating – can be quite different to the overall group credit profile.
7 A consideration of the carrier credit profile for a carrier carrying its parent group level rating (or a rating at least enhanced by its group membership) does not imply at all that the assignment of the group level rating level to it by the rating agency is somehow incorrect. Rather it provides and extra context for a cedant’s risk management when considering its specific counterparty exposure in tandem with the level of the assigned rating.
8 Two examples of “top-down” reinsurance asset stress-test approaches would be to either uniformly reduce the ratings of all reinsurance counterparties within whatever capital model the cedant focusses on for its own capital adequacy management or to reduce the absolute amount of the asset by a given percentage.
9 References to “S&P” throughout mean “S&P Global Ratings”.
10 LitmusQ financial scores are “model outcomes” which purely reflect the financial data inputs and the model selected. As such they are not in any sense an “opinion” and do not reflect any case specific perspective of the organisation being scored. Ratings, by fundamental contrast, are indeed case-specific – prospective – opinions that reflect very important qualitative factors and non-public financial and other information. The international rating scale mapping therefore just reflects Litmus’ experience and judgment on how historical financial profiles typically align with ratings overall. We review this periodically via comparisons of LitmusQ financial scores to the rated universe. The provision of the mapping should not be interpreted in any way to suggest that LitmusQ financial scores are somehow analogous to ratings.
11 “Failure” in this context would mean that the post-stress level of shareholders’ funds becomes negative. See Also note 13 below.
12 Reasons for non-suitability for inclusion included: 2020 financial data were not available; there were insufficient data to produce a five-year LitmusQ financial score; available data could not be reliably scored using LITMUSQ (e.g., too high a proportion of Life business); or, available data did not appear to include the full consolidation of all subsidiaries.
13 Data treatment when running the standard (pre-stress) LitmusQ models also includes some standard credit analyst adjustments to reported data: among others these include the non-allowance of goodwill as an element of risk absorbing capital but the inclusion of equalisation reserves for that.
About Litmus Analysis
Litmus Analysis specialises in helping the insurance and reinsurance industry understand credit risk and the ratings agencies. Its main areas of work are Ratings Advisory (helping companies manage their relationships with the rating agencies), cedant analysis, insurer and reinsurer financial profiling and stress testing, market analysis, peer reviews & benchmarking and market security/counterparty credit management (including the InsurTech application LitmusQ).
The team of consultant analysts all have a rating agency and/or broker market security background and includes the former heads of S&P Ratings Europe (insurance), A.M. Best EMEA and A.M. Best Asia Pacific.
Call: +44 20 3651 5044
Registered address: 6 Bedford Road, Barton-Le-Clay, Bedford, England, MK45 4JU
Registration Number 07306070