06 September 2021
In 2016, Solvency II's standard formula set out capital charges insurers must accept for their general account investments. Five years on, some underwriters argue the levels do not reflect their portfolio, or the reality of certain asset classes more generally, as David Walker finds out
'They want how much?' It is a cry of dismay one might expect to hear at an auction, but one might also have heard it from chief investment officers, when investment capital charges for Solvency II's standard formula were being debated in the lead-up to 2016.
Those charges designate how much qualifying capital an insurer on the common calculation method must hold, to protect its solvency while making given allocations.
A charge 'below' an asset's 'actual' risk could funnel general accounts dangerously into a proverbial firetrap, which could then ignite in a crisis. A charge 'above' what a history of risk/return data suggests would be 'reasonable' for a given asset class, could extinguish interest from CIOs abiding to the standard formula.
Since the 2016, some charges have been modified, and new ones have been introduced, for example for 'qualifying' infrastructure and long-term equity investments.
But some of Solvency II's charges still disgruntle some underwriters - and what's more, they're doing something about it.
The case for non-standard shocks
In scouring the 2020 solvency and financial condition reports (SFCR) of all Europe's insurers for its annual report on outsourcing of investments by CIOs, Insurance Risk Data has unearthed numerous standard formula users testing some of their investments against non-standard risk charges.
A common instance is property, which carries a 25% charge based on historical analysis of IPD real estate benchmark data for UK property. For a swathe of standard formula users, from France to Austria, having a charge whose basis is the price volatility of UK property does not fit the risk profile of their own property investments, which are not in the UK
Some mention a different risk charge entirely, in their solvency and financial condition reports, while others use a different model – a more flexible 'partial internal model' (PIM) for their property calculations, at least.
Vienna Insurance Group and its PIM fall into the latter camp because "the geographic specifics of [our] real estate portfolio, and in particular the Austrian real estate market, are not considered in the standard formula".
Generali Versicherung is another Austrian undertaking trumpeting the ability to adjust the capital charge it applies for property using a Solvency II internal model (IM). By the standard formula, it notes, "a flat rate of 25% for all property investments would be used," whereas its IM allows the charge applied for the capital requirement to be "calibrated country-specifically [and] since the real estate portfolio of Generali Versicherung AG is predominantly in good and very good Austrian locations - especially Vienna – this can reduce the risk due to the country-specific approach [so the] real estate portfolio can be valued more appropriately with the help of the IM".
Charles Moussier is no stranger to Solvency II risk charges (and stress-testing), having spent five years at Axa, three of which where he was deputy head of group investment solutions responsible for the optimisation of the asset liability management (ALM) position of the group. Moussier is now head of EMEA insurance client solutions at Invesco, the asset manager, since 2019.
In regards to property investments made by standard formula users, he says: "The Solvency II property shocks were built on the available liquid indexes, the UK IPD. However, the UK property market is very levered and volatile compared to the European property markets. Hence, French or German insurers could document their own internal modelling on this strategic asset class if they find enough data points."
UK IPD indices are not representative of the property markets in Europe, the market shocks are too severe, he emphasises. "Real estate exposure is an efficient way to diversify insurance general accounts and to increase the profitability of the general account."
More favourable capital treatment is often documented in internal models, he adds, while real estate debt is also documented in internal models "to better take into account the collateral".
VIG and Generali Versicherung are not alone in applying different capital charges to their property investments, in SFCRs.
Groupe des Assurances Mutuelles de l'Est (GAMEST) in France reduces the 25% standard forumla charge by 10 percentage points, "in order to reflect the real risk of GAMEST [property] which is concentrated in France". It also retained a risk of default and non-zero concentration on government bonds - all outlined in detail to France's regulator (ACPR) in GAMEST's Own risk and solvency assessment (Orsa). The insurer still wants property to invest in, it says, "to increase overall returns and generate additional income".
Another French mutual reduced the charge for its property by one point more than GAMEST did – showing the reductions, when they happen, are not uniform. It, too, said its own 14% shock factor reflected a portfolio "concentrated on very specific French regions".
But in the Netherlands, the Onderlinge Levensverzekering-Maatschappij's-Gravenhage says: "the standard [real estate] shock based on UK data is assumed to be appropriate for the average European market [and] there is no reason to assume that the standard shock would not be appropriate for the portfolio of Gravenhage."
The Dutch insurer's portfolio consists of lettable homes "with a good geographical spread throughout the Netherlands".
For some EU govvies, says Moussier, a 'nil' charge for spread risk may not match what really happens in times of crisis. He mentions BTPs that many of Italy's underwriters are trimming with quite some vigour. For instance, BTP spreads over Bunds in Q1 2020 ballooned past 300 basis points, though no shock was applied to the BTPs
Moussier says: "There is a huge difference between the standard capital charges and the economic reality of managing a diversified portfolio. For instance, Italian govvies are a very volatile asset, but this is still the most attractive asset class under Solvency II [with] no credit spread capital charge and positive returns."
Moussier says BTPs do not offer a good diversification for the insurance investment portfolios, "which are essentially made up of corporate bonds".
Still, the volatility of the yields on nil-charge BTPs played havoc with Solvency II ratios of some Italians in Q1 2020, because BTPs form part of their eligible own funds – the numerator of an insurer's solvency ratio – and CIOs either want out, or want greater control over the impact of such a 'risk-free' investment.
Insurance Risk Data has identified CIOs variously selling down their BTPs, outsourcing mandates for investments to replace BTPs, or even using derivatives structures to limit the volatility that seems part and parcel of lending to Rome.
Moussier says introducing a capital charge on Italian govvies is "a good way to add a buffer of own funds for holding a lot of BTPs [and] an effective financial communication tool for listed Italian insurers".
Some insurers apply different capital charges in 'what if' stress tests of their debt investments. Provinzial Rheinland, for example, treated its EU sovereigns as if they were non-EU holdings in a stress test that "leads to a minor uptick in our spreadrisk," its SFCR notes.
And a fellow German composite stress-tested its balance sheet as if it contained more equities than it actually does. "A fictitious volume of shares at risk that represents a possible build-up of the equity exposure in the coming year, after a large part of our actual stocks were sold in 2020," the underwriter explained. "The aim of mapping such a fictitious equity exposure is that the company's risk capital, which is determined on the basis of a one-year horizon, does not under-rate [by using] the current situation."
Moussier notes a certain flexibility has been introduced into some SF capital charging, in part due to societal ambitions of the regulations' designers, for qualifying infrastructure and long-term equity (LTE) investments, for instance.
Each class has strict criteria to meet, and reduced capital charges for those that do – for instance 30% or 36% depending on the form of infrastructure investment, and 22% for LTE.
Insurance Risk Data's outsourcing report has identified numerous CIOs making use of these benefits, commonly with help of external advisors/managers to originate, certify and invest in the assets.
PensionDanmark is one such insurer, using Copenhagen Infrastructure Partners, to allocate to a mixture of wind farms, gas infrastructure, and even train carriages that all win the cut charges for 'qualifying infrastructure'. There are more Danish, German, French, Belgian and even Croatian insurers seeking out LTE and/or qualifying infrastructure.
Moussier says there are a limited number of investments whose standard formula shock is out of kilter with how the investments, and/or insurers buying them, work.
"For private equity the volatility could be seen as important, but since insurers and general partners are clearly managing their assets in the long run, it is way too punitive to impose a very high capital charge on these assets on a one year basis only. In addition, over one year, the risk could be limited [with] 2020 a good example."
Moussier adds securitisations are tranched to provide different types of credit protection to the investor [and are] a good way to access leveraged loans markets with a protection for insurers. "But the liquidity squeeze during the [financial crisis] fuelled a very punitive Solvency II capital treatment for this asset class in the EU and UK," he adds. "[Although] this is not the case in the US."
What are called 'simple and transparent' assets may obtain better capital treatment than before under Solvency II, but for Moussier the capital charge "is still too high to favour any allocation in this asset class by insurers".
There is arguably a further inherent shortcoming in the Solvency II framework and the design of its risk charges, "calibrated on historical data, namely the deflationary environment of the last 30 years [which] cannot reflect the future risks for the investment portfolios". But that is a story for another day.
Names of underwriters in Europe that are seeking qualifying infrastructure and long-term equity are in Insurance Risk Data's forthcoming annual insurer outsourcing report, as are CIOs using the standard formula, but detailing non-SF shock levels to allocations, in their solvency and financial condition reports. For more details on the value of IRD's outsourcing report, contact email@example.com